HarryC

My Feedback: (1)

Its not harder at all, it just asks was this servo's data shorter than X or longer than Y. It does not check frames, it checks every single servo for validity. Taking into account the entire process of error checking, then PCM having to convert the signal back into PPM for the servos, and all the extra data that has to be sent on PCM as well as servo data, PPM is the faster system across all channels. To try and compensate, PCM systems can use a refresh every cycle for channels 1-4 and refresh every second cycle for channels above that. Well that's not so good for the aileron and elevator servos on channels 5+ which lag behind a tiny bit.

.
Any signal which is valid overruns my signal and produces an invalid signal so it pops straight into failsafe. It can't accept another signal over mine, whether valid or not.

I challenge anyone to an aerial down range contest of their PCM against my Multiplex PPM, because I, and every Multiplex user knows we will win easily. Maybe not for S/N reasons, maybe because Mpx reputedly pumps out the max milliwatts allowed while others don't, but Mpx users know their range with Mpx PPM kit is simply awesome.

H

HarryC

My Feedback: (1)

You can install a programming button on a flylead. Programming could not be simpler and does not require programming skills. The only death due to interference in the UK was not on a PPM model, it was on a PCM where the user had not programmed the fail-unsafe on his easy to get to Tx because he found it too difficult.

On what basis is this judgement? I have been flying IPD for a few years, our airfield has a known glitch alley where plain PPM model glitch like clockwork every time around the circuit, mine goes through without a care, and it's a Synergy 3D (or similar CAP etc) with 80 degree elevator travels and very rearward CG, the slightest twitch and I would know all about it. On 3 occasions in all those years it has popped into full failsafe for the blink of an eye then come back into control. Is that totally equal to PCM?

H

JohnW

My Feedback: (6)

HarryC - You are correct, I was wrong on the details of IDP, but the end result is still the same. This is what I found at Modelspot.com on IPD.


Start-------->
The advantages of IPD technology

1. Only valid signals “get through�

The IPD receiver only considers a signal valid if its length lies within the range 890 µsec to 2350 µsec. These are limit values which cater for most radio control transmitters - even those not made by MULTIPLEX.

2. Invalid signals are replaced (HOLD MODE)

If an invalid signal is picked up, the receiver rejects it and passes on the last valid signals received. It continues to do this until a “good� signal arrives again. This action alone suppresses much interference, or at least reduces its effect.

3. “Safety setting� (FAIL-SAFE) in the case of total signal failure

If no valid signal is picked up for a particular period (varies according to receiver type), the IPD receiver moves the servos to a freely programmable position (Fail-Safe settings), e.g. throttle closed and/or control surfaces to neutral.
End------------<

However, this sounds just like what PCM does, but PCM works off a checksum as opposed to looking for a valid PPM pulse width. So, unless I am missing something, IPD locks out just like PCM with mild interfernece. PCM will revert back to the last know good commanded position under mild interference. And under heavy interfernce, both PCM and IPD go to full failsafe. While my facts were off a bit, the jest of my post is still correct, both IPD and PCM act very similar with interference.

You are correct that PPM can typically send a valid frame faster than PCM, but this varies from PCM to PCM. Futaba PCM seems to be the fastest as they use some form of differential PCM frame system. But then we need to ask our selfs, how many frames per second is enough? Can anyone really feel the individual frames in the air with PPM or PCM? The average person only needs 15 frames per second to detect smooth motion. I believe both PPM and PCM are above 15 frames per second.

No, IPD and PCM are not totally equal. Overall I'd give IPD a slight edge. I base this mainly on the ability of IPD to work with any standard PPM TX, not that it is a superior technology. IPD performs the same functions as PCM, it just goes about it in a different matter. If it can be proven that IPD offers more control compared to PCM under typical interference one might experience at the field, then I'd give IPD the edge it deserves.

As for PCM hits, I went into full failsafe once in 2003. The problem was someone turned on in the pits on my freq. I yelled out and as soon as they turned off in a I was back in full control. The PCM entry/exit speed from failsafe varies from RX to RX, but at least on mine, I can't detect any noticable lag. Equal to IPD? From what you described I'd have to say yes.

