I know subframe focusing has been requested previously, but I would like to propose something a bit different which addresses the root problem. This kind of request arises because some optics are not perfectly flat at best focus as determined by HFR. For example, with my FSQ85 after autofocus the corner stars are slightly oblong, which can be made round by half a stepsize worth of focuser movement inward. This minute focuser movement will make the stars round across the field but does not severely impact the star size in the centre. It would seem that the best focus by HFR and best focus for field flatness is different with this kind of optics. One way of solving this problem is to focus using an area half way from the centre and edge, hence the previous request for user selectable subframe focusing. However, I am wondering whether it may be more direct to add eccentricity as a measure of focus quality, such that optimal focus is achieved when both HFR and eccentricity are minimised.
I guess if that’s too much to ask for, user selectable subframe focusing would be an alternative / workaround and I would like to up vote that.
As my autofocus is not working until now, due to hardware problems (Focusser not moving to correct possition - equipment too heavy? - Auto Focus - Main Sequence Software) I can speak only from my “sem-autofocus” experience. But it corresponds to the description of HenryNZ. If the focusser is at the lowest HFR position, the stars of my FSQ85 in the corners are more or less oblong. A slightly correction away from best HFR position produces a much better overall result, regarding the star shape.
Therefore I would like to support this request (or any other solution, which helps with this problem - hoping that after upgrading to a FeatherTouch Focuser with a PosiDrive motor I finally can start using autofocus
i think you guys should upload some AF packs for ken and jared to look at. this has been discussed before in the context of reflectors - some people like to focus on a star that’s somewhere away from the center of the field, to compromise between coma at the edges of the field, and perfect focus at the center. in those discussions K&J felt that computing HFR across the whole field essentially accomplished the same thing. but it would appear your OTA is “proof” that this is not the case across all types of telescope.
I have performed a series of tests last night to document the problem. When using lum filter or no filter, there were discrepancies between the best focus as determined by SGPro using HFR and the best focus for global round stars. The differences between the two focus points were in the order of -8 to -24 clicks of the focuser, when the step size in the autofocus routine was set to 24. So it is approximately 1/3 to 1 step off. Interestingly the discrepancy was not seen with my RGB or NB filters. However with these filters the corner stars were very dim and I was not sure if the test subs of 20 seconds were not long enough to make the corner stars bright enough to show the elongation in that setting.
I think the problem with petzval design is that star roundness is linked to the focus, whether more conventional design they are decoupled. So for the conventional design focusing by HFR is great, but this is not necessarily so with the Petzval design.
It would seem to me the difference between best focus by HFR and best focus for star roundness is under 1 step size. In theory if I lower my step size this problem may be alleviated however I found that with my focuser motor I cannot get a reliable V curve if I reduce the step size further. May be if I upgrade the focuser motor to something super premium to enable smaller step size this problem will also go away. But I don’t know and premium focusing motor is not cheap.
I have uploaded two AF packs to my drop box. One is with L filter and one is with no filter. In the L filter folder there are also the corner star images taken from the worse corner.
Unfortunately the amount of offset was not consistent. For example in three consecutive runs using the L filter I got three different focusing points which differed from the “ideal” by 0-24 clicks.
AF packs below.
No filter, difference in focus was 8 clicks:
L filter, difference in focus was 24 clicks (but I ran the test three times and the differences recorded were -24, -12 and 0):
