Bug: Full throttle despite not being armed in bidirectional dshot300 mode.

[maxikrie]

Hi,

Today we had a fairly scary incident where one motor went full throttle despite not being armed. We were testing bidirectional dshot300 mode and forgot to invert the checksum being sent to the ESC (Tekko32 F4 45A ESC). Hence the ESC wasn't armed, which we confirmed as the motor wouldn't make the 'armed' sound. When we set the throttle for testing to 150, one motor out of six suddenly went full throttle.

We couldn't reproduce this, hence we are suspecting a weird coincidence. However, we are quite worried that the motor could spin without being armed.

This issue from some time ago seems very similar - #353.

Thanks!

[sskaug]

Sorry to hear about your scary experience. And glad that there was no injury. We did quite a few changes to Rev32.10 to address unintentional spinups, and have not heard many reports since. But unfortunately fails do occur, it is impossible to prevent completely.

[maxikrie]

Thank you for this quick and honest answer.

However, respectfully, I must disagree. I cannot think of a reason why this would be impossible to prevent completely other than a bug in either the state machine or bad memory access. Clearly, the function spinning the motor should turn off the motor safely if the ESC is not armed and nothing but the arming command should be able to arm the ESC.

We have decided to move away from BLHeli for our application due the fact that the code is closed source and you are confirming that the issue, which even was reported several years ago, may occasionally still occur.