Frame rate testing
 

Knocked together a little diagnostic and tested the gear I had on hand. No idea if settings/bindings are optimal, but would love to hear if this is typical. I tested:

• Sanwa M17 with RX-493
• Futaba T3PV with R314RB
• Sanwa M11X with RX-451R

Shows pulse width, framerate and voltage on the receiver. Ignore the voltage readings - they're just a consequence of the way I tested.

1) Do people get different numbers from these with the same gear?
2) What other gear should I check?
https://cimg3.ibsrv.net/gimg/www.rct...b94e90c355.png

Does your instrumentation also have the ability, or can it have the functionality added, to also test off center reaction time, and total response time?

Also, also, are you running these with SSR/SXR/SUR and SR modes enabled?

Because frame rate only tells part of the story. If the response time isnt quick, then the added resolution of a higher framerate doesnt add much benefit.

It would be great to really drill this down to the base numbers, and definitively unpack how these fast response systems work.

Thanks for taking the time and effort to do this.

So you have tested the main modes and no SSR, SXR modes with the M17, that would be more interesting

I have done similair with my Futaba and 3 different receivers.
https://www.rctech.net/forum/showpos...4&postcount=21

Not in its current form. Measuring overall latency requires getting into the transmitter. Doable, as other threads have shown, but I'll add that to the backlog.

The M17 is borrowed, so I'll need to dig up the manual. Will do the same for the T3PV.


Not yet, but as above, I'll see what I can learn from the manual, unless you have a quick guide/cheat sheet and flipping between settings.

Quick test with SUR mode. Frame duration ranges from 110->513 microseconds. Will try other modes later.

https://cimg6.ibsrv.net/gimg/www.rct...1af9a847a8.jpg

Those numbers look in line with typical SHR mode or regular digital servos. I believe SSR centres around 300. SUR I think has the same centre pulse duration as SSR but ups the refresh rate. Not sure about SXR.

In practice SHR and SSR shouldn't have a significant difference in latency since the refresh rate is typically similar despite having the shorter pulse duration.

Here (I think), all of the mode combinations between M17 and RX493, in no particular order.
https://cimg5.ibsrv.net/gimg/www.rct...b39ea48323.png

You're missing SXR. Might need a firmware update on the radio if it's not showing as an option.

SHR and SSR I did measure the same om the scope. SSR is faster in latency because the the pulse is 1.2 sec shorter on center but even about 1.4 sec on one outer edge.

NOR looks like the analog servo mode but can be very tricky with older servo's. Also SHR is searching the limits of digital servo's which normally are specified up to 333Hz. I wonder if you can test UR from the Futaba 10PX (with latest firmware), I think it is a 1000Hz SR (SR is normally 833Hz)

Not my radio, so might wait for the owner to do it. It's on 1.01.06. What's different about SXR?

I just need to find someone with that radio. There's an event this weekend, so fingers crossed there might be one there.

Hello PDR,

SXR is faster than SUR.

Cheers.. :)

Ok, got the nod to do the firmware update and gave it a whirl. As Socrates once remarked: "un beacoup de frames"
https://cimg9.ibsrv.net/gimg/www.rct...ae4c6f420d.jpg

The latency will be variable, as the source signal may be too late for the current frame and need to wait for the next frame. SSR should have lower lows, but the average response time will be about half the frame rate for all modes. If anything that'll make SSR feel worse to some people as the difference in latency between best case and worse case will be wider, feeling less consistent.

That is SXR?
Pfew 1500Hz, I wonder if someone can feel a difference between SUR and SXR. Althougd I did read some people can feel a difference between SSR and SHR (same frame speed but only a shorter pulse) but I think it is more like a placebo effect.

There's another factor not many people consider in the SSR vs SHR debate. Microcontrollers have a finite timer resolution. I built a USB dongle adapter for VRC Pro using an Arduino, which by default has a 4us resolution on the timer function in the standard code library. I wrote a custom timer that knocks the resolution down to 0.5us. The SHR range is 1000-2000us, a 4us resolution means there are 250 steps between the min and max values. Each step has an error of 0.4%. SSR has a range between 50us and 550us, which means a 4us resolution has 125 steps, with an error of 0.8%.

Even when switching to the 0.5us timer in my code, there was still quite a bit of jitter in the recorded interval in SSR mode. It was visible in the software calibration screen. That suggests that the interrupts of the micro weren't consistently firing on time, or the signal from the receiver itself had some timing resolution errors. An oscilloscope would be able to verify the latter, which I don't have access to.

In short, assuming equal resolution timers in the servo and receiver etc, SSR (and by extension SUR and SXR which have the same pulse timing) has half the resolution of SHR. A cheap servo could easily support the SSR protocol with a firmware change, but if the timing isn't up to scratch then it might lead to a worse experience.

https://github.com/gigaplex/RC-USB