Hello,
I have been measuring the latency of various transmitter/receiver pairs. This test is designed to measure how long it takes from the time a stick is moved until the corresponding output is seen at the receiver's output. For this test, I will be using the collective/throttle stick input and monitor the output of all three swash servo channels in 120 deg eCCPM mode.

For a quick look, here are graphs of the 2.4 GHz radio systems I have tested in eCCPM mode.
Sorted by minimum latency, average, and range (max minus min)

Here are graphs of PCM and 2.4 GHz radio systems I have tested in eCCPM mode.
Sorted by minimum latency, average, and range (max minus min)

Graph showing the distribution of the latency samples collected during testing. This shows even more detail of the latency characteristics of a system than the numbers below.

I will add each result to following table as I get them.

DSM, DSM2, FASST, XPS in eCCPM mode
|
Transmitter |Receiver|MLF|MLL|MaxLF|MaxLL|Avg
---------------+--------+---+---+-----+-----+-----
ATX RD8000 |92824 |14 | 19| 34.3| 39.3| 26.8
ATX SD-10G |92104 | 4 | 10| 10.3| 16.3| 10.3
ATX SD-10G |92104 | 4 | 4| 16.3| 16.3| 10.3 (10)
Futaba 8FG |R6008HS |11 | 11| 17.7| 17.7| 14.1 (9)
Futaba 8FG |R6008HS |11 | 11| 24.8| 24.8| 17.6
Futaba 10C/TM10|R6008HS | 8 | 9| 15.2| 15.2| 11.7 (9)
Futaba 10C/TM10|R6014FS | 7 | 7| 28 | 28 | 18
Futaba 12Z/TM14|R6014FS |19 | 19| 57.8| 57.8| 38.4
Futaba 14MZ/DM8|AR9000 |20 | 23| 74 | 76 | 47 (4)
Futaba T6EX |R606FS | 8 | 15| 54 | 60 | 34.25 (5)
Futaba T7C 2.4 |R617FS |11 | 13| 39 | 41 | 26.4
Hitec Aurora 9 |Optima 7|41 | 46| 62.6| 72.6| 55 (11)
Hitec Aurora 9 |Optima 7|41 | 43| 62.6| 66.6| 53.5 (12)
JR 12X |R1221 |20 | 21| 43.5| 45.5| 32.5
JR 9303/Spektru|AR9000 |40 | 45| 63.8| 68.3| 54.4
JR 9303/XPS Tol|XPS 8-ch|34 | 52| 74 | 81 | 60 (7)
JR 9303/XPS Now|XPS 8-ch|33 | 33| 61.7| 61.7| 46.9 (8)
JR X9303 |AR7000 |24 | 26| 45.5| 49.5| 36.2
JR X9303 |R921 |24 | 26| 45.5| 49.5| 36.7
Spektrum DX6 |AR6000 |59 | 63| 80.6| 88.6| 72.8
Spektrum DX6i |AR7000 |23 | 25| 47.6| 49.6| 35.74
Spektrum DX7 |AR7000 |14 | 16| 36 | 40 | 27.5
Spektrum DX7 |AR6100 |14 | 16| 36 | 40 | 27.5
Spektrum DX7 |AR6000 |31 | 35| 53 | 61 | 45
Spektrum DX7SE | |12 | 13| 20 | 22 | 17
Spektrum DX8 |AR7000 |11 | | | 37 | 24
Spektrum DX8 |AR8000 |12 | | | 26.7| 19.2 (11 ms frame)
Spektrum DX8 |AR8000 |12 | | | 37.2| 25.0 (22 ms frame)
WFLY WFT09S |WFT09S |30 | | |105.2| 70.2 (13)
WFLY WFT09S |V8FR-II |30 | | | 79.4| 53.4 (14)
|
|
PCM in eCCPM Mode
|
Transmitter |Receiver|MLF|MLL|MaxLF|MaxLL|Avg
---------------+--------+---+---+-----+-----+-----
ATX Stylus |92186Z |13 | 16| 27 | 30 | 21.5
