Propeller Application

Erik Skarman


This is a new version of this page. Many things are updated here. Most of all, I am using a new language, called MPS. Many of the applications are new or updated too. Many of the files can be downloaded from the chapter MPS objects

The old version is available here

The Propeller microprocessor

The Propeller microprocessor is a computer developed and produced by Parallax Inc. It is a 32 bit computer running at 80 MHz. Its most important feature is that it has eight individual processors, called cogs, which operate in parallell. When all these processors are working, the total capacity of the machine is thus 640 MHz.

More importantly, however, the eight parallell processors allow you to work easily and nicely with real time systems, controlling and interacting with "real things". You can avoid the interrupt concept, with all its pedagogical difficulties and pitfalls. Instead, you assign a dedicated processor to monitor events from, say, a sensor. When it detects the event it can grab some external signals and then invoke computations in the other processors in a very versatile and transparent way. For the monitoring there are instructions that monitor an event in a single instruction, so that the computer stays within the instruction until the event happens.

The processors communicate with each other over a global memory. This memory is connected to the processors in a round robin fashion, thus avoiding the risk that two processors try to access the memory simultaneously. To allow constistent transfers of data chunks with more than one word, there is a semaphore concept.

The processors communicate with the external world over 32 I/O pins which are programmable to be either input or output pins. There is a simple scheme for resolving conflicts between several processors accessing the same output pins. All input pins are simultaneousy available to all processors.

The processor also has a real time counter, running through all the values of a 32 bit word in a few minutes (after which it overflows and restarts), and wait instructions that halt the processor (cog) until the time counter has reached a specified value. In this way it is possible to contruct very accurate real time functions.


Programming languages

Propeller Assembler

The Assembler language of the processer is available to the user through the Propeller Programming Environment (even though in some sense it has to be encapsulated in a Spin program). Here is a sequence of four instructions which can illuminate some special features of this assembler language:

sub x,y nr,wz if_eq mov x,y if_eq jmp #zero if_neq jmp #nonzero Here we subtract the variable y from the variable x, The result would be stored back in x, but not in this case, due to the nr flag (no result). Instead the wz flag commands an update to a z-register. This z register becomes true if the result was exactly 0 and false otherwise.

The next two instructions are gated with this. They are executed only if the z-register is true. First we move the value of y to x (which is kind of meaningless, because in this situation, x and y are already equal). Then we jump to an instruction labeled with "zero".

If the z-register were false, the two first instructions would be executed as no-operation-instructions (nops), and then the program would execute a jump to the instruction labeled with "nonzero".

The z-register has a sister, the c-register, which indicates whether the MSB of the result is 1, but it can also detect overflow.

This gating of instructions with conditions allows the building of "if-then-else" constructs without using jumps.


The Propeller computer is supplied with another language at a higher level, the Spin language. It is an interpretive language. This means that the programmer is not running his own code. He runs an interpreter, resident in the propeller, which interprets the Spin code (somehow encoded into 1:s and 0:s).

This interpretive principle facilitates the design of a nice high level language. But the whole method is slow.

Given a high level Spin language and a fast Assembler language, the natural approach would be to write time critical functions in assembler, and connect them through a Spin program. To my understanding, this is not the way Spin and assembler code work together. The Spin "call" to Assembler code amounts to that a cog is started to run from the machine adress of the indicated assembler code. But there is no return instruction to return to the Spin code.


After some time with the Myra language, that you will find here, I decided to go back to an older language idea. The purpose was to get smaller and faster programs, by working against a computer model that was closer to the real architecture of the Propeller. Speed may be important, but program size is even more important in Propeller applications, because of the relatively small memories in the Propeller cogs. In fact, the size is only 496 longwords. It may seem pretty hopeless to solve any task with such a small program size. But, after all, you have 8 cogs of this size, and in these types of computers you can do a lot with small programs.

You find a more or less complete documentation of the MPS language, and also the standard object file std.mpo here

For Myra, I decided to make an interpretative version, to overcome the progam size limitation. For MPS I haven't done that. It seems that I can solve my problems with the standard language. Myra also made use of an assembler resource file written in assembler, but for MPS I don't need that. I can write everything I want in MPS, except for a few lines with so called in line assembler code.

The MPS system

Here is the software you have to download in order to use the MPS language:

(Sorry for that some of java files are commented in Swedish. I don't think yo need the comments, but if you know Swedish, they may be of some help.) starts with a comment block, which describes the hierarchy of methods. In the right collum, there are designators for the methods, and these will reappear at each method, so you can find the methods by searching for these designators.


An interest I had in stack programming several years ago, led me to try to design a stack based higher level language, that could be compiled into assembler. I was also inspired to this by someone, who had made a FORTH programming language system for the Propeller. FORTH is a stack based programming language, and so is my language, called MYRA. A documentation of this language is found here. I had some introductory comments on Myra here, but I have now removed them.

The Myra System

Here is the software you have to download in order to use the Myra language:

(Sorry for that some of java files are commented in Swedish. I don't think yo need the comments, but if you know Swedish, they may be of some help.) starts with a comment block, which describes the hierarchy of methods. In the right collum, there are designators for the methods, and these will reappear at each method, so you can find the methods by searching for these designators.

"A Propeller ecology"

As a little apetizer for the rest, I will present the following figure, which displays a Propeller surrounded by devices, from which it can get sensory data and operator commands, and on which it can act. In short - an ecology around the propeller:

Let's start in the upper left corner, and go around in counterclockwise direction.

In the upper left there is an analog to digital converter. To it, we can connect potentiometers, and thus control the program with one or more knobs. There is also a great selection of sensors, that give analog outputs. In the figure we have shown gyros, which measure rotation, and accelerometers which measure acceleration and tilt (due to gravity). Under the A/D converter, there is an RGB LED on which we can create colors. Its main use is perhaps for debugging. You can display a certain color, when you come to a certain point in the program.

The next item is a keyboard, which can be used to control a program. Here I show a numeric keyboard, which is smaller and handier than a full alphanumeric keyboard. Touch controls, which play such an important role in smartphones, are available to amateurs today, but perhaps only in the older form of resistive touchscreens, which require using a stylus. With a touchscreen, you can make a universal mode selector, by drawing "buttons" behind the screen, or you can input analog data. You can relatively easily make a program to input simple sketches to your computer.

RFID is a way to recognize simple and cheap "tags", when they come into the vicinity of the RFID device. Each tag has a unique serial number, and you get that, when the tag is detected. One, maybe bad, idea I had about this, was to use RFID for robot navigation. You placed out tags, and when the robot encountered a tag, it would know where it was.

"TCP/IP" represents a little device called WiFly, which lets you connect your Propeller to you WLAN at home. From there you can go with your Propeller data out on the world of internet. ZIGBEE represents a wireless device, so that you can send data (and commands) between your Propellers wirelessly.

The EEPROM is an integrated part of the Propeller system. That's where you store your program, if you want it to be started after power down/power up. All that is programmed into the Propeller. But you can write you own code, to talk to the EEPROM, for your own purposes, and you can stack more EEPROM chips to the same wires, to get more space for data. I have also tried some other components with the same functionality, but with more capacity. They are usually called flash memories, but they are functionally EEPROMS (Electrically Erasable Programmable Memories).

RAM is a Random Access Memory, i.e. a standard memory, where you can store data at high speed. But you loose the data at power down. A RAM has many pins, so I have chosen to connect the RAM via an interface Propeller.

SD is an SD-card (Secure Digital). Technically it is a flash memory and a small interface computer, built into a "card" that you can plug into a slot. They have a huge capacity, and it is practical that you can plug them in and out.

ESC is an Electrical Speed Control, which belongs to the world of Radio Control Aircraft. You connect a powerfull battery to the left, and a brushless motor to the right, and then you can control great mechanical power with pulses from the Propeller.

Over that, there is a more conventional way to control standard DC motors with a so called H-bridge. The diodes further up belong, together with the striped strip, to an odometry system. With that you can measure displacement along the strip.

The, maybe, camera like object at the right top, is - a camera. It is a camera with a serial interface, so that you can get the pixels of the picture into the computer. The rectangle with the sailboat is a graphical display, with color depth good enough to display photographs. But, with some programming, you can also display alphanumerical data on it.

Finally, the USB box, is the standard umbillical through which the Propeller is loaded with programs from a PC. But you can also use it to up/download data from/to the PC. You need some software at the PC-side to take care of the data. Such a program will be presented here. Concretely the USB box contains a little device called a Propstick. With that and the Propeller development environmens, you can load programs into the Propeller.

Given all this, you can build different combinations. For example, if you take a picture with the camera, you would like to store it on the SD-card. But for most cameras, the camera produces more data than you can handle in time with the SD-card. So you may need to store the picture in the RAM. After you have transferred the picture to the SD-card, you can display it on the display at any time.

Hardware considerations

Analog input

The propeller computer doesn't have any analog inputs. Its cousin, the Basic Stamp, handles analog inputs by means of RC circuits, which convert analog voltages to times. Times can be measured accurately with both Basic Stamps an Propellers. So I guess, one can use a similar design for the Propeller. However, one needs one pin for each analog input, and probably one more input to discharge, or precharge the RC-circuits.

So my alternative has been to use separate A/D converters. My favorite is Microchip's MC3208. It converts 8 channels to 12 bit digital data. It has a serial interface. You need to handle a chip select pin, a clock pin, one data-in-pin and one data-out-pin. So for 8 analog channels, you need 4 pins. If 8 channels aren't enough, you can get 16 channels for 5 pins etc..

You send a conversion order to the converter's data-in-pin. This order announces which channel should be converted. Then you can clock your digital result from the converter's data-out-pin. All this is done in a function adc.get in the file adc.mpo. The call is like this:

( channel )adc.get ->x The result appears in the 12 least significant bits in x. "channel" of course is the number of the desired analog channel to the MC3208. The system needs to be initialized with )adc.init That function is also in adc.mpo. This function assumes that the converter is connected to the first four pins of the Propeller in the order they appear on the chip.

Unlike what I have said before, MC3208 doesn't need 5 volts. It can run on 3.3 volts. In case you connect it to 5 volts (which gives you a slighly lower noise level) you need a resistor of some kohms between the converter output and the Propeller. Otherwise you will destroy that pin on the Propeller.

Analog output

A parallell input Digital to Analog Converter (DAC) would require many Propeller pins, but I have a serial interface DAC called MCP4822, and I have made an mpo-file for it:DAC.mpo. It is a small power device, so you may need to add some power amplifier to it.

But another popular way to make analog outputs is with PWM, pulse width modulation. There is a pulse repetition frequency (prf) for the pulses. Then the pulse is on for some percentage of the total period, and off for the the rest. Afterwards, you can remove the pulses with a RC-circuit, and just keep the average, which will be an analog signal. But if the output goes to, for instance, a motor, the motor can filter out the pulses with its mechanical inertia. In that case, you may need to raise the prf to well over 10 kHz. Otherwise you will hear the prf from the motor. The time constant of the RC circuit or the motor will limit the bandwidth of the output. You need a margin between the desired bandwidth and the prf, but for mechanical systems, this is usually not a problem.

Here's an example. We want to control two motors, and we want to drive them in both directions. Assume that we connect motor 1 between pins 0 and 1, and motor 2 between pins 2 and 3. We can drive the motors with codes like this:

In order to control two motors simultaneously, we can use two processes, which is simple. But if we need to save processes, we can control the two motors with a single process. Here comes some MPS code for this.

We have two velocities v1 and v2 for the two motors. They are scaled so that 255 means (almost) full speed, and 0 means no speed. In some cases we have to avoid full speed for hardware reasons (more about that later). Now let ttot be the time of the full period. Then we can compute an on time like this: ttot *v >>8 ->ton The shift by 8 is a division by 256, to remove the scaling of the v variable. Multiplication and shifting must be done in this order. Now, we can save computation time here, by letting ttot be an integer power of 2. We can choose ttot as 4096, which is the 12th power of 2. So, now the multiplication can be replaced with a leftshift with 12. Then the whole first equation can be replaced with: v <<4 ->ton This is good, for it reduces the time, we have to keep the motors idle. The value 4096 gives a frequency of 20 kHz which is not hearable for (most) humans. The direction of motion is given as two masks m1 and m2. We need a program that turns the motors off at the right time. Now as we control two motors at the same time, we can't use the wait instructions, so we have to check the computers real time clock cnt and use if-statements instead. Here's the code process pwm ttot = 4096 begin ( $F )setdir :loop v1 <<4 ->ton1 v2 <<4 ->ton2 ( m2 <<2 |m1 , $F , )outpins now ->t0 :tloop if(now -t0 >ton1){ ( 0 , $3 , )outpins now +ttot ->ton1 } if(now -t0 >ton2){ ( 0 , $C , )outpins now +ttot ->ton2 } if(now -t0 >ttot):loop A:tloop The mask $3 directs the outputs to pin 0 and 1 where motor 1 sits, so we can turn motor 1 off with that. The mask $C does the same for motor 2. And the mask $F is used to turn both motors on. Note that the instruction now loads the real time counter cnt. v1 and v2 and m1 and m2 are global variables computed elsewhere. m1 = 1 means that motor 1 is driven forward. m1 = 2 drives the motor backwards. We could also use m1 = 0 if we want to hold the motor entirely. Note that the motors are held while we compute ton1 and ton2.

Power drivers

We now want to connect power drivers to the Propeller output pins, and connect the motors between the outputs of the power drivers. This structure is sometimes called an H-bridge.


Here's a solution based on an integrated circuit called TDA2050:

TDA2050 is marketed as an audio amplifier, but it is actually a universal operational amplifier, with capability to deliver high power. It amplifies the difference between the + and --inputs with a gain of one million or so. Then we can reduce the gain with a feedback through the resistors R2 and R1. We would like to create a gain of about four, to increase the 3.3v span of the Propeller to an output span of 12v. But with the TDA2050 you can't reduce the gain that much without getting oscillations. A gain of about 40 is minimum. So we live with that, and reduce the input signal first with RI and RT.

RC and CC make up a capacitive load, also to enhance stability. The capacitor CV shall eliminate spurious feedback between the different steps of the amplifier over the 12v supply rail. It should be placed physically close to the circuit.

The capacitor CT together with (essentially) RT helps remove the pulses from the PWM-system, and just keep an average. We have time constant of appoximately 500μs, which should filter out the pulses with a period of 100μs pretty well. At the same time, such a time constant shouldn't cause any stability problems for the mechanical system.

The question is still, if the capacitor CT should be there. With the capacitor in place we avoid hearing a tone from the motor. At least cats may appreciate that. Perhaps we save the motor from some wear. And we avoid disturbing the electrical environment around our device. The penalty for all this is heat dissipation.

Without the capacitor, we drive the TDA2050 in full saturation all the time. The output transistors are either fully on or fully off, and hence consume no power. However, such operational amplifiers are not so fast, so they spend some time in the transition phase, so some power is still dissipated.

With reasonably small motors, both solutions should work, but you need a heatsink for your amplifier, and free air flow round the heatsink. TDA2050 and many other modern circuits have the advantage that the cooling tab is connected to ground, so you can without risk screw all the circuits firmly to the same heatsink.


An alternative for even higher power is to use a IR2103 circuit, which is a FET-transistor driver from International Rectifier. Here's a quick sketch of how it is connected.

To drive the upper FET to open, you need a gate voltage a few volts over the drain. So to bring the output all the way up to the E volt rail, you need more voltage than E somewhere. In fact, the upper FET is driven by a little circuit, that sits between the terminals O and X. X should be 12 volts higher than O. The question is, how we get that voltage there.

A hint is to think about a so called charge pump, as in this figure:

C1 and D1 will clamp the voltage at P never to be under the rail voltage. (12 volts). Consequently the voltage at P will vary between 12 volts and 17 volts (as 5 volts is the amplitude of the oscillator). D2 and and C2 will rectify and smoothen this signal to give a voltage 5 volts over the 12 volt rail. This works fine, but it probably doesn't work unmodified, if we are going to pump a voltage over a rail, that moves all the time. That is the situation we have with our IR2103 circuit.

The conventional way to provide the X voltage is to use the systems output signal as charge pump oscillator. Here's the way it is done.

When the output O is all the way down to zero, the capacitor C will be charged through the diode D. Then, when O flies up, C will fly up with it. It will gradually discharge, though, so O will have to go down to zero every now and then. This is a constraint for the programming of the device. With the PWM, the pulse width must never be 100% of the full period. This is OK, but it may be easy to forget, when you make your programs.

Hence it would be nice if we could design a system with an independent oscillator for the charge pumping. It's open for invention. With a little luck, we could maybe use one oscillator for all IR2103-circuits around.

IR2103 is a purely digital system. The input is truly digital, so there is no way, that you could smooth away the pulses with capacitors. The circuit is fast, so the time in transition between the two saturated states is short, and the switching is also synchronized, so that both FET:s are on simultaneously only for very short periods.

The most dramatic property of this circuit is, however, the bound on the voltage E. E can be as high as 600 volts! With 600 volts supply, these 600 volts penetrate right into the IR2103. It's a tiny 8 pin dip-circuit, or an even smaller surface-mount version, but it is said, it can take it. Beware however! Your circuit board design might not tolerate 600 volts. Maybe you dare to work with 200 volts. You easily find FETs that can handle 7 A (with some cooling). Then you can control 1.4 kW power from a 3.3v output from a Propeller pin. That's pretty awesome.

Integrated power drivers

Lately I have had some problems with the IR2103 driver, in that some of the FET:s have been destroyed. I don't know why, but it always hard to remove and replace components.

Because of this, I have tried another alternative. It is an integrated driver called L298. It contains two full bridges (4 half bridges), so you can drive 2 DC motors (or 1 stepper motor) with it. It comes as a so called Multiwatt package with 15 pins. The circuit works fine, and I have been surprised that the component stays so cool. You can mount it on a bit of aluminum profile, if you just drive moderately sized motors. The circuit can drive up to 48 volts and a few Amperes.

Another circuit is called L293. It is an ordinary DIL20 capsule, which is a little more difficult to cool. It has four ground pins, and you can make a larger suface on the circuit board connected to the ground pins for cooling. I haven't tried these circuits yet, but they are part of an Arduino motor shield, and also of a Raspberry PI motor interface.

Protection diodes

Electrical motors have an inductance. Then, when you stop the current through the motor, the inductance will respond with a very high voltage. This could destroy your electronics, so you need to have protection diodes to absorb the voltage peaks. The choice of these diodes is a little mysterious. International Rectifiers driver FET:s (called among others IRF510) contain protection diodes, so the problem is solved there. For L298 you have to add diodes yourself. I have the impression that they are built in into L293, but I am not sure. An important factor is that they are fast. When you reverse the voltage over a diode, you expect that it would stop conducting. But before that, the charge carriers have to leave the diode. In a standard diode, there is no electric field at this occasion, so the charge carrier leave by diffusion only, and this takes a loooong time (a few microseconds). (There is the same mechanism in bipolar transistors). Shottky diodes have another desigh that overcomes this. The measure of this is called trr (reverse recovey time). You rarely find this information for slow diodes, so look out when this is not mentioned. As there weren't many enough fast diodes in stock with my supplier, I tried a fast but small diode called 1N4148 (a classical diode). I put two in parallell, and it seems to work fine. My recomendation is still to look for somewhat bigger Shottky diodes. Here's a circuit diagram for how to connect them:

Here's a picture of a motor driver circuit based on integrated power drivers:

The integrated circuit is to the left. I have chosen to connect it in a pretty primitive way with wire to make it easier to replace when necessary. Then you see a 5v stabilizer for the logic level of the circuit, and then the protection diodes.



(Most of this applies to AC-motors as well, but DC-motors are more popular in control applications.) As just mentioned, a motor has an inductance (it has a coil in it), and this means that the motor produces high voltages if you try to stop the current through it quickly. So use diodes as presented previously.

Secondly, a motor is also a generator. So it produces a voltage proportional to its rotation speed: ug =a·ω.

It is the difference between the applied voltage, and this generated voltage, that drives the current through the motor winding. So the current i = (u-ug)/R, where R is the resistance of the motor winding (usually a small value).

Then that current determines the torque M of motor: M = b·i. This torque is then reduced by external torques Mext, like friction. This gives an acceleration of the motor as dω/dt = 1/I·(M-Mext), where I is the so called moment of inertia of the motor + what is connected to it (this becomes a bit complicated, if there is a gearbox).

Now, let us think that Mext is small, so that motor is almost free wheeling. Then the motor will keep accelerating, and ω will grow, and so will ug. Finally, the current will be very small, but the torque M will manage to balance, the small Mext. So we have the equation u-ug ≃ 0, or u - a·ω ≃ 0, or ω ≃ u/a. So, the voltage over the motor controls the motor's speed.

Now, let's grab the motor axis and hold it firmly, so that the motor stalls. Then ω = 0 and ug = 0. So the current is u/R, which is a high current. This current controls the torque of the motor. This is the torque, that we feel in our hands, when we try to stall the motor. So now the voltage of the motor, or the current through the motor controls the torque.

Hence the motor acts somewhere between two extreme modes, one where voltage controls speed, and another where voltage controls torque. We can also see that the motor is not a passive receiver of electric signals. It responds by requiring high currents if it experiences mechanical resistance. Finally, if you make some kind of feedback to your motor, from, say, the position of the motor, the generator effect in the motor will contribute with some damping, which you probably very well need.

Brushless motors

Brushes in a motor serve two purposes. They bring the electric current over to moving coils in the motor. And they serve as so called commutators, which reverse the current, when the motor has gone a half turn. To get brushless motors, you have to do two things. First, you build a motor with stationary coils, and a rotating permanent magnet. Then you have to solve the commutation problems.

You can simply drive the motor with alternating current. With some luck, you can have the motor rotate in sync with that alternating current. In order to drive the motor in one, and only one, direction, you should have a three phase system, but you can create a false third phase with a capacitor.

Another idea is to measure the position of the motor with a sensor, and then reverse the voltage with electronics at the right moment. This is a kind of "software commutator". But then you can maybe avoid the separate sensor, by registering the motor voltage. There will be a small pulse induced, when the moving magnet passes one of the electromagnetic poles, This is not quite easy, because you are trying to measure some small induced voltages, while you are, yourself, imposing much higher voltages. But it can, obviusly, be done.

After all this, the motor is simply a three phase motor. And the electronics is a machine for creating three phase alternating voltage based on

This is an ESC, an Electronic Speed Controller. It's admirable stuff! You can get high speed and hight torque out of such a motor, and it is easy to control. You tell it what you want in terms of a regular pulse train, where the pulse length declares what you want.

The question is: What is it that you want? The specifications for these devices are not very clear on this. That pulse width doesn't determine the speed only. At high mechanical resistance, the motor goes slower. Another thing that you would like to see specified, is the dynamics. How fast does the motor go up to the final speed? There is also some logics, that may make the motor wait a little while before starting. All this was crucial in my application: A helicopter with four rotors. I failed with this project. Since then I have tried to measure the dynamics of these devices, and maybe I will return to the project after having analysed this.

But if you have less dynamically critical applications, brushless motors together with ESC:s can be a good alternative.

Stepper motors

A stepper motor moves in steps. These steps are steps in position, so they are in themselves position controlled devices. You can control their speed, but hardly torque.

