HobbyKing.com New Products Flash Sale
Reply
Thread Tools
Old May 19, 2014, 07:10 PM
Xtreme Power Systems
Lake Havasu, AZ
Joined Jun 2005
15,882 Posts
Discussion
JR Module in Taranis - latency testing

Well, I decided to test the new XJT mode for the JR modules. The firmware works great, and we will be releasing a new firmware update shortly.

For latency testing I used a logic analyzer/scope combo (RIGOL DS1102CD), the same setup used to test our Aurora 9 module.

I connected the trigger for the capture to the momentary switch. I set that switch to control the output of the channels. Basically, when the switch changes states is gives us a reference point to watch the interaction between the switch change and when the data changes at the receiver. The results were interesting...

Using our JR module in XJT emulation mode and a 6 channel Nano, I got a worst case latency of 40.4ms, and a best case of 24.6ms.

Using the Taranis' internal module and their X8R receiver, I got a worst case latency of 52.1ms, and a best case of 28.2ms.

Because the FrSky receivers do not output all of their channels at nearly the same time, their latency from channel 1 to channel 6 can be as much 10ms, so the latency for channel 1 is much better than channel 6 with the Taranis and X8R receiver. There is no difference (well, at most a few hundred microseconds) for our system.

I might be able to come up with an internal module for the Taranis that uses our RFU. It would require some custom wiring (power, ground, and a line to the Taranis' internal module port), but that could cut latency by another 5ms, due to the transmission from Taranis to the module, and then module to output latency. I don't know if it's worth it at this point. I don't have a FrSky XJT module to test to compare the difference between the internal and external module.

So, it looks like our module doesn't introduce any real latency, which was my concern.
JimDrew is offline Find More Posts by JimDrew
RCG Plus Member
Last edited by JimDrew; May 20, 2014 at 11:15 AM.
Reply With Quote
Sign up now
to remove ads between posts
Old May 20, 2014, 03:50 PM
Registered User
joshgluck's Avatar
United States, NJ, Middletown
Joined May 2012
419 Posts
Jim thank you for all the work and testing on this. I know it was not originally on your list of things to do right now, but I will feel much more confident running XPS w/ Taranis on my first 100cc plane. I am a bit curious though about how latency above compares to other systems so I have a couple of questions. If I am not understanding something correctly, please feel free to set me straight

I found the following on runryder which has comparison for latency across a bunch of different TX/RX combos:
http://rc.runryder.com/helicopter/t1...r+latency+test

From those results, I pulled out the following for Spektrum (I also own a DX9) for comparison

Transmitter |Receiver|MLF|MLL|MaxLF|MaxLL|Avg
--------------------+------------+---+---+-----+-----+-----
Spektrum DX8 |AR8000 |12 | | | 26.7| 19.2 (11 ms frame)
Spektrum DX8 |AR8000 |12 | | | 37.2| 25.0 (22 ms frame)
Spektrum DX18 |AR7200BX|12 | | 24 | | 18.3 (11 ms, 200 Hz)
Spektrum DX18 |AR9020 |12 | | 23 | | 17.5 (11 ms frame)

If I am reading this correctly, the first number after the RX model is MLF or Minimum latency to first change at receivers output, which for the above is 12ms, and MaxLF or Maximum latency to first change at receivers output which varies and Avg which is the average of min and max for each. Would it be fair to say that with your testing, the following would be added given your results?

