ftomteen4cat

Keep in mind that the PCM signal is frequency-modulating an RF carrier. It is the RF medium that is subjected to interference. In it's digital form, PCM allows for redunancy checking, error correction, (though not used in our implementation) and interleaving.

I'm not trying to make a point for PPM over PCM. I'm trying to clear some of the misconceptions about these methods.

Phil Cole

My experience is that PCM receivers will operate without holds when a PPM receiver in the same situation is jittering quite badly. I can't honestly say that the PPM model would be unflyable, but you'd certainly want to land as quickly as possible. When I do get PCM holds, it's quite obvious to me while I'm flying. The model doesn't lurch like it would with a PPM receiver, so it may not be obvious to someone else. I get single hold events a lot less than I used to get PPM lurch events when I flew mostly PPM. The PCM update rate is low enough that one or two missing updates to a major flight control servo channel is easily noticeable.



I've also done some simple comparative range testing. I was comparing the failsafe range for PCM with the complete loss of signal for PPM. PCM came out with two to three times the range consistently. I have not compared the onset of PPM jittering with the onset of PCM holds. The question, of course, is how much jittering do you tolerate with the PPM case?



So much for my observations. Now for some opinions:



In our context, the definition of PCM includes any error detecting or correction that may occur. Performance through the air of the RF channel itself is irrelevant. The error detection is necessary, since data errors could occur in the MSB of the data fields, or in the channel address. These sorts of error, if passed though to the servos, will result in far worse jitter than at the output of any PPM receiver. If they do error correction, that's a bonus.



Given the bits/channel of information that need to be sent for eight channels with 1024 quantizing levels, and the data stream on the DSC output of my 9Z, it's very possible that they are using an error correcting code.



In any case, given that I see PCM operating without holds when, under similar conditions, I see PPM operating with significant jittering (though not necessarily fatal) I have to conclude that the PCM, as used for RC, is more robust.



When there is a problem, I prefer to deal with a model going in and out of hold over a jittering model. I have never had a complete loss of control with either system, so I won't comment on that. With PCM, the holds may occur when the model is much further way than would be the case with PPM jittering. This would be the case if you forget to extend the Tx antenna, for instance.

ftomteen4cat

There is an excellent article that delves into the encoding for both schemes. If you enjoy technical literature, you'll like it.

http://www.aerodesign.de/peter/2000/...M_PPM_eng.html

Phil Cole

Now, there's a decent site. There was an interesting article on brushless motors too.



The details of the Robbe/Futaba PCM coding he gives don't match with what I saw coming out of the DSC output on my 9Z. I didn't see any evidence of the differentially encoded second frames. By varying each channel in turn I was able to see which bits changed and which didn't. There was only one group of constant length for each servo channel. Also, ch 9 did not have its own group of bits. Ch 9 appeared to be encoded by changing the coding of ch8 somewhat. Ch 9 can only send on/off, and things like ATV, mixing, etc., are not available.



It may be that the Robbe version is different to the 9Z, and the R149DP can decode either. Or, the encoding through the air is different from the DSC encoding. I doubt the latter is the case, though.



Also, if you are prepared to do the math, you can correct bit errors using the CRC16. It does require quite a bit of calculation, maybe beyond the processor they could put in the receivers.



I can't fault his range test methods. The results are more quantifiable than anything I've done. Maybe I'm less sensitive to the holds than I thought.

bob_nj

My Feedback: (62)

I'm going through a situation where a 1/3 gasser I bought had a tangled mess of wires from the receiver (located at the wing tube area) to the rear of the plane. This guy made extensions but I don't know how they ever worked. I emailed Roger about some optic replacements and hope that will take care of the short range and jitters. A couple of questions I have are;
1. Some of you have gone from 6v back to 4.8_Explain
2. Some have eliminated 6v regulator?
3. Where is the Andy Low website?
4. How does an individual make his own 30" extensions lets say, with 22 gauge wire when at the plug end, it's still 26 guage for instance. In other words, does the small wire at the plug negate any of the good stuff that goes along with the bigger wire?
How effecitive is twisting? So many questions, so little time.
Thanks and take care_Bob

Cley

Nice web site!

