I’ll explain how I have my setup working and maybe it might help you with yours.
A. Yes, my focuser, N-Step is setup so the primary mirror moves in and out to obtain focus.
B. Exposure Time = 5 seconds
C. Binning = 4 x 4
D. My Backlash is set to “IN”.
E. Backlash is set to 500 steps. I would recommend something higher to start with. I actually started out with my Backlash set to 1,500 but over time I found 500 steps work well for my setup.
F. Data points set to 9
G. Step Size = 325. Be careful I think this is were folks get a little confused. the “Step Size” is the number of steps SGP tells the focuser to move for each data point.
H. I have my crop set to 20%. It seems the stars on the outer edge of my scope don’t play well. So I found if I tell SGP to ignore those outer 20% it helps with my focus.
I hope I can explain this next part correctly.
For my setup of 9 data points Ideally we want the 5th data point to be the lowest point in “V” curve, “best focus” point. Therefore the purpose of the step size and Data points is so to get far enough out of focus to start with so the focuser can “Walk” back into focus stopping at each data point, in my case every 325 steps and takes an image and determines the focus value. So when the focus routine starts it moves out (4 data points * 325 steps) + my 500 Backlash steps. Lets say my focuser number when starting out is 1,000. It will move out (4 * 325 + 500) steps for a total of 1,800 steps. the focuser will be reading 2,800 since I actually started at position 1,000 plus the 1,800 steps it moved out. Once it walks out 1,800 steps from where it stated it then walks back in, “Backlash” 500 steps, stopping at the 2,300 step position and takes the image and determines the focus value. It now moves in another 325 step, another data point, and repeats. If you started out close to focus your 5th data point will be close to the best focus value. At this point the line should be dropping down and to the left. Higher step numbers on the right and lower step numbers to the left. and the HFR values will be lowering as the focuser walks in. once it reaches the 5th data point, again in this example if you started out in somewhat good focus this will be your lowest HFR. It then continues moving another 325 steps, step position numbers are going lower, for another 4 data points but the HFR values are going back up. Thus you should see something that looks like a “V” or depending on the seeing maybe something that looks like a “U”. Anyway the bottom of the “V” is the best focus step position. After all of that, using the the 9 data points SGP determines the step position that will give you best focus. Lets say it determines the “Best Focus” position is not the 1,000 that it started out at but 1,150. In my case it will move out to position 1,650. (1,150 best focus position + my 500 Backlash) then back in 500 step to land at the best focus position of 1,150.
One added note, determining Step Size is trial and error. but from best focus position to poor focus position you want the HFR value to be something around 4. So if at best focus position the HFR value is 1 you will want enough data points and step size so when the focuser moves out the HFR value is around 4 times bigger then the best focus HFR. Nothing really scientific about it just a rule of thumb to shot for. some folks use 7 data points with a higher step size. what matters is you move far enough out of focus so you can get a steep looking line when moving back toward best focus.
I know this was a long explanation and I hope it helps to make sense.