Transmitter |Receiver|MLF|MLL|MaxLF|MaxLL|Avg
--------------------+------------+---+---+-----+-----+-----
Spektrum DX8 |AR8000 |12 | | | 26.7| 19.2 (11 ms frame)
Spektrum DX8 |AR8000 |12 | | | 37.2| 25.0 (22 ms frame)
Spektrum DX18 |AR7200BX|12 | | 24 | | 18.3 (11 ms, 200 Hz)
Spektrum DX18 |AR9020 |12 | | 23 | | 17.5 (11 ms frame)
Taranis w/ XPS DJT | XPS NanoII | 24.6ms| | 40.4ms| | 32.5ms
Taranis | X8R | 28.2ms| | 52.1ms | | 40.15ms

In comparison does this really mean that the Taranis "feels" twice as slow as a native dx18? Also, when you said the FrSky system has a better latency from ch1 than ch6 did you mean ch6 is 38.2 and 62.1 (added 10ms) where it is the same across the board for XPS across channels.

On the XPS system, do the latency numbers change using 16 channels w the x10? What would an entry above look like for a system that is using xtremelink natively?

thanks!
joshgluck is online now Find More Posts by joshgluck
RCG Plus Member
Reply With Quote
Old May 20, 2014, 04:51 PM
Xtreme Power Systems
Lake Havasu, AZ
Joined Jun 2005
15,882 Posts
Your chart and assumptions are correct - the Taranis is slow by comparison to modern transmitters. Keep in mind that the comparison chart you got from RunRyder is for only the first 6 channels. Latency goes up the more channels you add.

The X8R receivers output the channels sequentially. This means that every channel is output one after the other. If all of the channels are 2.00ms, then each channels requires 2.00ms, holding off the output the next channel until the previous channels has been output. By the time you get to channel 6, you are looking at 5*2.00ms (10ms) of extra time required to output channel 6 compared to outputting channel 1. This is a REAL problem for anyone using a rotorary type aircraft. You really want to make sure that ALL of the channels on the cyclic are operating at the EXACT same time, otherwise you get swash plate interaction. Spektrum systems group servo outputs together in groups of two, which helps, but does not eliminate the "wobbly" swash syndrome.

Also, sequentially output channels results in latency that is constantly changing every time you move the stick because the channels "stretch" and "shrink" depending on the stick position. This makes the radio seems "spungy" by comparison. When I switched our system from sequential outputs to simultaneous outputs the difference in flying everything was night and day for me.

The X10+ adds a maximum of 1.184ms to the total latency with 16 channels... ~74us per channel. Even with the X10+ being used, it's faster to use our module and receiver than the Taranis' built module and X8R receiver.
JimDrew is offline Find More Posts by JimDrew
RCG Plus Member
Last edited by JimDrew; May 20, 2014 at 05:01 PM.
Reply With Quote
Old May 21, 2014, 12:32 AM
↓↘→ + (punch)
theKM's Avatar
central PA.
Joined Sep 2004
20,189 Posts
Quote:
Originally Posted by JimDrew View Post
Well, I decided to test the new XJT mode for the JR modules. The firmware works great, and we will be releasing a new firmware update shortly.

For latency testing I used a logic analyzer/scope combo (RIGOL DS1102CD), the same setup used to test our Aurora 9 module.

I connected the trigger for the capture to the momentary switch. I set that switch to control the output of the channels. Basically, when the switch changes states is gives us a reference point to watch the interaction between the switch change and when the data changes at the receiver. The results were interesting...

Using our JR module in XJT emulation mode and a 6 channel Nano, I got a worst case latency of 40.4ms, and a best case of 24.6ms.

Using the Taranis' internal module and their X8R receiver, I got a worst case latency of 52.1ms, and a best case of 28.2ms.

Because the FrSky receivers do not output all of their channels at nearly the same time, their latency from channel 1 to channel 6 can be as much 10ms, so the latency for channel 1 is much better than channel 6 with the Taranis and X8R receiver. There is no difference (well, at most a few hundred microseconds) for our system.

I might be able to come up with an internal module for the Taranis that uses our RFU. It would require some custom wiring (power, ground, and a line to the Taranis' internal module port), but that could cut latency by another 5ms, due to the transmission from Taranis to the module, and then module to output latency. I don't know if it's worth it at this point. I don't have a FrSky XJT module to test to compare the difference between the internal and external module.

