Autofocus failure -- too many range shifts


Was doing testing a calibration on 3/15. Ran into a problem with the autofocus routine (after focusing with Bahtanov mask). Focuser stepped in and drew a curve, then reset itself and started over further in, which did not converge as it was too far in. Then routine terminated with message that focus had failed due to too many range shifts.

Attach function for logs is a great idea, but you might want to include the autofocus package files.

Aprox time of issue: 9:30pm

Link to Logs

Useful Info

OS: Microsoft Windows 10 Pro 1903 build 18362.719
.NET: 4.8

There seem to be a couple things off here. You are producing asymmetric graphs which do not provide SGPro with enough confidence to calculate a focus position.

  1. Step size seems too small. At 100 steps, a full 9 data points (900 steps) moves HFR from ~8 to 3 and then back up to 4. Double it and see if that helps.
  2. It looks like you may have some backlash you need to compensate for…

Note that SGPro 4 is going to improve how data is captured. In this case, it would have been best to continue going instead of recentering (but, on the other hand, it would hide inadequate settings).

Thanks, Ken.

You may get SGP4 out before my sky clears again… will try your suggested values at my next opportunity.