XJet

The problem with IPD (or any receiver-end system) is that it can only detect an error if the pulsewidth received falls outside the acceptable range of 1.5mS +/- 600uS or so.

So, smaller errors, which may still be enoughb to drive your control surfaces to full deflection, will get through.

For example, if a burst of noise results in series of pulses that are 0.950mS, 1.10mS, 1.05mS, 0,99mS, etc -- that's enough to drive your elevator, aileron, rudder or throttle servo to full deflection -- even if it's supposed to be at neutral. The receiver-side system won't flag those pulses as errors because they're within the "acceptance band" of 1.5mS +/- 600uS.

By comparison, PCM can detect an error of 1 bit -- ie: 1/1024th of the servo's true position and block it.

In fact, a PCM system treats an error with a magnitude of 1/1024th of a servo's movement exactly the same as one that might be the full movement of the servo -- and blocks both.

So, while the receiver-end systems are an improvement, they're not as good as PCM when it comes to actually allowing a user to maintain control during a period of interference.

The problem with receiver-side stuff is that there's just no way for the receiver to know whether it was your transmitter or a burst of interference that sent all those short pulses -- although it could use a slew-rate detector to help discriminate between stick-movements and interference to some degree -- but then you have the problem if someone programs a sudden change in pulse-width into their transmitter -- such as a snap-roll button or something -- the receiver may well reject the inital pulses as noise and introduce some lag into the control-surface movement.

It's a complex setup -- and what surprises me most is how expensive PCM receivers are compared to smart PPM systems like the FMA M5. That's the main reason why many people don't fly PCM I suspect. Even the Hitec PCM receiver is significantly more expensive than its PPM units -- disproportionately I'd say.

HarryC

My Feedback: (1)

Ah, you've fallen into the elephant trap! IPD does see errors that are within the time limit!

Servo pulses are not allocated a complete 2.3ms in which they just fill the time they need and leave the rest blank. If that were true then your description would be correct. But instead, servo pulses are like wagons on a train, no matter how long or short they are, at the end of the wagon there is the same little gap and then the next wagon starts. The entire frame (train) changes its length every time you move the sticks and cause each pulse (wagon) to change its length. So interference that causes a pulse to lengthen will cause it to overlap the next pulse, the two pulses become one and added together they are outside the limit. So IPD sees errors that re within the time limit it allows because the error causes two pulses to merge into one invalid pulse.

When an invalid pulse is detected the rest of the frame is rejected, so all valid pulses pulses before that point in the frame, get through. In a few milliseconds the next frame starts and each pulse is again checked for validity, if the interference has stopped then each pulse is valid and gets through.

H

tadawson

Under most real world cases, it probably does. In the theoretical argument ***IT CAN'T *** it is technologically impossible for IPD to distinguish valid bad data from good data - period! PCM, as XJet noted, has the error correction built into the protocol, so it is inherently more capable than trying to do it after the fact. All the other "Well I saw it do this or that" arguments are basically crap - the theory cannot be argued! As I said a few posts above, in the real world, most would probably not notice the difference, but IN THEORY PCM will always be superior! As an electrical engineer degreed in communications, I think I can see this pretty clearly, and I disregard the "emotional" arguments as just extraneous crap. I am talking pure technology here - not what goober likes at the field. And if that aforementioned goober is too damn stupid to program failsafe in his TX, the problem is not the protocol, it is located directly between the ears!

- Tim

Crash90

My Feedback: (17)

I am talking pure technology here - not what goober likes at the field.
[/quote]


LMFAO

HarryC

My Feedback: (1)

You are wrong, as I have already explained a valid length but error pulse overruns into the following pulse making a merged invalid pulse.

H

tadawson

HarryC - since you are not the "lord god of interference" you cannot guarantee that ANYTHING will happen - that is why it is interference! What CAN be depended on is how a given unit will respond when the interference hits.

- Tim

Forgues Research

My Feedback: (7)

LMFAO
[/quote]


No goober, just what really works at the field.