Futaba 10C |R319DPS |13 | | | 43 | 28
Futaba 12Z |R5014DPS|14 | 14| 31.5| 31.5| 22.75 (1)
Futaba 12Z Md A|R5114DPS|11 | 11| 29.5| 29.5| 20.3 (1)
Futaba 12Z Md B|R5114DPS|15 | 15| 37.6| 37.6| 26.1 (1)
Futaba 14MZ |R5014DPS|14 | 14| 31.5| 31.5| 22.75 (1)
Futaba 14MZ |R5014DPS|14 | 24| 31.5| 41.5| 27.75 (2)
Futaba 14MZ |PCM1024 |29 | 93| 42 |106 | 67.5 (3)
Futaba 9C |R149DP |44 |110| 57 |123 | 83.5
Futaba 9ZWC2 |R149DP |29 | 93| 41.9|105.9| 67.5
Hitec Eclipse7 |QPCM |41 | 43| 62 | 64 | 52.5
JR 8103 |649S |27 | 59| 47 | 59 | 48
JR 9303 |649S |15 | 47| 36 | 47 | 36.3
JR 9303 |770S |15 | 47| 36 | 47 | 36.3
JR 10X |649S |34 | 66| 54 | 66 | 55
|
|
PPM and eCCPM
|
Transmitter |Receiver|MLF|MLL|MaxLF|MaxLL|Avg
--------------+--------+---+---+-----+-----+-----
ATX Stylus |FM |13 | 17| 37 | 45 | 28
Futaba 10C |FM | 2.5| | | 24.9| 13.9
Futaba 14MZ |PPM8 | 4 | 6| 39 | 41 | 22.5 (1)
Futaba 9C |FM |50 | 55| 72.3| 82.3| 64.9
Hitec Eclipse7|FM |22 | 28| 42.6| 50.5| 35.8
JR 8103 |FM |10 | 14| 32 | 40 | 24
JR 9303 |FM |10 | 14| 32 | 40 | 24
JR 10X |FM |10 | 14| 32 | 40 | 24
Multiplex Evo9|FM |10 | 14| 32 | 40 | 24 (6)
|
|
PCM, DSM2, and FASST in Normal mode (Channel 6)
|
Transmitter |Receiver|MLF|MLL|MaxLF|MaxLL|Avg
--------------+--------+---+---+-----+-----+-----
ATX Stylus |92186Z |13 | 13| 27 | 27 | 20
Futaba T6EX |R606FS |17 | 17| 61 | 61 | 39
JR 8103 |649S |38 | 38| 58 | 58 | 53
JR 10X |649S |35 | 35| 55 | 55 | 50
JR X9303 |R921 |25 | 25| 47 | 47 | 36
|
|
PPM and Normal (Channel 6)
|
Transmitter |Receiver|MLF|MLL|MaxLF|MaxLL|Avg
--------------+--------+---+---+-----+-----+-----
Futaba 9C |PPM |32 | 32| 59.3| 59.3| 45.65
Futaba 14MZ |PPM8 | 3 | 3| 40 | 40 | 21.3
JR 8103 |PPM |16 | 16| 38 | 38 | 27
JR 10X |PPM | 8 | 8| 30 | 30 | 19
|
All times are in milliseconds.
MLF - Minimum latency to first change at receivers output
MLL - Minimum latency to last change at receivers output
MaxLF - Maximum latency to first change at receivers output
MaxLL - Maximum latency to last change at receivers output
Avg - Average of MLF, MLL, MaxLF, and MaxLL
|
(1) Using channels 4, 5, and 6
(2) Using channels 1, 2, and 6
(3) Deduced numbers from observed behavior
(4) Channels 4, 5, and 6 on 14MZ and channels 2, 5, and 6 on AR9000
(5) Subject to collective input slew rate limiting
(6) In PPM 8, JR channel assignments, HELIccpm model and NOT an IPD receiver
(7) Units sold in Toledo
(8) Current tx module firmware v2.3 and rx firmware v2.4
(9) High speed mode
(10) Using channels 1, 4, and 7
(11) Using channels 1, 2, and 6
(12) Using channels 1, 2, and 3
(13) WFLY WFTRFS module with WFLY WFR09S receiver
(14) FrSKY DFT module with FrSKY V8FR-II receiver

- John

RR rules!

