rsilvers

I measured the data rate and latency of a Spektrum DX3 and Pro receiver system using a Tektronix TDS-220 100 Mhz bandwidth digital oscilloscope.



The data rate being sent to the analog servo from the Spektrum receiver is 20ms per frame, or 50 updates per second.

For comparison, JR PCM radios are approx 46 updates per second for Z mode, and 47 for S mode per channel. Standard FM PPM systems are about the same.


Latency of the Spektrum DX3 when using the channel-3 switch is 7-8ms.

Here you can see that the latency is over 6ms:



Here you can see that the latency is less than or equal to 8ms:




The Nomadio salesman said that the Spektrum system had about 15ms of latency inherent in the transmission, and another 30ms of latency provided by the transmitter encoding the controls. That would be 45ms. I was not able to duplicate those numbers.

Spektrum's claim of '3ms' is quoted as 'the figure added to your radio's processing time' so I have not confirmed or denied that specific claim because I included the radio processing time in my calculations. If the radio processing time was 4 or 5 ms, then their figures seem accurate.

Nomadio also claims 100 updates per second for their system, and does not specify a latency figure. I would like to verify if Nomadio means this for ALL channels combined or for EACH channel. Then I would like to specifically test to see if they update any given servo at 100 hz. I currently have no Nomadio system to test.

So while I have not tested the Nomadio as of yet, the Spektrum has very little latency (just 1/125 of a second).

rcxboy

Its good that people with the gear can do this stuff that the majority of people can't, so thanks a lot.

HaveBlue

thank you for taking the time and effort to do the test and then post your findings , seeing the results is second to none while others like to debate these sort of things without producing results first hand, does away with misguided information that some might act upon... I was happy to see this as I myself have recently bought a 3PK and updated to a spectrum pro module and Rx...again thanks for posting this information

SAVAGEJIM

If you ever come across any Nomandios and do the same tests, then please by all means post those results too! I'm hurting to see what those figures are like.

As far as one sales rep bashing another product, sales reps are simply liars and hacks. If you give a sales rep any hint that you are considering another brand, they will trash it (even if they prefer the other brand themselves)! They do this because they have $ $ in their eyeballs. Give a sales rep any hint of interest of buying their product and all of a sudden you hear ching cing! out of his ears and $ $ immediately replace his eye balls.

Nomadio_Sales

Would be happy to send you a Nomadio. Do you have a "non pro" version to test? You also need to need to take LOTS of measurements, not just one, and look at min, max, average.

rsilvers

Thank you for the offer. That would be great.

I did test the non-pro version. I got the exact same results. Mine is very new (just got it last week) and I have a suspicion that it is identical inside. I opened the cases and they had the same circuit boards. The chips had slightly different numbers engraved on them, so they may be revised. When I installed them in my cars, I put the 'pro' receiver in my best car. Having tested them, I no longer consider the non-pro one any different.

Latency did not change as I took several measurements. Since the oscilloscope's external trigger was electrically tied into the same switch that activated channel-3 to transmit, I did not expect any variation.

rsilvers

For comparison, I tested the standard radio that came with my E-Maxx. Since there is no digital processing, it has slightly less latency.





I measured 6ms, or about 2ms less than the Spektrum (1/500 of a second less).



According to my measurements, the '3ms' that a 'Pro' Spektrum 'adds to your radio's processing time' (their marketing claim) is likely a correct / conservative number. I say 'likely' because I did not use the same radio but it seems to be adding up.

Nomadio_Sales

Remember on the non pro their average has to be somewhere around 10ms probably a bit higher they send all their servos, in analog mode, to the inputs of the digital radio encoder each servo pulse is 1-2ms each (so that's per channel) that happens every 20ms

P2gee

I think a better test would be a dual channel display off of one of the analog inputs ( steering or throttle). With A and B waveform stacked one above the other.

Nomadio_Sales

Savage Jim please read the flame rules, I need your aopolgy asap.

Nomadio_Sales

When I get back from the 1:5 scale nationals I will send you out a unit.

rsilvers

I measured the point at which the receiver had received and decoded the signal and was ready to output it as I wanted to see if their system 'added' to the delay already inherent in RC systems.

But I see what you are saying and how you arrived at the ~15ms value for the Spektrum. I think when you tell people that value, it is best to note that it is not in conflict with Spektrum's claim of 'adding 3ms to your radio's own processing time.' However, I do feel it is a meaningful number that is worth comparing to the Sensor's number when measured with the same method, and that is what I shall do when I get your system.

By the way, if measured that way, a JR PCM would have about 15.6ms average latency and the Spektrum would have about 17ms. If I see that the Nomadio is sending 100HZ info to each servo, then that would effect the average latency in a significant way though I still need to measure your processing time.