It's going to take me awhile to be sure that I'm not misinterpreting it (maybe forever) so maybe the Electronics Professionals could give their interpretations of it.
DPCM: It seems that I've seen this relating to Futabas equipment years ago though I may be mistaken.
It looks to me like this guy is saying that Futaba 1024 PCM has the fastest Frame Rate of any of the PCM systems and that they send each of the main channels (that are being used the most)
information twice (at least most of it, some overlapping?) one after the other.. Wouldn't this cause a single bit error to only affect one of the frames with the next adjacent frame that follows it immediately replacing the faulty frame? so the hold would be nill practically?
How do the refresh rates compare with PPM by his information?
I have personally seen Holds in the air and on the ground easily!
But this was years ago. As the plane would roll for example you could literally see it stepping as the servo would stop then move more, lagging behind the stick movement. Gas Plane.
Also 2 conductive objects moving close to one another making and breaking contact or not? are said to create interference. I do not see how they could possibly be creating a signal by themselves just from picking up the current from a radio signal so wouldn't this mean that they are changing the radio signal as it is airborne? And what about this?
>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
(quote)Servo's are not damaged by pulses that are too long/short, which could happen with PPM.
>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
When I start to see the servos buzzing, that's when I start to see occasional glitches. Under the same conditions with PCM I don't believe that any problems are detected. It's been my experience that PPM on a gas plane with the range extended out a bit is dangerous to ones airplane.

ftomteen4cat

Wouldn't this cause a single bit error to only affect one of the frames with the next adjacent frame that follows it immediately replacing the faulty frame? so the hold would be nill practically?

I'm not sure about this. The key/delta method described is a little like mpeg video compression. Using absolute and difference data conserves bandwidth. Since every frame contains absolute and difference data on alternate channels, rejecting a frame would have a consequence on the subsequent frame, since the difference data in the subsequent frame would have no absolute reference from the previous frame. If they reference the last valid frame's data, the hold could last for just one frame. If not, the hold would have to last two frames on the channels that had absolute data in the rejected frame.


How do the refresh rates compare with PPM by his information?

Futaba PCM updates faster than PPM. (?)


Also 2 conductive objects moving close to one another making and breaking contact or not? are said to create interference. I do not see how they could possibly be creating a signal by themselves just from picking up the current from a radio signal so wouldn't this mean that they are changing the radio signal as it is airborne.

Sudden potential changes create noise all over the spectrum. You can see this by hooking a battery up to a file and scraping a lead across it while listening to an AM radio. (or run an AC power tool with a universal motor) A metal rod going by an ignition module is subjected to EMI, and will act as a transmission line/antenna for this signal, which can be orders of magnitude larger than the signal from the TX while the plane is flying. To make matters worse, when it comes in contact with another conductive body, the impedance, gain, and potential of this part all suddenly change and create noise.


(quote)This explains why , at the limits of the transmission range,with PPM (because of the overlaid noise) the servos are buzzing heavily, because the pulse flanks vary (up to+/-30 us), while PCM keeps them quiet, having half a bit (150 us) to play with, thus avoiding faulty interpretation of a bit.


He's saying that the PCM receiver is synched to sample a bit's value in the center of it's pulse, which is 300us long. Since his measured timing variances were only +/- 30us, they could never affect the center of the bit.

Phil Cole

Cley,

The PPM frame period is around 22 ms. It will vary between manufacturers and the number of channels a little. According to him, the Futaba PCM is actually faster than PPM, and the others update at about half the rate.

My 9Z appeared to send two groups of four channels each in PCM. Each group of four took 26 ms. So, it was 52 ms to send eight channels. Usually, the groups alternated between channels 1 through 4 and 5 though 8 if I didn't operate any controls. However if I waggled the airleron, I would see ch1 in both groups while the stick was in motion: the groups would be 1,2,3,4 and 1,6,7,8.

I was using a generic 4-ch setup. I didn't have time to get into flapperon, ailevator, etc. type setups.

As for the holds, I used to think the same thing as you: holds are easy to detect. I've certainly noticed them while flying and when I've done range testing by watching and feeling the servos while walking away with the model.

The test on that site showed quite obvious holds (over 100 ms) under the same conditions as PPM was showing about 10% jitter. 100 ms is enough for a servo to travel a third to half its range, so I would have thought that I would really notice holds that long. At the same range, PPM apparently jitters 10%, so your model should look quite jiggly as it flew along. The slowness of the servos will smooth the jiggles somewhat, but according to his test, you should see holds at about the same range you see PPM getting jittery. I don't see/feel PCM holds until PPM in the same situation is really bad.