The idea is that you can send a sequence of positive and negative voltages to (usually) two windings. Such a sequence will take the motor 4 steps, but you can stop anywhere in the sequence. Then the motor will stop where it is. You basically have full control of the motor. By running through the sequence at a defined pace, you can get a specified velocity.

A stepper motor has a hold torque. If external forces exceed that torque, the motor will slide away. If you thought that you would have full software control of where the motor is, this is a disaster. The holding torque is a function of the current you send into the motor. Higher current means higher holding torque, but at some current, the motor will overheat. Now, here's a disadvantage of the motor. If we want a reasonable holding torque, we have to send a hight current into the motor all the time.

For some applications, where the motor is connected to a big inertia, stability may be an issue. The motor might slip over a lock position and go to the next.

A stepper motor normally has two coils and a rotating permanent magnet with many poles. First one coil attracts a north pole, while the other is passive. Then you can remove current from the coil, and apply current to the other. Then it will attract the north pole. In the next step you reverse currents to attract the south poles instead. After 4 steps you can repeat the pattern, to attract the next north pole.

Obviously you need 4 power amplifiers here, to drive a stepper motor. With these amplifiers you could otherwise run 2 DC motors in both directions.

So the issues with stepper motors are:

The object file Stepper.mpo contains functions to be called in the following way: ( motorpin1 , motorpin2 , )step.init ) )step.drive2 motorpin1 is the lowest pin number for motor1, and motorpin2 is the lowes pin number for motor2. The rest of the 4 wires from each motor should be connected to the pins next to the lowest pin. The two wires belonging to one and the same winding, should be connected to consecutive pins. If you are unlucky, the motors might go in the wrong direction, and in that case you might have to change the wiring. The function step.drive2 is made for driving two motors. If you only use, it is unimportant which value you assign to motorpin2. The motors are controlled by one or two global variables xc or xc and yc. The motors move to the rotational positions xc and yc. The functions and step.drive2 are infinite loops, so the process can just call them, and then it won't be returned to. So here's what a process might look like: process drive begin ( xx , yy , )step.init )step.drive2 load from (stepper.mpo) *step.init *step.drive2 \



RS232 is a very old protocol for communication. It has changed name now to EIA232, but very few people seem to have noticed that. It is a standard for serial communication over only one line (per direction), where the bits are sent after one another in time. It was fairly specific about voltage levels, and it had several options for flow control, so that a receiver could inhibit data sending, when it was too busy. Surprisingly for a one line concept, the protocol was standardized for a 25 pin D-sub connector. The standard was also pretty open, when it came to different parameters like number of bits sent, parity, sign on start bit, communications speed etc..

Luckily a certain set of parameter values have become de facto standard, so now we communicate with a large number of devices and modules, with almost the same communication parameters. At the same time, the voltage level issue has become less important. Most devices use 0 volts as 0, and 5 or 3.3 volts as 1. But if you buy a true standard RS232 device, and connect it to a Propeller computer, the Propeller will be destroyed, as the voltages are too high.

If I can, I always avoid flow control, as it complicates things (more wires, more Propeller pins used etc.) Most devices can be configured so that flow control is not used.

The standard allows parity control, but the most common choice is "no parity", which simplifies your software. You don't need to set parity bits, and you don't have to check them (you never have to check them, even if they are there).

The most common communications speed nowadays is perhaps 115200 bits per second, which is about what today's processors can manage. But the default setting is often 9600. Conventionally these bit rates are called baud rate, and the unit is baud, and not bits per second. So we have 115200 baud etc. (The reason for this is that the term bits/second is reserved for the effective speed of a line with disturbances. If there are disturbances, you have to check and resend, and you loose speed on that.) Some devices can tell you which baud rate they are set to, but if you don't know the baud rate, you can neither put the question, nor interpret the answer. Most devices have some sort of reset button, which restore default parameters, like 9600 baud. For the eventuality that you have abandoned the default parameters, and happen to press that button, you have to have a little reconfiguration program up your sleeve. This is a maintenance problem for your system.

Then, the normal standard is that the line is 1 when it is idle. A new transmission is announced by the line falling to 0. This is the start pulse. Then you wait one and a half of TS, where TS is 1/(baud rate). Then you can clock in yor data every TS, least significant bit first. The standard is that one such package of bits is 8 bits long. Then follows an optional parity bit and one or more stop bits. When it's all over the line goes up to 1 again. It is not terribly complicated to build and read such a pulse train, if you have computer. To do it with conventional hardware was probably quite complicated.

As RS232 is such a common protocol, I've put the code for it in the standard object file std.mpo. The standard routines are called as:

( data , pin , pulse , )send ( pin , pulse , )receive ->data pin is the pin to which you have connected the wire to the other device. pulse is 1/baudrate, but expressed in Propeller tics. So with the standard setup for the Propeller with a 5 MHz chrystal and a PLL of 16, for a baud rate of 115200 you should set pulse = 80000000/115200 The receive program is blocking, i.e. you get stuck in there until some data comes. You also have to be in that program almost all the time, to be alert, when data arrives. This means that that function needs its own process.

There are versions for 9 bits called send9 and receive9, which are called the same way. And there are versions for 32 bits word length, but they are called without the variable pulse, as they assume a default baud rate of 500000.


I2C is a bus standard invented by Philips. I2C means Inter Integrated Circuits, and is thus a bus for communication between individual circuits in an electronic system. It is a master/slave concept, where a master controls the trafic on the bus. In the general case, there can be more than one master on the bus, which calls for an arbitration policy to divide the bus between the masters. However, in most cases, there is only one master.

The bus has two wires, one data line and one clock line. The slaves and the masters are connected with open collectors. Consequently, when a device tries to send a '1', the collector is open, and thus, the impedance is infinite. The voltages are pulled up with a set of pull up resistors, which belong to the bus.

The role of master in our application, will most likely be played by a Propeller computer. The Propeller doesn't have open collector outputs, but we can mimic this by using the direction register. We permanently set the output to the pin to zero. When we direct the pin to be an output pin, zero with low impedance will be visible on the pin. When we direct the pin to be an input pin, a high impedance is visible, and the pull up resistor of the bus will pull the line high. When we expect inputs from the line, we should set the pin to 1, i.e. high impedance. When we send data to the device, we should expect an acknowledge signal from the receiver. At that moment the master should have high impedance, but there is no harm if we forget that (except that we mislead ourselves to beleive that the acknowledge signal has come).

Data are clocked by the clock wire in the sense that the slave can trust the data, when the clockline is high. Hence, the data line should not change, when the clock line is high. If it does, this is taken to be a start pattern or a stop pattern. When the start pattern appears, the slaves should expect data on the bus. The first data after the start pattern is a device adress (plus a read/write bit). Those slaves which do not recognize the adress as their own, can now go to sleep, and they don't need to wake up again, until a stop pattern appears. Communication is now open between the master and a single slave, and the way the communication proceeds is pretty much an agreement between the two. What normally happens, is that the master sends data to the slave. If it wants to receive data from the slave, it normally sends a new start pattern and the adress, but now with a read bit. When one of the sides receives a piece of data, it acknowledges this with a single zero on the data bus.

The I2C protocol has an object file called iic.mpo. Its functions are called in the following way

( clockpin , datapin , )iic.init )iic.start )iic.stop )iic.ack )iic.wack ( data )iic.send )iic.rec ->data iic.ack will clock out an acknowledgement signal (=0) as a master is expected to do it. iic.wack will clock in the acknowledgement from the slave. The result will be returned. It should be zero, but you may skip checking that. But you do have to run iic.wack after you have sent data, because the slave needs the clocking to go on.

This is the basic protocol. You use iic.send also to send the slave adresses. But you have to consult the specification of your device, to find out which adresses and data you should send.

The iic-object is intended for making master software. To be a slave is actually more difficult. because you have to recognize start and stop patterns, and this is not supported by this object.

We will come to an application of the I2C bus in the radio application. Also, the Propeller standard EEPROMs use the I2C bus, and there are many sensors that use it too.

Formally, when you use the I2C bus for comercial use, you should pay a license fee to Philips.


SPI means Serial Peripheral Interface. Unlike RS232, it has a separate line for clock signals. Normally the clock signals are generated by a master device, so the master clocks in data from a device. This has the advantage that a computer can acquire data in a pace that it can handle, while in RS232 the receiving computer can be flooded with data. Data can be sent in both directions, normally over separate wires. You can tie them together, but with the risk that hardware is destroyed if both devices try to send simultaneously. Finally there is a chip enable wire, which enables a device, normally when that signal is low. All together, then, there are four wires. But for one wire more, you can communicate with two devices. One of the devices can then be disabled, so it doesn't listen, and it doesn't send. With seven wires, you can communicate with 16 devices, if you buy yourself a one of sixteen decoder circuit.

SPI is simple to program in both ends. The normal standard is to send 8 bits serially, but you can easily go to 9. A typical case of that, is when you send both commands and data. The data is normally 8 bits, but you can use the 9:th bit to distinguish commands from data.

SPI functions are on the spi.mpo object file. The functions are called like this:

( dipin , dopin , clkpin , )spi.init ( data )spi.send )spi.rec ->data ( data )spi.exch ->data ( data )spi.send9 )spi.rec16 ->data So, there is one function for sending 9 data and one for receiving 16 data. spi.exch shows one of the talents of SPI. You can exchange data with a device in a single session. The data you give to the function is sent over, and in return you get what the device has to offer. Technically this works as two shift registers tied together in both ends.

As you can see, the chip enable signals are not handled by the object. You do that separately in you application files. There is a newer "objectoriented" mpo file called Spio.mpo which contains standard "constructors" or init functions, associated with my standard pin association to each device.


If RS232 is a very old communication standard, Morse is even older. The good old Morse alphabet with combinations of long and short beeps, could that be something for our time? Perhaps. In the first place, I simplify the alphabet to this: 0:short pulse, 1:long pulse. Important in Morse signalling is also the pauses. There are short pauses between the individual pulses. If the pauses are longer, it means the end of the character. If the pauses are even longer, it means the end of a word (there is no code for "space"). Here, a slightly longer pause means "end of message". With that we have a variable length protocol. Away with the standard 8 bits. If some message requires 32 bits, we can send 32 bits. If in some case 4 bits is enough, we can send 4 bits. If we want, we can go to the original Morse alphabet, to send common characters with fewer bits. "e" was only a single short beep (a single 0). We have a self synchronizing system, no separate clock pulses, no hard requirements on common frequencies. All you require is that you can distinguish a long pulse from a short, and a long pause from a short. I have tried the system successfully up to an average pulse frequency of 500 kHz.

There aren't many Morse type devices around, so the only thing you can use this for, is for communication between your own propellers. I used it for communication with my RAM computer.

The functions are in the objct file morse.mpo. It has functions called like this:

( sendpin , recpin , )morse.init ( data )morse.send ( data , n , )morse.vsend ( )morse.rec ->data sendpin and recpin are the pins on which the Morse communication takes place. n is the word length. morse.send is a function for a default length of 32. The receiving function detects the wordlength itself, by observing the longer pause.


USB, Universal Serial Bus, is an almost indispensible protocol, as it is almost the only way to talk to a PC nowadays (specially a laptop). It is a complicated protocol though. It is a master slave concept, where the master is called host, and the slave is called device. Writing a host software for USB is certainly not easy, but we get it for free, when we by a PC. But writing device software is not easy either.

Fortunately, there is a Scottish company, called FTDI, that has made the job for us, in the form of integrated circuits, and hybrid modules. Most of them wrap an RS232 interface into a USB interface.

One of these FTDI modules is built into a device called Propstick, which is used for loading and running Propeller programs under control of a PC:

In the other end, in the PC, the RS232 reappears again as something called Virtual Com Ports. They behave as good old RS232 ports, but there is an arbitration of port numbers, that can cause the programmer some troubles, specially when the port number exceeds 10. As a Java programmer, I regret that there is no USB object available (it has been anounced), so I have to use Java Native Interface and C code.

An alternative to Virtual Com Ports, which is said to be faster, is called D2XX, which is a series of DLL:s to access the data.

FTDI also produces a module, where 8 bits a time are loaded in parallell. This allows a rate of at least 1 MByte/s, where the limit seems to be on the PC side, at least if one uses the Virtual Com Port concept.

There is no usb.mpo object file, as usb from the Propeller's point of view is nothing but RS232 (thanks to FTDI). But there is PC-communication object called pc.mpo. It uses the fact that the PC interface goes over standardized pins, 30 and 31, and with a standardized baud rate, 115200. The functions are called like this:

)pc.init ( adress )pc.ssend ( data )pc.longsend ( data )pc.bsend ) )pc.rec ->data pc.ssend sends a null terminated string, that it finds at the given adress. pc.longsend sends the four bytes of a longword. sends a carriage return. pc.rec receives a byte from the pc. To do all this you need software on the PC-side. For example, you can make a PC program that sends over a file byte by byte.

Wireless, Proto-Zigbee

The standard IEEE 802.15.4 describes a wireless communication protocol for the 2.4 GHz band, with checks, resending and acknowledgement, so that data are safely brought from one point to another. It more or less acts as an RS232 connection, where the cable is replaced with wireless communication.

On top of this standard, there has been built a higher level standard for communication in networks, called Zigbee. That's why I call IEEE 802.15.4 "proto-Zigbee". Zigbee is an advanced standard for building networks, where units beyond range can talk to one another by relaying over intermediate units. Important is also means for letting units go down to sleep mode, from which they can be wakened up on command. This is intended for battery driven data logging equipment and the like.

All these advanced possibilities are certainly usefull, but I have found much of it difficult to understand in detail, and handle. That's why I have stayed with "proto-Zigbee". Parallax have gone through different communication concepts and have chosen "proto-Zigbee" to be very easy to work with. Units are manufactured by Digi inc. previously Maxstream. But buying from them can be a little confusing, because it's hard to tell if you are buying Zigbee or "proto-Zigbee". Proto Zigbee is mostly called "IEEE 802.15.4 OEM-modules".

Proto Zigbee units are gathered in networks call PAN:s, Personal Area Networks, and each such PAN has its one PAN-ID. In the same place, there can be another PAN with another PAN-ID, and in that case, they are isolated from each other. Furthermore you can set which 2.4GHz channel you work on. You can set the role of the unit in the network. The simplest role is as an end-point. If your units are end-points, then they can communicate point to point with one another. You can also set the baud rate for communication between the unit and a computer. This is about what you need to configure in the first place.

When you buy units, they are configured to the same default PAN-ID, the same default 2.4GHz channel, and they are all set to be end-points. Then you can just connect your units to power and to two Propeller pins, and then you are ready to communicate between computers. What you send from one computer to one unit, arrives to the other computer via its unit. That's what we call cable replacement.

What you might like to reconfigure is the baud rate, and maybe move from default 9600 baud to 115200.

Once I bought three units, and later two more. But in the meanwhile, for some reason, that I have forgotten, I had changed the PAN-ID for my old units. So I had to spend some time reconfiguring. It took some time, but when I finally succeded, I had done it exactly as I had thought it would be done. So now all my units can talk to another. Among the object files there will be a simple program that may be helpful for reconfiguring units.

Say that you want to command a robot wirelessly. At the same time you want to upload sensor data to a PC, (which can also talk to Zigbee, it is RS232, but nowadays, you have to take the way over USB). There may be many ways to solve that. The simplest way, but maybe not the cheapest, is to have two PAN:s for the two tasks, command and data uploading. A unit can not belong to two PAN:s so you must have two units onboard the robot. This is a simple example of how you can use the simpler concepts of "proto-Zigbee".

Beside the simplicity ("point to point communication direct out of the box") proto-Zigbee has another important advantage: The units are cheap. You get them for something like 30 or 40 dollars.

The zigbee communication is controlled through a zigbee object called zig.mpo. Its functions are called in the following way:

( outpin , inpin , baud , )z.init ( inpin , baud , )z.initin ( adress , )z.ssend ) ( byte )z.bsend ( longword )z.lsend )z.rec ->byte ( adress )z.exe z.initin can be used if you just want to receive data with your zigbee-unit. In that way, you don't have to reserve any outputpin. z.ssend sends a null terminated string from the given adress. sends a carriage return character. z.exe combines a z.ssend, and z.rec.


Bluetooth is a complicated standard with a stack of different profiles, for transmission of images, music etc. One such profile is SPP, Serial Port Profile, which is "RS232" in both ends. Like Zigbee, Bluetooth has checks, acknowledges and resending to guarantee safe arrival of data. Parallax marketed devices called Embedded Blue, which you accessed through a RS232 interface. I found them easy to work with, but they are now out of production, and their replacers seem more complicated to interface.

The main drawback of bluetooth is however that the modules are expensive, about four times as expensive as the proto Zigbee modules.

Parallax' RF-modems

Parallax markets some simple transmitters and receivers. They are truly cable replacements. When you set the input to the transmitter high, the output of the receiver goes high. So you build your own RS232 link with that. The RF modules don't contribute with any security checking. I found that the transmission failed now and then. But I never found out how to pair transmitters and recievers in this system. There are new modules now which combine transmitter and receiver, and this could make it easier.


I have tried several displays over the last years. But right now I have a favorite. It is from Adafruit, and they call it "lovely". So I agree. It is pretty small, and the communication to it is not superfast, but it is easy to use, and it has a colorspace rich enough to show photographs. It is purely graphical, so if you want to show text, you have to do some programming. It is a 2.2 inches TFT-display with 220x176 pixels. More information is at". It is sold in Sweden at

(This product is now discontinued. I will look for their replacement, and study how it is driven as soon as possible.)

It is controlled with the SPI protocol. What comes on the SPI-bus is characterized as commands and data, where data has the 9th bit = 1. Hence, we use the spi.send9 function. Data are normally 16 bit data. For color data, all 16 bits are used. For things like coordinates, only 8 bits are used, so the first 8 bits are filled with 0.

Now, the principle for drawing is the following: You send a set window command, where you define a window on the screen, by the x and y coordinates of the lower left and upper right corners. Then you send color data. Pixels with these colors fill the rectangle collum by collum. That's all.

To draw text, you need a font, that you place in a global data block like this (so this is done in your main program after the global variables).

global . afont globaldata font @afont from(deascii.myd) (the variables "afont" and "font" must be exactly like that.) Then you have to set the colors of the font, typically white and black or yellow and black. Black is zero, and white is $FFFF. Yellow is typically $FFE0. You declare these as, say black and yellow, and then you make ( yellow , black , )gra.setcharcol Now you can draw text with ( x , y , char , )gra.type ( x , y , adr , )gra.stype In the first function, char is the ascii code of the character, so if you want to draw an "a", you just write ( x , y , "a" , )gra.type. In the second function, adr is the adress of a null terminated string, which is typed at the position. The easiest way to make such a string, is to write: . hello = "Hello World!" begin ( 50 , 50 , #hello , )gra.stype

Before all this, the display has to be initialized with: ( clockpin , datapin , cepin , resetpin , )gra.init This is a pretty long program, that among other things, sets the colorspace to be 16 bits. This is assumed by the other programs. The bits are used so that the first 5 bits represent red color, the next 6 green color, and the last 5 bits blue color. As this is such a long program, you normally has to run it in series with the programs that use the display. For this, the program starts like this: system something global s0 sd . . exec ... 1/initdisplay/usedisplay Then, the process initdisplay should call (...)gra.init, and then set the semaphore sd to 1. That will load the process usedisplay. Now, as the cog has been restarted with this new code, it has lost connection to the display, so first usedisplay has to reconnect to it with: ( clockpin , datapin , cepin , resetpin , )gra.connect This is code of a type you always have to use, when you run processes in series.

Here now, is a summary of the functions of the object file grafa.mpo: ( clockpin , datapin , cepin , resetpin , )gra.init ( clockpin , datapin , cepin , resetpin , )gra.connect ( xa , ya , xb , yb , )gra.setwindow ( color )gra.clear ( xa , ya , xb , yb , color , )gra.fillrect ( x , y , color , )gra.paint ( color )gra.paintcolor ( foregroundcolor , backgroundcolor , )gra.setcharcol ( x , y , char , )gra.type ( x , y , adressofstring , )gra.stype gra.paint draws the pixel at (x,y) with the indicated color. gra.paintcolor draws the color, wherever the display's cursor happens to be.

There are some other functions not mentioned in this list. There are some gra.connect-versions that have been set up for particular connections of the display to the computer. There are some internal functions, and there is a function to display a curve. curve is a global array, that you can fill with data elsewhere, and there are also ways of scaling the curve by modifying the parameters in the two functions gra.tscale and gra.fscale.

At the file grafa.mpp there are some "prefabricated" processes, that you can import to your program with lines like these: extprocess initdc from(grafa.mpp) extprocess typedec from(grafa.mpp) These processes also have to be mentioned in the exec line. initdc initializes and clears (in black) the display. You may have to modify the code, to adapt to how you have connected the display. For typedec you have to declare four global variables u1, u2 ,u2 and u4. Whatever you write to them, will then be typed out on the screen. There is also a process typehex, which can type in hexadecimal form, which can sometimes be more informative. It plays with two variables more, so you have to declare u5 and u6.

Replacement display

As mentioned, this 2.2" display has been discontinued. With Lawicel, the replacement is a 1.8" 18 color display, also from Adafruit. Adafruit may have some display of bigger size. So this display is slightly smaller, and it also has fewer pixels, which may be just as important, when it comes to modifying code. The image resolutions is 160 x 128 pixels The color quality is still so good, that you can use the display to display photographs. I have the impression that this display is slightly faster than its predecessor.

Information about how to program this display comes, as with the previous one, as C++ code at Adafruit's webpage. The code is written as common code for several different microprocessors, with a lot of #ifdef:s and #endif:s which makes the code a bit complicated. Furthermore, the designer has chosen to write the initalization codes, as a big field of constants interpreted by an interpreter. I don't think that really pays. A programming language is already like an interpreter for computer code, and it doesn't pay to wrap another layer of interpretation around it. So my code is just a straight code, sending commands and data to the display. Furthermore, some of the code is just sending values to registers who already have these values by default. So I have been able to eliminate parts of the code. There is code for two different versions of the display driver circuits, a B version and an R version. But to my understanding, only the R version is sold by Adafruit.

I also tried to remove the gamma correction code, to do with default settings. But then I found that the image become rather dim. So it pays to use the values provided on the web page. The gamma corrections somehow have to do with how transparent the pixels appear. I guess, if you are really interested in this, you could go on doing experiments to get an even better picture. Right now my picture has very good contrast, but it is a little dark.

The code provided is not for the raw display, but for a breakout board, which also contains the driver circuit and a micro SD card reader. You can run this board with just 3.3 volts, which you also connect to the pin marked "lite" for the LED background light. The new display, unlike the old one has a separate d/c pin to distinguish between command (0) and data (1). (In the old display, the d/c information was sent as a 9th databit). The rest follows the SPI protocol. So there is Clock and MOSI (Master Out, Slave In) and MISO (which you can skip) and Chip Enable. Finally there is a Reset pin, but I have the impression that you could do without that. With these pins connected, the following code should be equivalent with the code in the previous section. But remember that the display resolution is different. The code is packed into the object file grafb.mpo. I also provide a grafb.mpp with prefabricated processes.



