Some ideas for improved SCT autofocus

  1. Put a option for settling the mirror prior to taking autofocus exposure.
    At least with my system it takes about 10 seconds for the mirror to settle into a new position. I’d likely get much more accurate focus if there was a way to wait until you take the exposure.

  2. Average a number of exposures for a given point on the V curve

  3. Acquire the V curve data just a little differently.
    Collect the right hand points as before to generate a linear regression line for that side. When the curve begins to tail out at the bottom move the focuser to the the extreme left-hand position and move down the V curve on the left. The advantage of this is that by the time you have identified a good focus position you’d be near focus. This would eliminate the large (problematic) movement of the focuser at the end of the current autofocus regime.

As it stands right now I’ve spent more than 40 hours of sky time trying in vain to get autofocus to work with my SCT. When I implement the above recommendations manually I can nail focus repeatedly. Thanks for all your work!!!

Glad to hear I am not the only one unable to get auto focus to work with a SCT. I think I commented before that the current approach of moving the focuser to the edge of the range and working toward the center does not work well with a SCT because you quickly get into donut stars which completely confuses the algorithm. It rather sounds like if you have a SCT, it would be best to stay away from auto-focus for now.

I’ll add my two cents to this. Through Beta 9, I don’t recall having a focus failure. If it occurred it was very minimal.
However both Betas 10 and 11 have suffered catastophic failures (i.e. donuts) roughly halfway through a night’s session.
FWIW, the focus failures have so far occurred at an AF, during a target sequence, rather than following a slew and AF.
I may yet got back to Beta 9 to see if there is any improvement. If I do, I’ll comment on the results.
Mark

While we are not committing to implement exactly the features as outlined above, we are committing to address long focal length and / central obstruction (mirror) focusing in 2.5.

1 Like