ESC to RPM sensor processing

[LupusTheCanine]

Feature request

Is your feature request related to a problem? Please describe.

• AM32 ESCs sometimes report RPM values that are significantly out of range (50k+ RPM on a motor that spins at 4k RPM). Thist mostly happens during spool up causing heli governor using ESC RPM telemetry to fail to engage)
• BDShot RPM telemetry (on AM32 ESCs) is jittery ±10RPM jitter at 4k RPM which probably messes with the Ardupilot governor, causing throttle jitter.

Describe the solution you'd like
Adding lowpass filter to RPM library and respecting RPMx_MAX when taking data from ESC telemetry would alleviate these issues. At the cost of slightly decreased governor tracking accuracy

Describe alternatives you've considered

• Checking that the governor is engaged prior to takeoff
• Kickstarting the motor with the interlock to avoid the issue
• Cycling the interlock to re-engage the governor
• Using throttle curve (not feasible due to large voltage range and apparent lack of voltage compensation)

Platform
[ ] All
[ ] AntennaTracker
[ X ] Copter (heli)
[ ] Plane
[ ] Rover
[ ] Submarine

Additional context
Will add screenshots soon.

[andyp1per]

I've seen this. I have to say that in my case it was a bad motor, so its worth checking this. The ESC data already has a CRC so really don't want to add a filter if possible - would prefer to fix the problem

[LupusTheCanine]

I observe the issue with such erroneous RPM only at very low RPM during initial spool up and at the end of spool down so I would be more inclined to believe the issue is caused by the ESC losing track of the rotor than the motor. I think the ESC should handle such slip ups more gracefully, but it is what it is.

Spool up event, this one wasn't caught by the RPM library 🤷‍♂️.

The filter I want mostly to deal with the ESC telemetry being limited to the resolution of eRPM frames for with the LSB represents 8RPM at 4000 RPM, 16RPM above 4088RPM according to the documentation and progressively getting worse. I think smoothing it down a bit could improve governor jitter (which shakes the tail).

Unfortunately, I do not have experience with tuning the ESC to make the spool up more reliable and the AM32 documentation doesn't really help either.

Main motor ESC settings for good measure, the motor claims to be 800kV.

[rmackay9]

It's great to see use of the AM32 ESCs (I use them too on my copter). I agree with the comments that we should try and fix this at the source. The AM32 devs are close-by, some of them even in the AP dev team (FYI @tridge, @Huibean)

[LupusTheCanine]

@rmackay9
I reported the issue on the AM32 GitHub am32-firmware/AM32#242 (comment)
Though I haven't seen spikes during runing on 2.19

[rmackay9]

Hi @LupusTheCanine,

Thanks very much!