SGP curve looks great but ends with "...too many range shifts..."

SGP has been working beautifully!..thanks for producing such a great piece of software!.

Inexplicably half way into the imaging session I started receiving sequence warning messages for “smart focus failed…too many range shifts”…“focus cannot be determined”.

Happened a few times (but one) when focusing in between filter changes.

No clouds, clear night, happens when SGP is about to finish a normal “V” curve and actually achieving good focusing ( 0.7, 0.9, etc)…then right by the last couple auto focus exposures warnings start and the focusing sessions re-start all over again ( when actually good focus was already achieved).

Focusing as usual with my light pollution filter, clear night, same setup and imaging/focusing settings used w/o issues in prior nights, BUT I made a small change in the position of the scope/camera in the mount to re-balance the weight ( 1/4 inch or so). But I didn’t touch the focuser or moved the camera relatively to the scope.

Takahashi refractor, feather touch focuser, imaging at F/6.

AF settings obtained using the recommended procedure in the SGP help section. Focuser looks tight mechanically, no play, scope balanced, sub arc second tracking, etc.

What I’m doing wrong this time?..thanks!!

Annotation 2020-03-19 010605

Aprox time of issue: last one happened at 11:17:55 PM

Link to Logs

Useful Info

OS: Microsoft Windows 10 Pro
Ver: 3.1.0.457
.NET: 4.8

The range of your focus region is too large. The “wings” of you curve prevents the fitting routine to accurately evalute the focus position. If you look at your very first focus curve above, at that point the focus routine gets a good fit of your data for the points below HFR 2.4, when the routine continues taking the remainging data points to the left the quadratic fit function doesn’t work well anymore. It is now recommended to have your AF routine go out to about 3x the minimum HFR, which in your case looks to be between 2-2.5.

A reduction of the step size from your current 275 value to maybe 150 would probably improve your AF (with 9 data points as before). You probably have to experiment a bit.

Mikael

Success!, step size to 150 did the trick…thanks Mikael!

Those curves look really good, great!