Radio latency

[mnentwig]

Hello,

this is an interesting topic that was discussed at the field, but I haven't seen on Kopterit.net:
Radio latency, the time your radio needs to process, transmit, receive the stick movements.
Here is the runryder thread:
http://www.runryder.com/helicopter/t172571p1/

This picture is most interesting:
http://www.runryder.com/helicopter/gallery/8588/latency_17_avg.gif
* Futaba T6EX FASST: in average 34.2 ms
* Futaba 7C FASST in average 26.4 ms
* Spektrum DXx: anywhere between 27.5 and 72.8 ms

What does it mean? For example:
At 100 km/h the model travels ~ 1.40 m during 50 ms of latency, or 14 cm at 10 km/h.
When I have to react to something that cannot be anticipated (like, a gust of wind, a suicidal bird, the heli lost-and-found before the treeline etc), shorter latency means that the model responds earlier.
Fourteen centimeters isn't much... except if your airspace is 14 cm short ... I have had frequent almost-crashes that came closer, or where the blades actually touched the ground.

I believe it does not matter for a beginner, not at all. But you bet I'll keep this table in mind, when buying my next radio.

[kaardian]

I was witnessing a crash on friday afternoon. Pilot was doing a inverted flight, loop, flip, or so, but important was that heli was bit far away already 150 - 200m and he lost control, hit the throttle hold at 1 sec before ground, but still heli didn't react as fast as needed. Could be a range problem? Perhaps. Could be latency related problem?
Anyways, pilot said that he wondered for a while why nothing happend when he turned the sticks. There was about 5 seconds time to do something. I had enough time to look at my heli and put radio on, then I saw the last 2 meters into ground.
We discussed bit about what happend, and came into a conclusion that it had to be a radio malfunction somehow. He had a aluminium plate under the heli and if that can disturb the signal in a some specific angle, that could explain it...

[masu]

Here's a video related to the topic... I cannot comment to the subject because I really am not experienced enough to notice any lag issues.
http://vimeo.com/moogaloop.swf?clip_id=897956&server=vimeo.com&fullscreen=1&show_title=1&show_byline=1&show_portrait=0&color=01AAEA

[mnentwig]

I was witnessing a crash...

that sounds like a plain old radio problem, for example if one antenna of a two-antenna receiver is broken. Or, the battery is dead.
But latency is different, it means that the signal gets delayed by a fixed amount, under all circumstances.

I wouldn't trust any people who claim they can judge the difference "by feel". There are just too many factors involved. To get reliable data, one would need to perform thousands of tests.
But the video, or a simple measurement, tells it all.

BTW, when playing piano on a software synth via MIDI (which involves latency), you notice it at about 10 ms. With 50 ms you want to tear your hair out 

[masu]

Looking at the results of the graph it seems that the receiver plays a big part in this. For example if you compare DX7 with AR7000 and AR6000.

[Vispilä]

If a heli crashes due to a radio error the last thing to consern as the reason is the latency I think. It's a range problem if the heli doensn't react. On the other hand, if you hit thr hold 1 sec before ground, it's maybe too late because it takes time for the signal to travel to the model, most of us are using slow thr servos, and a combustion engine does not react immediately when the throttle is changed.

I wouldn't trust any people who claim they can judge the difference "by feel". There are just too many factors involved. To get reliable data, one would need to perform thousands of tests.

2.4GHz radio connection feels actually faster than a 35meg connection on the same setup and it's because of the resolution and the latency isn't it. I know because I changed to a Spektrum module without tweaking the setup in any other way. Though comparing different 2.4 setups by feel is not possible or at least hard because the differences are so small.


-op-

[Picofly]

Hello,

this is an interesting topic that was discussed at the field, but I haven't seen on Kopterit.net:
Radio latency, the time your radio needs to process, transmit, receive the stick movements.
Here is the runryder thread:
http://www.runryder.com/helicopter/t172571p1/

This picture is most interesting:
http://www.runryder.com/helicopter/gallery/8588/latency_17_avg.gif
* Futaba T6EX FASST: in average 34.2 ms
* Futaba 7C FASST in average 26.4 ms
* Spektrum DXx: anywhere between 27.5 and 72.8 ms

What does it mean? For example:
At 100 km/h the model travels ~ 1.40 m during 50 ms of latency, or 14 cm at 10 km/h.
When I have to react to something that cannot be anticipated (like, a gust of wind, a suicidal bird, the heli lost-and-found before the treeline etc), shorter latency means that the model responds earlier.
Fourteen centimeters isn't much... except if your airspace is 14 cm short ... I have had frequent almost-crashes that came closer, or where the blades actually touched the ground.

I believe it does not matter for a beginner, not at all. But you bet I'll keep this table in mind, when buying my next radio.

I am not a professional in this issue but I suppose that human reaction time plays a much more important role than radio latency. Your reaction time to a visual signal may be something between 0.1 - 0.25 sec depending how fast you are in gnereal, how tired or focused you are, weather conditions etc. A 0.025 to 0.075 sec. radio latency does not make a significant difference on top of that.

