July 7, 2009
Don Mathers and Dough Spreng developed the truly proportional system still in use today. Different modes of modulation were later introduced (FM) other than AM, PCM was introduced, but still, it is the Mathers/Spreng servo at work at the end. Even digital servos still use the system.
AM to FM. This is often confused, many thinking that we have to go FM once we are on the upper band. No, in fact, there were FMs on 27Mhz. FM is a modulation form to transmit information. It does not change the the Mathers/Spreng system in positioning the servo. In AM, the strength of the signal is varied to send info. In fact, in RC, the carrier wave(CW) is momentarily shut off to represent a signal. This is also practical since no extra power is use to send a short pulse. The receiver takes care of recognizing the blank and convert it to an ON signal at its output. In both AM and FM, a very short trigger pulse (.25MS) is sent at intervals. The intervals represent the time a port is ON at the decoder, which sets the servo position. This is called Pulse Width Modulation(PWM). PCM which stands for Pulse Code Modulation can be viewed by the layman as also a modulation form. Instead of the short pulse sent, the TX actually sends a binary code to the RX. The RX has a CPU which produces the command pulse for a servo (from 1 MS to 2 MS). In PWM, the TX sends the ON/OFF commands to several servos in sequence. Thus, usually,it can only send about 60 commands to each servo per second. In PCM, in as long as there is no new command from the TX, the RX keeps generating the same pulse width. So, an interference on the signal does not mean anything to the RX since it is unintelligible to it. PCM today is usually on FM. It remains a very good system against interference. But it should not be confused as to what the spread spectrum (SS) is doing. In PCM, The system is very good against unintellgible interference. But another TX on the same channel can of course interfere with one's RX. The spread spectrum takes care of this aspect. In frequency hopping (FHSS) for example, the signal is rotated among different spot channels and the sequence of the combination code is locked between the TX and the RX (called binding), and there are thousands of combinations possible, thus probability for two radios using the same code is very low. So, the SS is not an improvement on the modulation. It is an improvement on the use of frequency space. It could well be that PCM is used and transmitted the SS way on 2.4Gig. The 2.4G is just coincidental that the FCC opened up this band for wireless communication. Again, it could well be that SS were used at 72-75 Megs, or even 27 megs.
What is the Mathers/Spreng system?
Well, the best way to appreciate it is to look at the servo. In a way, this was the heart of their invention. Before them, the servo was analog. A motor is geared down to developed torque and there was a feedback pot on the final shaft. In the analog system, the pot adjusted the bias on a transistor. When an increased/decreased drive voltage turns ON/Off the transistor, the pot counters the move(opposite wise). So, the changed is canceled by the changed in the transistor bias controlled by the pot. The problem of the analog servo was delayed reaction, non-proportionality and overshoot, among others perhaps. You see, the voltage level moving was purely analog, meaning, it was on a curved path. There must be something discrete like On then OFF but sent continously and when a change is introduced, the interval between the ON/Off is changed.
The concept was an incoming pulse with a width from 1 MS to 2MS. A local pulse is also started by the incoming pulse but is inverted (positive incoming and negative local pulse). When one is longer than the other, a residual pulse remains. TX sends the trigger pulses which the decoder vonvrts to a positive pulse of a duration. The TX sends the trigger pulses on a stream (repeated). When the residual pulse is positive, the motor is turned one way. When the residual pulse is negative, the motor is turned the other way. Now, a pot (normally 5k today) is again, attached to the final output shaft of the servo and controls the width of the local pulse. It is wired so that the width of the local pulse matches that of the incoming pulse as the pot is turned which then stops the motor. By that time, the servo horn is at a different position.
How is the command pulse created?
It is created at the decoder. The TX sends two short pulses that we can call trigger pulses(.25MS). This is not the servo command pulse. It is the signal for a decoder to turn a port ON on the firsts pulse and OFF at the second pulse. This port is the servo channel port. In a hypothetical single channel system, the TX will just keep sending two trigger pulses with intervals controlled by a joystick. And since the max interval is 2 MS and there are 1000 MS in one second, the TX can send 500 command pulses per second to 1000 command pulses when the delay is at 1MS. But all of this changes when more than one channel is sent. First, the OFF trigger pulse of the fist channel can be made to be the ON trigger pulse of the second channel. Thus, only three trigger pulses are needed to create two command pulses at the decoder. However, to make sure that the first trigger pulse turns on the first port, a long inverval called a pause is introduced in the stram of trigger pulses. This must be at least twice the longest interval which is 2MS, thus 4MS. So, in a 2 channel system, the max intervals is 2x 2MS plus the 4 MS for a total of 8MS. Thus, a system can be designed to send 1000MS/8MS commands per second. A set of command pulses for the two servos is called a frame. That will be 150 frames per second. Mind you fellow modellers, if you are using an 8 channel TX and RX to control only two servos, the resolution is wasted.
PCM
Perhaps the above will make us appreciate PCM more. In PCM, the above computation becomes irrelevant.The TX in a PCM does not send pulses on merry go around such that we add all the intervals which becomes longer the more the servo channels. In PCM, it is like sending the juke box a number for a record to play and you do it only once. In PCM, the TX tells the RX-CPU to generate a pulse for a given channel only once until you had moved that joystick for a new command. There is no stream of commands to a set of servos. In PWM (or PPM), a cut in the stream sends the servos chattering. In PCM, a cut in the transmission is merel;y recognized as silence and does not mean much to the RX-CPU which is waiting only for intelligible commands to shift the pulse widths to a set of servos. PCM is truly intereference proof but not from same channel TX. That is where the SS comes in. By the way, PCM was not developed for the RC industry. It was developed for general communications works but was particularly used to send commands to the Mars probe and to send images from the Mars probe to Earth.
A short explanation of
Spread Spectrum
Let say we have 3 channels only on 72 MHZ. The way to allow more that 3 TX to be used at the same time is to make a system wherein the TX hops channels. Let us call the channels A,B, and C. Tx1 will hop ABC,TX2 ACB, TX3 BAC, TX4BCA, TX5CAB,TX6CBA. This of course is a very simplified semblance on the principle used. The pattern that the TX will adopt is copied into the RX when they are bounded(binding). Thus it is nearly impossible for two systems to cross.
The 2.4Ghz band is very high frequency of course and is not limited to RC. Wireless phones, cell phones and other services are in this band. The antenna is very short indeed as the wavelength of the signal is 300/2400 or just .125 meter or 1.25 CM. The problem at this frequency, are wires around the antenna such as servo extension cables which robs the RX of range. And that is one reason 2.4's suffer momentary cut offs. SS is good but the band gives this problem. One remedy is to wrap the extension cables around a ferrite ring.
To follow is a detailed explanation of the "hardware" of a typical TX and RX with decoder. Complete circuitry will be explained.
BK
