Send PWM_MIN value after a PWM_MAX value while calibrating ESCs

[dusan19]

When the ESC calibration is performed through QGC, the PWM_MIN value is not written appropriately. This can be verified by connecting to the ESC directly through BlHeli Suite.
ESC calibration should be done using the following procedure: Send PWM_MAX value for at least 3 seconds, followed by the PWM_MIN value for at least 3 more seconds. Instead the current implementation sends PWM_MAX value followed by the PWM_DISARMED value.
On the ESCs that I tested, this results in setting the PWM_MIN esc setting to 1012 regardless of the PX4's PWM_MIN value.
The following fix is proposed. Note: PWM values are being sent for 5 seconds, since 3 second long signals weren't detected reliably.
Using this fix, the values written in the ESC (checked with BlHeli Suite) matched PX4 PWM_MIN and PWM_MAX parameter values.
Following plots provide PWM output values during calibration, before and after the fix

[LorenzMeier]

Thanks! I have a couple of questions:

• For common ESCs, is the min where they do not spin or where they spin? Because the "min" in PX4 is the minimum where the motor should still spin, "disarmed" is the value where the propeller should stop spinning.

• The change goes deep into the PWM driver architecture and lacks generality with the calibration sequence. This needs to be changed to not change the output driver. However, given the first point, I'm not even sure it does what is required.

[dusan19]

Yes, you are right, they dont' spin with min pwm, you would have to increase the throttle just a bit to make them spin. So probably something smaller should be sent as a min pwm, like 0.9*min_pwm.
Anyway the way it works at the moment is not right because, the ESC min-max and px4 min-max ranges are not the same, so the mapping between PX4 outputs and motor inputs is not correct.

Regarding the driver changes, I didnt change the way it is used. It already checks if the calibration is active and sends the pwm max signal. I just added sending the pwm min signal next to that, depending on the calibration phase. Perhaps instead we should use set min and set max pwm ioctl definitions, and remove any calibration logic from the driver.

[dusan19]

With the latest change, there is a small range of pwm values between the one sent to the ESC and the PWM MIN value, to allow the motors to spin when armed with PWM MIN setting.

[mcsauder]

Hi @dusan19 , if you can resolve merge conflicts here and rebase I would be happy to take a look at your code!

[dusan19]

Hi @mcsauder, I could do that, but check out the discussion here #12759
In the end a suggestion to solving the problem was setting the PWM_MIN, PWM_MAX and PWM_DISARMED such that they match the ESC input range. Basically the disarmed is the low calibration end, and the pwm_min is a bit higher such that it spins the motors, but it doesnt set the esc min calibration setting.
This PR handles the problem in the opposite way, it calibrates the low end to be a bit smaller than pwm_min such that the motors still spin at pwm_min, and the disarmed is lower than all of that such that it is lower than minimum configurable value for min pwm in the esc. This way there is a clear separation between disarmed and min-max range configured in the esc.

If you still think this change is worth pursuing, I could solve the conflicts, rebase and push

[stale]

This issue has been automatically marked as stale because it has not had recent activity. Thank you for your contributions.

[MaEtUgR]

I assume this is superseded by #21711
Motor calibration now always runs between 1000 and 2000us independently of the PWM configuration.
Please reopen and state the use case where the implementation on PX4 1.14 does not work then we can improve further.