Let's say you you have exceptionally short reaction time (0.1 sec.). Your 'Total latency' will be 0.125 sec. with a fast radio and 0.175 sec with a slower radio. If you are an average guy whose reaction time is ~0.20 sec., those figures will be 0.225 sec. versus 0.275 sec. Those 0.1-0.25 sec reaction times have been measured in laboratory conditions. In complicated situations like road traffic it is closer to 1-2 sec, depending on speed of the vechicle and other circumstances.

[mnentwig]

Well, everybody draw their own conclusions. As far as I'm concerned, I won't buy me a new radio only to reduce latency (and my F7C is pretty good). But -if- I buy a new radio, this will be a key argument for the purchase decision.

Definitely, the operator's reaction time is longer than the radio delay. But that argument can be misleading:
Imagine a radio that is only a little bit faster than the average reaction time, say 150 ms. You can argue it's less, it does not matter. But that radio would be absolutely un-flyable. 
If the radio were an order of magnitude faster than the operator (~ 15 ms) I'd say let's forget about it and "sweep it under the carpet". But to my surprise, it isn't.


>> 2.4GHz radio connection feels actually faster than a 35meg connection
I have no doubt that you can tell the difference between two radios in "feel", but I would be surprised if it can be nailed down to a single issue like latency.

[Vispilä]

The following is not my opinion but just though to share this quote from a thread "10 things" by Dood on RR:



"I'm sick of hearing about "LATENCY". It's probably the most over discussed, overrated topic there is as of late.

"JKos" has a very cool THREAD on this forum comparing just about all of the modern radio sets. Now I appreciate the time & effort Jon has put into compiling the results, and I have checked it out.
Personally, I don't feel better off that I have reviewed those results. Why? Because we are dealing with measurement of time.
Time in such a small scale that no human can relate to, or detect with any of his natural senses. I start to wonder, "whats the point?"

The thread is cool, don't get me wrong.
What bothers me is that when someone is looking for a little advice about what radio set to buy, there is always at least one or 2 guys that spouts off, "Buy Brand X over Brand Y. It has a faster latency!"

Funny thing is, most people in this hobby didn't know what latency was a little more than a year ago, now it's the first thing
someone brings up when someone else asks about what radio to buy.

Everyone is arguing about which is the better radio set. If one has a latency of 25ms and the other a latency of 20ms. That's a difference of 5ms. Does anyone honestly believe you'll feel feel a difference of 5ms?

Did you know that it takes a person about 250ms to blink an eye? With your natural senses, a human being has no way of measuring 5ms. Just remember that it takes a human being 50 times LONGER to blink one of their eyelids.

leaving the Futaba 9C out of the equation since it's an exceptional case, regardless of radio make & model, It's very unlikely you'll ever feel a difference.

To really put this in perspective, please try to blink your eyes at a rate of 200bps (blinks per second). That's if you were able to blink
once every 5ms.

So enough about latency already. I think it's moot."


-op-
 

[mnentwig]

OK, I hadn't followed to the discussion to the point where it's getting overrated. Or, if everything has been said on the topic on kopterit.net, so be it. But it was new to me, and search for "latency" turned up nothing.

I do disagree with the "reasoning" (or lack thereof) in the quoted text, but this is not the place to argue about it.

With software synths, 10 ms are the most I'd tolerate, and many people spend lots of money for audio interfaces to go below that.
Question is, does playing an instrument have anything in common with flying helis? In my opinion yes, because in both cases you're part in a feedback loop and try to perform as accurately as humanly possible. Now if somebody doesn't buy into that argument, also fine with me...

The reasoning "you can't perceive it" doesn't convince me, because it adds up.
First, you can resolve time differences of 2 or 3 ms easily with your senses. Then, the latency difference observed between products (30 ms) can clearly make the difference between blades scratching the ground a bit, or a solid crash.

[Vispilä]

I agree with you in that it's possible to notice small reaction differences if you've flown the same heli (or played the same instrument) for a long time and are good at it. That is one reason why pros notice problems in flight faster than beginners, they "feel" smaller differences and know immediately if something ain't right.

To answer your question on helis vs. music instruments, yes, in my opinion they are very similar to each other because I play one my self too (piano). Though, I ain't as good at playing piano as I'm at flying so I really can't tell the difference between the reaction times of individual instruments.

I think this is quite interesting  . Let's continue

-op-

[Jussi.]

Permission to continue granted from Vispilä  so I add my 0,05€ also...

I, as opposed to Vispliä, can handle Instruments better than Helis (not that I'm that good, but my flying skills suck still) and I do agree, that latency is very much an issue especially when _recording_ . You get that plenty with bad home-pc-based "studio" with crappy audiocard/drivers that add latency and for reason or another you cannot monitor sound directly. It's very hard to play correctly.

In a way, you still get used to constant latency if you really really want to, but it will never be the same as playing "dry".

I would quess that people naturally will adapt to latency and we all learn flying with it, but if someone learns playing  keyboard having 35ms latency (he probably would adapt hitting keys 35ms earlier in no time), the WOW factor would be enormous when trying hist first ~0ms latency.