So, it looks like our module doesn't introduce any real latency, which was my concern.

always fun to see new toys after disconnecting for a few weeks, awesome.

afaik there's no difference to the internal module, but either way, as soon as I get the firmware update I will be finding out if the JR module can replace the internal module... the stock internal frsky module has already been disconnected on my set.
theKM is offline Find More Posts by theKM
Reply With Quote
Old May 21, 2014, 03:48 AM
↓↘→ + (punch)
theKM's Avatar
central PA.
Joined Sep 2004
20,189 Posts
is there a speed boost if the channel range is reduced to 8 or less?...
theKM is offline Find More Posts by theKM
Reply With Quote
Old May 21, 2014, 09:39 AM
Xtreme Power Systems
Lake Havasu, AZ
Joined Jun 2005
15,882 Posts
No. In fact, it doesn't work with anything less than 9 channels and every packet all 16 channels (used or not) are transferred to the module.
JimDrew is offline Find More Posts by JimDrew
RCG Plus Member
Reply With Quote
Old Jul 21, 2014, 10:57 AM
Xtreme Power Systems
Lake Havasu, AZ
Joined Jun 2005
15,882 Posts
Anyone using the JR module in the XJT *must* upgrade to the latest version. Besides module startup issues, it is possible that if the Taranis' data output hiccups at all, that the module could lock up - which is unrecoverable!

The new JR firmware is a complete re-write of the experimental XJT mode, and I am confident that this code will handle any situation where data could become corrupt with full immediate recovery.

See our website (under downloads->XDP software) for the updater (v4.9).
JimDrew is offline Find More Posts by JimDrew
RCG Plus Member
Reply With Quote
Old Jul 21, 2014, 12:52 PM
↓↘→ + (punch)
theKM's Avatar
central PA.
Joined Sep 2004
20,189 Posts
many thanks Jim... looking forward to playing with it.
theKM is offline Find More Posts by theKM
Reply With Quote
Old Jul 21, 2014, 03:46 PM
Registered User
west suburb of cleveland oh
Joined Nov 2004
197 Posts
to bad we cant get a complete broad the put in the tranis tx running all jims code....
I would spend the 185 plus $110 or so to have a pure xps system with new tx mine are starting to show their age.
don
dvk44039 is offline Find More Posts by dvk44039
Reply With Quote
Old Jul 21, 2014, 07:19 PM
Xtreme Power Systems
Lake Havasu, AZ
Joined Jun 2005
15,882 Posts
Be careful what you wish for!
JimDrew is offline Find More Posts by JimDrew
RCG Plus Member
Reply With Quote
Old Jul 21, 2014, 07:25 PM
Registered User
west suburb of cleveland oh
Joined Nov 2004
197 Posts
well I would prefer you got yous to market but baring that the above would be a good stop gap
don
dvk44039 is offline Find More Posts by dvk44039
Reply With Quote
Old Jul 22, 2014, 07:46 AM
Registered User
joshgluck's Avatar
United States, NJ, Middletown
Joined May 2012
419 Posts
Second that
joshgluck is online now Find More Posts by joshgluck
RCG Plus Member
Reply With Quote
Old Jul 22, 2014, 09:07 PM
Registered User
United States, FL, Port St Lucie
Joined Aug 2004
68 Posts
As intriguing as it sounds I'd rather have the best with less than second best with more , come Jim I'm wishing for it :-)
Jacksmobile is offline Find More Posts by Jacksmobile
Reply With Quote
Old Oct 01, 2014, 01:08 AM
Registered User
United States, AZ, Queen Creek
Joined Aug 2004
784 Posts
Quote:
Originally Posted by JimDrew View Post
Anyone using the JR module in the XJT *must* upgrade to the latest version. Besides module startup issues, it is possible that if the Taranis' data output hiccups at all, that the module could lock up - which is unrecoverable!

