Sensorless homing (selector) and travel delta when moving.

[cmroche]

Sensorless homing (selector) and travel delta when moving.

[cmroche]

I was able to track this down.

Basically it's caused by the fact that homing_move is used (which makes sense), Stallguard doesn't work very reliably near 0 velocities and basically what happens when the selector decelerates to stop at the target position, if it goes slow enough due to a low acceleration (I was using 600), it causes Stallguard to trigger a bit early.

With 600 m/s^2 I was getting the 1.2ms - 1.4ms I mentioned above.
With 1500 m/s^2 I am getting around 0.5ms each time.

Since this can vary a lot based on configuration, perhaps the default 1mm tolerance should be changed to something a bit larger? When using sensorless it might be good to advocate for a higher default acceleration on the selector too.

[moggieuk]

Wow. Don't know how I missed this message before. Very interesting detective work. I was never able to find the correlation between different deltas. Well done 👏 .

I will experiment with increasing the default acceleration. 1500 seems a little high but can ship with higher and definitely document this finding.

BTW I have a major rework of Happy Hare in a new repo. I'll accommodate this feedback there..

[moggieuk]

Ok. For Happy Hare v2 (which will replace this software) I have made default acceleration 1200mm/s^2 and increased tolerance to 1.5mm. I this brings everything back into the happy zone.

I'm going to close this issue on this release (because I'm unlikely to update given the HHv2 replacement).