RAM mean Random Access Memory. It means that, whatever random adress you want to write to or read from, the process of doing that, takes equally long time irrespective of the adress. This is not so for the other memories, where you have a block structure, that makes reading or writing take longer time, every now and then. Right now, RAM's forget everything, when power is removed from them. Nobody has come up with a design that would make a memory both RAM and non-volatile.

My RAM has 512 kBytes capacity. I think it is called LY625128SL from Lyontek, but I'm not sure. It comes in package called SOP32 which is for surface mounting, and not quite easy to deal with for an amateur. 512 kBytes requires 19 bits of adress. Data takes up 8 bits. Then there is a write enable and an output enable signal. So this chip uses up almost all the pins of a Propeller. That's why I have decided to use a dedicated Propeller for the RAM. Some sort of serial communication is the only choice for the communication with the Master Propeller, as there are only few pins left. So, I decided to use Morse communication, where I can manage with a single pin to communicate in both directions.

Then the two Propellers communicate with a command language, with the following commands:

  1. set adress (code 1)
  2. set adress 2 (code 2)
  3. set logg period (code 3)
  4. set logg size (code 4)
  5. write (code 5)
  6. write 2 (code 6)
  7. read (code 7)
  8. start logg (code 8)
So, the codes for the commands can be contained in 4 bits. When we send these commands, the code occupies the 4 least significant bits. Parameters occupy the higher bits. write and write 2 require 8 bits for the data and 4 bits for the command, so 12 bits are sent. read and start logg requires nothing but the command, so there, 4 bits are sent. For the others, the arguments to the command are allowed to take up 28 bits, so 32 bits are sent.

Note now, that we use the memory as a memory of bytes. We don't try to lump 4 bits together into a long word. When reading and writing, the RAM-Propeller will increment the adress by 1 at each operation. It is then natural, that we would like to work with two independent adresses, so that we can move data from one part of the memory to another. That's why we have two set adress commands and two write commands.

Some of the commands are associated with automatic data logging. We connect some external data source directly to the data pins of the RAM, and then we can set a period time for the logging (sampling time) and a total size of the logg. Then we can just start a logg. Afterwards we can retrieve the data with a sequence of read commands. To have the RAM data pins directly connected to external sources, can of course be a little dangerous.

We have a similar danger, when we connect the memory to the RAM Propeller. The datapins are bidirectional. Output Enble turns them into output pins. We shouldn't let the RAM Propeller pins be output pins at the same time.

The commands are encapsulated into functions in an object file ram.mpo. This object file is supposed to be used in the master computer. The RAM computer has its own software, which responds to the commands. Except for the function corresponding to the commands, ram.mpo has a function for writing data into an SD-card. )ram.init ( adress )ram.setadress ( adress )ram.setadress2 ( period )ram.setper ( size )ram.setsize ( data )ram.write ( data )ram.write2 ) ->data )ram.startlogg ( RAMadress , SDadress , n , )ram.tosd

The serial RAM

23LCV1024 is a static RAM, in a 8 pin chip, controlled through SPI. Thus the pins for communicating with it are only 4, but even that was too much, so I decided to use a dedicated SRAM computer. The interface is contolled by code in the MEMI.MPO object file. It has functions for different type of memory actions, and they send Morse signals for invoking these memory actions. Such a command consists of some data, and in the last four bits a command. We have:

write writes to adress a1, and write2 writes at adress a2. In this way chunks of data can be moved from one memory area to another. decradress moves the adress a1 backwards, so that the data can be read out directly after having been written.

These commands are then received by the memory computer, which impements them by calling functions in the object file SRAM.mpo. These are directed towards the serial RAM hardware, which uses SPI interface. There is other software and hardware for handling a RAM with parallell interface. Functions for logging data will be provided in the future.

With serial communication both between the computers and from computer to memory, one can't aspect very fast memory handling. What I get is still enough for my current needs. I can write a single bit, as part of a 8 bit byte, in 29 microseconds which corresponds to a baud rate of over 34000 baud.

As mentioned, the memory contains 128 kBytes, but I have four chips, so I can stack them up to 0.5 MBytes.

This idea of controlling an SRAM through a Propeller comes back in "Terminal project" which is presented here


An SD-card is an electronic component, which is built into a "card", that can be pushed into a slot. It contains a non-volatile flash memory and an interface computer. These cards have a huge capacity, and they are handy to use.

Here's a picture of a small interface box for the SD-card. You can see the card pushed into its slot:

The SD-standard is a semi-open standard. You can join an SD-society, but it probably costs a couple of milions. If you are not a member, there is a rough public spec. It is actually not bad, but some questions remain unanswered.

There is useful information on this site:

There are two protocols for the SD-card, one called SD-bus, and the other is SPI. Amateurs probably prefer SPI; the SD-bus seems unnecessarily complicated. A special issue is the so called CRC, Cyclic Redundancy Check. With CRC, you are supposed to add a so called CRC-code to the message you send. If there is any error in the transmission of the message or of the code, the receiver is able to detect this. CRC-codes are generated by a very interesting algorithm, but with the time I have spent on it, I haven't been able to penetrate it, so I can't compute CRC-codes in real time. Fortunately there is a web-page, that can compute CRC-codes off-line:

Fortunately, the CRC-checking is disabled in SPI-mode, though not quite.

So, we all want to use the SPI protocol. However, SanDisk, which is probably the major supplier of SD-cards, have abandoned the SPI protocol for their SD-cards, which means that they break the specification of their own society. I've tried Verbatim-cards and Kingston-cards, and they work.

SD-cards come in a newer standard called SDHC (High Capacity), and I thought it would be wise to adapt to this new standard, as it might be difficult to find regular SD-cards in the future, specially if I have to keep away from SanDisk. The point is this: You adress an SD-card with a 32 bit adress. That's good for adressing memories up to 4GB (232 is approximately 4 bilions). But SDHC cards are bigger than that. So instead of letting the adress be the adress of a single byte, you let it be the adress of a block of 512 bytes. When you read and write, you can only read and write full blocks. When you read you can find your interesting byte somewhere in that heap of data. This is in a way good. When you write, you always write a whole block, which you can do without problems. Block writing automatically opens the memory for writing. And you read almost the same way as you write.

But there is a job for you in this context. You have to tell the card, that you have understood, that it is an SDHC-card. For some odd reason you have to send this message with a correct CRC-code. But this is a static message, so the CRC-code is static, and you can find it on that web page.

The cards are controlled by numbered control commands. The message starts with binary 01, and then follow 6 + 8 bits of command. Then follow three bytes of parameters associated with the command, then the CRC code and a single bit 1. In the sequel I give the command number as including the 1 in the second bit. That means that all commands are greater than $40 (in the way that hexadecimal words are written in the Propeller world.)

Card initalization

Now, let's take it from the beginning. The SD-card needs clock cycles to get through its own initialization procedure. It seems that the card computer doesn't have a clock. A simple way to achieve this is to send 8 bit data with the SPI protocol. I send 20 such bytes, which is probably more than enough.

Now, the card wakes up in SD-mode. It is comfortable but strange, that we can still send an SPI-message to it. That message is $40 (hexadecimal 40), with a zero argument. And it has to have a correct CRC-code. But many amateurs have found out that this CRC code. is $95. (formally it is $47, but after you've squeezed in that final 1, the last byte is $95.) This command puts the card in idle mode, but it also tells the card, that we will talk to it in SPI-mode

Then we must give the message "I know, that you are an SDHC card". The command is $48, but this is a kind of a group command, so it needs a parameter $1AA. The CRC-code is $43.

Then we need to leave the idle mode. It is a command $69 but it belongs to an expanded group of commands (so called app-commands) so we have to send the command $77 before it. The command $69 has a parameter which is 1 in the thirtieth bit. The card always sends a response to the commands. There are different categories of responses, but this is the simplest of them. In it, the last bit represents that the card is still in idle mode. So we will keep sending these commands until the response is 0. With that, the card is initialized.

Once you have downloaded the software, take a look to see if it fits with this description.

Writing the card

To begin with, writing is simple. We just write the data to our own writebuffer in the Propeller. We build the writebuffer in longwords, and there are 128 longwords in the buffer (=512 bytes). We strip off the last 7 bits of the adress as in-buffer-adress. In every such write operation we increment the adress by 1. Initially, we have set the adress through a set adress function.

But when the in-buffer-adress reaches 127, we have to do more. We have to write the data out to the card itself. The command for this is $58, and the parameter is the block adress, which is the adress except the 7 last bits. We can then put anything as CRC code. This command should then be followed by a so called token, which in this case is $FE. Then we send the data byte by byte, so we have to extract these bytes from the longwords in the writebuffer. Then we send two arbitrary numbers as CRC for this whole data packet (this is supposed to be a longer CRC of 16 bits, while the CRC codes mentioned before are 7 bits long), and then we wait for a response from the card. The card now has all the data, but it takes a while to actually write them to the card. In the meanwhile, the card will send a busy signal.

Reading from the card.

The read system tries to find the data in the read buffer first. This supposes that no one has written new data to the card since the buffer was downloaded from the card. This fits with the assumption that we don't write and read the card in anarchy.

For this, the system has a current block number (called rbl), which is the number of the block most recently downloaded. From the read adress, the block number can be extracted as the bit pattern except the last 7 bits. If this block number equals the current block number, the data can immediately be read from the buffer.

If not so, we have to download a new buffer from the card. This is done with the command $51 with the block adress as argument. Then we wait for a response from the memory, and then we clock in the data by using the spi.exch command. We also use it to clock in two bytes of CRC-code for the package from the card.

Now the buffer is updated, so we can read the desired datum from the buffer.

The object file

The SD-card is handled from an object called sdh.mpo ("h" for "H" in SDHC). The functions there are called in following way:

( dipin , dopin , clkpin , cepin , )sd.init ( dipin , dopin , clkpin , cepin , )sd.connect ( adress )sd.setadress ( data )sd.write )sd.readbyte ->data ( longadress ) ->data Except for these functions, there are some functions used internally. sd.write and sd.readbyte write and read sequentially from the adress set in sd.setadress. requires an adress to read from in each call.

I have been using a breakout board for an SD-card holder. It is marked with pin designations according to the SD-bus standard, so here is a translation to SDI-designations. Note that SO means "out of the card" so it is for traffic into the computer. The opposite holds for SI.


The standard program memory that Parallax uses for the Propeller is called 24LC256, but if you can find it, it pays to buy a 24LC512 instead. You get twice as much capacity for almost the same price. 512 means 512 kbits, organised as 64 kBytes. (we will mention a version with even more capacity here)

But you can do more than that. The chips have three pins called adress pins. Parallax advises you to ground all these pins, to make for the adress 0. In fact you don't need to. The pins have a pull down circuit, so they "ground themselves". But if you add another chip, you can give it another adress, by pulling some adress pins high. Then, alltogether, you can have 8 different 24LC512:s with unique adresses. So, the maximum capacity of such a system is 512 kBytes or one half Meg. We call these subadresses, set by the adress pins, chipadresses.

These memories are pretty straight forward I2C-components. In the I2C world, each circuit type has its own device adress, though these device adresses can hardly be distributed in such an orderly fashion as IP-adresses or Bluetooth adresses. Anyway, these memories have the device adress $A0. But then the above mentioned chipadresses fill the three bits after the 'A'. The final bit is a read/write bit; 0 for writing. Then the chip with the chipadress 3 is adressed for writing with $A6.

The protocol is then as follows:

If you want to write to the chip with chipadress 3 (as an example), you start with the start pattern. and continue with the code $A6. Now the appropriate chip will listen, and all others will go to sleep. Next you send the adress within the memory, which takes up 16 bits for 64kBytes, so you have to send this adress in two bytes (the most significant first). Then you can just send your data in bytes. After each byte, you have to wait for an acknowledge signal from the memory. (it's no really just waiting. You're the host, so you have to send the clockpulse to get any response.) Finally you send a stop pattern.

If you want to read from the memory, you start the same way, sending device adress, and the adress within the memory. But then you have to send a new start pattern, and then again the device adress but with a read bit. So in the example, the code to send is $A7. Then the component will send you the data byte by byte, encouraged by the acknowledge patterns you send. You stop the traffic at any time by sending a stop pattern instead of an acknowledge pattern.

These memories behave pretty much as RAM-memories in the sense that the time it takes to read and write, is independent of the adress. But they are slower to write than to read. When you have sent data for writing to the memory, they have just been buffered so far, so you have to wait some extra time, before they are acutally written.

The object file eeprom.mpo contains functions for handling the EEPROM. You can initialize the memory system, and then you can set the chipadress (called chipcode), to adress a particular chip. There is no particular support for writing over the boundaries of the chips. Then you can set a start adress for reading and writing, and this adress is then incremented, when you read and write. You can read and write bytes and longwords. But you can also move chunks of data between an array called ee and the EEPROM. If you use that, you have to declare ee as a global variable. Here's how the functions are called:

( clockpin , datapin , )ee.init ( chipadress )ee.setchip ( adress )ee.setadress )ee.readbyte )ee.readlong ( data )ee.writebyte ( data )ee.writelong ( from , to , n , ) ( pageadress )writepage from is the adress in memory from which reading starts. to is the adress in the array ee where the first data are written. n bytes of data are transferred. The concept of page is an area of 64 bytes in the memory. pageadress is a numbering of the pages, which is made by stripping off the 6 last bits of the normal adress. writepage writes the 64 first data in the array ee to the page in memory.

You can upload data for a program to the EEPROM. Then you need an upload program, using the write functions. And then the actual program has to use the read functions. On the other hand you can use the EEPROM to logg data with the write functions. Then you have to start a download program using the read functions, to get you data to, say, a PC.

Flash memory

The flash memory I have used is from SST and is called SST25VF032B. It has a capacity of 4 MBytes. It comes in an 8 pin surfacemount package, which is not all that hard to work with. It's a pretty straightforward application of the SPI protocoll.

Two things are a little more difficult. One of them is the write protection mechanism. It is nice of SST to be careful about my data, but after all, I would like to write something at some time. So we have to break open the write protection mechanism. First there is a WP-protection pin, that has to be high. In that situtation, it is in principle possible to write to the status register, that controls the write protection. You can do that with a Write Status Register command WRSR. But before that, you have to enable the writing of the statusregister with an Enable Write Status Register command. But that command is not allowed, unless a Write Enable command has been sent before it. For all these instructions you have to handle the chip select bit, that belongs to the SPI protocol. When you have found all this out, it is not terribly complicated, but after you've read the spec, you couldn't feel sure about it. It is about trying. In the object file Flash.mpo there is a function called which does all this. It involves a call to fl.wren, which is used in many places.

The other difficult thing, is that this is a flash memory, and you don't write to flash memories just like that. You have to erase it first. You can erase the memory in blocks of 1024 longwords (there are 1024 of these). When the block is erased, all bits are just 1. While writing, you can change any bit to 0. If you try to write to the same cell again, new bits are changed to 0, but nothing ever gets 1 without erasing. There is also a command for erasing the whole memory. It is the fastest way to erase the memory, but you erase all your data. All this, of course, calls for some planning.

The protocol is like this. You send a command number, and then the adress, split to 3 bytes (which would be good for 16MBytes, but 2 bytes of adress is insufficient for 1 MByte). And then you send your data, or receive your data. You terminate this traffic by pulling the chip enable bit high. Here are the command codes used:

And here are the functions, and the way they are called: ( dipin , dopin , clkpin , chipenablepin , )fl.init ( byteadress )fl.bread ->byte ( longadress ) ->longword ( longadress )fl.erase )fl.chiperase ( byte , byteadress , )fl.bwrite ( longword , longwordadress , )fl.write ) You can see, that we look at the memory in two different ways, either as a memory of bytes, or as a memory of longwords (32 bits). When we regard it as a memory of longwords, we adress it accordingly, i.e. with a longword adress. The next longword in memory then has adress + 1, even though it is 4 bytes up. In fl.erase, longadress is any longword adress. The system itself will erase the whole block, that contains that adress.

If you look at the code for fl.write, you can see that it uses fl.bwrite four times, which is far from efficient, so there is room for improvement (You can send all four bytes of a longword in a single write command).

You can also see that writing a byte takes a little more than 15 microseconds. But to erase a block takes 40 milliseconds, and this will effect the speed of writing much more. In my applications, the data rate has been lower than one data per 40ms, so then there is no problem, but if this is not the case, you have to buffer data, and erase the next block while buffering data.

A File System

I have made a file system for the flash memory. I tried to do it for the SD-card, but I haven't been successfull yet.

The key issue for a file system is that you should be able to erase files. After you've done that, you have some space left. If you try to squeeze in a bigger file there, you have to give up. If you have a smaller file, you will get an even smaller space left. As you write and erase, the memory space will be more and more like a Swiss cheese. The trick to handle this, is called fragmentation. You divide all files in equally sized fragments. When you have erased your file, a number of fragment spaces will be free to fill with fragments from your new file. If the new file is smaller, then some fragment spaces will still be free for future files.

The fragments are kept together by links. So when a fragment space has been filled, you have to find a new fragment space elsewhere, and then you write a link to that fragment space, last in your current fragment. When reading the file, we have to follow these links through the file.

In operating systems like Windows, there uses to be a program for defragmentation. This might seem as a cure for the fragmentation disease. But fragmentation is not a disease. It is a necessary part of a file system. But - for physical memory devices like hard disks, it's better if the fragments are not physically too far away from one another. To move from one part of a hard disk to another, far away, may take maybe a millisecond. It doesn't sound like much, but if you have to do that thousands of times, when reading a big file, it gives you a very noticeable delay. Hence, defragmentation is about moving the fragments of a single file as close to one another as possible on the surface of a hard disk.

You may think that a file system is much about creating a hierarchy of files with so called directories. But all this is just a name issue. Say that you have a directory c:\propeller\mps and that you have a file there called flash.mpo. Then the actual real name of that file is c:\propeller\mps\flash.mpo. Now, if the systems thinks you are "in the directory" c:\propeller\mpo (and informs you about that with e.g. a prompt), then the system will automatically add c:\propeller\mps\ to the filename you write. So if you write the filename flash.mpo, you will automatically get the right filename. This gives you the impression that the file flash.mpo is in the directory c:\propeller\mps\. In fact, all files are all over the place on the disk, specially as they are fragmented.

My file system is much simpler, when it comes to names. A file name is only 3 bytes long, i.e. three characters. It doesn't allow for any system of directories, but it is enough for my purposes. The fragments of the files coincide with the blocks of the flash memory, i.e. they are 1024 long words. That is really a little too much, if you have many small files.

Then there is a file catalog, which takes up the 64 first longwords of memory block 0. Each file here is represented with its name in the first three bytes of the longword. The last byte contains the block number, where the file begins. This allows us to let the files begin in the first 255 blocks. But then the files can continue in blocks with higher block numbers. After the file catalog, in block 0, follows an account for which blocks are busy in the memory. Each block has a bit for this. So we need 1024 bits, which takes up 32 longwords. Block 0 is always marked as busy, as it should contain the file catalog - and this busy accounting structure. The busy accounting structure is called occu.

In each block, words 0 to 1022 are sequentially filled with data, but the last word is the link to the next block. When we try to write data there, we should observe that this place is reserved for a link. Then we should find a new block, by inspecting occu. We prefer to look for free blocks above number 255, as the lower numbers are good to have as first block in each file. We don't need to erase the new block, because we assume that this is done either initially, or when we delete a file. After we have chosen a new block, we just write its number in cell 1023, and then we move to the new block, and write our datum there.

The file catalog and occu belong to a block in the memory. Such a block can not be handled easily, as we cannot change data in the block without erasing the entire block. Due to this, we have to create a mirror image in the Propeller of this block. We keep the two mirror images updated by writing mutually, whenever anything changes.

There is a danger in this for battery driven equipment. The computer will stop working when the battery falls under a certain level. This is likely to happen when a block erase of the memory occurs, as a block erase consumes pretty much current. If so, the computer is lost in a precious moment, when it should restore the file catalog after it has just been erased. I've seen this happen. The file catalog is gone, but with some detective work, I can still read the files. But with the file catalog gone, it is impossible to manage further writing to the memory. Of course I can't be quite sure about the causes of this loss of the catalog.

The state of an individual file is just, except for the data we have written into the memory, a pointer to where we stand, and are ready to read or write. (In this system, a file is either being written or being read, never both.) The pointer is practically three variables. One is block, which is the block where we stand, and the second is pos, which is where we stand in position within the current block. The third variable is called bit, and tells us which bit we are currently reading or writing. It would allow us to write and read the memory bit by bit, but right now it is only used to read the memory byte by byte. But normally, the file system regards the files as files of longwords. (Slightly unconventional, but I prefer to view the Propeller as a longword machine, and then it is practical to regard the memory as the same.)

The file system is built according to the principle of a class, and instances of a class. This means that the variables that belong to the file, i.e. block, pos and bits are arrays. The index to the arrays is the last argument to every function. (This is what was called unit numbers in Fortran). So you can open a file with index 3, and write data to it with that index. At the same time, you can have another file with number 2, which is completely independent.

Files are files of longwords. The file system, as mentioned, is kept in the flash memory. Thus, the file system serves as an interface to the flash memory, where the flash memory is completely invisible to the user.

Here's how the functions at file.mpo are called. The file index (Fortran unit number) is called k:

( dipin , dopin, clkpin , cepin , )f.init ( dipin , dopin, clkpin , cepin , )f.connect )f.grandreset ( name , k , ) ( name , k , ) ( data , k , )f.write ( adressofstring , k , )f.swrite ( k )f.close )f.eof ( k ) ->data ( k )f.readbyte ->databyte ( name )f.delete ( c1 , c2 , c3 , ) ->name f.init sets up the communication with the memory and f.connect reestablishes this communication, if we have changed process code in a cog. f.grandreset resets the whole file catalog, and erases the whole memory. In the PC-world this is something you do "once in a lifetime" for a computer, but for these smaller devices, I guess you do it more often. A Propeller is not a part of an archive. opens an existing file for reading. opens a brand new file for writing. f.swrite writes a null terminated string to the memory. The format is the usual one, that each character takes up a full longword. f.close closes a file by writing an "end of file"-symbol at the current pointer position. The symbol is "e","o"."f",0 packed into a longword. For comparison, you can retrieve that symbol with the funcion f.eof. f.readbyte reads a single byte from the file, and then moves the pointer one byte forward (on fourth of a longword) by manipulating the bit variable. In the f.delete function, you refer to the file by name, and not by unit number k. Finally is a "name factory". It packs the three characters c1, c2, and c3 in the three least significant bytes in name. This is the form names have, when you call and

When you use this file system, you have to make the following declarations of global variables: block[5] pos[5] bit[5] cat[64] occ[32]

A File System for the EEPROM

Lately, the EEPROMs as described above have come in a version with 1024 kbits, which could maybe be enough for some applications. Problems I have had with a GPS logging system, encouraged me to make a try to use these memories for GPS-logging. In that application, I store three longwords per seconds. That is 12 bytes, and the memory has a capacity of 128 kBytes, so I could store 10000 samples, which is well over 2 hours. If I would like more, I could change to sampling every 2 seconds.

