Independent tests prove lack of frequency hopping with XPS

[buttface]

I had a coil go bad in the ignition of my DA 150 and caused my XPS to lock out , falling for 8 seconds.I landed it with no damage.I was one of the lucky ones.I sold my XPS and will never buy another product from Jim Drue again.

[pilot232]

I believe that would have affeted Spektrum, Futaba, Airtronics, and XPS. Just my belief.

Travis

[4*60]

...

[pilot232]

4*60,

I feel the same way. If the noise level is that high, I don't believe there will be a clear channel for any system!! I am no giant here either, but 40+ years in computers leads me to this conclusion.

Warm Regards,

Travis

[XJet]

	Quote: Originally Posted by pilot232
	4*60, I feel the same way. If the noise level is that high, I don't believe there will be a clear channel for any system!! I am no giant here either, but 40+ years in computers leads me to this conclusion.

Sorry but this is just not true.

The 2.4GHz band is nearly 80 times wider than the 72MHz band and it's not at all uncommon for segments of the 2.4GHz band to have very high levels of noise -- while other segments are completely clean (ie: virtually no noise).

That's why frequency agility/redundancy is so important. If the channel your RC set is using is suddenly hit by a strong signal from elsewhere (not all that uncommon in some environments), it may only require a shift of 10MHz or so to move into a totally clear piece of the spectrum. If your gear is unable to hop under such conditions (XPS) or has no alternate channel already established then you're stuffed.

This could explain at least some of the growing number of "I flew XPS for months then suddenly..." reports that have appeared.

Both Spektrum/JR and Futaba have deliberately designed and implemented systems that can cope with such events -- XPS can't.

All systems work well while most of the time -- but Spektrum/JR and Futaba are far better able to cope with this kind of interference. If you're a gambling man then you may be happy to trust a $5K giant scale model to XPS.... ultimately it's a personal choice.

[Leseirf]

Xjet,
I posted this on RCG looking for an opinion from you but I think you were typing a post at the same time and missed it.

It occurs to me that there could be a simple software solution. The way I understand it (and I've followed pretty much every thread for the beginning) when the TX doesn't get acks, it retransmits. No acks for x amount of time it alternates looking on other frequency's with retransmitting on the original freq. Why couldn't the rx after x time with no signal just hop to clean air. The TX would then find it on its own. With the transmit-ack sequence being (IIRC) in the 2ms range and only 12 frequencys it shouldn't take long to re-establish connection.

That said I'm just a lowly RC hobbiest. No software or RF engineering background at all. This just seems logical given my understanding of how XPS works.

Mike

[XJet]

	Quote: Originally Posted by Leseirf
	It occurs to me that there could be a simple software solution. The way I understand it (and I've followed pretty much every thread for the beginning) when the TX doesn't get acks, it retransmits. No acks for x amount of time it alternates looking on other frequency's with retransmitting on the original freq. Why couldn't the rx after x time with no signal just hop to clean air. The TX would then find it on its own. With the transmit-ack sequence being (IIRC) in the 2ms range and only 12 frequencies it shouldn't take long to re-establish connection.

The problem with a "hop on demand" system is two-fold:

First you have to determine *when* to hop, then you have to determine *where* to hop.

If you suddenly find that the link with your receiver is lost -- do you wait and hope the interfering signal goes away (as it often does on 2.4GHz) or do you instigate a hop?

Just how long do you wait before deciding that it's better to hop?

If you hop too soon then you may be start introducing unnecessary latency and the risk of lost synchronization between transmitter and receiver.

If you wait too long you run the risk of the resulting "lockout" causing the model to crash.

A second's wait-and-see may be a reasonable for a sports-type model but totally unacceptable for someone flying a jet or doing low-level 3D flying -- yet a quarter-second may introduce a greater overall latency by allowing relatively low levels of noise to repeatedly trigger the hopping sequence.

Then there's working out where to hop. You can't assume that *any* other channel will be noise-free enough to allow continued control and you can't simply pre-program in a default -- or a single burst of noise could send all the XPS systems operating on a number of channels straight to the default channel where they'd effectively knock each other out.

The best way might be to maintain a constant spectrum analysis at each end of the link and negotiate an alternative channel *before* any interference even appears. Both ends of the link *must* be analyzed because the spectral profiles of each can differ significantly. A row of trees or buildings can completely shield the transmitter from detecting a noise source that effectively knocks out one or more channels at flying altitude and the close proximity of another weaker 2.4GHz signal at the transmitter end could also render different channels marginal at that end.

Unfortunately the fastest that the XBeePro modules can perform a spectral analysis is about 16mS and that's a very limited scan that spends just a few mS on each channel so may miss some very strong but bursty (typical of 2.4GHz) noise. All you need are a few retries thrown in there and you can rapidly run out of time -- since each frame of data is just 20mS apart. So the very act of scanning could introduce random levels of latency to the link.

If I were doing this, I'd regularly (but not constantly) scan between frames and build up a rapidly moving weighted average of the noise levels on the unused channels (at both Tx and Rx) then use that weighted average to negotiate a "jump-to" channel and pass that info back and forth once per frame. This would allow the Tx and Rx to agree on where to go if the current channel got clobbered.

Unfortunately this isn't bullet-proof either -- since there's no guarantee that the agreed alternative channel would not also be hit by the same (or other noise) at the same time.

Fortunately, the risk of this happening could be minimzed by using some simple algorithms (ie: give channels further from the current channel a higher weight when calculating the best alternative).

However, no matter which way you look at it -- any solution is going to be a kludge at best. The XbeePro simply isn't designed to be frequency agile.

This botch-job "hop on demand" is still not as good as FHSS or automatic redundancy though IMHO.

I can see why JD went with the XBeePro modules but it has locked him into some less than optimal technology (for RC purposes) on the 2.4GHz band.

The XBeePro modules are *easy* to use but that doesn't make them the *best* solution to the 2.4GHz RC problem. Various people (including myself) have knocked up XBeePro-based 2.4GHz RC links in just a few hours and they work well -- but just don't have the reserve of safety that other solutions offer in a busy 2.4GHz spectrum.

If the XBeePro solution was so good you could bet that lots more of different brands would already be using them due to the low barrier to entry and the incredible simplicity of building transmitters and receivers around them. The fact that nobody else has, speaks volumes me thinks.

[ChrisHays]

	Quote: Originally Posted by ChrisHays
	hows that new china system you are testing xjet?

?

[XJet]

	Quote: Originally Posted by ChrisHays
	?

Well one of the systems I'm testing seems to work okay but it's not suitable for giant scale -- no failsafes -- however it is dirt cheap (under $50 per 8-channel receiver *with* satellite).

There's video here:

iMax 9X

Full review will be posted shortly.

[exeter_acres]

	Quote: Originally Posted by XJet
	Full review will be posted shortly.

teeeheeee!

[ChrisHays]

xjet.. please include a pic of the tx!..thanks

[zoomer260]

	Quote: Originally Posted by XJet
	Well one of the systems I'm testing seems to work okay but it's not suitable for giant scale -- no failsafes -- however it is dirt cheap (under $50 per 8-channel receiver *with* satellite). There's video here: iMax 9X Full review will be posted shortly.

Xjet, I'm NOT a basher on this thread I am following with accute interest. But I am wondering why you said a particular system needs failsafe for one airborne craft but not another ? They both can be lethal. Is it because meet rules require it on one but not the other or is it a personal requirement ?