Roger

Lynx

Tadawson, if it's only better in theory, it's not better. This is not an emotional responce so please don't just throw it off as 'heresey' <G> The practicality of the matter is that no true application of PCM vs PPM can ever be argued inteligently because they're fundamentally different. Argueing one is better over the other is a complete and total waste of time. And PCM can't tell the difference between valid bad data and good data either. If it's valid it's valid period, PCM or PPM. It happens less with PCM and no user will probably have ever had it occur long enough to notice, but the error detection method that PCM uses is not 100% reliable, certain patterns of interference can produce false 'good' packets. Thing is they probably only occur for 1 frame out of like 100,000 or so. IPD is not better than PCM either, it's just a method of filtering out the most common error signals in a PPM frame. What redundancy and 'correctness' PCM offers it also hide's information that a user experianced with PPM/IPD glitches can interpret as a serious problem and react to. Just like there are people out there swear by PPM and will die by it, and there are people out there that will swear by PCM and will die by it, keep in mind they're BOTH usefull, you just have to be aware of the good sides and bad sides of each and know which to use in a given circumstance or environement.

JohnW

My Feedback: (6)

Lynx: I agree than standard PPM and PCM/IPD are fundamentally different... but in some respects that is what makes one better than the other. It is the ability of both PCM and IPD to enter fallback and failsafe modes that makes them "more advanced" than standard PPM RXs. Is more advanced "better?" In this case I think so because properly setup, a PCM/IPD plane has less chance of causing property damage or personal injury. Of course, there is always an example of a plane with a poorly setup failsafe that casued damage, but this is not the fault of the technology; instead the user misued the technology. Yes, it is more difficult to detect minor problems with PCM/IPD becasue of the inherent filtering they perform, but it is still possible to detect small errors with some extra work. The only reason I can see for not using either PCM or IPD is cost, which is a valid reason. I see no problem with those on tight budgets to use PPM because of the lower cost.

Crash90

My Feedback: (17)

So can somebody please explain how to properly set up failsafe?

mglavin

My Feedback: (31)

Have we touched on resolution afforded or realized of IPD verses PCM? Signal processing/filtering will be at the expense of resolution, right? What about PPM and its inherent pulse jitter? Does pure or true digital technology skew the results?

JohnW

My Feedback: (6)

Crash90: The "proper" setting for failsafes will open up another bag-o-monkeys. I think most would agree that an idle setting for throttle is best... after that, we all depart. I suppose ideally, the failsafes should be set for spin. The idea here is that if there is enough interference that you don't have control over the model, i.e. you've entered failsafe, it is best to spin the plane into the ground at the point. If the plane was where it was supposed to be in the first place, this should put the plane down in a clear overfly area. If you have a very large overfly, I think one could justify idle and neutral aileron/rudder/elevator. Most planes at idle it can't get too far before they would hit the ground. If your overfly is huge, this would also typically be safe as long as you weren't flying at yourself at the time. Of course, if the interference leaves, you could regain control in either case and save the plane. Nothing is perfect, and no matter how you set up failsafes, there is still the risk of property damage and/or personal injury.

Crash90

My Feedback: (17)

I didn't mean to take this thread off subject. I will repost the question elsewhere. Sorry guys.

HarryC

My Feedback: (1)

Do they matter? I think they don't. The reason is that so many of these "improvements" are only visible if you write the number on a bit of paper. They are not visible to a human being in the real world. I have demonstrated my PPM system resolving to greater than 1400 steps yet I will guarantee that no-one will be able to detect an improvement at the servo, the control surface or in flying characteristics compared to a poorer 1024 PCM and quite frankly I think anyone who claims they can tell the difference between a 512 and 1024 system is just convincing themselves their money was well spent. The difference at the servo output between a 512 and 1024 system is approx 0.12 degrees which for a typical sport model translates to an angular difference of around 0.05 degrees at the elevator or aileron. That assumes that the servo is perfectly precise, which it certainly is not and that sort of angular difference is far smaller than the resolution lost in the servo's pot and is fraction of the angular slop in many servo's gearboxes. In other words, you can't see it even if the system was perfect and the system loses that difference in all its other errors anyway so their is no difference for you to see. We could go on and on making higher and higher resolution PCM, try a 16384 system, wouldn't that be impressive? Not in the slightest. Why? Because what matters is what humans can tell is different, not what is a different number on paper. So what if PPM frames suffer from analogue jitter, you can't see it because servos can't respond and servo output arms don't jitter because of frame jitter. It exists but if you can't tell the difference between a steady PCM and a PPM jitter, does it matter.