First a note about these components. The 1024kbits is organized as 128 kbytes. To adress that space requires 17 bits. But in the protocol to control these components over I2C, we only send 16 bits. Rather than changing this, the manufacturer has chosen to send the 17th bit in the device adress. That device adress consists of the device code (hexadecimal A) + three free bits + a read/write bit. The three free bits were supposed to to match values set to three input pins on the chip. Now, the first of these bits is instead used for the 17th adress bit. Previously we had 3 bits, so that we could build a battery of 8 chips. Now, with only 2 bits left, we can only build a battery of 4 chips. The chip still has three adressing pins, but now we can only use two of them. For some reason, it has been decided that the third bit has to be connected to the voltage supply line. If you forget that, the memory won't work. This is described in the data sheet.

The EEPROMS have the advantage over the Flash memories I've used, that I can write to individual adresses. That means that I don't need any mirror areas in the computer. As I had some problems with that (see here), I decided to build a file system for the EEPROMs, as an alternative, and that is now what I use in the GPS logg application.

Writing to individual adresses is fine, but it comes at a cost. Firstly, writing an individual byte takes as long time as to write a whole page of bytes, and this is about 4ms. If you write individual bytes, this is the average time it takes. If you could buffer your data, and write a whole page at once, that again would take 4ms, so your average write time would be much shorter. But the maximum write time would still be 4ms. But there is more to it than that. EEPROMs, which are in fact flash memories, wear out after a milion writes or so. When you write individual bytes, the memory chip still has to write the whole page. So the number of writes is much bigger giving the EEPROM a shorter life. But for the applications I have in mind, I don't think this matters too much.

Then, another thought has caught me. In my other file systems I had a global array of pointers to point at current block and current position. This was fairly costly, so I decided to have only single pointers, and then I could make them local, by letting them be field variables in my file system object. That would mean that I could only hold one file open at a time. But this is not quite true. If I manipulate files in separate processes, I can handle them individually. The pointers are local variables, and so there are instances of them for each process (cog).

The end of this, is that there are no global variables at all. The file catalog and occ are in the EEPROM only, and the pointers are local field variables.

The EEPROM has pages of 128 bytes (the smaller EEPROMs have pages of 64 bytes). I like to see the memory as a memory of longwords, and then the pages are 32 longwords. The first page, Page 0, is used for the file catalog. With that I can maintain 32 files. The next page is used for the correspondent of occ, i.e. as an account for which pages are occupied by files. There are altogether 1024 pages, and these can be represented by the 1024 bits of the 32 longwords in page 1. Page 2 is reserved for a single filename for the GPS application, so actual data are stored from the Page 3 and up.

Then, files are stored in "fragments" filling up one page each. In the file catalog, each entry consists of a three byte (character) file name, and then a pointer to the first page. Then, when a page is full, a new free page is sought for, and a link to that new page is written into the last word of the page. The EEPROM, unlike the other flash memory, is never erased completely, so to be sure, whenever a new page is reserved for use, a pointer 0 is written to its last word. This 0 will be overwritten by a new link, when the page is full. Specially, in the GPS applicatione, there is never time to write an end-of-file-symbol, because logging ends when power goes away, and then it is too late to write the end-of-file-symbol. Then, the end of the file is noted when a 0-link to the next fragment appears.

To read the files, one just follows the links through the fragments, until an end-of-file-symbol is encountered, or until a 0-link is detected. In the latter case, we have usually read a little too much, but it is not always a problem. When deleting a file, we have to follow the links, and set the corresponding bits in the words in page one back to zero again. And we erase the entry in the file catalog. But no actual erasing of data is necessary.

Here are the functions of the object in fileE.mpo, where the 'E' of course symbolizes the EEPROM:

The remaining functions are for internal use. There is no )f.connect function as such a function couldn't be smaller than the )f.init function. It wouldn't be necessary to initialize the file system in its own process.

The file fileE.mpo uses functions from the object file eepromm.mpo, which interfaces to the bigger EEPROM 24LC1025.

Human Interface Devices

Potentiometers can be good as human interface devices. Their outputs can easily be connected to A/D converters, and those are controlled by functions in the standard object file std.mpo. Some other interface devices require a little more.


The object file key.mpo contains a relatively simple collection of functions for handling a standard ps2 keyboard developed for the PC many years ago. The ps2 communication was a serial communication protocol not quite unlike the SPI protocol. The communication is handled by two function kb.rec and kb.send. Unlike SPI, the communication is clocked by the keyboard, so as a receiving computer, you have to be alert to catch when the operator hits a key.

The object normally only receives data from the keyboard, but there is a historical circumstance, that forces us also to send to the keyboard. The creators of ps2 decided that the PC and the keyboard should synchronize the following way. The keyboard would repeatedly send a character, but with the wrong parity. Synchronization would be established when the PC complained about this. So we have to send such a complaint message.

The user has to call the function key.init. He also has to declare two global variables key and keyp. Keys are scanned with key.scan. This is the function that has to be alert at all times. Thus, it has to have its own process.

The keycode for the key pressed the latest, will be deposited in the variable key. Keycodes are seemingly arbitrary numbers assigned at the dawn of computing, but they do reflect the position the key has on a standard English keyboard. The A-key has its number, as well as the shift key. The latter would enable us to distinguish 'a' from 'A', but we don't care about that here. When a key is released, key will assume a value called 'break'.

Now, we can call the function key.ascii. It will translate the key code to the corresponding ascii code and return that. As long as we keep a key pressed, a call to key.ascii will return the ascii code of the key. When the key is released, the function key.ascii will return 0, which is the ascii code for the null character.

Right now, I have a numerical keyboard, rather than a full size keyboard, as it is handier. Then, to produce figures, I use the keycodes, that the numeric keys have on the so called numpad part of the keyboard.

So the functions to call are the following:

)key.init )key.scan )key.ascii


As an amateur, you can buy a type of touchscreen formerly used by Nintendo. They are cheap and simple to use, but of course they don't compare with modern touchscreens used in smartphones. They are still usefull, but you need a little stylus to handle them; they don't work well with your fingers.

They are resistive devices consisting of two conductive sheets, which can be brought to touch one another, when you press them. One of the sheets has two horizontal metal bars up and down. The other has two vertical bars right and left. Say that you apply high voltage (3.3v) to the upper bar, and low voltage (0v) to the lower bar. This will create a potential field that varies linearly over the sheet. Then you connect say the left bar to an A/D converter, and let the right bar float. The high impedance of the A/D input means that there is no current through the second sheet, so the voltage over the second sheet will be constant. When the two sheets touch one another, that voltage will be the voltage at the contact point. In this way, we get a measure of the y coordinate of the point of touch. To get the x coordinate, we just have to shift the roles of the upper an lower sheet.

To put this under program control, we connect all four bars each to its own Propeller pin. For the active sheet, we make them output pins, and give 1 and 0 to them. For the passive sheet, we set the pins as input pins. Then we connect two of these pins to input channels of the A/D-converter.

We also have to detect the situation that the touchscreen isn't touched at all. For this we connect one of the A/D channels to 5 volts through a reasonably big resistor (10 kohms should be good). This resistor will pull out the voltage to values, where it couldn't be if there were contact.

Here are the functions of the object file touch.mpo, and how they are called:

( chy , piny1 , piny2 , chx , pinx1 , pinx2 , )touch.init )touch.get 0>x 1>y 2>valid )touch.get )touch.censor 0>x 1>y 2>valid touch.get delivers its result in the three variables arg0, arg1 and arg3, and they can be retrieved with the instructions shown. If valid is false (0), then the screen is not touched, so you should not use the values. The last line indicates the use of a function touch.censor. With that code, x and y will not be updated if there is a big jump in values. You can also download a program draw.mps (see under MPS Objects below) that lets you draw sketches on the touchscreen, that appear on the graphical display. Move the stylus out towards the right edge to erase.

Here's the setup with the touch sensor and the display, and the greek characters alpha, beta and gamma drawn.

And here's the result (There is an interference pattern called Moirée pattern, when you photograph a display with a digital camera):

You can mount this touchsensor on top of a graphical display. That's when the thing really deserves the name "touchscreen". But these touch sensors are sold separately, and they seldom have the same format as a display, and if you mount them together, there is an unavoidable distance between the two, that creates a parallax. So I never did that. Otherwise, the idea would be that you would draw icons on the screen, and point at them with the touch sensor. I have preferred to draw the icons on a slip of paper, and put that under the touch sensor. It's not so dynamic, but it gives you a multi mode mode control.


An accelerometer is not in itself a human interface device. It's a sensor. But you can use it to build a joystick.

Those who sell accelerometers use to say that they can be used to measure two things: acceleration and tilt. In fact, they can measure none of them, but just a confusing mixture of them. The deeper reason for this is that acceleration and gravity are indistinguishable, which was the basis for Einstein's foundation of the General Theory of Relativity.

But this means, that if we don't move the accelerometer with too much vigour, we can sense its tilt angle. Then I've built the accelerometer into a plasic tube with a heavy brassfoot, that makes it stand up: The accelerometer is padded in there between two pieces of plastic foam.

If I grab it and tilt it, I can read off the tilt position, and use that to control a program.

But you can also mount the accelerometer on a vehicle, and use it to measure acceleration. But remember: The signal is always disturbed when the sensor is tilted.

The code is made for an acceleration sensor called Memsic 2125, which delivers pulses, whose widths vary with acceleration, and with a width of 5 ms for no acceleration. These accelerometers are sold by Parallax. Electrokit in Sweden sells an accelerometer called DE-ACCM2G. It has direct analog outputs for the acceleration, so with an A/D converter, they are easy to use. It has a range of +-2g, which is more than you need for the joystick application, but there is a +-6g version also. SparkFun sells, among others, an accelerometer from Analog Devices called ADXL335, which is probably of very high quality. It is a three axis accelerometer, while the others mentioned ar 2 axis. The outputs are analog (I think),

The object acc.mpo, which works for the accelerometers from Parallax, has functions to be called as follows:

( xpin , ypin , )acc.init )acc.get 0>ax 1>ay There is a third function for scaling, but it's used internally. acc.get delivers the results in arg0 and arg1, but they can be retrieved with instructions shown. The output range is 0..4096 with 2048 representing zero acceleration.


Odometry comes from a Greek word for "way" or "road", and it is the art of determining travelled distance. To do it we have a striped strip and two optical sensors. We can put that strip around a wheel of a robot, to measure its travelled distance, or we can glue the strip on the road itself, that we travel along. Here's how it works:

We have two pictures of a strip and two optical sensors (the circles) in both pictures. The optical sensors are reflex detectors, which consist of a light source (IR-LED) and a light sensitive component. The strips move as indicated. In both these cases, the right hand detector switches from white to gray, which indicates a motion. But which motion is it, the upper or lower case? We can tell by looking at the left hand sensor. In the upper case it is white, in the lower it is grey. So now we can detect a motion "with sign", and we can accumulate the incremental motions to a distance travelled.

The Odo object odo.mpo handles two strips and two sets of optical sensors, to determine movements in the plane. The functions are called as follows:

( chx1 , chx2 , chy1 , chy2 , )odo.init )odo.get 0>xo 1>yo chx1 etc. are the analog channels to which we have connected the optical sensors. The result from odo.get is delivered in arg0 and arg1. The result is the total travelled distance since initialization.

Color Object

This is a very simple object, that handles a three color RGB-LED. The use is maybe most for debugging. You can get a certain color, when you come to a specific place in your program, or when a variable has assumed a specific value. To make it simple to use, the object has no init function. You call the only function in the object col.mpo like this:

( number ) The color emitted is controlled by number as follows: The current program assumes that the LED pins are connected to pins 5,6 and 7. The LED is an array of three diodes connected with the cathodes common. It is connected like this:

With the extra silicon diode at the green LED, we get a reasonable balance between the colors. The supply voltage of 5 volts is not critical, but 3.3 volts may be too low.

It's nice if you can dim the LED by letting it shine through a piece of white plastic. I let it shine through a nylon screw threaded into the computer box. With the diode in place, you can make other stuff, like mixing colors using pwm technique.

MPS Objects

Here are download opportunities for object files already mentioned:

A way to get these files is as follows. Click one of the links above. Then hit ctrl/A to select the full text. Then hit ctrl/C to copy the text to the computer paste buffer. Then open e.g. Windows Notepad, and drop the text into it by hitting ctrl/V.

Complex Number Object

Here's an object file for objects that are so called complex numbers.

In higher mathematics, a complex number is defined as a pair, (x,y), of real numbers x and y together with the following addition and multiplication rules:

(x1,y1)+(x2,y2) = (x1+x2,y1+y2)

(x1,y1)*(x2,y2) = (x1*x2 - y1*y2, ,x1*y2 + y1*x2)
Then, there is the style, to write (x,y) as x+i·y. We think of x + i·0 as just the real number x. We call 0 + i·y just i·y, and 0 + i·1 just i. Now, if we take (0,1)*(0.1) alias i·i alias i2 and use the above multiplication rule, we find that i2 = -1.

This has lead people to name i "the square root of minus one". And it is, but only if we introduce these very special pairs of numbers, called complex numbers, and with this very special mulitiplication rule, and only if we consider the pair (-1,0) as the same thing as the real number -1.

If we accept that i2 = -1, and write our pairs as x + i·y, and expand the product (x1 + i·y1) (x2+i·y2) as a normal product, we find the above multiplication rule.

We have these complex numbers for two reasons: for old missunderstandings, and because they are useful. Among other things, they help us understand n:th degree equations, and their solutions (roots). The main result is that an n:th degree equation always has n roots (though some of them may coincide). Hence, an object file for complex numbers may be useful.

So a complex number is a pair of numbers. We call the individual numbers components. In (x,y) we call x the real component, and y the imaginary component. We cannot pack both components into a single 32 bit number without loosing too much precision, so a complex number has to be two 32 bit numbers. So, to present a complex number to a function, we need to send two arguments, called xr and xi. And to send two complex numbers to a function, we need to send four arguments. The functions return the real component of the result in arg0, and the imaginary component in arg1.

Here, then, is how the different functions in the object are called: ( xr , xi , yr , yi , ) 0>zr 1>zi ( xr , xi , yr , yi , )c.times 0>zr 1>zi ( xr , xi , )c.norm ->squarednorm ( xr , xi , )c.conj 0>zr 1 >zi (Remember that retrieving arg0 and arg1 with 0>... and 1>... doesn't work if the receiving variable is global! Write explicitly arg0 ->... and arg1 ->... in that case!) represents the complex number addition rule, and c.times represents the complex number multiplication rule. The function c.norm actually computes the square of what uses to be called the norm of a complex number. So, the function returns xr2 + xi2. c.conj returns the complex conjugate of a complex number, i.e. the number where the imaginary component has switched sign.

Complex numbers in application usually have decimals, but are limited in magnitude. To allow this, they are scaled so that the binary point is sitting at the 16:th bit. To make this work, the above functions use a multiplication function called c.rmul to multiply the different components. It is internal to the object, but you have to consider this when you set values to the components of the complex numbers.

Download complex.mpo


PC communication

We start with this abstract and somewhat involved matter, as it is used by some of the applications below. It is about communication between a Propeller and a PC. Once you have that, you can use it for building human interfaces, for data collection and for debugging.

The PC communicates with the Propeller over an RS232 link connected to pins 30 and 31. This link is used for uploading and starting programs, but it can also be used for sending data to and from the PC. As PC:s don't have RS232 interfaces nowadays, the RS232 goes to a device called Propstick, which contains a module from FTDI, which converts RS232 to USB. Inside the PC the serial communication is converted to something called Virtual Com Ports. In some software sense, these virtual ports look like real RS232 ports, so you can connect serial port software to them.

Unfortunately there is no such software in Java. There was once, and there still is for Linux and Unix. But the Windows version has been abandoned by Sun. So I need to use Java Native Interface together with C code. Having done that, I have almost made the work that Sun and Microsoft could have done together. If you study my C code, you can see that I am not exactly a C expert programmer, so you are welcome to improve my code. I have a 32 bit Borland C compiler, so I have to compile my Java code for 32 bits, which I do by using an older version of JDK (Java Develpment kit). Here now are the basic pieces of code for communicating.

When you compile Port.c, you should make it so that you get a .dll file. Then, if you write a main Java program, that uses the object, it should be possible to compile it.

On this, we build a basic program to communicate between the PC and the Propeller: You need some subobjects for this too. These are This is a fairly general program for communicating with a Propeller (or anything that supports RS323) over a Propstick. When you connect your Propstick to the PC it is assigned a port number for a Virtual Com Port. Write this number into the code of at this line: {port = new Port(3,115200,(byte)0,(byte)8,(byte)1); (In this example, the port number is 3.) You also have to dive into your PC and set the speed of that port to 115200 baud.

When you start the program you get a window with 4 textfields:
  1. The 1:st textfield presents everything, that the Propeller delivers, as if it were characters encoded with ASCII-code. "Numerical" data that you get in this way will look like nonsense, but you can also choose to let your Propeller send messages with ASCII-coded characters.
  2. The 2:nd textfield presents the same data as hexadecimal or decimal figures. Refer to the 4:th window, to see how you change that.
  3. The 3:rd textfield is you input field. Everything that you type there, is sent over to the Propeller
  4. The 4:th textfield is a mode field. Normally you should erase this window, so that it only contains a single character. Here's what these characters mean:
    • h means that the input is presented in the 2nd window in hexadecimal form. This is the default
    • d means that the input is presented in decimal form.
    • s means that everything that comes from the Propeller is stored in an internal array in To see when this transfer is finished, you have to look at the LED on the Propstick. But this communication goes much faster, than when you send text to the textfields.
    • A filename folowed by a "<" means that that array is sent to the file. The data are transferred at the moment you type "<".
This is a fairly general way to get data from the Propeller. But then you can make variants, where you actually take care of the data, and do something with it directly. You will see that in conjunction with the USB Oscilloscope application.

A "Universal" Computer

Once I made a box containing a Propeller Computer, A/D converters, a bluetooth modem, an alphanumeric display, an SD card reader and three power amplifiers. Unfortunately the box became full of wires, and it wasn't easy to keep all these wires intact.

So then I made a more modest project. It was a smaller box containing a Propeller computer together with the standard EEPROM and the standard 5 MHz crystal, an A/D converter, an RGB LED for debugging etc. and circuits for voltage supplies. Originally it had D-sub connectors, but later I've switched over to RJ45 network connectors.

These are nice to click in, and they work well. On the female/chassis side they are a bit difficult to work with. They are connected to wires like a flat cable through soldering, but it's a bit tight there. Some suppliers sell breakout boards for these connectors, and that's much better, but it makes it a little more difficult to mount the connectors mechanically. On the box I've mounted the connectors on a plexiglass bar, but that requires some machining. The male connectors are much easier to work with, and you might consider using male connectors throughout and use male-to-male-adaptors. Note then, that these are not so easy to mount. You can try to drill and thread a hole in them, but that almost inevitibly destroys the device.

The male connectors are cheap. You get five or so for a dollar. And you connect the wires with a simple crimp tool in one step without soldering or stripping cables. But you need a flat cable with a division of 0.1 mm, and they are not all that common. More common is 0.127 mm (1/20 inch) but that doesn't work. You can also push single wires into the connectors, if they are thin. That's good if you only use few of the connections.

Here's the circuit board itself:

The picture is a little old. Since it was made, I have mounted a few more EEPROM:s to contain more software, though I don't use that right now. There are also some more connections to the card.

Here, the board is put in its box, and connected to the RJ45 connectors:

Here is the "programming side" of the box: There is a power supply input, and a reset button, and then three RJ 45 connectors. The leftmost of them is for programming through the FTDI USB circuit. The middle is for connecting to a similar box containing the RAM memory. The last connector is for 6 of the analog inputs. On top of the box, you can maybe see a grey little spot. That is the upper end of a nylon screw, threaded in the lid, that acts as ray guide for the light from the RGB LED.

On the other side of the box, there is just a row of 5 RJ45 connectors. The two leftmost are universal outputs giving acess to 6 Propeller pins each. The second of them also connect to two analog inputs, for use with the touch screen. The remaining connectors use fewer of the outputs. The rightmost connector is for connecting the keyboard.

Bluetooth has been replaced with Zigbee, but the Zigbee modem is now outside the box, and connected through one of the RJ45 connectors when needed. The power amplifiers are separate units also connected through RJ45.

The FTDI programming unit ("Prop-stick") is built into its own little box with a USB connector and an RJ45 connector.

The white button has to be pushed while programming to enable reset of the Propeller. When it is released, the box can be used for RS232 communication between the Propeller and the PC.

The Robot

Everybody who has a Propeller, or any other microprocessor, wants to make his own robot. Here is mine:

It has two relatively strong 12 volt motors with gear boxes (1:21), two strong rubber wheels and a caster wheel in front, and a led battery, that I charge with a CTEK led battery charger.

It has three power amplifiers. They are built with discrete components, but they don't work at very high PWM frequencies, as they get hot. It's OK for this application to work with low frequencies, but I don't recommend these amplifiers. The alternatives mentioned previously are better. Anyway, you can see the coolers for the transistors as three aluminum bars on the circuit board.

The computer is a Spin Stamp, which is a surface mounted Propeller mounted on a hybrid circuit, which also contains a crystal (10MHz unfortunately, perhaps), voltage stabilizers, and an EEPROM. On the card, there is also a 3208 A/D converter, and two voltage stabilizers. One of them reduces the 12 volt battery power, not to exhaust the stabilizers on the Spin Stamp. The other provides 3.3 volts for some sensors. Unfortunately, the Spin Stamp doesn't make its own 3.3v voltage available externally.

In the front, there is a Lego block, on which I can put sensors and other stuff, mounted on small Lego blocks.

In the picture you can see a compass, mounted on a high Lego tower, to keep it away from the magnets of the motors, an infrared distance sensor, and a "proto Zigbee" unit. The whole robot is controlled over proto Zigbee from a remote control. I also have an infrared receiver, adapted to an infrared LED modulated to 38 kHz with a Propeller computer. So, there is an infrared control for the robot, but it requires that you direct the LED pretty accurately towards the receiver.

One thing I don't have is odometry sensors on the wheels. I once had that. You can read more about that in the description of the milling machine.

I am a specialist in autonomous control of unmanned vehicles, but as this is a hobby, I have made a robot completely free from all autonomy. It is a remotely controlled vehicle, basta! But I did some experiments with a software, that would drive the robot to the nearest object. But that is not ready yet.

Otherwise the robot is pretty robust, fast and strong, though it can't compete with modern radio controlled model cars with brushless motors.

Here's the software RobotZ.myr

The first thing you notice is a "10 MHz" tag. That's because the Spin Stamp has a 10 MHz crystal, while the standard is rather 5 MHz. The Spin code, that is generated from the MPS code has to know that. So the MPS code uses the "10MHz" tag to modify the Spin code.

The process input receives the input from the Zigbee unit. It's a single number comm whose bits signal which buttons on the remote control are pressed.

The process motor's main program decodes comm and sets two motor masks and a velocity. The motor masks are each of two bits, and determine whether the motor will run forward (01), backward (10) or be stopped (00). The velocity is set lower if we turn the robot left or right, and higher when we drive forward and backward. This facilitates the driving. In the main program, there is also a call to a heading control function, heading, which drives the robot to a given heading, as measured by the compass.