Well ? how did the Stylus do ?

www.liorzahavi.com

> Well ? how did the Stylus do ?

Haven't had a chance to do it yet. Maybe tonight.

- John

RR rules!

This just in...

Test results for the Airtronics Stylus with a 92186Z PCM receiver:

Minimum Latency to First Change: 13 ms
Minimum Latency to Last Change: 16 ms
Maximum Latency to First Change: 27 ms
Maximum Latency to Last Change: 30 ms

Holy cow! The Stylus, a 10 year old radio, is, in fact, faster than the 14MZ on the maximum latency end. Minimum and average latencies are very close, but the maximum for the Stylus is lower because of the faster frame rate.

Again, this is the time it takes a collective stick movement to go through the eCCPM mixing, get encoded, be transmitted and received, get decoded, and output to the servo.

Next is a Futaba 9C, maybe. I must admit I'm not sure how to measure this one since it takes 4 to 5 frames to transition from one collective setting to another. Any ideas? To be fair to the two combinations tested thus far, it would need to be to the end of the transition. That will automatically put it in the 60+ ms range. This would also then be true of the 14MZ when used with a PCM1024 rx.

I really need to get my hands on a JR system or two.

- John

P.S. To be fair to the 14MZ, it is transmitting a lot more information in about the same amount of time. This is still a breakthrough for the RC world.

RR rules!

Testing

Hello, could you explain your test set-up. What are you using to move stick, are you capturing output on an Oscope or ? Thanks , Ken

Ken,
Sure, I'll spill the beans.

My test setup consists of the transmitter, the receiver, a custom built micro-controller circuit, and a digital oscilloscope.

What the circuit and code does is watch for a rising edge on an input, delay an adjustable amount of time, and then output a 200 ms pulse on an output. The input is a channel from the receiver which is basically used for timing synchronization. The output is fed through a resistor to the center pin of the collective/throttle stick potentiometer. The stick is placed at 50%.

The resistor value must be chosen carefully for each radio. The correct value allows the radio to see a stick "input" over a usable range, say 10% during the low state output and 90% during the high state output. Adjusting the pitch curve then gives a clearly different pulse width for each stick "position".

By adjusting the delay time of the circuit, it is possible to home in on the minimum and maximum latency. On the scope is displayed the output of the delay circuit (i.e. the change in stick position) and the output of the receiver. By monitoring the time between the stick input and a change in the receivers output, one is seeing the latency of the tx and rx system.

- John

RR rules!

Very interesting, I'm messing with a small Basic Stamp microcontroller that has four programable input pins that can move a standard servo. From your experiments, is a 14 channel radio going to have a higher latency time than say a 6 channel radio because it has to send position info for more channels ? Thanks , Ken

Latency is one thing but on its own it does not mean much...

More important is the latency between the channels...

ie if you move the collective stick, whats the latency between the 2 cyclic and single collective channels...

This is what gives most of the interaction in fast pitch pumping manouvers which the pilot must correct... the lower the latency between these channels the lower level of corrections the pilot will have to make...

Did you measure all 3 channels (sorry its not clear if you did) as different radios prioritse the different channels (for a collective movement the minimum latency could be observed on the collective servo and the maximum on say the roll servo...)