If I make you watch a servo, and ask you to tell me if it is being controlled by a PPM, PCM512 or PCM1024, you could not answer that question. You can not tell the difference. PCM will only produce a steady number on controls that you are no holding off centre, such as steady throttle, main controls at neutral, flap switch etc. Once you hold a control off neutral such as up elevator, your thumb jitter will cause the numbers to fluctuate just as it will cause the PPM frame to fluctuate, and that fluctuation is so much greater than PPM frame jitter that the frame jitter disappears inside it.

Don't forget that the signal from stick to servo starts and finishes as an imprecise, uncertain analogue signal, even in digital PCM sets. The output from the stick pot is analogue and imprecise and uncertain. It jitters due to analogue uncertainty. That uncertainty is transmitted precisely by PCM. At the Rx, it has to convert the PCM back into PPM because servos are PPM, so it introduces a new layer of analogue uncertainty all over again. PCM is not digital from end to end, it is only digital between Tx encoder and Rx decoder. The signal from stick to encoder is analogue varying voltage, and from decoder to servo arm is analogue PPM.

PCM has a lovely mathematical certainty compared to PPM. But take those certain numbers off the piece of paper and apply them to the real world of cheap carbon track pots, in sticks as well as servos, gear slop, thumb jitter, servos unable to repond to the frame jitter, poor human resolution, and then the mathematical certainty of PCM over PPM disappears.

H

Geistware

My Feedback: (16)

HarryC, I have to agree~!!!!
Hey that rhymes!

Highflight

My Feedback: (4)

Atta' way Harry. That's what I was trying to explain but you did it so much more elequently.
AT SOME POINT, there is a departure from the technology where real world application allows you to forget about the technology.

If at breakfast, I am presented with two different waffles that were made with different ingredients, I don't care what the ingredients are if they both taste good.

Highflight

mglavin

My Feedback: (31)

1400 steps? How is this possible? The maximum you can get is 1024 and the maximum swing in PW is 1200 microseconds. Therefore each step is 1200/1024 = 1.1719 microseconds. In practice you can't get 1024 steps out of the PPM because noise in the form of pulse jitter exceeds 1.17 microseconds. I have observed the EVO and IPD RX operate at 5-6usec, this appeared to be the minimum pulse width or resolution at the RX I could note.

Other factors which are common IMO, such as TX's programmed with limited end-points or ATV functions in the realm of 300usecs, which exasperate the resolution conundrum...

What happens when the driving pulse width signal is wider as realized with IPD? These are valid points nothing less... In some circumstances this argument is valid, especially with less precise low end sport servos. Step up to the high end precision digitals offered of today and you may note and see the differences, IMO...

Alas, if you can see and feel the difference with high end precision digitals it does matter, IMO. I know "IF" is a big word in this assertion.

Interesting analogy. Truthful, nothing less.

We could bring up things that happen in route or not from the encoder to decoder that shines in the face of PPM.

True but all things being equal as you note PCM is superior on paper... Some of this superiority may actually make it to the surface and be realized.

Were presently fitting the Multiplex RX 12 DS IPD in a new 40% model and using the Royal EVO 12 to manipulate it. Looking forward to drawing a conclusion to this debate with first hand experience... From what I have read and heard of others has all been good. This if nothing else supports your assertions.

