Modem roundup (Latency)

[reedchristiansen]

I thought I would throw some data out that we have gathered in the last few days about modems with respect to latency, specifically for sending RC control signals to an autopilot. Maybe this will be of interest to those who have tried or contemplated trying to control an airplane (pilot-in-the-loop) using commands sent through a radio modem.

Since most frequency hopping modems transmit on every other frame the hop rate has a direct impact on the maximum latency. Most of the modems will not accept data on the tx frame unless it is given very early on in the frame. This means the average latency is around twice the hop speed. Here are a few of the common modems and the hop Periods:

Worse to Best:

1. Aerocomm ac4868 (European) 56 Ms = Around 100 Ms of latency most of the time. This is the worst by a long shot. Unfortunately I have been unable to find any other low cost alternatives that are legal in Europe (And are not 2.4 Ghz).

2. Aerocomm ac4490 (918 Mhz) - 20 Ms


3.Aerocomm ac4424 (2.4 ghz)- 8 Ms.

4. Maxstream 9Xtend (918) - 9.4 Mhz - this is the total latency - Maxstreem does not show hop period in the manual. These modems seem to work the best of all the modems in the low cost category ( < $500).

Higher cost:
1. Microhard MHX 320 (350 mhz military band - Around $2000) - 20 Ms hop.

2 Microhard mhx920a (918 mhz) - 20 ms average - but it is adjustable.



I have flown several different types of airplanes using only stick-to-surface commands on all the above links. All the links expect the AC4868 provide reasonable control with some feeling more responsive then others. The AC4868 has significant latency - I have measured up to 250 ms - making it difficult to manually fly anything but a lighter then air ship.

One thing to note about the microhard modems - by default they buffer up data during RF link drop outs - the fun part is when the link is re-established they belch out all the data they stored. So your airplane gets a re-play of the control surface commands you issued while trying to "fly" the airplane during the com drop out. Not a good scenario. After much on the phone with Microhard they sent me a couple of register settings and a new binary to fix the problem.


If anyone has any alternatives for UAV control in Europe, please let me know.

Thanks,
Reed

[workshop]

This supports my research. :cool:

I know that several long range RC systems are soon to be released. Tron, AnthonyRC and perhaps Daniel Wee? are all making or about to release a long range RC system.

Has anyone created a system based on the XTend family of products? I'm ready to invest in a modem based system. :D Especially one that can manage to spit back a few lines of NMEA code from onboard GPS.

Jeff

[zik]

Really interesting stuff Reed. I'm impressed by how much the a4868 sucked. You quote the Maxstream at 9.4MHz - I presume you mean 9.4ms?

[AnthonyRC]

Interesting numbers, although I have to disagree that the ac4868 is not suitable for controlling a UAV.
All of my planes now are equipped with these, including a cularis, an EasyGlider, and even an 1/8 scale truck.
The firmware to drive them is not trivial, but certainly possible. You never want to drive them in full-duplex mode, stick them in broadcast mode, and reverse the link direction when required to send a limited amount of data from the plane to the ground. I do this during a period of no stick movement.

I'm seeing latencies of the published 53ms, which with most low-cost servos is indistinguishable from the usual 20ms frames.
(one thing to keep in mind is that the first generation of PCM radios were around 50ms, and were used to fly all kinds of interesting models).

For use in Europe, there aren't too many possibilities which have the appropriate certification, even when integrated into another product.

[reedchristiansen]

Really interesting stuff Reed. I'm impressed by how much the a4868 sucked. You quote the Maxstream at 9.4MHz - I presume you mean 9.4ms?


Yes, 9.4 ms - from their manual.

Reed

[reedchristiansen]

Anthony,

Very interesting.

So, you keep them in transmit only mode (ground to air) while the pilot is moving the sticks. During periods of no stick movement, you keep them in transmit only mode, but reverse the direction (air to ground) to get telemetry back.?

Of course, that means you need to use the CC commands reliably and have a comm protocol which is tolerant of one way communications.

Thanks for the ideas.

Reed

Interesting numbers, although I have to disagree that the ac4868 is not suitable for controlling a UAV.
All of my planes now are equipped with these, including a cularis, an EasyGlider, and even an 1/8 scale truck.
The firmware to drive them is not trivial, but certainly possible. You never want to drive them in full-duplex mode, stick them in broadcast mode, and reverse the link direction when required to send a limited amount of data from the plane to the ground. I do this during a period of no stick movement.

I'm seeing latencies of the published 53ms, which with most low-cost servos is indistinguishable from the usual 20ms frames.
(one thing to keep in mind is that the first generation of PCM radios were around 50ms, and were used to fly all kinds of interesting models).

For use in Europe, there aren't too many possibilities which have the appropriate certification, even when integrated into another product.

[AnthonyRC]

Yup, basically that is how it works. The ground-bound packets in my application are lower priority, and include the Rx RSSI data, along with a bunch of diagnostic info.
Basically, the sticks always get priority, and that lets you get a consistent 53ms latency.
The protocol sends absolute servo positions in every packet, so missing a few doesn't really matter.

[sectrix]

Nice work Reed. I had already decided on Digi (Maxstream) XTend modems for my plane, but this proves I was not mistaken. I'm considering full-duplex mode so I can piggyback an RCATS telemetry unit to it; I hope this doesn't kill my bandwidth.

BTW: Digi does sell 'International' versions of some of their modems (the 900 MHz ones), not sure if they are legal where you are, but hey.

[reedchristiansen]

The Maxstreams are a good choice. The end to end latency is about 64 Ms. on our system. This includes:

1. decoding first 5 channels of multiplexed servo output from RC trainer port
2. creating packet with servo data
3. sending packet to modem over UART at 115200
4. Over the air latency - Maxstream says 9.4 ms - probably a tad higher depending on where in the hop sequence the data arrives.
5. shifting in the packet on the autopilot UART - 115200
6. decoding the packet (we check the uart every 20 ms)
7. adding rate damping (again 50 hz main loop)
8 writing to servos

All this in about 60 ms. Not bad - I am sure if we did not gaurantee the main loop timing and did not have to do all the other autopilot related tasks we could speed this up a bit.

When the maxstream modems are replaced by a wire, the latency does not improve much - Confirming that most of the latency is in waiting for the 50 hz main loop to finish, etc., not the over the air transmition.

BTW, we run the Xtend modems in full duplex, broadcast mode with 2 broadcast attempts.


Thanks for the info on the Digi modems.

Reed



Nice work Reed. I had already decided on Digi (Maxstream) XTend modems for my plane, but this proves I was not mistaken. I'm considering full-duplex mode so I can piggyback an RCATS telemetry unit to it; I hope this doesn't kill my bandwidth.

BTW: Digi does sell 'International' versions of some of their modems (the 900 MHz ones), not sure if they are legal where you are, but hey.

[sectrix]

Thanks for the info on the Digi modems.


No, no, thank you!

I'm happy to hear the modems could handle all you put it through. This was one of those systems that I was worried would not show issues until I bought it and tried to use it.

Nice!

[Jack Crossfire]

Have been very happy with the XBee Pro, the only reasonably affordable system for a private individual after taxes. Can't use 2.4Ghz video with it, it probably has over 50ms latency, but it's tiny enough to stick anywhere.