Hence my statement about not being as sensitive to holds as I thought: If can't feel the holds when PPM would be starting to jiggle the servos then either the test shows PCM to not be as good as it really is, or I'm not feeling the holds until they are really bad.

However, my Tai-Ji 40 takes less than a second to get around in a snap roll. After a bit of practice, I can stop the rolls within about 30 degrees of horizontal (I'm no threat to Chip Hyde!) - this means I have to time it to 30/360 * 1000 ms = 110 ms. This is about the same as two complete frames of hold. So, I should be able to detect holds of 100 ms or longer while I'm flying.

So, I don't really know whether to believe in his test result, or my ability to feel PCM holds.

What his test was put a signal into the transmitter that corresponded to moving a stick all the way across and back every two seconds. He then recorded the signal the receiver sent to the sero. The record would have been the pulse waveforms, between 1 ms and 2 ms as the stick moved. The graphs show the pulse durations as a function of time.

The top graph shows what the receiver should have got. The holds are shown by the flat portions of the bottom graph - the receiver was in hold, maintaining a constant servo position. Most of the holds were only for one frame, but two lasted for what looks like more than 100 ms.

Phil Cole

I think what Cley's asking about is 'metal to metal' interference caused by making and breaking a connection between conductive objects in the model. Apart from intermittent wiring and ignitions, there is no potential source large enough to generate the interference. Electro-chemical and thermoelectric potentials are much too small to affect the received signal, particularly at close range.

All the conductive objects in the model will have some effect on the radiation pattern of the receive antenna. If the pattern changes, it can change the phase of the received signal, which looks just like FM. If two or more conductive objects are large enough to significantly affect the phase, then they can modulate the phase if an electrical connection between them changes. Since the 'power' for the interference comes from the signal itself, signal strength does not matter. The effect will be just as bad at 10' or 100' range. I've seen this in helicopters with bad bearings going into hold at 10' range. Another case was a frame that cracked while in flight. It was a PPM receiver, so the heli started glitching badly. The glitching really didn't change much as the model was flown back to where it could be landed safely. It stopped when the model was on the ground, and not being vibrated by the engine.

A large object and small object (e.g. a loose screw in a single piece of metal) won't cause a significant phase change.

If chemical or thermal potentials were involved, the problems would be worse as signal strength decreased.

If there is a locally generated potential involved (e.g. ignition, RFI from a digital servo or gyro) then the problem does get worse with range.

Phil Cole

1. I will be doing it in one model I have to save weight. One less cell to carry.
2. See above - no regulator either.

Other reasons would be changing to servos that work well enough on 4.8 V, or installing gyros, etc. that aren't rated for more than 6.0 V. Five cells can get up to 7 V fresh off the charger.

4. The 26G lead with no extention is better than any extention of any gauge. A 22G extension is not as bad as a 26G extentsion. So, if you need the extension, thicker (within reason) will always be better.

Twisting can be very helpful in some cases. The ideal is that each wire sees the same induced signal on average. If they are not twisted, then one wire will pick up just a bit more than the others, and this could get where it's not wanted.

Things work the same for transmitting and receiving, so the twisting will also help reduce interference being radiated by the leads to the receive antenna.

bob_nj

My Feedback: (62)

Has anyone used the Aerografix fiber optic jobbies?

Cley

It will be interesting to see what the future holds for our radio systems...
Too bad the manufacturers don't provide more technical reference regarding their systems..

RCPilot100

My Feedback: (4)

Do you guys realize that in the time that this subject has been discussed on this subject - on RCUniverse alone - I have built two MiniLasers and finished my CGM Sukhoi inspite of the lousy parts and instructions they provided with their new and improved kit. And, I will still use my Futaba 9Z WC2 transmitter with Futaba and Hitec PPM receivers - two receivers in my Giant Scale - and will be a frequent flyer this summer - and not worry about my radio systems. Enjoy!

Dan

ftomteen4cat

Maybe that's why none of my planes are done... hrmm

Just to clarify, I wasn't talking about galvanic coupling. I was thinking more EM.

BTW, I think twisting in unbalanced operation adds shielding.

