rrTV-PHOTO   New HD TV
HOME   rrTV-PHOTO   GALLERIES   MY GALLERY   HELP-FAQ
myHOME PM pmRR MEMBERS 905 ONLINE 73 EVENTS SEARCH REGISTER  START HERE
 
44 pages [ <<    <     1      2     ( 3 )     4      5     NEXT    >> ]62859 viewsPOST REPLY
Autography FlightPower . Advantage Hobby . Revolution Models

.
.
Radio - Servo - Gyro - Gov - Batt > TX/RX eCCPM Latency Test Results
 
 
JKos
Elite Veteran
Location: Town of California in the state of Maryland

Well, I tried to measure the latency of the 9C Super and the 7C, but they have a very strange behaviour which is not making it easy or even meaningful to measure the latency. It seems that these two radios smooth the collective/throttle input. Here is a plot showing the response of the 9C Super to a step inputs on the collective.



Just to make sure it wasn't something PCM related, I also did the same test with an FM receiver and noted the same results.



I'm not sure why Futaba has elected to do this. I looked through the manual and also through every selection on the menu and can't find any option or way to turn this off.

Compared to other radios which transition in one or two frames, this doesn't look so good.

- John
07-05-2005 Over year old.
 
 
w.pasman
Elite Veteran
Location: Netherlands

JKos such interpolation would be disastrous for the latency!!
16 frames that is more than half a second latency !?! I dont think you can fly with that

This looks like you have the smoothing option turned on. This is especially useful for electro motors, where you don't want full power at once, which might ruin the gears and other parts. This is called 'delay' , p. 50 of the FF8super manual

FF8s manual is on my website

(
http://graphics.tudelft.nl/~wouter/...8/ff8manual.pdf )
07-05-2005 Over year old.
HOMEPAGE  
 
 
JKos
Elite Veteran
Location: Town of California in the state of Maryland

w.pasman,
I checked the Delay function thinking that is what was causing it. However, it is set to 0 with no further way to turn it off. Further, the delay function acts between flight modes and should have no effect while in a given flight mode. The smoothing seen above occurs while staying in one flight mode.

I will be asking Futaba about this behavior.

> 16 frames that is more than half a second latency

No, it is just over 200 ms. To be noted is that it didn't reach its true stick-position value before being turned around to go the other way; Therefore, the latency is actually even longer.

- John
07-05-2005 Over year old.
 
 
w.pasman
Elite Veteran
Location: Netherlands

Are we talking about PCM1024? I remember the frame rate was 35 Hz or 28.5ms per frame, that gives 16*28.5=456ms latency for 16 frames?

Furthermore "The Delay function provides a smooth transition between the trim positions whenever offset, revolution mixing, or throttle hold functions are turned on and off.". So that's much more cases than just switching flight modes?
07-05-2005 Over year old.
HOMEPAGE  
 
 
JKos
Elite Veteran
Location: Town of California in the state of Maryland

w.pasman,
My frame time measurements are shown in the thread on eCCPM output behaviour. It is approximately 12.9 ms per frame for Futaba PCM1024. JR SPCM is about 20 ms, Airtronics PCM is about 12.7 ms, and Futaba G3 PCM is about 16 ms.

About the delay function... Again, it should only come into play when something on the transmitter is switched, just as the instruction manuals state. When in a steady-state mode, it should have zero effect.

- John
07-05-2005 Over year old.
 
 
JKos
Elite Veteran
Location: Town of California in the state of Maryland

I would like to thank Stan Johnson of Hobbies 'N Stuff (home of Callie Graphics) for allowing me to borrow the Futaba 9C Super and the 7C for testing purposes.

I would also like to thank John Pompa for allowing me to use his JR 9303 and Tom Medland for allowing me to use his Futaba 9C.

- John
07-05-2005 Over year old.
 
 
w.pasman
Elite Veteran
Location: Netherlands

Quote 
My frame time measurements are shown in the thread on eCCPM output behaviour. It is approximately 12.9 ms per frame for Futaba PCM1024. JR SPCM is about 20 ms, Airtronics PCM is about 12.7 ms, and Futaba G3 PCM is about 16 ms.


Mmm I wrote it down so compact that we need to do a little calculation to get the frame time. I will repeat some details for clarity.
Transmission of 1 full + 1 delta-coded channel takes 16 bits plus 8 bits CRC = 24 bits. On top of that 6to10 coding gives 40 bits. We need 4 of those packets for a full frame = 160 bits + sync bits = 190 bits per frame. Every bit takes 150us. So that makes 28.5ms per frame.
So I guess you must be misinterpreting the signal somewhere and taking the halfway point somewhere as a sync pulse for the next frame?

Or is the PCM1024 even more flexible than we thought?

What's going on here??
07-06-2005 Over year old.
HOMEPAGE  
 
 
JKos
Elite Veteran
Location: Town of California in the state of Maryland

w.pasman,
The frame times I have quoted are from the leading edge of a pulse on a given output channel until the leading edge of the next pulse on that same channel. The numbers for PCM1024 are very similar for the 9C, 9C Super, 14 MZ, and 7C and were verified both on the logic analyzer and on an oscilloscope. I am not looking at the PCM encoded signal but rather the receiver outputs.

Could there be a difference in operation at 72 MHz versus 35 MHz?

- John
07-06-2005 Over year old.
 
 
w.pasman
Elite Veteran
Location: Netherlands

Ah I got it !