Instant is instant

[mnentwig]

Now I can only guess, but my feeling is that it would matter (if at all) for beginners, and on small helis. Or then, maybe on top competition level - No experience with that on my side, but I have seen scores of "1000" and "997 point something" on first and 2nd place.

The example that comes to my mind is the Walkera 4#3, where there is no direct mixing between swash and blades. All cyclic control goes through the flybar, the flybar needs time to react, and as a result the heli is very, very difficult to fly (probably also for other reasons). It's mechanical delay, not radio delay, but the outcome is the same.

[Halfelf]

Once upon a time I used to play 3d shooting game. If you put up two equally skilled players and other has.. lets say 20ms slower connection it makes a difference sometimes. Thats what I tought and still do. Does it have anything to do with flying helis? Dont know

[Vispilä]

It's true that you kinda have to learn everything again when you change the setup, in this case get a faster radio. Pros get it in no time, but for a beginner changing a parameter (e.g. expo value) can be a dramatic change leading to a crash. But that's really something we have to deal with. When one decides to buy a new radio, it surely can't work exactly the same way that the previous radio did.. If it has got a better latency, then you'll have to pay the price and fly a couple of test flights before doin anything that you used to do on the previous setup. I believe that latency difference can be noticed, but thinking of it when choosin a radio is completely unnecessary for 99.9% of all RCers. Related to the topic, someone mentioned a long time ago that Szabo got his DX7 radio tweaked as an experiment, and the result was that raising the transmitting resolution was clearly noticeable, so yes, it should be possible to tell a difference between latencies too. But referring to the grand master Dood, let's just go out and fly and stop wanting a faster radio because it would help us fly better  .

btw, I'm using a 9c which is the one at the end of the latency comparison graph, but still, for some reason all my helis stay in the air  .


Big cheers and G'night.


-op-

[Jyri]

Vaihdan kielen kotimaiseksi kun sitä kaikki kuitenkin osaavat täällä lukea ja ymmärtää

Asia joka on minulla ainakin mennyt ohi silmien näissä latenssikeskusteluissa on servojen nopeus, jos tuo latenssi on niin suuri ongelma, niin miksi ei sitten kiinnitetä enempää huomiota servojen nopeuteen. Esim. hyvin yleinen ja arvostettu 450 luokan nuppi servo, Hitec HS65 jonka arvot ovat:
Speed (4.8V): 0.14sec/60, (6.0V): 0.11sec/60
Ja sitten esim JRn uudet digimikrot (JR DS287MG):
Speed:.08 sec/60 @ 4.8V; 0.07 sec/60 @ 6V

Eli nopeuseroa on 4,8V 340ms ja 6V 410ms / 60 deg eli aika paljon tulee eroa jopa pienillä korjausliikkeillä, joten jos tuo latenssi on niin iso asia, niin eikö ensiksi kannattaisi laittaa nuo servot kuntoon?

Edit. Niinhän siinä sitten kävi että tuli pilkkuvirhe Eli erot ovatkin:
nopeuseroa on 4,8V 34 ms ja 6V 41 ms / 60 deg

[mnentwig]

Of course, it all adds up. But keep in mind, 60 degrees is a lot of travel. Especially cyclic movements are pretty small, because two servos act in tandem.
Take for example 1/4 of 60 degrees, and you're in the same ballpark with radio latency (one quarter of 0.14 s is 35 ms).

If you want to see how fast they accelerate, the slow-motion video earlier in this thread is interesting.

On one side, always "needing" the best equipment can be ridiculous. On the other hand, it's only fair and reasonable to ask value for the money.

The stock Align 35X ESC for the T-Rex 450 gives 6 V. But I think there is something funny in the above numbers. 0.14 s is 140 ms, 0.11 s is 110 ms.

And, I don't know who Dood is, maybe he flies better than all of us together but from his writing I doubt that he has any technical background. It seems more like a rant (which is the right of a "Guru") than an argument (as appropriate for ordinary human beings).

[Jyri]

Jeps, eli kun korjasin tuo pilkkuvieheen ja otetaan huomioon että tosiaan cyclicin liikkeet ovat käytännössä joitakin asteita niin ei tuo sitten enää olekaan niin iso asia, mutta toisaalta kun katsotaan peräservon toimintaa ja eroa erinopeuksisten servojen välillä niin taitavat ne millisekunnit merkitäkin jotain

[kaardian]

Yleisesti kun puhutaan ajasta, niin... Aika kuluu nopeasti mutta vuodet hitaasti. Ajatelkaapa millainen aika tuo 1ms taikka 100ms on tähtitieteellisessä mittakaavassa? Kärpäsenkakka itämeressä taitaa olla melkosesti enempi

Mielenkiinnolla seurannu tätä keskustelua. Paljon tullu uutta mietittävää ajatellen kaikkea radio-ohjattua. Loppuviimein taitaa olla enempi omien aivojen synapsien hitautta kun ei osaa. Lisää reenagee omilla laitteilla, niin hyvä siittä tulee, oli viivettä eli ei.