Question -- does the Nomadio send servo pulses at the same rate, whether to an analog or to a digital servo?

Nomadio_Sales

Also look at input to output total time because it is total time to movment from movment is how we drive.

Digital servos get 100 pulses per second you can set it analog 50Hz but you really only need to use Digital 100Hz output all the time because most Analog servos can handle the faster rate.

rsilvers

Because the timer started when I flipped the switch on the transmitter, it is taking the input time into consideration. At least for the channel I measured.

mkincy

great work, thank you.

SAVAGEJIM

The rules specifically say
"NO bashing of other members "
Not once did I specifically target you, just a group of people I have had bad experiences with. I always take a sales pitch with a grain of salt.

I do apologize to you if I offended you, but know I never targeted you nor intended the collateral damage to affect you. I however still have a bad opinion of sales reps in general; but don't take that to mean I have a beef with you. I do not. Your posts do seem honest, and so long as your posts do not fall into a sales pitch of bashing and exagerating the flaws of other products, I will always hold you opinion with respect.

Anyway, I don't need any convincing that Nomandio makes a great radio (as a matter of fact, I really want a Sensor). After skimming over hundreds of treads and posts concerning Sensor, I already have it in my mind it is a great system even with the flaws it currently has.

rsilvers

http://www.rcdriver.com/articles/spektrum_1.shtml

Here Spektrum is claiming that DSM has less than 1/2 the latency of an M11 radio. That seems unlikely.

SAVAGEJIM

Yeah, 5.6ms seems really fast. I wonder if Spektrum reads any of these posts? Maybe they can shed som light on these advertised numbers.

rsilvers

The 5.6ms is the duration that the original Spektrum added to 'your own radio's processing time.' That seems likely. In fact they got it down to 2-3 ms as far as I can tell in the latest version. The article was saying that for the purpose of comparing to their early experiments where their FHSS system added 24ms.

My problem with the article is that they then go on to say that the M8 has a rating of '12ms.' How does an M8 add to its own processing time? Remember, that is what the 5.6 was for, so there is no point in quoting another number unless they are apples/apples comparison. If they mean 12ms total latency of the M8 -- ok, that could be correct, but then they they need to quote the total latency of the Spektrum -- which is more than 12ms if you use the same method of measurement. So really they used two standards in the same paragraph which is like saying 'Car X can do the 1/4 mile in 12.3 seconds and car Y was 15 seconds' and fail to mention that car Y actually did the 1/2 mile.

The article is misleading in that any reader who relies on them and does not do their own experiments would be under the impression that a DSM equipped M8 would have less than 1/2 the latency of an FM M8. But that is not true. The Spektrum will always be slower than a standard M8 (by 2-3ms which is just 1/125 of a second).

SAVAGEJIM

Thanks for shedding light on these numbers. I am getting the idea that advertisers will do anything to twist and lie to make the product they are working for look better. This is the kind of S... I dislike in life.

Even though I am for DSM and other 2.4 GHz technologies in RC, I hope Airtronics does come back with a rebuttle and puts Spektrum's advertising company in check. Better yet, Spektrum could even fine or fire that ad company for this lie. I feel strongly against false advertisement.

rsilvers

It is not an ad company but a Spektrum employee who seems to explain but not clarify the specifics to the magazine reviewers who happily write what they are told without doing their own performance tests. No doubt it is frustrating to their competitors such as Nomadio to have to read that stuff in reviews.

When I get the Nomadio system I will just post what I think the final word is in latency of each system. I will also test range but will do it as a sort of worst-case rather than best case. What I mean by that is I will hold the radio at a 45 degree angle rather than straight up and down which is the kind of thing that happens in real life but not on a manufacturers range rating test.

SAVAGEJIM

I do notice range has a drastic effect on DSM systems. When I run my RC down the street, I notice my commands take considerable more time for the car to react. But then, the ranges I amd talking about are so far that I can hardly see the vehicle.

rsilvers

I have the Nomadio Sensor that I was loaned and it does output 100HZ to the sero, so it will almost certainly have less latency than the other systems. I will give a report at the end of next week.

Nomadio_Sales

SO sorry for being off RCU for the last weeks I was at a race came home got sick for 3 weeks and then vacation. If you have a Sensor already cool or I will ship you one.

BTW http://support.nomadio.net/viewtopic.php?p=6027#6027

Nomadio_Sales

By the way as you are testing a DX radio you are not getting total numbers for the vast majority of spek systems which are plug in units working RF for other radios. The spek plug in systems add latency on top of the systems latency as pointed out in the posts above.