sg_logfile_20191013143124.zip (690.3 KB)
sg_logfile_20191013143630.zip (720.1 KB)
Ken, I can post all the focus run images if you need them.
Here are the log files from my AG12 and TV127 from last night using version 295 to test the new quadratic focus routine. There are 160 focus runs in each log file. My three targets for the night each contained the following sequence of events, with auto-focus running for each frame, the frames exposing for only 10 seconds, thus 160 focus runs on each scope. Basically just focusing all night long.
The events run in this order, repeating until end time for target:
L R G B L Ha Sii Oiii
The quadratic focus routine itself ran perfectly. I watched both SGP programs for about 3 hours.
However, there is a very serious failure to move the focuser to the correct new position.
Since I was doing continuous focusing, I turned off temperature compensation of both scopes, wanting to make the night’s runs as simple as possible to analyze.
Both scopes have backlash compensation set to IN 150.
My filter offsets of both scopes are very good, refined over the past 2 years.
Here is the correct sequence of operations when a focus run starts with a new filter, and temperature compensation is turned off: (this is logically what is required, but SGP appropriately combines these operations to minimize movement)
- change filter to new requested filter
- move focuser in or out to correct for the filter offset delta between the old and the new filter.
- take requested number of images to build focus curve
- run quadratic fit routine to determine best focus
- if quality of run not acceptable, move to new starting position and start over, or quit
- move focuser to this best focus position
- take image
All focus movements in the above sequence must apply the 150 step backlash for all outward moves.
Here is what happened:
- the quadratic focus routine worked perfectly to determine best focus, based on the focus data points
- in about 75% of the focus runs, the above sequence was followed perfectly
- in the other 25%, after coming up with a good best focus position for a nice centered curve, step two did not work as required, usually moving the focus position to a position that was about 2 or three intervals off. Step 3 then had a very off centered run, giving either a straight line, or only one or two data points above the low point on the low side, thus forcing a rerun.
- the final focus position after these reruns was often not correct resulting in the target image being badly out of focus. Here is an example:
Unfortunately, I have been unable to determine what set of circumstances is triggering this failure. Filter deltas both + and - both work or not work. Cycling past the first event is not a trigger. The best I can figure for what is causing this is perhaps an uninitialized variable or some random overwriting of a variable this routine depends on. Perhaps the routine does not like performing a focus run on every frame and with a new filter on every frame.