The motor is finally controlled by the function drive. It sets the PWM duty cycle according to v (16 is full speed, 100% duty cycle). The motor masks merged together with shifts and an or instruction are sent to the amplifiers during the on period, and 0 is sent during the off period. The total cycle time of the PWM signal is 80000000/100 ticks, which gives a repetition frequency of 100 Hz.Fortunately we don't hear any 100 Hz tone from the motors, probably because of the inertia of the wheels.

The compass process catches the signal from the compass, which amounts to measuring the length of a pulse, that the compass electronics sends regularly. The scaling i 100μs/degree.

Finally a word about the heading control function heading. The control error is basically psi - psiC where psiC lies on the stack, and the measured heading is a global variable. But just subtracting angles is a little dangerous. 1 degree and 359 degrees are only 2 degrees away from each other, but the subtraction would give 358 degrees. So we have to "normalize" the result to within ±180 degrees, by adding or subracting an appropriate multiple of 360 degrees. Then we control the motor mask with the sign of the control error, and we control the velocity with a ramp function of the absolute error. (We use the || assembler function). The ramp function is difficult to trim, due to the friction of the motors. The motors don't run at all if the signal to them is too low. That is also why I haven't succeded yet with my "drive to the nearest object function".

The remote control

The remote control is a box with buttons, a Propeller computer with its crystal and EEPROM and a proto Zigbee unit. It is driven with a 9 volt battery. At the back side there is an on/off switch and a programming connector. (This figure is slightly old, so it still has D-sub connectors rather than RJ45 connectors)

There are 7 buttons, which is less than 8, so the whole button pattern can be sent in one byte. For that, each button is connected to its own Propeller input pin.

Download remote.mps.

Download zig.mpo.

The Wheel Steamer

This picture shows the wheel steamer

I built it together with my son, when we were in the countryside with limited technical resources. You can see, perhaps, that we are not exactly any expert naval architects. The ship is built in fur wood, and not plastic foam, so it is rather heavy. To keep it afloat, it is not a katamaran, but a pentamaran.

We found a small DC-motor as a main motor. To keep it away from the water, you can have a propeller with a long shaft, or a belt drive. But a wheel steamer was also a way to bring the motor up. We have only one drive wheel, placed in the middle of the ship, and built around a square hub.

Control is made with a rudder, which is driven with a little gearbox kit (for those who want to learn what a gear box is). It gives a slow enough movement to make the rudder controllable.

We also found a waterproof lunch box, in which we mounted the electronics, which consists of a proto Zigbee module, a Spin Stamp module, and a TDA2050 based power driver.

I don't publish any software here, as it is essentially the same as the software for the robot, although we have one drive wheel and a rudder here, while the robot has two drive wheels and a passive caster wheel.

Stepper Motor Robot

A robot equipped with stepper motors has the advantage that it moves with precision - as long as the hold torques of the motors are not exceeded by obstacles on ground. Hence, it is possible to build a navigation system based solely on the commands given to the robot. The simplest case is if the robot moves in either of two modes: forward motion and turning. As long as the robot turns, you just to update its estimated heading. As it moves forward you can update its position in the direction of its heading. But you can generalise this to any type of control. Then when the robot knows where it is, you can design control algorithms to go to given points, or to follow given tracks. Right now, I have just designed the robot, and connected it to the Zigbee receiver to receive commands.Here it is:

The stepper motors expect a voltage of 7.4 volts and pull 280 mA. For that the hold torque is 650gcm. It is not terribly much, so the robot behaves best on a plane floor. The robots are driven by two Lithium batteries connected in series. They are wired in such a way that they are charged in parallell which seems to work. Don't try to charge them in series. If you do that, you need a balancing cirquit. Also don't charge them with anything but a Lithium battery charger. Otherwise, you risk overheating and exploding batteries. The wheels are made in plexiglass (in fact the other wheel is made in delrin, as I had ran out of plexiglass), which is cut round and the smoothed in a turning lathe, and surrounded by O-rings. I use my standard motor driving cards (you need two for two stepper motors) and a third card with the Propeller and the Zigbee receiver mounted in an old Kodak film can. As I use my standard cards, the wiring looks pretty messy. The robot balances on two brushes. which makes it an excellent home cleaning machine.

Here are the programs to download:




Note how simple the main program is, as it uses the object file stepper.mpo so effectivly.

The Force Arm

This is a project that requries a good deal of mechanical skills. Nevertheless it is quite interesting.

A force arm is used in virtual simulators to provide mechanical feedback to a user, who otherwise gets his feedback from looking at the computer screen. If the simulator is for education, the force feedback can enhance the learning a lot. The situation is that the operator holds some sort of "pen" in his hand. On the screen he can see in a perspective image the tip of the pen together with different obstacles. When the tip touches an obstacle, a force should be fed back to the operator through the arm.

Such an arm is not an easy thing to make, and the ones on the market are quite expensive.

The normal principles of force arms, is to provide forces from torques of motors. Big forces require big torques, and that requires big currents, even though the motors have a special design to provide high torque. The high current problem also induces a cooling problem. If you try to use a gear box to provide more torque, you get a new problem: friction. You would drive the gearbox backwards, to speed up the motor, and that gives a lot of friction.

The story has a connection to the Swedish Gripen aircraft. Some experts argued strongly that the pilot must be given a force feedback in the control stick, so that he could feel what he did. But the cooling problems forced the decision to leave out the force feedback in the stick. I don't think that was terribly important, but after an accident, some people claimed that the lack of force feedback was a major cause of the accident. I think they were wrong.

My idea is to use geared DC-motors! I think you can come around the problem of friction, by measuring the torque, and creating a feedback from that to the motors. The result is an arm, that is worse than professional arms, but after some work, it is still quite good, and I think it may be useful for many things,

If you have a force arm, you also have a robot arm, which can move around to different positions automatically, and you can do this with different methods, and also combine position motion and force control.

Here it is:

Maybe you can see that it is only a two dimensional arm. Everything moves in a plane. The way to create 3 dimensions is probably to mount the whole arm arrangement on a plate, and move that around a vertical axis.

You can also see the stick or "pen" sticking out from the arm. It is connected to a joystick, so it would be possible to measure the angles of the pen relative to the arm, but I haven't done that yet.

So our first task, is to measure torque. What is torque (moment in most other languages)? Well, if you mount a lever arm on a motor axis, you will note that the length of the arm is important. If you have a longer arm, you get smaller forces. So any motor is characterized by the product of the length of the lever arm, and the force generated at the tip of the arm. This product of arm length and force, is called torque. If you have a fix lever arm length, measuring force is the same thing as measuring torque. There is only a fix scaling between them.

I argue strongly, that the only way to measure force, is to give after for it. You let the force act on a spring, and then the spring is deflected, and you can measure the force by measuring the displacement. Ideally, you can have a very stiff spring. That allows you to measure forces with very little displacement, which may be a point. But then you must be able to measure very small displacements. So, probably there is a balance to reach. In my arm I use a flat spring, consisting of plexiglass! It may seem to be a pretty odd material to build springs of, but I think it works. So, the torque bends the plexiglass spring, and you want to measure how much, by measuring a displacement caused by the bending. How do you do that?

Not long ago, I tried a so called pressure sensor, which is a thin film device, which changes its resistance depending on pressure. In that case, the displacement is very small. But I found that they weren't very stable. For the same constant pressure, their output varied pretty much over time. Further, they don't measure negative pressure, so you have to make some mechanical arrangement to bias them with a constant pressure.

Another idea I had, was to use an optical reflex detector. It has an LED as a light source, and picks up the reflected light with a photo transistor. If you have a white surface, they should give a high signal. But not if you place the sensor too close to the surface. Then efficiency goes down, and you can measure the deviation from the ideal distance. It works fine, but the noise level is pretty high.

The next idea is to use a magnetic field sensor, and a magnet. When the magnet comes closer to the sensor, the sensor gives a stronger signal. It is a very non linear relationship, but you can find a range, where the relation is linear enough. The sensor is based on the Hall effect. The whole sensor is a little integrated circuit. You need to find a linear circuit. Most circuits are logical circuits, which can only distinguish between magnet and no magnet. The magnet can be a neodynium magnet if you want something small, and as long as you can afford one; it is said that the price for neodynium is going up. The signal has a reasonably low noise level, and as long as you manage to build a stable mechanical system, they are stable over time. It is important to mount the sensor firmly. If you put it on a small circuit board, and let it hang in the wires, you will get stabiltiy problems.

Here's a primitive drawing of the arrangement:

Distinguish between the arm and the bar! The arm is a U-shaped aluminum profile. It hangs over the axis, which is a brass cylinder. The axis has a diameter of 12 mm, so this is the size of the holes in the arm, but they should be made with some precision, to avoid backlash. The motor's axis enters the axis through a hole, and is locked to it with a stop screw. The axis is flattened on two sides. To it, the plexiglass spring and the bar are fixed with a common through-going screw. Note, in the left figure, how there is a u-shaped hole in the arm to give place for the screw. The other end of the plexiglass spring is attached to the arm, but with a washer or two to avoid that the spring lies flat on to the arm's surface. The bar is a square brass rod. This is nice and stable. Aluminum might work, but maybe not plexiglass. At the end of the bar, there is a steel screw, which the magnet likes to adhere to. The magnet can be moved to an optimal position with the screw, and then the screw can be locked with a nut. The bar moves relative to the arm, as the spring is bent. This is measured by the sensor. The sensor I found is a trapetz shaped little thing. I've mounted it firmly in a hole in a piece of plastic, which is screwed to the arm. The hole should be close to the surface, so it may be easiest to make the hole first, and grind away material afterwards.

The sensor has three pins. Solder a ribbon cable to them and lead that down the arm to the electronics.

This is one arm unit. To get a double arm, to allow movement in the plane, we could put the next arm at the tip of the first. But the weight of the arrangement would probably be forbidding. So we mount the second unit at the same base plane, and then connect that unit to the second part of the arm through a mechanical link. This is easy, if we mount the two units so that they have common axis line, i.e. the two motors are coaxial.

The link arm is a round aluminum rod. It is tied to an aluminum plate, which goes out from arm #2 in a right angle. For these links I use screws, that are tightened to one of the members, and threaded into the other. You can see this arrangement here:

The motors axes are somewhat unsteady, so I have built a little support block in plastic where the two brass axes can rest. You can see this in this pictures:

Schematically, we have this arrangement:

The little circle marks the end of the arm at position (x,y). From there, the arm acts outwards with a force F with the two components (Fx, Fy). The position of the arm is controlled by the two angles α and β. Note that with the link arrangement we have, the second arm unit controls β directly. If we had put the second arm unit at the tip of the first, that unit would control β-α, which would be more difficult.

Now we realize, that we need to measure α and β, which we do with potentiometers (if not anything more advanced). But as the motor are coaxial, there is hardly space for any potentiometers directly coupled to the motors, so we have a link arrangement to drive the potentiometers. Look at the photographs to see these link arms in white plastic. A somewhat open issue is whether the potentiometer should be connected to the motors or to the arms themselves. The good side of connecting the potentiometers to the motors, is that we entirely eliminate the friction of the potentiometers. But there is a slight difference in position between the motor and the arm. When we build feedback loops around the arm, this difference can easily destroy stability. Here we have connected the potentiometers to the arm, and hence we should look for potentiometers with low friction.

Now, some mathematics. We can compute x and y as:

x = l1cos(α) + l2cos(β)
y = l1sin(α) + l2sin(β)

So this is a function that we would implement in our system, so that we could base the computation of forces on position.

But you could also differentiate this equation, to get the relation between the rotational rates of the motors, and the velocities of the arm tip. But it is even more interesting to invert this relationship. The differentiated equation is linear, and so it is described by a matrix. That matrix can be inverted. In doing so, you need the determinant of the matrix. It is (after using one of these trigonometric rules):

D = -l1⋅l2sin(β-α)

When this is zero, you can't do the inversion. When is that?. Well, it is when α=β. In that case, the arm is fully streched out. You can wave with the arm, but if you want to move the tip of the arm closer, it takes a great change in α and β. Here, then is the equation:

ω1 = (1/D)(vxl2cosβ+ vzl2sinβ)
ω2 = (1/D)(-vxl1cosα- vzl2sinα)

So now, given a velocity (vx,vz) that we would like to give the arm tip, the equation gives us the rotations we need from the motors. Typically we control these just by the voltage supplied to the motors. We may think that we have a goal point of the arm tip. We can compute (vx,vz) so that it is directed straight from the current position of the arm tip to the goal point. We have to scale the velocity, so that it depends on the distance. Now, when we start the feedback, the arm would ideally move from any point to the goal point along a straight line. It doesn't quite, because of the transient behaviour, when the motors speed up. Also saturations in the motor control amplifiers will alter the behaviour. But sometimes it works quite nicely. This is called kinematic control.

Before we go to dynamic control, we should mention shortly geometric control. Then, given a position that we would like to move to, we can use our first equation to compute corresponding angles α and β. It is a computation, that involves arctangents which is a little complicated, but it can be done. Then we construct a feedback from the α and β potentiometers to the motors, to bring α and β to the desired values. The motion of the tip is not along a straight line and it is hard to compute the exact track over long distances, but of course it works.

Dynamic control is about controlling forces, and that is what we started to talk about. So the question is: What is the relationship between the torques of the motors, and forces at the arm tip? It doesn't sound all that trivial, but it is almost trivial, if we think about a situation of equlibrium. Let's grab the tip of the arm, and hold it firmly, to prevent any motion. The forces it takes to hold the arm are exactly the force by which the arm presses onto my hand. If I released my hand, the arm tip would move just as if it were affected by the same forces. The forces from the motors, and the forces from my hand are the only forces present, except gravity, which is negligible. So we let my hand be there, and compute the balance between the hand forces and the motor torques. Here's the result:

Fxl2sinβ+ Fyl2cosβ = M2
Fxl1sinα+ Fzl1cosα = M1

Here, we don't need any inversion. The equations gives us exactly what we want. Given some force (Fx,Fz) that we want at the arm tip, we can compute the torques immediately, and we can achieve these torques through our torque feedback mechanism.

Then we can make a position feedback by computing a force from the deviation in position. But that't not necessarily a good idea. But, let's go back to the original idea, of pointing at a "virtual object" on a computer screen. When we touch the object, we would get an outward force. But that is a feedback from position to a force. And back from the force to the position, there are two integrations. So, there is no damping. Fortunately, the inner mechanisms in motor control helps contribute with some damping. The motor current controls torque. And the motor current depends on the applied voltage and the winding resistance. But not while the motor moves. Then the generator effect of the motor is in charge, and it reduces the current. This helps up stability.

But what happens in real life? Well, if you drop a metal ball on a table, the ball suddenly experiences an upward force. So it accelerates upwards, and then it bounces. But the bouncing stops, because some energy is absorbed in the table and in the ball at each hit. And this has to do with the characteristics of the forces that repell the ball away from the table.

With a force arm, we can attain stability, by designing the repelling forces somewhat smoothly. But it doesn't work unconditionally. If we hold the arm firmly, we can usually get a good behaviour, but if we hold the arm more loosely, it can start oscillating. Manufacturers of expensive force arms claim, that they have solved this problem, but I remation sceptical. The problems are very fundamental consequences of ordinary mechanical laws.

Now to an application: The window cleaning robot. The normal engineering approach to this would be to have the robot arm tip move over the plane where the window is. Then it would have some brush or so. But if the window is just slighly closer than we thought, we would crash the window. And if it is just slighly farther, we would just clean the air. What we want is a given force normal to the window, to press the brush against the window for a good result. In the remaining dimensions, we would move over the surface of the window. So we want dynamic control in one direction, and we would perhaps prefer kinematic control in the other two. But as far as I have understood, it is difficult to mix dynamic and kinematic control. So we must work with dynamic control throughout. It's OK. We can control position through force feedback. The friction of the window will provide us with damping. We keep imagining that we have a three dimensional arm, so we have a force vector of (Fx,Fy(x-xc),Fz(z-zc)) where (xc,zc) is a position that slowly moves over the surface of the window, while Fx is a constant force suitable for window cleaning.

As you know, I just have a two dimensional arm, so I can just clean a one-dimensional window. Furthermore I have laid the window down, so this is maybe a factory facility for cleaning windows before delivery.

Here's a picture of the setup:

As you can see, the window is not even a window any more, but a Java book. And the pen and joystick arrangement has been replaced by a bullet. (There is a workshop technique for making nice smooth metal sufaces, by pressing a bullet against the surface. In Swedish, it is called kulpening, which sounds funny even in Swedish.) Here are the downloads:

I use the keyboard to control the arm, and some data are presented on the display. For the keyboard, '4' turns the motors off, and they are turned on with '6' and '7'. Before turning them off, you should use '0' to remove biases from the torque sensors. While you do that, you should leave the arm in peace.

The main program, Arm2D.mps, needs a little overview. With the exec line, as it is written, the arm runs in dynamic mode, through the process posforce which means force as a function of position. Before that, the signals from the torque sensors are collected in an which also collects the angles α and β. All this comes from the A/D converter. The function debias finds the 'null-value' of the torque sensors by low pass filtering the signals. The low pass filtering requires a temporary upscaling of the signals with a factor of 26. The process pos computes the position of the arm tip, but the process also computes some goal positions for the arm tip. There is a circle, and a track between some given points (a function called bana, which means track). Posforce then computes a force in terms of two components fxc and fzc. From them, the process computes the corresponding commanded torques m1c and m2c. There are some functions to choose between. Brushforce is for the window cleaning task. That function also maintains the position over the "surface" of the window, where the arm should go, and a force is produced to move it there. Noforce is good to start with. It produces a zero force. If everything works, it should be very easy to to move the arm. Boxforce produces forces to force the arm tip into a box. This is the major test of the arm as a force arm. Refforce is the way to move the arm around the series of points produced by bana. The torque feedback is made in the process momfb which compares m1c and m2c with the torque sensor signals with bias removed, and sends the results to the motor in form a pulse width signals. 256 here means full signal to the respective motors (100% on). The pulse width modulation is handled by the process pwm, which can handle both motors simultaneously. The process is compact and fairly cryptical for speed. The total period time (on and then off) i 4096 tics, which is an integer power of 2, which helps making the program simpler.

The second alternative exec line is for kinematic control, which is mainly handled with the process posfb. One of the tasks is to move the arm through the goal points of bana. The other task is more challenging. I have failed with it, but I leave it to anybody to make a try. Here's what it is about.

The chua circuit

The chua circuit, built by Leon Chua, was the first successful try to build an electronic circuit, that could produce chaos. Until then, chaos had only been made in computers, though in some cases these computers were analog computers. I tried to mimic what Chua did in his circuit with my arm. but so far, I haven't made it.

You have an unstable oscillation around a center. So the tip of the arm moves in a spiral with increasing amplitude out from the center. On its way out it will come close to a second center, and when it comes close enough, it will switch over to spiral about that center. As it is now close to its center, the amplitude of the oscillation will be small. But it will grow. In this way, the spiral will come close to the first center, and will oscillate around that. And so on.

The information about which center the system shall spiral, will have to be stored separately, so it is a (boolean) state variable. The other two state variables are the position coordinates. So altogether, there are three state variables. One can prove that it is impossible to make chaos with fewer state variables than three.

So, one would like to see the arm move erratically around the two centers. What failed to me, was that my motions around the centers were always much more unstable than I expected. Probably that is because of the long sampling timed. Those who know about Runge Kutta's integration method might do something.

Exact linearization

The dynamic control of an arm is actually a special case of a procedure called exact linearization. There is an established theory for how this is done and when it is at all possible by the Italian differential geometer Alberto Isidori.

Recent developments

Recently, I rebuilt the arm electronics, so that it has its own dedicated computer. Then I replace the keyboard with two manual switches, A push button controls the debiasing of the torque sensors. and microswitch activates the motors (through software). Furthermore, I have done away with the display.

I have also built a second arm, which copies the position of the first (through kinematic control). That second arm has a little force sensor, whose force signal is fed back to the first arm. Then you get a force feedback when the second arm hits something. I'll come back to this at some later time.

The Milling Machine

I am the happy owner of the milling and drilling machine you can see here:

It has a coordinate table with two cranks to move the workpiece in two orthogonal directions. I have mounted striped tape on the two cranks, and then I have mounted a pair of reflex detectors nearby. The reflexdetectors are from Polulu and intended for odometry. The following pictures show the arrangement more in detail:

Then i have coupled the reflexdetector to a box containing a Propeller computer and an alphanumeric display.

The software consist of odometry computations based on the odometry object as described above. With the stripes I have made, the resolution in table movement becomes 1/14 mm, but I have recalculated this, so that I get resolution of 0.1 mm. On the two first lines of the display, I show the coordinate table positions in tenths of milimeter. All coordinates refer to a reference point, that you set by going there, and then pressing the right hand red button.

The third line shows the distance to the reference position. This is computed with Pythagoras' theorem and the itsqrt function in std.mpo for computing the square root. This function is of course usefull when you want to mill out a big circular hole. If the hole should have the radius R, and your milling tool has the radius r, you should move out to the distance R-r everywhere, but no longer.

Another thing that you maybe would like to mill, is an oblique straight line. You could define such a line by moving to two points on the line, and press the two red buttons. Then the display could show the distance to the line. This is a function I haven't made, but it would be possible.

The equipment is good for milling, but also for drilling where you could place your holes with good precision. In all, this is a very nice little equipment, that raises the value of the machine considerably. Of course, from this point one can go on and add motors, and make a numerically controlled machine.

The following picture gives a little instruction for how to make a hole for a D-sub connector.

Download Odo.mps

Download Odo.mpo

Download std.mpo

Download Display.mpo

The last object file is for an alphanumeric display sold by Parallax. It lets you type characters on the LCD display just by sending the ASCII-codes and some command codes. But, of course, you could also use the Adafruit display presented earlier.

The USB Oscilloscope

This is a new version of a USB oscilloscope. It's based on an A/D converter from MAXIM, called MAX118, which is capable of sampling at more than 1 MHz. It has 8 bits resolution, which isn't so much, but it simplifies programming, as the whole result can be delivered in a single byte. The result is delivered from the converter in parallel form, which partly explains the relatively high sampling frequency.

To use that speed. requires some efficient programming. This is done in the panalog.mpo object file, and involves some preparatory works making masks for adressing the converter rapidly.

Additionally, the device can log 6 bits of data sampled directly from the input pins of the Propeller.

Data are then logged in the Propeller global RAM. There is space for 7000 longwords, which means 28000 bytes. This is pretty much, so you have good opportunities to follow what happens in your system under investigation. Data are sent over the USB to the PC at the end of logging, and then they can be displayed immediately with a plot program.

Here's a scetch of the cicuitry:

Note that the inputs to the A/D-converter are not drawn in the order, they appear on the chip. You have to consult a data sheet.

To the left, there are 4 different analog circuits with different properties.

