Help Needed Setting Up Auto Focus

I’m trying to set up SGP for the first time to auto focus.

I’ve read through the documentation and created a step size using the method described several times and keep coming up with the around the same step size (450-500). That seems like a huge step size compared to what others use.

When I run autofocus, I do get a “U” curve however the final focus is off when compared to a Bahtinov mask. I have to adjust the current position by 200-300 points to get focus perfect with the Bahtinov mask. Prior to using the mask the HFR is aroud 1.2 After using the mask it’s around .8 They vary slightly with each image I take.


  • SGP
  • CGEM with StarSense
  • WO Star 71
  • QHY23
  • RigelSys Focuser using nStep


  • Bin: 2x2
  • AutoFocus Data Points: 9
  • Step Size: 450 (If I make this number lower it becomes more of a backwards “J” graph)
  • Minimum start Diameter at 1x1 (px): 2 (tried several settings with no difference)

Any suggestions on how to improve my focus?

You really want to get your Max HFR to be around 2-3x your min HFR. For instance using your min of 0.8 you want to shoot for a max of 1.6-2.4 as the top of the auto focus graph.

Essentially adjust the step size to get there. Focus with refractors is pretty linear up to a point so this should be fairly easy to find with a little math if you care to do it, or with some trial and error. If you can post your SGP log from some AF runs we can probably get you pretty close. I wouldn’t concern yourself too much with the step size…but I do agree it seems higher than most, but that can be because of a handful of different factors.

Also 5-7 data points is generally sufficient. However 9 will allow you to “push though” for poorer seeing.


I may have been going a little high then. The directions said Max HFR for 3-5x.

Seeing where I’m at isn’t really good most of the time, but I can try fewer data points. Was just following the directions for this setting too.

Attached is a copy of the log file.

sg_logfile_20160802215432.txt (469.3 KB)

I spent some more time last night trying to get the focus correct. I tried several different setting and step sizes, but ended up pretty much where I started with the settings. I took a few screen shots in case it helps.

I’ve never used autofocus before, so I’m not sure exactly what to expect when compared to focusing with the Bahtinov mask.

Before I started auto focus, I focused using a Bahtinov mask.

After it finished, I got this message.

Here is what the graph looked like.

After autofocus I checked it again with the Bahtinov mask. Focus is off compared to before autofocus.

Here is the log file from last night.
sg_logfile_20160804190754.txt (382.3 KB)

Are you measuring HFR with the bahtinov mask on? That would throw off the calculation I would assume. Your AF graph looks excellent. I would trust the AF graph much more than the mask. After using AF did you examine the stars? Did they look out of focus to you?

The HFR was measured without the bahtinov mask. I was just using it for comparison of the focus.

The stars looked focused to me after I ran autofocus, I couldn’t see a different between the autofocus run and focusing with the bahtinov mask.

This is my first time trying to do autofocus of any kind, so wasn’t sure how else to compare. Perhaps like you said I shouldn’t put so much trust in the bahtinov being perfect focus.

I’ll try an imaging run with the settings I have and see how it turns out.

Thanks for the help.

The mask position seems to be 23314 which is within one focus position of the HFR determined position, that probably counts as the same for all practical purposes.


Instead of comparing with Bahtinov mask, I would just repeat the autofocus routine a few times and see what the numbers are. If the curve looks nice but it isn’t repeatable, there is probably something mechanically loose. It looks like you have a step size of 400 and the focus position would need to be repeatable to about 100 or less. If it isn’t, then even if the curve looks good - when the focus winds back to the calculated focus position it won’t be exactly where it should be.

Do you have backlash compensation set? You should probably set it to a good value and make sure it is invoked when moving from a low number position to a high one. It should overshoot when going that direction and come back.