Focus and Platesolve problems

I would very much appreciate any help to try to improve the reliabilty of my autofocus runs. Sometimes they are flawless, bust mostly I end up choosing a low point from the autofocus graph and using that manually with a visual quality check. I have provided a Log at the end of this post.

My other problem which badly affects my abilty to do timed runs is using a timed sequence e.g. I set the sequ. to start at 2 a.m. but the Platesolve process fails, and its odd in that the other night with clear sky, I watched PS (ASTAP) solve the first iteration but fail on the second. I caught that one and just used the first iteration as the target - started PHD2 guiding quickly so as not to lose the target.

Part of the problem (and I think this is significant) is that the platesolve image sometimes looks ‘rough’, e.g the stars are smudged, example below. What causes this to happen? I have a MESU 200 mount which easily manages 20 minute guided subs. If someone asked me what the problem was I’d probably say ‘something jogged the mount’ or ‘cable snag’ , these would explain one off events, but not a persistent problem.

The other problem related to platesolve is that SGP took a platesolve image whilst my dome was still moving - that one didn’t solve :slight_smile: and if this is happening that might also explain the variability I’m seeing. I was in the observatory when this happened. Because I have a Sitech II mount controller, I visit that forum too, and recently someone posted that they use SGP and had a platesolve image happen whilst the dome was moving to target.

I’ll try to confirm the platesolve whilst mount moving and post back, that one can be checked in daytime.

Any help much appreciated.
Log link


For the elongated stars problem the solution it to enable/increase mount settle time.

For autofocus i suggest you change from 7 to 9 autofocus points, you might also want to try increasing step size to raise the highest HFR measured to around 10 or maybe even higher, watch for where it starts to fail detecting stars.
Backlash is something you want to look into testing different settings too, try change the direction, if that doesn’t help try increasing it more (i do see that it’s already pretty high)

I cannot see exactly which scope you use from your settings, but it seems like it’s an F/11.3 scope?
Slower long FL scopes are really hard to get consistent autofocus from.
If i’m correct on the FL you could also try 3x3 binning during autofocus.

thanks for this, much appreciated. I’ll experiment along the lines you’ve suggested. It’s a celestron c14 with a 0.8X reducer , so FL is 3128 ish.

the focuser is a feathertouch with microtouch controller which moves the primary mirror.

I’ll post back my results.



Just a note from a long time SGP user. I suspect your imaging setup is testing the limit of reliable auto focusing. The small FOV can result in having only a small number of stars to work with and the central obstruction produces donut star images when the stars are noticeably out of focus.

I image with a 12" Meade f/8 scope reduced to f/6. I use 9 data points and set the steps to have the most out of focus stars have no more then 3x the HFR of the best focus HFR. It is also important to have a pretty good (manual) focus before you even start the auto focus process. SGP has always had the goal of maintaining good focus but, typically, it is not able to reliably take a highly out of focus image and bring it to focus.


The number of stars detected in the fov can be small I agree. How many does SGP need to run the algorithm effectivley?

Perhaps a better approach might be to AF in an area of sky with high star density and then slew to my target.

I’ll try doing some AF runs in such an area and see if it improves reliability.

I’ve done quite a lot of experimentation with step size and about 90 seems good. 120 gets big dounuts for the initial point.

thanks for help,

my mount just developed a fault which means it can’t slew/ track so I’ll be a while before I can come back on this.


The ASTAP routine probably needs a dozen or so good stars for auto focus. Keep in mind that when out of focus, the stars get larger and dimmer, so having as many as possible is always good.

SGP supports auto focusing at a specified RA-DEC that is different than the RA-DEC of the target. I have not used it, so don’t know the details of its implementation. One issue I can think of is having the focus RA-DEC being on one side of the meridian and the target RA-DEC being on the other side. Depending on how the routine works, it could introduce a meridian flip in the middle of the focusing routine.