Strange RCOut to the ESC

ArduCopter Copter 4.1
1

Guys
today I went for short test of my quad, because I changed ESCs and Props so I hoped I can do autotune. It was not possible since quad didn’t respond well on throttle lever. So I looked at the log and found strange graph when looked at RC3in and RC9-12out, where I connected new ESCs. I connected them to these out because I switched to Dshot 600 ESC protocol. Could this be the reason of the strange RCout and very sluggish response to throttle? Look at the image, the green curve is RCout on channel 12 and the red is RC3in - throttle.

obrázok
obrázok1362×536 28.6 KB

Now have loaded v4.1.3 on it, will try 4.1.4. Auto pilot is Pixhawk PX4.

2

These are not F4 based ESCs are they?

3

Racerstar RS30A V2 30A Blheli_S ESC OPTO. Are they?

4

Nope. F4 would be BLHeli32

5

Does it play some role if is it or not F4 based?

6

There is an issue with F4 and BLHeli32 related to a faulty temperature sensor. That is not what you are seeing here.

7

Yep, indeed.
Anyway I’m just interested if the output to the ESCs is just OK like this when the MOT_PWM_TYPE is set to DShot or there is some problem with my Pixhawk.
Sluggish reaction to throttle is perhaps different story.

9

Have you turned off Low RPM Power Protect in BLHELI settings?
Also it might be handy to see a log file or at least param file. For example it’s easy with an overpowered quad to have your MOT_THST_HOVER lower than MOT_SPIN_MIN , when it shouldnt be of course.

10

Here is log file Dropbox - 2022-02-12 10-55-31.zip - Simplify your life

11

What size is this craft?

12

wheel size is around 650mm

13

So ~15" props? Those outputs are oscillating terribly as you say.

14

No I put 11’’ 3 blade folding props on it , because I couldn’t find 12’’ folding props for reasonable price.
Motors are Sunnysky V3506-23 kv:650.
What could be the oscillation?

15

I think you have the stab P pit and roll gain number set too high, reset them from the 12 13 you have now both to 5 and try again.

16

Thanks for hint. Those were from previous autotune process guess, though not quite good.