Phil - I checked with the RF guys here, (I'm in an advanced radio technologies group) and they confirmed that my line of thinking on intermittent mechanical connections was on track. Slight potential changes will swamp the receiver at the microvolt levels that they are sensitive to and either make it difficult to resolve the signal or get the AGC to crank down.

regards

SteveBoy-RCU

I have been watching this thread with interest and am still undecided whether to go PPM or PCM. It seems there are arguments for both.

Is it true that you can program out "holds" on PCM as an option? So then does it behave like PPM and show the glitches in flight?

Thanks, Steve

FCC

I know many of you will become angry on this comment... but just think of it.

How about a Transceiver Sytem with auto freq(channel) shift ability? The explanation of this system is this that there will be a Tx in the plane and Rx in the hand set (full-duplex). And in case of intereference or common channel problem, the system will sift automatically to some other free channel or hold the plane at constant altitude and after recovery (signal loss) must be in control.

I think now R/C has evolved much and many of us carry $100 to $2000 planes in th air and nobody wants his ship to the ground...
Nodoubt the price of this system will be very high but as the cellphone techology have become mature... why don't R/C?

ftomteen4cat

SteveBoy -

You can change what it does when it goes into failsafe. I haven't heard of a setup that allows changing the hold behavior. That doesn't mean one doesn't exist though.

I'm sticking with PPM for my glow planes. It works well, and I'm happy with it.

I'm putting a PCM receiver in my 1/4 scale gas-powered extra. (still building) I've had people I trust tell me that they are better for planes with ignition modules in them. Considering the cost of the model, the cost difference between a PPM receiver and a PCM receiver is pretty small, and I don't want to take any chances.


FCC -

That's not a bad idea. I'm not sure why you think it would anger anyone. The TX would need a synthesizer, but the RX could rely on a single back channel to xmit link quality info. Each TX could have an ESN or a MAC address. Think about it though, if you were able to do full duplex, you could do packet data and even share channels using a multiple access method like CSMA. To squeeze all this info into 3kHz, we'd need to use another modulation like viterbi - FSK wouldn't go very far. It would take some development, and probably everyone would need to switch for it to work, but it wouldn't have to be expensive in the long run.

Companies that develop cellular spend enormous amounts of money doing it. It's not unusual to have several hundred engineers take two years to develop a new standard, but cellular is a lot more complicated than what we're doing. The infrastructure is pretty deep. The mobile receiver is a piece of work too. Cellular's not a good paradigm for RC though. Sorry if I've rambled on this.

Phil Cole

>>>>>Phil - I checked with the RF guys here, (I'm in an advanced radio technologies group) and they confirmed that my line of thinking on intermittent mechanical connections was on track. Slight potential changes will swamp the receiver at the microvolt levels that they are sensitive to and either make it difficult to resolve the signal or get the AGC to crank down.


If there was a voltage source for the potential I would believe it. Intermittent servo wiring (within itself, or shorting to a conductive structure) can cause the problem and falls into this category. I've certainly seen this happen.

I just don't see where the potential comes from in a lot of the cases I've seen. When I think about it, the intermittent contact usually seemed to open and close a conductive loop. Opening a closing a shorted turn with some mutual inductance WRT the antenna could do it. I figure an 8 degree phase change in 150 us will give an apparent 1 kHz frequency shift. This would be enough to cause PCM bit errors, or PPM decode errors.

How easy it is to get such a phase shift at the receiver input? In a helicopter loops of several square inches an inch or two from the antenna are quite common. The antennas are fairly high impedance (being short), so could easily be affected. It's been so long since I did this stuff that I've forgotten how.

ftomteen4cat

The potential I referred to is from local sources, through induction from fast current spikes going to servos, and the previously mentioned ingnition modules. This interference could even come in through the IF. I'm not sure if intermittent galvanic coupling producing several millivolts would have any effect, even at sensitivities near -100dBm. I suppose I wouldn't rule it out if the antenna was sitting on top of it. At close range, even waving a magnet by can cause interference.

I think I understand what you were getting at now. At first I thought you were referring to multipath. I honestly don't know what effect rapidly changing the reactive properties of the antenna would have on phase at the base. I can see if anyone knows.. I'll email you if I find something.

In any event, it I think that either swamping or phase shifting can create bit errors in a linear system, and the effect of either is a product of each model.

regards