You must be assuming that the frame rate you see to the servos is the same as the frame rate of the signal going through the air

So I guess you have to straighten up some conclusions
07-06-2005 Over year old.
HOMEPAGE  
 
 
JKos
Elite Veteran
Location: Town of California in the state of Maryland

> You must be assuming that the frame rate you see to the servos is
> the same as the frame rate of the signal going through the air

That is how the systems seem to be behaving. For example, if this wasn't the case then my latency measurements would have been much harder due to the inconsistency of how the system would respond. As they are, the latency measurements are 100% repeatable and consistent down to very low levels of timing.

If tx to rx frame times were mismatched with the servo frame times, then what I observed on the scope and logic analyzer would be much harder to interpret. As it was, it was fairly easy to hone in on the latency and watch the receivers output flip from one frame to the next while going just a few microseconds over the minimum latency.

- John
07-06-2005 Over year old.
 
 
w.pasman
Elite Veteran
Location: Netherlands

That's a quick reply!
Quote 
That is how the systems seem to be behaving.


Why do you think I'm wrong here? Did you ever look at the signal between radio and receiver?
07-06-2005 Over year old.
HOMEPAGE  
 
 
JKos
Elite Veteran
Location: Town of California in the state of Maryland

Seems I was editing while you responded. Please see above.

I did look at the PCM stream of the Stylus out of curiousity but did not attempt to decode it. The structure appeared to stay aligned time-wise with the output pulses from the receiver. Again, the systems sure do seem to respond such that the frame times are aligned.

Perhaps I will see if I can capture the PCM stream and servo pulses. That would help us see how long each bit is. I only have an Airtronics Stylus available right now.

- John
07-06-2005 Over year old.
 
 
noahb
Senior Heliman
Location: Carlasbad, NM

9303

So how does the 9303 compare with other radios that you have tested?

Is it on par, better, or worse.

Thanks
07-06-2005 Over year old.
 
 
JKos
Elite Veteran
Location: Town of California in the state of Maryland

noahb,
If you look at the first post in this thread the numbers are posted there. The JR 9303 appears to be about three times faster than the Futaba 9C and just a tad slower than the Airtronics Stylus and Futaba 14MZ (with G3 rx).

- John
07-06-2005 Over year old.
 
 
noahb
Senior Heliman
Location: Carlasbad, NM

9303

Thanks.
07-06-2005 Over year old.
 
 
davehour
Key Veteran
Location: Guayama, Puerto Rico

9C

Thanks
07-06-2005 Over year old.
 
 
w.pasman
Elite Veteran
Location: Netherlands

John

First, note that you don't need to decypher all this to find the latencies. You can measure both minimum and maximum latency by just measuring. Check my report on the PCM1024 latencies [if you need details on how to do that] [I write this because I'm not sure now what exactly you assumed and what you measured].

But to answer your question.
Just knowing the pulse lengths is not enough to estimate the frame time. I could do it for PCM1024 because I exactly know all the format details.
For PCM1024 the structure also probably "appears to stay in line with the servo pulses". That is because the frame length is exactly TWICE the servo drive speed.
For PCM1024 you could easily check that I'm right, by just hanging a scope with a loose pickup wire next to the Tx antenna. I bet you can see the sync pulses of PCM1024 coming by at just the rate I mentioned.
With Airtronics this might not work easily as I think you said that you could not find sync pulses. My guess is that there are sync pulses but that they are hard to recognise. We could try to figure that out if you put some ATX wave samples here.
To sample this without a digital storage scope, you might try using the computer audio in attached to the Tx trainer port output and grabbing with your favourite audio sampler. For PCM1024 that is the same, although unmodulated, signal as on the antenna. For ATX you might want to check that.
07-07-2005 Over year old.
HOMEPAGE  
 
 
JKos
Elite Veteran
Location: Town of California in the state of Maryland

w. pasman,
> You can measure both minimum and maximum latency by just
> measuring.

That is exactly what I did. I did not look at the PCM format at all. I simply injected an input on the collective stick, varied the delay between a pulse on channel 1 and a the stick input, and watched the receiver output as the delay was varied.

- John
07-07-2005 Over year old.
 
 
w.pasman
Elite Veteran
Location: Netherlands

Great I think this issue is resolved.

I just reread the first part of the thread. This is what made me think you assumed some things to get the max latency

Quote 
I found it more difficult to test this setup due to the odd nature of how the channels change so I only went for the minimums. In theory you would add one frame time to get the maximums.


That comment gave me the impression that you just assumed that for the results you gave (without making measurements), but found it to be failing in the 9C case.

So you ment here that normally what you measure as max latency is the min latency plus one frame. Apparently you just found this to be true for most cases. But as we discussed that might fail, first because "radio frame time" may not be equal to "servo frame time" and second because multiple "radio frames" may be involved in transmitting a signal.
07-07-2005 Over year old.
HOMEPAGE  
 
 
JKos
Elite Veteran
Location: Town of California in the state of Maryland

Finally, results for the Futaba 9ZWC2 have been added. Please see the table in the first post of this thread.

A little better than the 9C, but still way behind the 9303, Stylus, and 14MZ.

Thanks,
John
08-03-2005 Over year old.
 
 
44 pages [ <<    <     1      2     ( 3 )     4      5     NEXT    >> ]62859 viewsPOST REPLY
CarbonXtreme . Midland Helicopters . HeliProz

.
.
Radio - Servo - Gyro - Gov - Batt > TX/RX eCCPM Latency Test Results
  UPDATE SCREEN   PRINT TOPIC Advertisers 

Subscribe to This Topic

Tuesday, May 13 - 3:03 am - Copyright © 2000 - 2008 runryder.com | email | link to rr | runryder needs cookie