So with my latest rig, a 12" TPO RC, and AP focal reducer, Auto focus has been working well until suddenly the last few nights I seem to have doughnuts, and even had one sequence where I could see them getting larger as the sequence continued.
So, loose focuser assumed but it seems solid as a rock, so I’m a little stumped.
SGP is set to autofocus every 0.8 degrees, and I see in the log it can see the focuser temp but I don’t think the temp has ever changed that much in an event/filter (yet).
In the meantime I do see that SGP (and ccdstack2, for that matter) seem to be struggling with giving good, or ZERO HFR numbers for doughnuts, so it seems I should try to get that sorted as well as looking for a mechanical issue.
I have “Disable Smart Focus” checked, since I have a center obstruction. Should I try with it unchecked?
I have, left over from previous rigs, “Minimum star diameter at 1x1 (px)” set to 4 but not sure what this should be. I’m imaging, via an CCD67 focal reducer and 2x2 bining at 0.92 arcsec/pixel and assuming seeing is no better than 2 arcsec. What should my minimum star diameter be and will this help detect doughnuts?
Anything else I should try to avoid ZERO HFR results with doughnuts?
I have a 10-inch RCT - a few things I have noted for ensuring good AF performance.
Good collimation. If the alignment is out, even slightly, it messes up the V-curve around the focus point. It appears to be the case that a good V-curve with equal slopes is a very sensitive measure of collimation.
Don’t go too far out. With shortish exposures, seeing noise will create hotspots on a tiny donut and confuse the HFR result. I try and restrict the focusing range to prevent any turn-over of the extreme HFR values and use a starting point which is no more than a stepsize out.
My RCT is a CF truss and from one night to the next, is quite consistent over a good temperature range.
Focus with one filter - especially since you have a flattener - this means that each AF event starts at a known good point - rather than hunt for something different (see 2). I used a GoldFocus mask on my RCT to establish the different focus points
Yeah last night the solution seemed to be going with 7 points instead of 9, as a way to “not go too far out”. I was reluctant to reduce the step size as I had originally determined that it needed to be that big to get "V"s vs. "J"s.
I do have good shaped V curves, with all filters and had recently collimated.
I will consider using offsets instead of focus per filter. I too have a truss (TPO 12") and have not determined the temperature stability yet, but perhaps the offsets would be consistent across ambient temps?
I guess we won’t get anymore tweaks to SGP 2.x, but it seems like it should know that an HFM of ZERO is bogus and not use that part of the curve. For instance last night I saw four points at zero, before the remaining points generated a good V curve, but still SGP used one of the ZERO HFM points as the focus to return from the AF process, instead of the bottom of the V.
I’ll certainly buy 3.0, when it comes out, however.
Another forcus related challenge I had last night was in resuming the sequence (after abort) after realizing that the AF had failed (due to ZERO HFM). It insisted on starting right in with the Ha filter (first event) without redoing the AF, even though I have AF per filter and AF at start of sequence. I tried reseting progress and that didn’t help. I had to manually run the AF on the Ha filter to get things going again.
OK, a “smoking gun” seems to be that my camera settings were not what I normally use so the gain was all the way down (guess I need a check list if SGP isn’t going to remember these ascom settings).
I am unfamiliar with the camera, but with such healthy exposure times, I expected considerably more star detections, unless you are doing this through NB filters maybe. Do you use the dark frame subtraction feature in SGP?
With all my scopes now, I focus with Lum and apply offsets. I use an exposure time of about 8-10 seconds and with dark subtraction. I only rarely get a poor result and so, for that reason, have not only a temperature trigger for AF but also an AF after every 10 frames (RGB) or 5 frames (NB), just so that there is little possibility to ruin an entire night when the temperature is stable.