The new JR firmware is a complete re-write of the experimental XJT mode, and I am confident that this code will handle any situation where data could become corrupt with full immediate recovery.

See our website (under downloads->XDP software) for the updater (v4.9).
Are you referring to the JR module with XPS?
dirtybird is offline Find More Posts by dirtybird
Reply With Quote
Old Oct 01, 2014, 06:22 AM
If it has wings it will fly...
ckleanth's Avatar
United Kingdom, England, Birmingham
Joined Mar 2011
1,425 Posts
Quote:
Originally Posted by JimDrew View Post
Well, I decided to test the new XJT mode for the JR modules. The firmware works great, and we will be releasing a new firmware update shortly.

For latency testing I used a logic analyzer/scope combo (RIGOL DS1102CD), the same setup used to test our Aurora 9 module.

I connected the trigger for the capture to the momentary switch. I set that switch to control the output of the channels. Basically, when the switch changes states is gives us a reference point to watch the interaction between the switch change and when the data changes at the receiver. The results were interesting...

Using our JR module in XJT emulation mode and a 6 channel Nano, I got a worst case latency of 40.4ms, and a best case of 24.6ms.

Using the Taranis' internal module and their X8R receiver, I got a worst case latency of 52.1ms, and a best case of 28.2ms.

Because the FrSky receivers do not output all of their channels at nearly the same time, their latency from channel 1 to channel 6 can be as much 10ms, so the latency for channel 1 is much better than channel 6 with the Taranis and X8R receiver. There is no difference (well, at most a few hundred microseconds) for our system.

I might be able to come up with an internal module for the Taranis that uses our RFU. It would require some custom wiring (power, ground, and a line to the Taranis' internal module port), but that could cut latency by another 5ms, due to the transmission from Taranis to the module, and then module to output latency. I don't know if it's worth it at this point. I don't have a FrSky XJT module to test to compare the difference between the internal and external module.

So, it looks like our module doesn't introduce any real latency, which was my concern.
Jim, im not sure how you got the 28.2ms best case measures but latency in d8 mode, at least for the d type rx (8 ch & telemetry capable) depends on each transmited chanel pulsewidth. If the maximum pulse is used on all chanels including the synch frame is less than 22ms, the min is around 9ms for high speed mode, 18ms in normal mode.

X type rx also suport sbus, this is digital counts that detetmine the position of the servo and includes failsafe... Anything but PPM isn't subject to pulse width, thus doesn't change the latency. The channel positions are literally a number in a packet, the packet transmission time is fixed, 8 channels per packet for the Taranis, 9ms frames.



I also didn't like defaming comments made by others regarding people who buy the taranis and the developers of opentx...
I'd expect from others not to jump to conclusions because of prejudice/missinformation. Anyone can buy anything they want, we all fly rc at the end of the day, I have absolutely no issues for someone to spend a fortune buy/use whatever system they desire but I can assure you its not just foamies that get flown using opentx/taranis. ..
ckleanth is online now Find More Posts by ckleanth
Reply With Quote
Reply


Thread Tools

Similar Threads
Category Thread Thread Starter Forum Replies Last Post
Discussion JR module with Taranis? butchalline Radios 3 May 17, 2014 03:33 PM
Sold OrangeRX DSMX/DSM2 2.4Ghz Transmitter Module (JR/Taranis/Turnigy) JimG_AZ Aircraft - General - Radio Equipment (FS/W) 4 Feb 17, 2014 05:42 PM
Found Taranis JR orange DSM2 DSMX module: Hawk600 Aircraft - General - Radio Equipment (FS/W) 1 Jan 07, 2014 11:40 PM
Discussion Taranis Centering Testing comparison to JR XP9303 Fly2High Radios 12 Aug 20, 2013 01:19 PM
Question Spektrum module in Futaba 10C - analog or digital connection? Latency? AndyK Radios 2 Oct 12, 2009 01:15 PM