JR/Spektrum DSM2 Weakness Revealed
WHEN YOUR REDUNDANCY ALL BUT DISAPPEARS
Dated: 23 Feb 2010
While tidying up some of the final reviews for the great 2.4GHz Spread Spectrum RC system shootout, I discovered something very disappointing about the Spektrum/JR 2.4GHz DSM2 system.
As stated in the review, the DSM2 system is generally quite good. It uses two separate parts of the band and has a good degree of "spread" for each of its chosen frequencies, thus ensuring that under normal conditions, DSM2 provides a robust and resilient link between transmitter and model.
However, while doing a little more testing, so as to get a couple of extra screen-shots I discovered something surprising.
The 2.4GHz environment inside my metal-clad workshop is very clean. Being located some distance from the nearest population center and with the metal walls/roof providing excellent screening against external signals, the noise level is extremely low.
In such an environment, any 2.4GHz system turned on will see a band that is entirely free from potential interference and should therefore seek to take advantage of this.
In the case of the DSM2 system, that means I would expect it to allocate itself two channels that were separated by a good wide gap -- such that if one were hit by interference, the other would likely be well clear of the offending signal.
Well the spectrum analyzer scan above shows what I got when I turned on a JR DSX9 in this "clean" environment.
As you can see, on this occasion it has randomly chosen two channels but they are so close that they actually overlap.
Now, in a perfect world and in 99.9% of real-world situations, this won't be a problem and I should emphasize that the sky is NOT falling
for JR/Spektrum users.
The reality is that because of the resilient nature of a spread-spectrum transmission it would still take a reasonably strong signal to knock out the link created by the DSM2 system, even with these closely-spaced channels.
In order to see what would happen when just such a "reasonably strong signal" was about, a video-transmitter was turned on and tuned so that it clobbered both of the narrowly spaced channels. A spectrum analysis of the video-transmitter on its own is shown above.
Using a 600mW 2.4GHz video transmitter, I found that the receiver only needed to be equidistant from both the DSX9 and the video transmitter's antenna for loss of control to be experienced. This surprised me a little -- I'd expected more resilience from the DSM2 system but, having said that, analog signals such as video transmitters are really not spread-spectrum-friendly.
The above image is a scan of the video sender and the original closely-spaced DSM2 signals interfering with each other. As you can see, the video transmitter's signal has almost completely obliterated both channels of the DSM2 system because they were too closely spaced.
When the two channels were spaced further apart (as is usually the case), so that one was not covered by the video transmitter's signal, the range of the DX9 radio was unaffected. However, on those rare occasions when they are close together (as in the image above), the effect of a modestly powerful video transmitter is significant.
This is not so good. In fact I think it's fair to say that it's not the kind of poor performance you'd expect from a "brand name" 2.4GHz radio system.
In effect, this means that if you were unlucky enough to have your Spectrum/JR radio select two closely spaced channels and someone was using a video transmitter on the same part of the band (effectively clobbering both DSM channels) then, should your model flew closer to that transmitter than it was to your own, you'd risk a lockout.
When the Spectrum/JR system selects two channels this close together, your expensive radio performs no better than a super-cheap FlySky radio, offering virtually no redundancy in the event of strong on-channel interference.
And, should this situation occur in the real-world, no number of satellite receivers, voltage regulators or other "bits" will help. Once the interfering signal overpowers the one from your transmitter, your model is at risk.
In testing, it appears that when turned on in a low-noise environment, the DSM system selects its operating channels at random and very occasionally (because those assignments are random), they happen to be almost on top of each other.
My question to JR is: Why?
Any sensible designer would have included a check to make sure that when the band allows for it, the two channels should always be spaced by a sensible amount. This clearly is not happening.
So, if you're a JR/Spektrum flier, should you sell all your RC gear and switch to something else?
Probably not. As I have said, most of the time the DSM system does select channels that are sufficiently spaced to provide a good level of redundancy and resilience. What's more, if you're flying in a relatively benign RF environment then the close proximity of the two channels that can occur on occasion won't have any effect. On the rare occasions when the JR/Spektrum radios do choose two adjacent channels, the effect is simply that your risk of lockout is increased but only if there just happens to be strong noise on that small part of the band.
However, if you have already encountered previously "unexplained" lockouts and failures at your field when flying DSM-based systems then perhaps this could be part of the cause.
As a JR owner (I have a 9XII/9303) I would not use Spektrum as my upgrade path to 2.4GHz unless they fix this flaw. While the odds of this actually causing a crash are low, they still exist and, given that it could be fixed with a minor software revision (that mandates a sensible minimum for the spacing of the two channels used), I see no reason not to suggest that such a fix be implemented.
With the onslaught of cheap (and increasingly good) Chinese-made equipment, much of which is now fully certified for use in the USA and EU, "brand name" manufacturers are going to have to step up their game if they want to retain marketshare. Even other companies with a strong global presence (such as Hitec) are now nipping at the heels of a technology that, in the case of DSM2, could do with some tidying up.
If you're a satisfied (or dissatisfied) JR/Spektrum user who'd like to have your say on this article, please email me and I'll append the comments below.
If JR/Horizon have an answer to the puzzling question as to why the system should significantly nullify the benefits of a dual-channel operation by allocating channels so closely spaced, I'd like to hear (and share with readers) that explanation.