Wanted to open a thread up for folks to share their experiences with Voyager focus. This is not a thread to discuss SGP, SkyX, FocusMax, etc... It is to talk about and share experiences with Voyager focus.
I'll kick it off by showing my most recent focus tonight. This was taken with a FLI ML16200 camera, AP 92mm refractor, with the G filter, in 86%-ish moon. Unguided, using APCC Pro/APPM.
Thats in HFD, so the HFR of this was 0.835.
Very, very impressive. The focuser is a 2.5" Feathertouch focuser using a Starlight Instruments Handy Stepper Motor. I have no backlash compensation enabled (even though the focus motor does have backlash). The controller here is the Pegasus Astro Powerbox v2 with a custom cable I made. The lack of care about backlash compensation comes from my familiarity with the ML alg used in Voyager focus. In my experience, over time, you dont need to bother. It will figure it out on its own.
The focus does indeed look very good but I'm scratching my head over your numbers here. The Airy disk size for your 92 mm, F/6.65 scope is 8.93 microns in the focal plane, which is 1.48 pixels on your camera and 3" in the sky (@ 550 nm.) You are sampling at a rate of 2"/px with your camera. The number that concerns me is the medial FWHM of 1.584 pixels using a Moffet4 fit. That number is right at the limit of your optical system, which would be pretty good (and maybe possible with a 92 mm scope;) but, it's a red flag--even considering the HFD value of 1.67 px computed by Voyager. The Moffet4 fit is a pretty good model for a stack of very high quality data and for for star images taken under pretty poor seeing; however, it is a poor indicator of FWHM on high quality star images. So, your data jumps out as suspect--in more than one way, which I'll discuss next.
In order to achieve an accurate measure of FWHM, it's important to fit to the correct Moffet order. You can find the nearest Moffet order by using the DynamicPSF tool in PI. Sample a number of well exposed stars and look at the "beta" value for each star. That's the exact "best least squares fit" Moffet order needed to minimize the fitting errors shown in the MAD (Mean Average Deviation) column. The higher order Moffet functions tend to be less sharply peaked with a much wider "base". When you use the FWHMEccentricity tool, you can't use the best fit order so you have to use one of those provided. In your case, I suspect that the best fit Moffet order might be either 2.5 or maybe even 1.5. When you get the fit right, you'll find that the FWHM as measured by the FWHMEccentricity tool will decrease a bit. In your case, I bet that you'll find that you'll see FWHM measures somewhere between 1.1 and maybe 1.3 pixels. Combine that estimate with the fact that you have the Upper Limit set at 1.00 and that you measured 4042 stars tells me that you are most likely measuring mostly hot pixels due to dark current. Setting the Upper Limit to say 0.8 excludes using over-exposed stars, which will bias the results toward too large of a value as well as eliminating saturated hot pixels. Setting the star detection sensitivity to a higher value will prevent including too many extremely faint "blimps" in the data. You have a lot of stars in that image but using over 4,000 stars is red flag that the algorithm is picking up too much garbage in the image. Try adjusting that value until you get 300-500 stars with the upper limit set to 0.8. More is not better! You want a good sample of properly exposed stars only. Then examine the support image to make sure that you aren't picking up warm/hot pixels. That's how to get a more accurate reading of the performance of your system.
I'll be really impressed if you are indeed operating at the diffraction limit of your system but don't be surprised if the FWHM is closer to 4" or 5". The HFD number shown by Voyager looks really good but it does not represent the final result in your image.
Finally, I have one other comment. Seeing a good result from any autofocus routine is great but it's always a good idea to cross check the result against another method. The simplest way to confirm both the accuracy and repeatability of autofocus is to cross-check it using a Bahtinov mask. It's not hard to judge focus to within about 1/10 wave using a B-mask and that's a very good way to make sure that the final focus is spot on. Run the autofocus routine say 10 times and carefully check it each time with the B-mask. If it's good each time, you have very high confidence of the result--both in terms of accuracy and repeatability.
Edited by jhayes_tucson, 10 October 2019 - 08:37 PM.