  1. This input accepts an AC signal through a capacitor. It has a gain of 10 and an input impedance of 10 kΩ. The range is moved up from zero to the range of the A/D converter. Hence the input range is +/-0.25 volts.
  2. This input has the range +/-2.5volts. So the gain is one. The impedance is 10kohms.Not super much, but good for most purposes.
  3. This input also has the range of +/-2.5 volts. but it drives current out to the source, which isn't always acceptable, specially not if you have a series capacitor in the source.
  4. This input has the same range as the A/D converter itself, i.e. 0 to 5 volts. You might think that this input has a high impedance. but it isn't quite true. The way the sample and hold circuits in the converter work, your source can't have a high impedance. But a couple of kiloohms is OK.
The amplifiers in the circuits are not operational amplifiers. They are far two slow. Instead I have used videoamplifiers from a circuit called AD8013 from Analog Devices. To my surprise, it was possible to give them a strong feedback without stability problems.

The converter is driven with 5 volts. Its reference inputs are supplied with 0 and 5 volts, which gives the conversion range of 0 to 5 volts. But the reference inputs need to be well stabilized.

The A/D converter is controlled with the CS and RD pins. When they are set to zero, conversion begins, and when it is ready, this will be visible in the INT-signal. During conversion, the adress bits have to be set, so that the right input signal is converted. The converter is a 5 volts device, and the Propeller is a 3.3volt device, so the outputs of the converter have to be delivered through protection resistors.

There are also digital inputs directly connected to the Propeller. As they are not protected, you should be carefull.

But, on the other hand, you could consider to build this device with just the digital inputs, and skip all the analog circuits. A digital data logger is a usefull thing.

The Propeller has a 3.3 volt stabilizer (not shown in the diagram), crystal and EEPROM connected the usual way, and the Propstick. That's where we take out data to the PC.

The program that catches the data builds on the same principles as, which is presented here, but it is more specialized, and takes care of the data from the Propeller automatically. It's called

When you start the program, you get a square window, from which you can control the oscilloscope. You do so by having the window in focus, and typing characters into it. These command characters are as follows: The result of these choices is displayed in the window. When you then hit the spacebar, the program will put together a two byte command, which is sent over to the Propeller. When the Propeller gets these two bytes, it starts logging the data in its global RAM memory, and when the logging is complete it will immediately send the data over to the PC.

The PC program will capture the data in a byte array, and then process the data depending on mode. The data will be brought to a file, and then a plot program window will appear, where you can look at the data.

Here's a little how you command this plot. It's mostly done with the keyboard. In the upper right corner of the plot window, there is numbered list of the variables including time. You choose a variable by typing its number and then Enter. If You type two numbers before Enter, you will get the the one variable as a function of the other. While you look at one variable, you can type the number of another, followed by a p and followed by Enter. Then you see both variables in the same plot with a common scaling. Then you can go on in the same way with a third variable. When you plot digital variables, the variables will be somewhat separated vertically, so that you can distinguish them.

In the lower right corner, there is a little menu of words, of which one is marked by being red. You can move in that "menu" with the arrow key, while holding the control key down. Then you can change the scalings with the arrow keys. Try this out, to see how it works.

The plot has a time scale, which should be accurate, because it's based on actual time measurements in the Propeller before and after logging. Here are the files for the Propeller program: Here are some Java files: But that's not the whole story, for the main program uses USB, which requires some C programs too. Refer here for more information about that.

Plotting from programs

Here's a process and an mpo file for make the same thing for variables inside your own Propeller software. This allows you to plot your own variables. Put the process into your mps program, and mention it in your exec field. Then you should declare 4 global variables u0, u1, u2, u3. Then you can write values into these variables anywhere in your code, and then these values can be plotted. To do this, open on you PC. Hit the V-key on your keyboard. The control window will display the word "Variable", which means that you display Propeller variables. In this mode you can't set anything else. The sampling time has to be set "hard wired" in you Propeller logging process. Now, hit the spacebar, and you will see your variables plotted.

The values are stored and processed as bytes. These are the least significant bytes of u0 ... u3. In your Propeller program you should also declare the area where you logg your data as

area[5000] This area is slightly smaller than the one in Panalog.mps. The reason is that the program you connect it to may take up more space in global RAM, and in that case there is less space for area. In the process itself there is a variable called size. It should now exceed the declared size of the array area. Lately, I've made a new Java program for displaying variables, sent over USB, visible as figures on the screen. It also creates a file, that you can plot afterwards. It is controlled by a control file, that gives variable names, and information of how the different variables are packed into the bytes. I will come back to this at a later time.

The Telescope

When you look at a star with a telescope, the telescope will rotate with the rotation of the earth. So to get a stable picture, you have to rotate the telescope backwards, relative to the earth. Amateur telescopes usually have something called an equatorial gimbal system, which is a mechanical arrangement, such that one of the degrees of freedom falls along the rotation axis of the earth. There is a knob there, and by rotating that slowly, you can counteract the rotation of the earth. Equatorial gimbal systems are a bit clumsy, though. They require a big counterweight, which makes the whole telescope heavy, and the telescope tube sometimes makes a pretty awkward journey, when you direct it against different parts of the sky.

So the whole project here is about making a virtual equatorial gimbal system. The starting point then, is what I would call a "fork gimbal system". The telescope tube is hung up between the arms of a fork, and can be moved up and down there. This is called the elevation movement. Then the fork can be rotated around a vertical axis. This is called the azimuth movement. The problem now is to distribute the counterrotation against the rotation of the earth to the two axes, azimuth and elevation.

Here is a picture of the fork gimbal arrangement:

First of all, the rotation of the earth is directed towards north, and it is tilted up from the horisontal plane with the latitude of the place. So on the equator, the rotation axis is lying down, and at the North Pole it stands right up, so the direction towards north doesn't matter, which is lucky, because there is no notion of north at the North Pole. Then, there is a more tricky mathematical problem. The azimuth angle and the elevation angle are so called Euler angles. There is a third Euler angle around the axis of the tube. So what we want to control is time derivatives of the Euler angles, and what we have is the earth rotation vector. But the relation between rotation vectors and Euler angle derivatives is a little bit involved, and it is easy to make mistakes there.

But I provide you with a Java program that solves the problem. The result shall be a function with three inputs:

... and two outputs The motions involved are very small. The earth rotates 1 full revolution in 24 hours (or to be more exact, 366.25/365.25 full revolutions in 24 hours). We can not hope to use a speed servo to achieve such small motions. Instead we work with small fixed angle increments. The Java programs produces the time interval between these incremental movements. Concretely the Java program produces a definition of an array written as MPS code. This MPS code is then copied into the complete MPS program.

The Java program assumes one latitude angle, which is the latitude angle where I live. If you don't happen to live at the same latitude, you should change the latitude angle. Furthermore, the azimuth angle rate is virtually constant in elevation and azimuth angle. It is essentially the vertical component of the earth rotation vector. In my system this rotation corresponds to an interval between incremental movements of 3.4 seconds.

In elevation, the angular rate varies with both azimuth and elevation. When a star is seen strictly south, it doesn't move in elevation. It is on its peak on the sky. To the left of south (i.e. east) the stars are moving up with varying speed depending on elevation. On the right side of south, they move down, and so should the elevation angle. These effects are seen in the table.

Negative elevation motions are encoded as negative time intervals. The Propeller program has to observe this, and use the positive time interval, but at the same time reverse the motion direction.

Technically the angular increments are made with a combination of a motor, a hole disc, an optical fork, a long scew and a leverarm with a threaded joint. In the computer there is a motion state variable called x (xel for elevation and xaz for azimuth). x flips between 1 and 0, once for each time interval. When x is 1 the motor moves until light passes through the hole disk. When x is 0, the motor moves until no light passes through the hole disk. The hole disk has 12 holes. Consequently the disk moves 1/24 of a whole revolution per x-flip. This makes the screw threading move outwards or inwards 1/24 of 0.5 mm (because it is a M4 screw). This pushes the lever arm outwards this much, and as it is a 1 dm lever arm, the angle increment is:
1/24·1/2/100 radians = 2.08333·10-4
This value appears in the Java program.

Here is a picture of this arrangement in azimuth. I painted the hole disk with opaque color to avoid reflexes, but it turned out to be unnecessary.

You can see the motor, the hole disc, the optical fork (in black), the screw, a threaded joint in brass, and the lever arm in aluminum.

This is the correspoinding arrangement in elevation.

The disc is slotted rather than a hole disc, which makes the mechanical mounting of the optical fork a little easier. You can also see the arrangement that locks the lever arm to the elevation axis of the tube, with its wing nut. Inside the aluminum tube profile, there is piece of brass with an oval hole, which is pressed towards the elevation axis.

It is worthwhile also to have an electric focusing. If not, you have to touch the oscilloscope focusing knobs, and it is hard to make the gimal arrangement so stable, that it doesn't shake a little. Focusing is something you have to do, after you have changed ocular. Fortunately I have two amplifiers left for driving a focus motor, and i have two buttons left on the remote control. The software is simple, but of course you have to find out how to couple a motor to the focusing knobs. This picture shows how I have done it:

The computer has four input signals. Azimuth and elevation signals come from potentiometers. They are right now handled manually. The user is supposed to direct the elevation potentiometer parallell to the telescope tube. A link arm could make this automatic. The user is supposed to direct a bar fixed to the other potentiometer to the south. An electronic compass could do this, provided it would be calibrated. The other two signals come from the optical forks. All these signals are fed to the A/D converter.

The output from the computer is three pairs of motor signals connected to the two poles of the three motors through power amplifiers. These amplifiers are the integrated drivers mentioned here. As mentioned, the integrated circuits are mounted together with protection diodes and a 5v stabilizer on little modules. Each module can drive two motors, so to drive 6, I need two modules.

Now, we think of motor control as setting a six bit code, which controls all three motors. A code of 101010 would drive all three motors forward. A code of 000100 would hold the first and third motor and drive the second motor backward etc. The toggle process sets these codes based on the state variables xel an xaz, and the fork positions. In turn, these computations rely on the time intervals dtel and dtaz, and on the directions direl and diraz.

The process motor sends the code further to the amplifiers, which amounts to shifting the codes to the right pins, and clocking the shift register. It also has to satisfy the amplifiers requirements to get zero input now and then.

The times dtaz and dtel and the directions diraz and direl are computed by the times process, which contains the table from the Java program. In fact dtaz and diraz are constant. The computation of dtel is based on a table, where azimuth and elevation have been discretized to tens of degrees.

This discretization is made in the anal process together with the A/D conversions. The discretization of the elevation angle is based on measurements. The optical forkpin signal are discetized to boolean values, using the G assembler function, which maps any number greater than 0 to boolean true (1) and any other number to boolean false (0).

Finally, there is a manual control of the telescope from the standard remote control used for the robot. The remote control signal is received through Proto Zigbee and is called comm. It is a mapping of the buttons pressed, and it requires that only one button is pressed at a time. The manual process computes motor codes similar to those from the toggle process, and they also control the focus motor. If no elevation button (the upper and lower) are pressed, an ignore code is set. Similarly the azimuth motor code and the focus motor code can be set to ignore. The merging of the "automatic" motor codes codeel and codeaz, and the manual ones, mcodeel and mcodeaz is done in the process toggle. If a manual code is not ignore, then it overrides the automatic code.

Download Teleskop.mps

Download Display.mpo


Download (used by

Here is finally a short manual to the telescope:
  1. Make sure that the lever arm threaded joints are in appropriate positions along their screws. The azimuth joint could be towards the end of the screw (It will move inwards). The elevation joint could be near the middle of the screw. Arrange this with the remote control.
  2. If you feel you want a cup of coffee, or expect that the next step will take some time, disconnect the power to the electronics.
  3. Loosen the gimbals from the lever arms. In elevation you could see a wing nut for that in one of the pictures above. There is a similar wing nut for the azimuth.
  4. Direct the telescope tube to what you want to look at using the manual sight on the tube. When ready, tighten the lever ams to the gimbals again.
  5. Align the elevation potentiometer along the telescope tube, and direct the bar on the azimuth potentiometer towards south.
  6. Reconnect the power to the electonics.
  7. Look in the telescope ocular, and if you see your astronomic object there, center it with the remote control. Otherwise you have to scan around with the remote control to try to find it. This can be quite difficult, so you may prefer to go back and use the sight on the tube instead.
  8. Once the object is in the center, it will stay there for a good while. You may have to adjust after a while with the remote control.
  9. Now you can also start changing ocular to get a closer look, and you can do that without having to worry where the object goes while you don't observe it. It stays where it is.
  10. After more than a good while, you will have run out of threading of your screws. No more to do than to go back to point 1 again.

The Digital Camera

A commercial digital camera is an impressive thing. It has an image sensor that gives something from 0.5 Mpixel to 12 Mpixels. Then it has a display with maybe 1 MPixel, and there is electronics that transfers the pixels between the two so that you can see the image in real time. That calls for an update rate of at least 20 Hz. 1 MPixel in color is at least 2 MBytes (16 bits per pixel in color information). So we have a data rate of at least 40 MBytes per second. That is far more than you can achieve with standard serial protocols. And it is not only to transfer the bits; you have to get them out from the camera sensor and into the graphical display. Furthermore, there is the JPEG encoding of the pictures, which is far from trivial.

If we would like to do something as amateurs, the best thing is probably to get some sort of camera sensor with a standard RS232 or SPI output. That means that the manufacturer of that sensor has solved most of the problems for us. But it gives us a camera, which we can use for things like robotics applications.

There used to be such a camera called C328. It had an auto baud rate detection, which could maybe be nice, but I never got it to work reliably. The newer version is called C329, and it works better. I still have some difficulties syncing the data, so I have a special fix for that.

A somewhat odd alternative is the latest version of the CMUcam, developed by Carnegie Mellon University. This version uses a Propeller processor, but it seems to be a missuse of a Propeller. The point is, that, like all microprocessors, the Propeller has far from enough memory to store a whole picture. The solution would be an external RAM. But CMU's solution is to download the entire picture several times, and screen out a single row each time. This gives an exposure time of a picture of 3 seconds (!) of more.

Back to the C329! It comes in two versions, RS232 or SPI. Probably it is a good idea to buy the SPI version, but I didn't! The point is, that with SPI, the master, which is the Propeller, controls the clock pulses, and can thus get itself the data at a rate, that it can handle. With the RS232 alternative, the camera pushes the data, and there is always the risk, that the Propeller is flooded with data. This rules out direct writing into SD-cards. While the SD-card shifts write page, data are unavoidably lost. So with the RS232 camera, data have to go to the RAM first.

The C329, like the C328 can send you a JPEG encoded (compressed) picture, which is nice, if you just want to send your image to a PC, which has facilities to make JPEG decoding. But JPEG decoding in a microprocessor, like a Propeller, seems to be an overcourse.

For the uncompressed image, you first have to choose a color format. My choice is the 16 bit format (two bytes per pixel) where red green and blue take up 5, 6 and 5 bits respectively. Then we get "files" with two bytes per pixel.

The "camera" comes as a breakout board with an optical sensor and electronics, but the lens has to be ordered separately. There is a lens holder which fits to the breakout board. But the threading of the lens or the holder was so poor, that I needed to make my own lens mounting. The threading of the lens is very fine grained, so I didn't think I could make such a threading, so instead I mounted the lens in an O-ring, which worked as lens holder.Then the whole thing was mounted in a little box.

Here's the camera mounted on a "Gorilla pod":

The system now consists of the following: The camera and its lens, a Propeller, the RAM, an SD-card and a keyboard for control of the system, and the RGB-led. When turning on the system, there will be an initialisation of the camera. Unfortunately, before doing the initialisation, the camera has to have its power off. Then, the successfull initialisation will be signalled with a green LED signal. Then, when the operator hits the enter key on the keyboard, a picture will be taken and will be stored in the RAM, and then displayed.

Now to the mysteries. The camera sends some data before the actual picture. Many data. It seems to vary a little too, and I haven't found out how that works. Safest is to put everything into the RAM. If the numbers of bytes is odd, the color bytes will change place, so the colors look really odd. To get it right, I shift the adress to the RAM a little, under control of the keyboard. In that way, by displaying the picture a couple of times with different shift, I can get a nice picture. Then I'm almost ready to store the picture on the SD-card. But to do this, I first have to choose a place on the card. As mentioned, I don't have a file system on the SD-card, so I have to choose an adress space manually, again with the keyboard. Then I store the picture, and the RGB-led shows again, when the storing is done. Then I can use the picture display program, presented next, to show the picture. And it is possible to write a program that uploads the picture to a PC.

Here now some information about initialisation of the camera. This is done by means of the cam.mpo object file.

The camera wakes up with an expected baud rate of 14400. We will change that to 155200, but it we want to initialize the camera again, we have to bring it back to 14400 by removing power. Then the first to do is a syncronisation procecure. This is done with a sync command.

All commands start with $FF $FF $FF (three bytes with only 1:s). Then follows the command number in one byte, and 4 bytes with parameters. Unused parameters are coded with 0. The sync command has number 13 ($0D) and no parameter. The receiver of such a sync command is supposed to respond with an acknowledge command. This has number 14 ($0E) and the first parameter is the number of the command responded to. So we expect that response from the camera ($FF FF FF 0E 0D 00 00 00). We wait for it 25 msec, and then we give up, and try again, and we do that for a maximum of 200 times. The administration of the waiting i done in the function cam.wack. Now when sync is hopefully successfull, the camera will send a sync command to us, and we should respond to that with a $FF FF FF 0E 0D 00 00 00.

Then it is time to set the camera parameters. This is done with command number 1 and the following parameters:

  1. Baud rate. 7 is 14400, and 4 is 115200. These are the ones I use, but it is possible to go up to 921600, which has code 1.
  2. Colour type. Code 6 gives me what I want, i.e. the colour information is encoded into 16 bits (2 Bytes) with 5, 6 and 5 bits for red, green and blue.
  3. Preview resoution. Code 3 gives me 160 x 120. Not terribly much, but it is what my display can handle. 5 gives 320 x 240. 7 gives 640 x 480.
  4. Compressed image resolution. The codes are the same as for the preview. I'm not going to fetch any compressed images, but 0 might be offensive, so I use 3.
Now the camera is ready, and we can get a picture by sending a get command. The command code is 4, and the first parameter is 2 for getting the preview image, i.e. an image with the resolution set above. The bytes come flying now, and we grab them with the cam.grab which we start in its own process.

Furthermore, cam.mpo contains the following functions: Some of these functions use the function which has a timeout function to detect, that the whole image has been sent.

Here's a picture of the whole setup:

These are the functions of the buttons at the keyboard: When restarting the computer, the RGB LED will first show yellow, and then hopefully green. If it shows red, a reason might be, that you have restarted the computer without restarting the camera. Turn the switch on the camera off and on, and then restart the computer.

Here are files to download:


Picture Shower

So far, this program is a little bit scetchy, but it allows you to look att pictures stored on an SD-card. So far, the program only uses single hits on the keyboard, which controls the showing. Hence the number of pictures to show is limited to 16. The pictures shown are of three different formats, and the encoding of which format each picture has is put directly into the program. So improvements are possible.

The three formats are the following:

  1. A format from pictures fetched from the PC. They have the same resolution as the display; 220 x 176 pixels, and they are stored row by row. The Java program that generates these pictures fill with black up and down or right and left to maintain the proportions of the pictures.
  2. A format for pictures from the CMU-cam, from which I have a few pictures. The size of the pictures is 160 x 120. Every pixel is stored in two bytes and there are no "padding zeroes" on the SD-card. The pixels are stored row by row.
  3. A format for pictures from the C329 camera. The size of the pictures is the same as for CMU-cam, put the pixels are stored collum by collum, which requries some extra programming.
The procedure for getting the pictures from the PC is this. The Java program generates files from the pictures. The program name is of course a little cryptical, but it resides in a directory for Photo processing, which perhaps makes it more understandable. The program is also quite primitive. You have to go into it and write input and output filenames manually. The files generated are simple accounts for the pixels of the image. The color information for each pixel takes up two bytes.. These bytes for a pixel are stored in text format style, with each byte represented with hexadecimal characters encoded in ASCII. This file format is called .hmp. (that is my name for it).

Then these .hmp-files are transferred to the SD-card through the cooperation of one Java program,, and one MPS-program, StoreH.mps, sitting on either side of the "PC-umbillical" to the Propeller. ImSend converts the hexadecimal representation back to plain bytes, and these are then stored on the SD-card. As we regard the SD-card as a memory of longwords, these bytes are stored in groups of 4 bytes. waits for a request from StoreH.mps in form of an arbitrary charactes, So you should start before StoreH.mps. Having run these programs, you have your picture in the SD-card ready to show.

The program in PictureH.mps uses 4 parallel processes. One (keyscan) just reads the keyboard. One (keychoose) reacts on the outputs from the keyboard and sets an SD-card block to read from. These blocks have the size of 215 bytes. Once a block has been chosen, keychoose resets semaphores, so that the memory and the display are reinitialized. Then the process get gets data from the SD-card, and writes it into the variable data, and the show reads data from there. This traffic is controlle by the variable hasdata. Both get and show are split to different routines for the different image formats. Some of these differences depend on differences between the formats, that are maybe a bit arbitrary. The main difference between showcmu and showc329 is that the pixels are stored row by row and collum by collum in the different formats. Here's an example, which is also a self portrait of the author:

The FM Radio

This radio is based on a chip from Airoha called AR1010. In a volume of 4x4x1 mm3 it contains a complete FM radio with stereo decoder, sound outputs and two different computer interfaces. I don't know if this is a so called Software Defined Radio, i.e. a completely computerized radio, or if there are any special circuits. Nevertheless, you don't see any traditional things like coils or capacitors. My work is based on a small breakout board, which includes a crystal. It is sold by Spark Fun Electronics. Apparently Telefunken has made circuits of this type, and there is another chip called SI4375. Read about that here.

The problem with Airoha AR1010 is that the documentation to it is confidential! All that is availble is a datasheet that describes the electrical connections, but it says nothing about the protocols, with which you speak to your radio chip. A software specification is avaiable on the internet, however, though it is pretty hard to find. The name of the document is ar1000F_progguide-0.81.pdf, which you can try on Google. One adress you find is which is a chinese site where you can see and read the document, but it is unclear to me how to download it. I have downloaded the document, of course, but I don't dare to make it available here. Some people have complained about the quality of this document, but I think it is quite good. As you can see from the name, it is actually about the circuit ar1000, which is a more advanced sister to ar1010, which has RDS, but it works for ar1010 also.

The breakout board is maybe intended for surface mounting on a circuit board, but it is easy to connect to a normal hole mount board with some small wires. The breakout board hides some details of the original cicuit, so all you have to think about is this:

If you ground the Busmode pin in the computer interface, you get I2C interface, and then all that remains is a clock pin and a data pin. We have mentioned about the I2C interface here and here will come some more detailed information.

The AR1010 chip has 18 registers, each with 16 bits, and you control the radio by writing into these. The procedure for this is the following:
  1. Send an I2C start pattern
  2. Send the adress of the chip, which is hexadecimal 10, but this should be followed by a 0 write bit, so what you send is actually hexadecimal 20 (in one byte)
  3. Send the number of the register, you want to write to.
  4. Send two bytes of data, to fill the sixteen bits of the register.
  5. Send an I2C stop pattern
All this can be done with the iic.mpo object file. With a similar pattern as above, you can also read any register, to find out about the status of the radio. I found that unnecessary. I trust that the radio will do what I tell it, and I expect that the corresponding status bits will be set, if I just wait a while.