All in the interests of technical accuracy and relevance of course

Don't Email me as I wont reply - PM Only (spam countermeasures)

G.Man,
Yes, I was monitoring all three channels. What you are asking about was covered in this thread: http://www.runryder.com/helicopter/t171302p1/. Bottom line thus far is that the Stylus and 14MZ (with a G3 receiver) are extremely well behaved in eCCPM applications. Futaba PCM1024 is not very good regardless of the radio driving it.

The Stylus has long been regarded as the "king of the hill" for both eCCPM "trueness" and for response time. These tests finally prove why that has been. The 14MZ, ten years later, finally trumps it by a hair.

G.Man, your question prompted me to go back and look at my data. I do need to add 3 ms to each of the Stylus numbers to account for channel 3 output starting 3 ms after the channels 1 and 6 pulses start. The chart and post above are now correct.

- John

RR rules!

Sweet

Thanks JKos...

I think PCM is always gonna be slower than a high response PPM system.

The same is true of the 3PK futaba in PCM and HRS modes, the latency is lower in PPM/HRS modes

Don't Email me as I wont reply - PM Only (spam countermeasures)

G.Man,
I plan to do some FM tests as well. So, we'll see just how much difference there is. One issue with the FM, however, is that you are guaranteed of a delay between the first channel to change and the last channel to change. In the case of Futaba which has always used channels 1, 2, and 6, that would be an inherent latency of about 15 ms just to get from channel 1 to channel 6. Using consecutive channels as can be done on the 14MZ would bring that inherent latency down to about 6 ms.

- John

RR rules!

Jkos, i would be very interested to see the results of the Stylus on an FM Rx !

www.liorzahavi.com

This just in...

Finally got around to testing a Futaba 9C with an R149DP PCM receiver. It's not good...

Minimum Latency to First Change: ~ 44ms
Minimum Latency to Last Change: ~ 110 ms

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.

Bottom line... The 9C is sloooowwww. The Futaba PCM1024 scheme is partly to blame of course.

- John

RR rules!

where you planning on doind test on JR equipment, like a 9303?

> where you planning on doing test on JR equipment, like a 9303?

Yes, I will be doing both the latency and the output behaviour tests on at least a 9303. I will be out of town for almost two weeks, so it will be a little bit before I can get to it. I also have to get a friend to let me borrow one.

- John

RR rules!

Hi John,

I think you've got a great thread going. Any chance of seeing a 9Z (WC and WCII) tested? Just curious as I'm a 9Z user.

Cheers,
Ken

Hi John,

Could you please post latency figures for the 14mz w/ a 149 (or 319) rx?

Thanks,
- Rob

Rob,
It is too late as the 14MZ has been returned to its owner. What I can tell you, however, is that based on the apparent behaviour of the PCM1024 process, the latency is going to be quite high. When it takes 4 or 5 frames for a transition to be fully made, you are talking about 52 to 64 ms right there!

When testing the 9C what I saw was that it took 3 frames before ANY change occurred at the receivers output which, quite frankly, amazed me. So perhaps if we can assume the 14MZ starts sending the change "immediately", then we can subtract about 39 ms (3 frames)from the 9C numbers and use those as the 14MZ + PCM1024 receiver numbers.

Looking at the numbers, 39 ms must be a bit too much to subtract since that would make the MLF only 5 ms which is not possible.

More realistically, I would by conjecture and some math say the 14MZ + PCM1024 receiver is somewhere around...

Minimum Latency to First Change: 14 ms
Minimum Latency to Last Change: 80 ms
Maximum Latency to First Change: 34 ms
Maximum Latency to Last Change: 100 ms

- John

RR rules!

Thanks John!

- Rob

I thought I would post that W.Pasman had found very similar results for the latency of the Futaba PCM1024 system. He noted that at times it took 5 frames(!) for the output at the receiver to change after the stick was "moved" on the transmitter.

He injected the stick movement signal the same way I did. Great minds think alike, I guess. Note that the his write up is dated November 2003 but I had not found it until tonight.

Here is his write up about it: http://graphics.tudelft.nl/~wouter/...s/pasman03i.pdf

- John