Improving quadcopter stability at very low throttle using Complementary PWM

This article documents a case study in use of Complementary PWM (COMP_PWM) to improve quadcopter stability at very low throttle.

An observation of two quadcopters of 450 size running several releases of Cleanflight and now Betaflight 3.01 is a loss of stability at very low throttle opening.

This is not uncommon for several reasons, and there is ‘airmode’ in both firmwares to address the problem that motors at minimum speed cannot be slowed further. Experience with airmode on Cleanflight up to v1.14.1 was that it raised throttle so much that descents were extremely slow sometimes, certainly never quick, and its use was discontinued.

I have since abandoned Cleanflight due to unresolved flight problems, lack of migration facility from version to version, and the quiet removal of the backup and restore facility.

The objective of this study was to explore the effect of enhanced motor braking with COMP_PWM on basic angle mode loop stability at low rpm.

Test scenario

The study uses a BC3540 1100Kv motor with 11×4.7 SF propeller, F-30A ESC with SimonK (1e4c01782eff85da3971f628a3bd599b7a0725eb) with COMP_PWM enabled.

Tests were conducted with a script that I use consistently with asrg and eLogger to capture current and rpm, and all tests conducted at similar pressure, temperature and humidity, altitude is 700m.

Test results

One of the effects of COMP_PWM is stronger braking of the motor when throttle is reduced. In multi-rotor application, the motor braking under COMP_PWM is dwarfed by the propeller load at maximum rpm, but propeller torque falls as the square of rpm and at lower speeds motor braking becomes more significant.

Above is a graph of the drive response with a non-COMP_PWM response feint overlay. It can be seen at 13s, that under rapid deceleration, the COMP_PWM response differs, lets zoom in on that.

Examining the response carefully, we see that there is negligible difference in the deceleration above 4500rpm (which is hover speed for the aircraft), but COMP_PWM enhanced braking does make an increasingly significant difference at lower rpm which are used in descent.

Flight tests were conducted, but using a single test quadcopter there was considerable time between before and after test flights which adds experimental noise. Nevertheless, a subjective perception was that whilst it was quite easy to get the quad to tumble out of the sky in quick descents low throttle openings without COMP_PWM, tumbling was much less likely at similar throttle openings with COMP_PWM.

Initial flight test comparisons of airmode on Betaflight v3.01 were that without COMP_PWM it was markedly better than with Cleanflight v 1.14.1, and with COMP_PWM it is stable and allows quite rapid controlled descent. More testing required.


Subjective assessment is that COMP_PWM significantly improved angle mode stability in quick descents at low throttle opening.

It is likely that improved low rpm angle mode stability will assist airmode, requiring less throttle injection and allowing faster controlled descents.

Initial flight tests with Betaflight v3.01 airmode using COMP_PWM are encouraging and more testing is required.