We also played with some low end IPD's type RX's of other manufacturer's the resolution realized was poor comparatively, but now one is complaining, go figure!

rmh

Mmmmmm.
I think PMS is at play here -

Lynx

Keep in mind the angular throw on a transmitter is something less than 45 degree's. I'd challenge anyone to move the control stick in as small of an increment as they possibly can through the full range and be able to count past 200 or 300 Higher resolutions can be 'felt' while flying but don't really impact flight at all unless you're pattern flying, and even that is debateable if you're a good pilot.

XJet

And let's not forget that the dead-band for most non-digital servos is around 8uS. If we assume that the operating range of a PPM channel is +/1 600uS then dividing the PPM delta by the servo deadband effectively produces a maximum resolution of 1500 steps anyway.

Digital servos (with their smaller deadband) have the potential to produce higher resolution -- but, as others have pointed out, we're talking about such small amounts of physical movement here that this begins to pale into insignificance when compared to slop and flex in the control linkages and other components of the physical control-system.

bwd

Highflight,I couldn't agree more.Had a pcm go into failsafe,couldn;t do a thing about.Got hit on ppm airplane went into hold,
Was high enough to turn off the trx and back on & got the plane back.

HarryC

My Feedback: (1)

My error, it was over a year ago I did the test, I found it resolved to greater than 1200, not greater than 1400 steps. But not 1200, greater than 1200 steps. The max is not 1024 steps in PPM. The max is 1024 in a 10 bit digital system. A true analogue PPM system would be stepless, it has an infinite number of steps. In practice the encoders these days are likely to be digital and therefore output the analogue PPM in discrete steps. The number of steps will depend on the encoder used. Since the output is not in bits it doesn't have to have 2^n steps as a digital system does, it could be made to output in 10 steps or 51 steps or whatever, though I would expect it will be much the simplest to output in a 2^n sequence. I don't know if the Tx I used in the test has a true analogue or a stepped output.

The max swing is 1200us, but that does not mean it steps in 1us at a time! It might step 10us at a time in which case its resolution is 120 steps. It might step 0.5us at a time in which case it has 2400 steps.

I used an Mpx Profi 4000 Tx, sending to an IPD 9 Rx and reading the pulse width at the servo sockets on a device that reads to 1us. Some frame jitter can be observed at that resolution. The 4000 allows for unbelievably fine step settings on its mechanical trims as low as 0.29us per click! Trim total effect set to 1% of travel, across 42 clicks, = (0.01*1200us)/42. This assumes that travels are turned up to the max allowable %, whereas at normal values (1100us for Multiplex) the steps are even finer.

Therefore it should take 3.3 clicks of trim to make the output on the pulse width monitor step up by 1us. Because the monitor does not read tenths of a microsecond I could not see if each click was stepping by around 0.3us or if the system was just accumulating trim clicks and then stepping by a full 1.00us. If it just accumulated trim clicks and then stepped by 1.00us then resolution would be 1200. If there is evidence that it did not step by 1.00us then even though that step can't be seen the evidence of it would point to a resolution greater than 1200. So much for theory, what actually happened? It took sometimes 1, often 2 clicks, and sometimes 3 clicks each time to step the output by 1us. I can't remember the exact ratio of each, it was over a year ago I did the test. So that showed the resolution was at least 1200 steps because the reading was moving in 1us increments after usually 2 trim clicks. Often on the second or third click the number would rapidly jitter up and back down by 1us indicating that there was indeed a fraction of a microsecond being read and that frame jitter was making it fluctuate around the 0.5us mark so the monitor was rounding up and down in rapid succession with the frame jitter. If I moved the trim back one click the jitter stopped, if I moved it forward again the jitter re-appeared and if I then it moved the trim forward another click the jitter disappeared as if the fraction was now far enough above 0.5us for frame jitter to always keep it in the rounding up area. Perhaps I am wrong, but I took this behaviour as evidence that the Tx -> Rx system was resolving to less than 1us and it was not shown on the monitor which did not output a reading to less than 1us but its behaviour indicated that it was seeing less than 1us and sometimes getting trapped in jitter around the rounding up/down fraction. On this basis I decided that it was resolving to less than 1us, greater than 1200 steps, the monitor output could not show it directly but the stepping behaviour was evidence that the fraction of a microsecond was being read.

H