Hi @jbiscan21 ,
The main issue was that the max_rpm_delta was set to zero. Even with PI gains (and max errors) set to zero, you could have had it working with non-zero max_rpm_delta. Let me explain.
max_rpm_delta term is used to cap the target rpm relative to the current rpm. This is done to prevent very aggressive motor behavior, especially if motor is spinning very slowly because it is tangled in grass, etc, allowing unlimited max_rpm_delta could cause motor to burn out or de-sync.
The RPM controller has a feed-forward term and feedback terms.
- feedback terms are Proportional and Integral with corresponding gains and error terms (k * e).
- feed-forward term comes from the calibration : for desired RPM, the calibration tells the ESC what PWM (duty cycle) to apply
- there is also battery voltage compensation, so that even if voltage is changing, the feed-forward curve will still be accurate.
The rpm error is computed like so : rpm_error = rpm_desired - rpm_current, which standard. But then it is capped with the max_rpm_delta, so if that is equal to zero, then rpm_error will be zero. One subtle detail is that the rpm_error is also used to look up the desired rpm feed_forward : rpm_for_feed_forward_term = rpm_current + rpm_error (roughly speaking), so having the max_rpm_delta set to zero, caused the motor to be always stuck at minimum rpm.
So you could still have zero P and I terms and the motor would use a feed forward term with voltage compensation and it would provide a nice and soft response (similar to traditional ESC). Adding P and I will make the response better than a traditional ESC.
In fact when using a new motor + prop, you should start off with P and I gains zero, just use a feed forward term to test at first and then you can increase the ESC's responsiveness by giving it more max_rpm_delta and also P and I terms. However, that is more advanced tuning should be done using voxl-esc tools first in order ensure ESC response stability (we could discuss that in another thread).
To answer your earlier questions:
- You could use pwm (power) control, but i have never used it with PX4. Perhaps @eric-katzfey can comment about that
- the ESC calibration for modalai ESCs calibrates the feed-forward term (as described above), and has to be done with a propeller. Yes, in flight, the air dynamics will change due to many factors such as air pressure, temperature, air speed, etc, etc, but the job of RPM controller is to keep a desired rpm. A higher level controller could potentially estimate the thrust vs rpm is changing due to air temperature or pressure difference from the calibrated response (outside of the scope of ESC's rpm controller).
Alex