The programming guide describes the contents of most of the registers, but it is obviously an incomplete description. There is a set of default settings to all the registers, and sometimes you set bits to registers that, according to the description, don't mean anything. Anyway, the default settings work fine. There is one set of default settings if you have a crystal on the chip, and another if you set the reference frequency from your computer, but here we do have a crystal. To initialize the radio you should set these defaults from register 1 and up, and last to register number 0. You will find the default values, and how to initialize the radio in the file radio.mpo.

The rest of the settings have to do with setting frequency, setting volume, muting and unmuting.

Radio.setfreq receives the frequency in 10:ths of MHz. This is also how the frequencies are distributed to the users. If you want to transmit something on the FM band, you are assigned a frequency in MHz with one decimal. So if you hit that frequency in the receiver, which you should be able to with a crystal, you should be exactly tuned. If you multiply the frequency with 10, to get an integer, and subtract 690, you get something called a channel number. That channel number should go into register 2. You first write into a mirror of that register in the computer, and then send that mirror over to the real register. But you also have to toggle a so called tune code. The radio retunes, when the tune bit switches from 0 to 1. Previously I muted the radio during the frequency switch, but it doesn't seem to make any difference.

Volume is set in radio.setvol which receives a volume number between 0 and 15. For some reason, the volume scale is turned up side down, so you should send (15 - volume) to the radio. This volume should be sent to register 3. There are some less significant bits for volume in another register. The programming guide shows how you could use these bits to get a smooth scale, but I thought that 15 steps of volume is enough, so I skipped the other register. The 4 bits of volume information should be shifted up seven steps into the register. But register 3 contains other information too, so you have to mask it a little to save the rest of the information in the register.

I chose to use potentiometers as input to the radio. They are A/D converted, and discretized to a number between 1 and 15 for volume and 1 and n for frequency, or rather station, where n is the number of stations I'm interested in. Then I have table for the frequencies of these stations. Somehow you can find frequencies for your favorite stations over the Internet. This is found in the main program in the file Radio.mps.

An aspect on this, is that when you move with the radio, you have to reprogram it, to feed in new frequencies. I do that twice every year, as I move to and fro the countryside. You can do things otherwise. You can vary the frequency continously, and display the frequency on an LCD-display, or you can use the chip's built in search facilities.

The Propeller with it's EEPROM, crystal and voltage stabilizers is placed in a box together with the radio chip itself, A/D-converters (for the control potentiometers) and pull up resistors for the I2C-bus. There is also a RGB-LED to indicate the current choice of station.

I was a little concerned about putting a 69 to 108 MHz radio together with a computer with a clock frequency of 80MHz, but it doesn't seem to be a problem.

The box is hung up between two PC-loudspeakers. I use the audio amplifiers there of course, and I have also broken myself into the box, to bring out 18 volts of power to drive the circuits. It turns out that the 3.3volt stabilizer doesn't heat up, despite the relatively high input voltage.

On the top of the box, there is an aluminum sheet, which serves as a ground plane for the antenna, which in itself is a metal rod, attached by means of a piece of plastic. The ground plane is connected to a wire, which is clamped on the outside of the power cable. For the 100MHz signals, this gives some connection to ground, but the 50Hz power signal can't penetrate the small capacitance between the wire and the power cable. This arrangement helps improve sound quality, when the reception conditions are bad. In my home they often are.

But the exact tuning, and the good sound quality of the radio chip, makes this an excellent radio. Here is a picture of my radio:

Download radio.mpo

Download radio.mps

Download iic.mpo

The Si4703 Chip

The Si4703 radio chip is an alternative to AR1010, and it is maybe more accessible than AR1010 nowadays. It comes on a breakout board from Spark Fun, which contains a radio chip, a few discrete components, including a crystal, an earphone amplifier and plug, and connections to a computer.

Programming Si4703 is about as confusing and difficult as programming AR1010. There is example code to fetch from Spark Fun written for Arduino by Nathan Seidle. But Arduino has I2C communication built in, while with a Propeller, we have to do it ourselves. But, we have already done that. Also I think that the Nathan Seidle has overdone some things a little.

Apart from the example code, there is a data sheet, and a programming manual. The programming manual is OK, but it has to be read together with the data sheet. The necessary information is spread somewhat randomly between the two documents.

The device can be controlled with two communcation protocols. One of them is clearly I2C, but this is never mentioned, so it's called "2-wire interface". There is a reset pin, that you bring down for reset. When you lift it again, certain states have to be present, to take the system to the 2-wire interface mode. There is an inverted SEN-bit, which has to be high, which it is on the breakout board thanks to a pull up resistor. Then the SDA bit (in I2C terminology; it is called SDIO on the board) has to be low. Finally a GPIO3 has to be low. That's a bit bothersome, becaulse that pin is also a part of a crystal oscillator. But, be calm, It is low. So bring the reset pin down, the SDIO pin down, and then lift the reset pin. Then the device wakes up in I2C mode, and we can start I2C on our side by calling iic.init from Iic.mpo.

As is usually the case, I2C devices are handled by writing to and reading from a couple of registers, in this case 16. But they are written and read in an unusual way. The device has default register addresses for reading and writing. It starts there, and then it increments its adress counter for every 2 bytes. It's comfortable, but it means that you can't write to a single register without writing to a couple of others too.

For some reason I failed in programming a register read function. It reads all right, but it somehow destroys the register contents. Nathan Seidle advices us never to write to registers without reading them first. The data sheet gives you the same message. But the programming manual says: "The data sheet tells you what the registers contain, and you can trust that, so it is not necessary to read the registers." So, I accepted that, and it made the whole task a lot simpler.

Now we come to the mysterious register 7. The first bit of this register controls the crystal oscillator. You want that, so you would like to write that register. The programming manual says: "Write hexadecimal 8100 there". Then the first bit is set, as wanted, and another bit for some unknown reason. While we write register 7 we have to write all registers from 2 to 7, but it's OK to write 0 to all of them. Now, the crystal oscillator wakes up, and we should give it about half a second to stabilize.

Now, the device is in power down mode, so we have to lift it to power up mode. We do so, by setting the last bit in register 2 to one. It is called Power enable. For some reason they want us to set the next highest bit too, so we do that by writing hexadecimal 4001.

While we do that, we can write to the rest of the registers too. Register 2 to 7 are writeable. Most bits in these registers are set to default values, that serve us well, but, at least, if we are Europeans, we have to adjust a few things. To reduce noise, the transmitters send the higher frequencies of the sound with higher gain, and you should compensate this in the receiver. This is called deemphasis, and there is a special bit for setting European deemphasis, while the default is for American deemphasis. The UHF band is different in Japan on one side, and America and Europe on the other. The American European standard is default, so we don't need to change anything. Then there is the channel separation, which is 0,1MHz in Europe and 0.2MHz in America, so I have to deviate from the default there. That's about all. Setting of frequency and volume, will happen later.

Now the radio is powered up. In 110 ms it is ready to use. For some reason, the mysterious register 7 now changes from 8100 to BC04, but that happens by itself. So no worry, but you shouldn't set the register to BC04 directly. Then the radio doesn't work.

All this is the content of the function radio.init in radios.mpo. It uses its own function radio.writeregs and various function from iic.mpo.

Then volume is set by writing the four LSB's of register 5.

Setting frequency is only slightly more complicated. First we have to convert frequencys to channel number. If we give frequency in the unusual sort of dMHz, which gives us all frequencies as integers, we only need to subract the UHF-band-bottom frequency, which is 875 dMHz, to get the channel. We write that to the LSB's of register 3, where we also write the first bit to one. That will trigger the tuning in the radio. Then we have to wait for the radio to react, and then we send the same information again, but with the tuning bit reset.

Here's the code for download:

As you can see, I haven't bothered about RDS, Radio Data System, which gives you information about the programs.

The Media Player

I won't give all the details of this application, which is a media player, which can play radio, MP3-files and TV sound through full size loudspeakers. Here's what it looks like:

As you can see, there are no knobs and buttons on the device, except a power switch to the right. Instead it is controlled entirely from a remote control. To see what you have done, there are two RGB-LED's. You can also see the MP3-player through a slot. I had to put it that way, so that the micro-SD-cards, that contain the music, could be pushed in. The back part of the box is a little higher, and black, so you don't see it (do you?). It contains the power supply and the cooling element for the power amplifiers. On top of the box, there is a smaller box, that contains the micro-SD-cards. They contain the content of a heap of vinyl LP-records, that is about 2 meters high.

Here's the remote controller:

It communicates with the player over proto ZigBee. It has two knobs, one for volume, and one for selecting MP3 files. And four buttons for choice of Radio, MP3, TV or nothing. And there is a power switch, to save batteries. It consumes pretty much.

The control computer in the player is, of course, a Propeller.

The radio is AR1010 as in the previous section. The way this device is placed, there is a better antenna.

The MP3 player is a little board, sold by Spark Fun Electronics (and resold in Sweden by Lawicel). The music is loaded in micro-SD-cards on files with standardized names TRKxxx.MP3. You choose file number xxx by sending a command over RS232 to the board, and then the board plays the music. There may be many of these boards around. Choose one that let's you control it from a computer.

Then we make the whole thing electrically controllable by using analog electronic multiplexers and electronic volume controls. The electronic multiplexer has the designation CD4051. It may not be easily available any more, but some sort of electronic multiplexers should be available also today. More correctly put, these circuits are called analog multiplexers, and they consist of analog switches. An analog switch is a digitally controlled switch. with either a low or high impedance from an input pin to an output pin. By the way, the device is symmetric, so you can swap the input with the output as you like. As long as the signal that goes through the switch is well away from the supply rails, the switch has a low resistance in the on state. There is some resistance, and it varies with the input voltage, but if you load the circuit with high impedance, the on resistance is negligable, so you can get an analog signal through with low distortion. I supply the circuit with 12 volts, and to bring the input signals away from the supply rails, I lift them from ground with 6 vots from a voltage divider. That DC component has to removed later on. CD4051 has three adress inputs, and with them you can choose between 8 different inputs going through to the output. The control computer has to provide these three adress inputs through 3 computer pins. So the computer interface is "parallel".

The so called digital potentiometer, MCP41010, also has analog switches. In fact it is a conventional chain of resistors, as in "a discrete potentiometer", and analog switches lead the signals out from some place along the chain. The digital circuitry that handles these analog switches is controlled over an SPI bus. (So now we have a parallel interface for the multiplexer. SPI for the volume control, I2C for the radio, and RS232 for the MP3-player, and ZigBee unit, so it is a good exercise in computer communication.) The digital potentiometer has to be loaded with high impedance to keep distortion down, so this section ends with an operational amplifier OP290 in a follower configuration. But the digital potentiometer, with its 10 kohms is good as a high impedance load of the analog multiplexer.

The power amplifier is based on the TDA2050 power operational amplifier as mentioned before. I use two for each channel. One is configured as a non-inverting amplifer, and the other as an inverting. Then I can connect the loudspeakers between the outputs of the two amplifiers. That gives me twice the range I can get with a single amplifier. Furthermore I can adjust the two amplifiers, so that the difference between the outputs is close to zero. Then I don't need any output capacitors. But do this adjustment well! It is said to be harmful for a loudspeaker to be driven with DC current.

The voltage supply is a switched 12volt supply of the type that is used for PC:s.

I enjoy this machine very much, and now I can play my records pretty much effortlessly. No more carrying of LP-records, and no more handling of pickups. Unfortunately, by the time I built this, for some unknown reason, my ears started working much worse than before, so I am not the right person to judge the sound quality, but it's good for me anyway.

GPS Logging

When I mention something about this application, some people say: "I have it in my smartphone". Such are the conditions for us "do it youselvers" nowadays. Everything is already in the smartphones. What is the meaning of life, then?

On the other hand: Doing the whole thing of GPS navigation is hardly anything for us amateurs. There are a number of satelites out there, and they send us their opinion of time. When we compare these times from some different satelites, we observe, that they don't coincide. But then we have the theory that this lack of coincidence depends on that it takes unequally long time for the time signal to arrive to us from the satelites. With that, we can determine differences in distance to the different satelites. As satelites move very regularly up in the empty space, we know where they are, and then we know we, ourselves, are. Down to an error of three meters or so! But that is not an easy computation! Errors are a little bigger if we move in a forrest (or a city) because the signals reach us after having bounced a little between the leaves of the foilage. This growth in error is quite fundamental, and there isn't much we can do about it.

Let me say something about giving positions. The earth is a ball. Then the surface of the earth would be a sphere if the ball were perfect. How do you indicate position on a sphere. Well, there would be many ways to do it, but one of the best is to give longitude and lattitude. First, then we would need a reference point. It is natural to let that point lie on the equator. But where? Well, it would be a point with longitude = 0. So where is longitude 0. Well, it's at the Greenwich Observatory, or at Constantinople, at the Cheops Pyramide or in Paris. In fact, for some time it was both in Paris and in Greenwich, but finally the French had to give up.

The story continues this way: Someone placed a metal plate at the Greenwich Observatory, and engraved a fine line in it. That was where the zeroeth longitude went by. But then the Americans made a radio navigation system, which pretty quickly gained great popularity, and become a world standard. It was not until after that, that someone came up with the great idea, that one should place a receiver at that engraved line at the Greenwich Observatory. The receiver said "108 meters east (or west, I don't remember) of Greenwich". So it was wrong! But it was much more difficult to correct the American Radio Navigation System, than to just move the zeroeth longitude. So now it is 108 meters off the Greenwich Observatory.

Then you move south along that zeroeth longitude point, until you hit the equator. As long as we have the rotation axis of the earth, that is actually well defined. The point we find is the reference point. It has longitude 0 and latitude 0. We can imagine a ray going from the center of the earth, out to the reference point. Then we rotate that ray first around the earth rotation axis, and then around an axis perpendicular to that, so that the ray points at some given point. Then, the two angles are longitude and latitude for that given point.

Now, when the earth isn't spherical, we have choices. The choice made has to do with so called Inertial Navigation, which in aviation is often combined with GPS. The result is, in practice, that these rays don't start at the center of the earth. The key idea is, that in each point on Earth, there is a plane (called the tangent plane of the earth's surface) whose normal is vertical. in the sense that it falls along the gravity vector. To be precise, that vector gives the direction of the sum of two forces: the gravity and the centrifugal force, that comes from the Earth's rotation. The net effect is that, if we materialize that plane, a ping pong ball put on it, wouldn't roll away, but stay where it was. Now we have a z-axis as the normal of the plane. We can fix one more axis, the x-axis, by letting it point North, which means that it points into the Earth rotation axis. Then the third axis is given as an East axis. So this is the North East Down system. It may seem unnatural to have one axis down, but that is required to make the whole system a right oriented system, which is desired. Now, we can make a similar system in the reference point at the equator. Now the story is to rotate the reference point coordinate system into the local North-East-Down system. After we have done that, we can make the two systems coincidence completely with a translation, but we don't really care about that. The longitude and latitude are now truly rotations.

This definition requires a model of the Earth. If the earth is so irregular, that we couldn't hope to model it as something "mathematical", we could still go along with huge tables and the like, but it would be very complicated. But luckily, the Earth can be modelled pretty well as a so called ellipsioid. The German astronomer Bessel made such a model in the 19th century, and up till a few years ago it was a basis for the Swedish geometric system called RT90 and its predecessor RT38 (Rikets Triangelnät 1938). But then, when GPS was to be born, one found it worthwhile to make av revision, and that lead to a new ellipsoid Earth mode called WGS84. From that the geometric system of Sweden has been revised to something called Sweref. These geometric systems are about defined reference points marked out in nature, and about map projections. But the notions of longitude and latitude are well defined, as soon as the ellipsoid model is given.

The problem with the longitude and latitude is that we get a very anonymous and long series of figures. A confusion is how "Babylonian" we should be. Things like degrees and minutes and seconds (arc seconds) are Babylonian in style, in that one divides by 60, rather than by 10 and 100. The compromise is that positions are given in degrees and minutes with decimals.

All this is delivered with figures (and decimal points, commas and the keywords E and N for East and North) encoded in the Ascii format and delivered as RS232. . This information is collected into "messages" called GGA etc. These messages belong to a standard called NMEA.

If you are connected with your Microprocessor to a GPS receiver, which delivers these messages, you have some work to do. One task is to throw out all messages you are not interested in. I took the choice to pay attention only to the GGA-messages. A good GPS may deliver about 10 such messages per second. You may want to keep just one message per 2 seconds. Then you might like to acquire some data in numerical form, rather than as text strings.

One unfortunate property of the latitude longitude system is, that differences in longitude (from one point to another) are not comparable to differences in latitudes. The longitudes run together as you approach the Poles. We can see that in a commonly used map projection called Mercator (from a Dutch merchant called Kramer, which means merchant, and whose name was latinized to Mercator). It just plots points on the Earth according to their longitude and latitude, but to keep proportions intact, the scaling in the northerly direction grows as you approach the Poles. Hence you can never show the Poles themselves. In such a projection, Greenland is about as big as South America, while in reality, Greenland is only as big as Venezuela and Colombia together. As another example, Norway stretches in longitude from Marseille in France to Alexandria in Egypt. Globally we cannot do much about this. Whichever trick we try, we loose something to gain something else.

But in a local neigborhood of some point, we can do a linearization of the Earth's surface, which amounts to working on the local tangent plane there. The trick is to measure differences in latitude and longitude relative to a reference point, and then scale the differences in longitude with cosine of latitude (This holds for a spherical Earth. For a true ellipsoidic Earth it would be a lot more complicated but no matter). At the same time, we could scale everything to meters. For that, we convert the angles (longitude and latitude) to radians, and then (after scaling of the longitude) multiply by the Earth's radius, which is approximately 6360 kilometers. With the format of the GGA-messages, we get the angles as an integer that represents the longitude and latitudes in 600000:ths of a degree.

Right now, I have one such reference point in Sweden. The data I get can be plotted, and the plot gives me an image of my track, which is correct in directions and proportions, and I can sum up my distance travelled. So these positions is what I log in my on board memory together with time in seconds since noon at Greenwich that day, and together with the number of satelites used. All this fits into 3 32 bit numbers, which I store in a flash memory.

So here's in summary what the program has to do:

  1. Grab the bytes coming from the GPS receiver in an RS232 form.
  2. Throw away everything that is not a message of the GGA type. This is a message starting with $GPGGA and ending when the next $-sign arrives.
  3. Count the number of GGA messages, and only keep one per two seconds.
  4. Convert the time part of the GGA message which is given in hours, minutes and seconds into an integer representing all the seconds since noon in Greenwich.
  5. Perhaps check for the characters N and E, so that you are where you believe on the earth. They could be S and W too. I don't care for this.
  6. Put the figures together with the decimal point in the latitude and longitude information into numbers representing the angles in 600000:ths of a degree.
  7. Convert all this to positions relative to some reference point somewhere near where you move about. Above all this involves a scaling of longitude differences with cos(latitude)
  8. Extract the number of satelites used by the receiver from the GGA-message.
To store the data, I now use an EEPROM with 1 Mbits, and a file system tied to it as described here. I store 3 longwords of data every 2 seconds, and then I can sample data for almost 6 hours. If that is not enough, you can decrase the sampling rate further, or you can buy up to three more memory chips.

I need a mechanism to generate an automatic file name. It is done so that the page 2 of the EEPROM contains an initial file name, namely "gp0". Every time the Propeller starts, it takes this name, increments it with one, writes it back, and then creates a file with that name. There is a reset program that restores the name to "gp0", when you think, the memory may be full.

Once the file has been created, and GPS position become available, time and position data are written to that file.

As a consequence, the files will be called 'gp1', 'gp2' and so on.

If I get a power interrupt, when I'm out logging, I get a new file. But I can paste that file and the previous one together afterwards, to get a complete account for my track.

Since I started to employ the EEPROMs for the logging, this has worked very reliably.

We need the logging program of course. And we need the reset program to clear the file catalog, and start over from the file name "gp0". And we need a program to read out the data from the flash memory. You will find these programs in the next section


This is the GPS logging program itself. The file also contains some processes so that you can verify functions over your PC. In the exec field there are som different alternatives. The one calling pcchar will show you the stream of data from the GPS receiver. The one calling pcmess will show you the GGA-messages at the rate you keep them. The version calling pcpos will show you the positions in meters. Use the program for this. (We will return to in a while) The version without comment marker (--) is the one that loggs data.

You also have to modify the position function in the evaluate process, to fit the system for the place where you live. It goes like this: Find yourself some good reference point and determine its latitude and longitude. Google Earth can serve you with this. Then you convert these angles to 600000:ths of degrees. Put these numbers as la0 and lo0. Then compute cosine of the latitude, using any pocket calculator. We call that coslat. Then you compute the factor 1/600'000*pi/180*6.36E+06*coslat The expression pi/180 is what it takes to go from degrees to radians. 6.36E+06 is the radius of the Earth. In my case this value is 0.0967723. Then multiply this with 2^16 = 65536. That gives you an integer, which is called sla. In my case that value is 6342

Then you need the following object files: Furthermore you need these programs for resetting and reading data from the device. They use the same object files as above.

Some Java programs and instructions for use

The program gpsread.mps communicates with the Java program It is presented here. The program lets you send and receive data to the PC. When data come from the PC, they act as commands to GPSread, and they make GPSread send responses to the PC. The commands (that you type in the third textfield in window) are these:

If the size of the file is not very small, or if you have seen that nonsense text, you have found some navigation data. To bring them to you PC for further processing, do as follows: Type s in the mode (4:th) window of Then go back to the 3:rd window again, and type g. Then the Propeller will send the data again, but now they will be stored in an array in Check with the LED on the Propstick to see when the transfer is complete.

Next go to the mode window again and type a filename followed by a "<". Now you have rescued your navigation data. It's easier in the sequel if you give a file name that ends with .gps

Unfortunately the USB communication can produce some undefined data initially, so you have to go into the file with an editor like Notepad to edit the file. The real traffic starts after the hexadecimal characters AA. Remove AA and everything before it. Then, in the end of the file, there are some data, that may belong to a previous logg. With some training, you can find a break in the pattern on the file, at its end. Otherwise you can remove these old data from the plot file generated later.

Using Google Earth

Then the perhaps most interesting thing to do with your data is to plot them with Google Earth. For that you need a GPX-file. GPX is a format which is a special case of the XML format. The file contains data strings, which describe longitude and latitude.

Isn't that sad? Didn't we just transform those data to relative positions? Well, take it easy! The mapping we used there is invertible, specially if we are less than 510 km away from the reference point. Look into the position function in GPSLogg.mps. It's a fairly simple linear function (as long as we don't have overflow). Find out what the inverse mapping is and write into the program (Download information will follow)

There are instructive comments on how to do that in the program. It just amounts to replacing some constants to values that you have used in your GPSlogg.mps. Now run GPXget with the filename witout '.gps'. This will give you your GPX-file, which has the same name but with '.gpx'. Now start Google Earth, and open the open menu. There is a choice to choose so called 'gps-files' If you do that, your new gpx-file will appear. Double click on it and answer OK to the next question. Then Google Earth will zoom into the area where you travelled, and show you trip.

Other uses

The program gives you a more straight forward plotfile of your navigation data. Its name is p[originalfilename].dat. It's a readable file, so you can look at how it is formatted in for example Notepad. Then you can maybe find a plot program for that format or a small revision of it. The plot will give you a plot scaled in meters with correct proportions and directions, if you manage to get the scaling in the two directions equal.

To get even more interesting data, you could run the program That will give you a similar file called p[originalfilename]V.dat, and it will contain two more variables: Distance travelled, and velocity. If you walk, the velocity variable will seem very noisy, but if you go downhill skiing, you can at least see if you ski faster than you go skilifts.

Here's a picture from Google Earth:

This is a kayak tour around some islands at my sommer place. I also made a slalom tour around the pylons of a bridge. You can see how the track seems to be a few meters east of the bridge. I also went up a bit into a small creek. It may seem that I paddle on land a few times but that's only because Google Earth has difficulties in separating shallow water from land.

Java downloads

Here are the Java source file downloads

The latter three are utility classes that are needed for all four programs. But needs more. Refer to this place for the details.

The Mandelbrot set

The Mandelbrot set is a set of complex numbers. Now, as we have learnt here, a complex number has two coordinates (the real and the imaginary), and a point in the plane also has two coordinates, and thus we can depict a set of complex numbers as a set of points in the plane. So the set becomes a "figure" in the plane.

The Mandelbrot set was discovered by Bénoit Mandelbrot, while he worked for IBM, but the Mandelbrot set has a kind of dual relative, called the Julia set, discovered by Gaston Julia and Pierre Fatou much earlier. But Mandelbrot had the opportunity to make his set into computer graphics, which quickly made it gain world reputation.

The Mandelbrot set is generated from a very simple equation (or rather function), but once it is there, and you can show it with your computer graphics, and you can zoom into its details, it reveals itself as an extremely complicated structure.

Here is the function:

f(z) = z2 + c
z and c are complex numbers. What you do now, is that you take what comes out of the function (f(z)) and plug that back into z. And then you keep doing that. What happens then? Well one of two things can happen, z can stay limited, and maybe converge to something, or oscillate in a limited world. Or, z can go to infinity. It depends on what we start with, but we always start with z = (0,0) = 0 + i·0 = 0. And it depends on the constant c.

Then, the Mandelbrot is the set of c-values, for which z doesn't go to infinity.

To draw this, we start with some picture coordinates (ix,iy). We map this with some linear mapping to a complex number c = (cr,ci). Then we just run the function iteration, and see what happens. We can't wait till infinity for divergence, but the fact is that if divergence happens, it happens pretty fast. Nevertheless, if we don't wait long for divergence, then we miss some precision in our drawing. OK, if there is no divergence, we draw a white dot at our picture coordinates, and then we go somewhere else. Here we wait for divergence 30 iterations. If z hasn't diverged in these 30 iterations, we assume, that it never will.

With this program we actually draw a three color figure. We say that if z hasn't diverged after 10 iterations, we draw it red, if not white. If it hasn't diverged after 5 iterations, we draw it blue, if not red or white. These multicolored accounts for the Mandelbrot set became quite popular after the publication of the Mandelbrot set. But, it is the Mandelbrot set itself (the white area), which is the most interesting.

The point now is, that by just changing the mapping from (ix,iy) to c, we can effectively move around and zoom out of and into the complex plane, and thus look for details in the Mandelbrot set.

However, with fix point arithmetics, and 32 bits word length (which is good enough for most purposes), it is hard to give justice to the Mandelbrot set, so I recommend anybody to repeat the experiment with, say, a Java program, where the variables are represented as double precision floating point values. In addition, you have a screen with higher resolution.

But this Propeller program is a little feasability test, which shows that you get a recognizeable Mandelbrot set. Here's what it looks like with the Adafruit screen.

The limited resolution of both the display and the camera gives the picutre a Moirée pattern, that you don't see in real life.

Download Mandelbrot.mps

I hope you can follow the program from the code. It has an outer loop over all the display pixels, and an inner loop of iteration, which goes on for n=30 iterations, unless there is a divergence. The pixels are generated in such an order, that you can fill the pixels by just sending the pixel color for each pixel.

Recent developements


After having spent lots of time on waiting for Windows to start up, I decided to switch to Linux. I bought a new computer with Linux Ubuntu installed. For the Propeller, I installed a Linux Propeller environment. It works fine, but I miss the possibility to run it from a command line call. Always with Linux, one has to consider, that Linux distinguishes between small and capital letters in filenames, which creates a great deal of work.

USB has to treated a little differently in Linux. USB has to be used, as serial communication goes over USB in the so called Prop Plugs. In Windows, Java doesn't have good support for USB, so you have to use C/C++ code and so called ports. In Linux, they have the view, that a USB port is a file, called /dev/ttyUSB0. This isn't quite true, though. For a file, you have a program, that has to read a file. A port, on the other hand pushes data into the PC. That difference causes some problems sometimes, but mostly it works to read USB-data in this way. It means that the read function is blocking, while it is waiting for data from the port. Sending data to a Propeller through a write to /dev/ttyUSB0 seems to work without problems.

The Linux program screen is also a usefull tool. Screen is a very complex program with an enormous help file ("man screen"). It is a screen or window, with a process behind it. This means that you can't close the program by "crossing" the screen, because that doesn't stop the underlying process. Instead you enter a command mode with ctrl-a, and then issue a 'k' for kill. You are asked if you really want that. So the whole command is ctrl-aky. Things you type in the window, goes out on the USB, and can be received by a MPS ()pc.rec function. Things sent with an MPS-call ()pc.bsend appear on the screen so you can see the results of your MPS efforts. Then you can enter the command mode with ctrl-a, and then a colon (:). This allows you to enter less anonymous commands like 'hardcopy -h filename'. The selector -h is important, because it gives you also the data that have been scrolled out at the top of the screen. Now you have your data in a normal file. This procedure may seem a little clumsy, but it gives you a good feeling for what happens.

This switching to Linux has worked out well, and I am not going back to Windows again. It is fortunate of course, that I have used Java, which works equally well and unchanged under both operating systems.

Instances of classes.

I have mentioned something about this already. In a fully object oriented language, an object is an instance of a class. You create and name such an instance with a constructor, and then this instance is available, independantly from any other instance of that class.

In MPS this is not so. If we call a constructor (usually named initsomething, we don't get anything with any identity, so that we can distinguish one instance from another. In Java, the identity of an instance lies in a pointer, though Java had some aversion against the pointer concept, so it never mentions the word pointer.

The Ada language was not originally an object oriented language, but in the later versions, they augmented it so that it could handle instances of classes. The idea was a little like this: An instance consists of two parts: variables, which are called field varables, and functions, which are called methods. For different instances, the field variables are different, while the methods are the same. Then Ada arranged the field varables in arrays, selected with an instance index (or pointer if you don't have the aversions). Then, when you call a method, you have to supply it with the correct index, for your desired instance, so that the method could work with the right set of field variables. So practically, rather than an abstract pointer, you have a concrete index.

I tried this idea in MPS, but it did become a little costly, so right now, I don't have that array of field variables.

The field variables are in a way a representation of the state of the instance. In many cases, this state is not very dynamic, but just static. You assign values in a constructor, and then they never change. Then, when you want a new instance, you run a construtor for that. Now, the original instance is gone, so if you need it again, you have to run its constuctor again. So now your second instance is gone, but you can run its constructor again if needed.

Then, if you use these instances in different processes, you don't need to worry. The field variables are physically different variables, if they belong to diffent processors.

An example of use of this is a class for SPI communication. Different instances of the object, correspond to different SPI channels, connected to different Propeller pins. So the field varables are the different pin numbers for data out, data in, clock, and enabling devices. And all these data are static. They do not evolve during the use of the bus.

The SPI class corresponds to a file called Spio.mpo and it contains

  1. field variables listed between 'fields' and 'endfields'
  2. methods, with dot notation like 'spi.send'
  3. some contructors called spi.initX for different Spi busses
  4. a destructor
The destructor may be meaningless. In the SPI case, I use do disconnect pins being used as output pins. Disconnect means to convert these pin into input pins, which wouldn't interfere with using such a pin for something else than that SPI-channel

Right now I have the folloing constructors: Here's a download opportunity: Spio.mpo

Litium batteries

Using batteries is sometimes handy, like when you build mobile instruments, remote controllers, robots etc. Modern LiPo batteries are good for these things. They contain a lot of charge per mass unit.

But beware: You shold only charge them with chargers designed for LiPo batteries. Also, these batteries shouldn't be charged in series. The reason is that some cells can have a lower voltage. The charger tries to give the cells the right total voltage. As a consequence, the good batteries become overcharged, The remedy to this is to use a balancer which bring charge to the bad batteries from the good ones.

I have preferred instead to charge the bateries connected in parallell, though I'm not absolutely sure that it is safer than series charging with a balancer. But so far, I have not seen any problem.

The ones you buy now contain built in protections against discharging and overcharging. If you discharge a battery too much it is no easy to charge it again. But the protection cirquits help you against that.

Here's my design for a litium battery power source.

The batteries are connected independently to four pins, to which you can couple plugs. The "use plug" connects them in series, and the voltage then goes to stabilizers which gives 5 and 3.3volts. A charge pump Intersil7660 gives you a negative voltage, which may be useful if you have analog operational amplifiers. I have omitted capacitors connected to the stabilizers, but they are now often built into stabilizer modules. For charging, you use the "charge plug" connected to an authorized charger for Litium batteries. As the batteries are now connected in parallell, you should put the charger in an 1S mode. Finally, if you want to save battery, you can connect a AC power supply with the "Supply plug". Then the batteries are disconnected from the circuitry.

Here's a supply box:

I use it for some of my designs. Others have the batteries built in, and other are driven by AC power supplies.

News for GPS

The GPS logg was down for some time. Without much hope I bought a new GPS receiver, of the same type as I had before. To my surprise that worked. The problem was that there had been very few position fixes for some time.

I've built the system into a little round plastic box, which I can carry in my pocket (it is not the worlds smallest plastic box, but its OK). Nothing electrical comes out of the box, and the box is waterproof.

I have also simplified the storage in the on board memory a little. The new file system for this is in the file filee2 .mpo. The files are no longer fragmented. The data are stored in a long array at consecutive adresses. The current adress for storing is saved in the EEPROM itself at a given adress (32). The data array then starts at adress 33. Below 32 is a "file catalog". The files don't have names; they just have numbers. The first adress for each file is stored at that number. Up to the reserve word 32, there are 31 such slots, which make up for the file catalog.

The point is this: If there is a power failure, but the system starts up again, the startup procedure will create a new file. It gets its place at the first free slot in the file catalog, and from there it points to the first place for data, which is at the adress after the adress stored at position 32.

We can now see that data have been distributed over more than one file, and we can paste these files together. But we can also pretend not to observe this, and just read all the data as a single long file. It will contain points of discontinuity, if the power interrupts have been long, but it may not be harmfull.

Like previously we need a reset program for clearing the file system. It resets the 31 first positions to 0 and clears the content at adress 32 to 0. In fact, it clears some more EEPROM words, just to make the action a little more visible.

More important than these things, is the switching to Linux, which requires a new procedure:

  1. Supply the device in the box with power from the batteries, by installing a little plug.
  2. Connect the Prop plug, and reset the system by loading the program gpsreset.mps
  3. Load the program gpslogg.mps, or interrupt the power for a short while, so that gpslogg.mps is reloaded from the Propellers own EEPROM.
  4. Close the lid, and go for your adventure, with the plastic box in your pocket.
  5. Back home, open the lid again, instert the prop plug and load gpsread.mps
  6. Open the screen program in Linux at your PC. Hit any key, and your data will come flying.
  7. Press ctrl-a and : and hardcopy -h filename, where filename is a filename you have just invented. You should run screen from a directory you know, so that you can find the files.
  8. Run the Java program GPSget filename. This will make a plottable file p[filename].dat.
  9. Run, optionally, the Java program Dist filename to get a file p[filename]V.dat, which also contains velocity and distance information.
  10. Run the Java program GPXget filename to get a file, [filename].gpx which you can plot in a geographic map.

On that last point, Google Earth seems to have abandoned its services for plotting gpx-files. I found the website, which works just as well. Open the website, and hit the "import" button, and choose your file filename.gpx, and the system will zoom in to your track.


Most files are unchanged from previously, but filee2.mpo is new, and the Java files may have changed.

Some of these files have an added '2' to their names, to distinguish them from earlier versions.

The three Java programs import a package called std. It is my own standard Java package. You have to create that package by importing the following three files, which make a package by mentioning the keyword package std.


Many languages allow you to define symbols, and assign values to them, without reserving space for any variables in the computer memories. The symbols are handled by the compiler only.

I found it worthwhile to introduce this in the MPS language too. Syntactically, the symbols are single lines defining a symbol name and its value, like

badluck = 13
Such definitions are placed in .mpo-files between the keywords defs and enddefs

The definitons are then loaded with a load-command

load from (Termi.mpo)

i.e. the same way as a function is loaded from an .mpo-file.

Then the mps-program can use the definition with the code ?badluck The question-mark, would mean "What is the value of this symbol?". The compiler will then replace "?badluck" with the value 13.

Here we don't need to store the value 13 in a separate variable in the Propeller. But the Mps-code is more readable, as we can figure out what the number is supposed to mean, or represent. The compiled code (Propeller assembler code), on the other hand is not more readable, but it is not supposed to be.

The Terminals

The Terminal

Here's the terminal:

It is a box that contains:

  1. Some serial RAM memory
  2. An SD card holder
  3. A numerical keyboard
  4. A graphical display
  5. A three color LED (RGB-LED)
These things are offers of services for a less equiped Propeller computer (the host). By sending orders over a link to the terminal, data can be stored and recovered or typed on the display, entries from the keyboard can be captured and the RGB-LED can be made shine.

The picture shows an example, where the Mandelbrot set has been drawn again, but now with a host using the terminal. After that, a program has turned on the RGB-LED with green light, but the graphical display still shows the Mandelbrot set. The camera has dimmed the light of the LED pretty much. It looks better in real life.

The orders are sent as Morse messages.

In the image above, you can see the link as "rainbow cable" entering at the front. It contains power, as 5 volts and ground, and a bidirectional Morse link. On this link, the host sends orders to the terminal, and the terminal sends acknowledgments back to the host. (You can also see how the link is also a bus - an order bus - which continues to some other terminal)

The host is supposed, not to send any new order, untill the acknowledgment has arrived.

Physically, the two Morse wires connect pins 26 and 27 on the respective Propellers.

The terminal software has

An order consists of a command which is given by a symbol in a definition in an mpo-file called TermI.mpo. Optionally, there is an argument associated with the command. The task processes split the order into command and argument.

The task processes are essentially loops around a series of if-statements, which check if an appropriate command has come. If so, the task is executed, usually with a call to a function. After that, the execution goes on to a common point, where the sending of acknowledgments is administrated.

Normally the acknowledgment is the command itself, but sometimes, when the host asks for some value, like the content of a memory cell, that value is returned instead. Right now, I let the acknowledgment process administrate this difference, but there is a global variable x for the value to be returned.

There is also a simple "internal acknowledgment" at the common point, in that the order variable is reset to 0.

The host helps itself with an mpo-file TermI.mpo ("Terminal interface"). It contains symbol definitions corresponding to the different commands. The terminal imports these same definitions, so that the two systems are coordinated. TermI.mpo further contains a ti.init function, which establishes the Morse communication. ti.send0 sends a command, and ti.send sends a command together with a single argument.

But TermI.mpo also contains some composite functions. This is due to the fact that a command can only have one argument. A typical example is the task to fill a rectangle on the display with a color. The terminal will need to know the boundaries of the rectangle as N(orth), S, E and W -limits, and the color. These arguments are set by orders like setN, setS,... and setcolor. A composite function

( N , S , E , W , color , )ti.fillrect
creates these suborders, and also the final order,which fills the rectangle. This makes the writing of host software easier.

There is an extra complication with all this. The reason is that a single process can't handle everything, that controls the graphical display. First of all, there is (as usual) a process for initializing the display. When its job is done, it leaves over to another process in the same processor, to do the rest of the job. But now we need yet another process to handlle the display.

The display in my case is handled through an SPI protocol. It means that pins are connected to pins on the display for clock, MOSI, MISO and enabling.

It is easy to see, that a Propeller has 32 pins (GPIO-pins). But in fact it has 32 x 8 pins, 32 for each processor. They are connected to the physical pins through a network, which essentially or:s the 8 different pins together. Now if a pin for some processor is 1, then no other processor can make itself heard; the output is 1 in every case.

So, what we have to do, is to disconnect a processor's output pin, when that processor doesn't want to communicate over that pin. This amounts to changing that pin into an input pin. For the display, we have a gra.connect function, which makes output pins. But now we also need a gra.disconnect function to make input pins

When the terminal software wants to make something with the display, it connects to the display. But the other process has to disconnect itself.

This is handled with an "internal order" in the form a global variable, called dproc (for "display process") When a process notices that dproc doesn't point to itself, it disconnects itself from the display. It also resets dproc to 0 as an "internal acknowledge". All this is an extra if-statement in the taskprocess' loop.

The connecting process, say number 1, does its job by calling a little function as ( 1 )setdproc.

The waiting is to allow the other process to disconnect before the SPI bus is used.

With this, the terminal works exactly as one would wish. After it has sent its acknoledgment, it is ready to handle any order. It doesn't need to be turned off, ever. You can reload the host with new code, and run it directly without resetting the terminal. (You can see this in the picture, where one host program has drawn on the display, and another has turned on the green light. Both these effects persisted, so that I could photograph them.)

The terminal hardware is displayed here:

At the left is the Propeller's own EEPROM. It needs a pull up resistor, which is omitted in the figure.

At the far left, you see the input from the host, including ground and 5v, and the bidirectional bus between the host and the terminal, which is connected to pins 26 and 27. (There is a diode at one of the wires. I will comment on that later). The host delivers 5v which has to be reduced to 3.3v. (I could deliver 3.3v from the host, but I fear that I could overload the host's 3.3v stabilizer.) Then, there are 3 SPI-buses for communication with The graphical display has the standard SPI-signal (clk, MOSI, MISO, chip enable) but additionally a d/c-pin to mark sending of commands/data, and a reset-pin. At the reset output, there is a little ring which represents an inversion, done with an external inverter chip. You might think, you could just as well do this inversion with software, but the problem comes, when the Propeller starts, and the program has not yet started.

The RAM and SD-card share a single SPI-bus with identical clock and datapins, which is OK, as these pins are driven by different processes. You can't use these two buses simultaneously, but the terminal software never does two things simultaneously.

The RAM consists of four chips, and thus needs four chip enable wires. The SD-card does with just one

The keyboard is a network of switches. The figure below shows one of the 12 switches, the one closed when key "0" is pressed. These switches connect one row line and one collum line. The numbering of these lines is shown in the figure. The numbering is a little iregular, but it seems to be an "international de facto standard".

The output from the collums send voltages to the Propeller, and these inputs require some pull down resistors, as is shown in the previous figure. In the software downloads, you can find an algorithm for finding out which button - if any - has been pressed.

Finally, there is a catchkey process running in parallell with everything else, and which maintains a value of the global variable key which is the current value of any key pressed on the keybord (if no key is pressed, it delivers 15). A getkey order will deliver this value as an acknowledgement to the host.

That same process also handles the RGB-LED, helped by an object file col.mpo. It contains two functions:

They are activated through orders colormix and shine. 'index' is a pointer into a limited set of standard colors. The LED is controlled with a PWM system. If the argument on is 0, then the light should stop shining, but I am not absolutly sure about that. The shine PWM-system is run in the shine function, which is called from the catchkey process.


The following orders could be found in Termi.mpo as definitions. I split them into some categories. Texts within [ ] are comments.

(In this list, I have placed the arguments of the command after the command, unlike my convention in MPS.)

sdflush means the following. When writing to the SD card, you first write to a buffer in the Propeller. The buffer is written to the card, when it is full. sdflush means that the buffer is written anyway. read, write, sdread and sdwrite include that the adress is incremented.

Text can be written to the display in the following way:

This is handled with the ( adr )ti.stype composite function in TermI.mpo. adr is the adress or the text to draw in the host computer.


Termtest.mps is a test or demonstration program. Look into it and try to guess what it tests.

Term.mps is the terminal software. TermI.mpo is the interface between Termtest.mps or any other host program and Term.mps.

The Analog Terminal

Analog inputs and output might be nice parts of a terminal. But the terminal box is full, and most of the Propeller's pins are used. So I made a new, analog, terminal. It is not so picturesque, but I will still show a picture of it further down.

Here, instead, a hardware diagram of it.

It has

To the upper left, you can see the orderlink to the host. Here I will explain the diode in the acknowledgment line back to the host. We have two terminals now. And both want to send acknowledgments. The Morse communication "rests" at zero. Hence, to be able to send messages from two or more sources, we need to or the signals together. The diodes are parts of an "oldfashioned" hardware or-gate. Another component required is a parallell resistor at the host's input pin. Without it, the impedance of the propeller's CMOS input, might very well exceed the backward impedance of the diodes.

The ADC and DAC are straight forward SPI-components. The ADC has 8 input channels. You choose them with the SPI protocol. In the order, the channel choice comes as an argument. The DAC has two outputs. You choose the channel with two different commands. Both take the value to convert as an argument.

There are four "free" analog channels, going to input pins of the terminal. There is one channel connected to a potentiometer with a knob at the panel. Two analog channels are reserved for the touch screen. All four outputs of the touchscreen are connected to Propeller pins, but two of them also go to two analog channels. They are controlled from functions in the Touch.mpo object file. How they work is described here. (In fact, I thought that the touchscreen was broken. These Nintendo screens are not easy to find nowadays. But, I tested it, and it worked!)

The power amplifiers are parts of so called H-bridges, each with two amplifiers. They come in pairs in modules. I have two such modules, so I have a total of 8 such amplifier, but I have only connected 6 of them. They should be regarded as "binary", so you feed them with either 0 or 3.3 volts. (What else could you? They come from Propeller GPIO-pins!)

They could be used to control motors through PWM (pulse width modulation). They could also be used to control stepper motors. This is done in separate paralell processes, which can be turned on and off with orders from the host.

The parameters to these processes are set in a task process. There are three pulse width fackors called pw1, pw2 and pw3 for three different motors. And there is an angle set value xc for the stepper motors. The stepper motor process has a state x, corresponding to xc. It is the angle of the motor measured in units of the motor step size. When x and xc differ, then the process steps the motor and increments x, which goes on untill x=xc. Hence xc is a motor angle command, though with an undefined null point.

Here now, a list of orders to the Analog Terminal:

  • Analog input
  • Analog output
  • Just a reminder: If the host wants to send an order adcget(channel), it makes the following function call:

    ( ?adcget , channel , )ti.send

    The output from the touchscreen can be called xy. If it zero, then the touchscreen isn't touched. Oterwise you can find x and y with:

    xy >>8 ->y

    xy &355 ->x