I have found that auto focus often under shoots the first move to a focus point causing it to have to go thorugh the routine a couple of times.
At first I thought this was due to filter offset but it seems to happen irregardless of which filter is in place. Besides the filters are very close to parafocal.
Pretty sure I have the step size correct as often this does not happen and the initial focus point is approx. 3x the prime focus point. When it doesent undershoot the first point I get a near perfect V.
Is this undershoot due to backlash? If so how do I figure out how much backlash and how to set the adjustment.
Bob hit on the common cause of this for me. As the temperature drops, the real focus point moves in, leaving the current focus too far out. This in effect reduces the first move relative to the (real) focus point. There are thrree ways to deal with this:
Focus more often.
Increase the step size. The intent here is to make the first move get outside of the nonlinear part of the curve and into the linear part. This will keep SGP from rejecting the first move point. I have mine set up for 7 points, with the HFR target for the most outward point about 6 times the in-focus HFR. This is more than the conventional wisdom of 3 to 4 times the in-focus HFR, but I find it to be more effective as long as enough out-of-focus stars can be found.
Increase the number of focus points. If you’re using 7 points, try nine. SGP needs 3 points in the linear range to fit a line. If you use 9 points, even if the first point is not in the linear range, the next 3 are more likely to be. As long as SGP has a minimum of 3 points on each side of the curve is a good, straight line, I have found that it will locate good focus reliably.
No I don’t halt guiding during autofocus. Never saw the need to.
I don’t think it was temperature shift as I was kind of ‘testing’ changing filters with each 3 minute frame. Running autofocus with each filter change.
The set started out at 13.7C and ended up at 12.7C
SGP doesent ‘reject’ the first point, a 'normal run it was running out to about 6 or so but sometimes it would only run out to 3 after 2 frames it was already past peak focus and starts running out the other direction till it is almost lost at like 8-10 or something.
Next time I will try increasing the step size but I had read that the first point should be in the neighborhood of 3x the sharp focus number. Then it gives up and starts over.
It usually works fine just once in a while the first point is not as far out as it should be.
You and I are going up the same learning curve. I just got my focuser running two weeks ago. I have been using nine points. I think that’s where you should go. I can see from my results that seven is not enough. With seven points you would have to be perfectly focused to get three dots on both sides of the focal point.
In terms of “turning of the autoguiding”, I found it made a big difference. That’s because I’m using an off axis guider. If you’re not then it’s not an issue. With off axis guiding it’s a must.
Thanks Bob but no I am using a guide scope (60mm f4) not sure why turning off auto guiding would make a difference but I will give that a try as well. Maybe someone can explain to me why turning off the guiding while focusing is a good idea.
It does often focus fine with 7 steps but I will try it at 9. It has been working pretty good with HFR’s in the 1.2 and sometimes lower range.
I just though it might be backlash and I have no idea how to incorporate backlash for a focuser.
560 total steps for the entire focus run, meaning that at 7 focus points you’d want 560/7 = 80 steps per focus point (rather than 140 or 110). This is just based on your graph. But lots of things come into account here. If your current setting is working just leave it as is.
I think Jared is saying 80 per step. 7 X 80 = 560 total steps.
With off axis focusing the location of the guide star changed when the focuser moves out. Then the guider struggles to bring it back which distorts the star shapes. With a guide scope this shouldn’t b e a problem.
I love when I get a pretty “V” curve like you posted.
A point about nomenclature – over the past year I have trained three people in using SGP and each has been confused by two of the labels used in the “Auto Focus Options” dialog box. The labels “Auto Focus Data Points” and “Step Size” don’t seem to convey their actual meaning very well.
For what its worth, I suggest changing them to:
“Number of focus frames”
and
“Focuser steps between focus frames”
I realize that the Tool Tips help as does the SGP manual but these are normally only read as a last resort.
Not implying that folks here have not read through this or that it is even needed (here), BUT since people search for stuff in the forums when they are having trouble with AF, I thought I would point out that we have a pretty concise guide you can use as a reference when setting up AF for your rig:
Hate to dredge this thread up again but I am still having the same issue.
Yes I went through the directions linked above and came up with an initial step size of 130.
When an autofocus is called for the first run out is not far enough so when it gets to the 3rd step (I use 7 steps) it runs past focus and continues out. So I never get the red line on the right half of the V.
Then it starts all over. On the ‘do over’ it runs the first point way the heck out to like an HFR of 8 and needs like 7 or 8 steps till it passes in focus point.
I have tried a number of step sizes from 80 to 140 and still have the same issue regardless of step size.
Seems like you want to use 9 steps then? Maybe I don’t understand what you are presenting. Seems like that would allow for a more granular measurement on the “right side” and provide a solid regression before it unceremoniously steps right past focus.
FWIW, it seems the ‘number of steps’, according to the logs, is the nmber of steps required before reaching smallest HFR and not as per tutorial page
"Set the “Auto Focus Data Points” field to represent the number of samples you want auto focus to take during the run. If you choose a value of 9, perfect focus would show a “V” curve with four point outside of focus and four points inside of focus. If you don’t have a specific value in mind for your rig, choose nine points here."
The log provides verification of this. Seven steps are shown configured, after 7 steps of the first try execute it says ‘getting larger’ and then restarts, moving farther out, versus the tutorial that says 9, 4 inside, 4 outside.
[3/15/2017 8:44:00 PM] [DEBUG] [Camera Thread] - Data Points: 7
[3/15/2017 8:44:00 PM] [DEBUG] [Camera Thread] - Step Size: 100
[3/15/2017 8:44:00 PM] [DEBUG] [Camera Thread] - Current Position: 4169
[3/15/2017 8:44:00 PM] [DEBUG] [Camera Thread] - Initial Move Position: 4469
[3/15/2017 8:44:06 PM] [DEBUG] [Camera Thread] Focuser position matches requested position (4469), continuing…
[3/15/2017 8:44:21 PM] [DEBUG] [Camera Thread] Auto focus HFR calculated at: 3.31
[3/15/2017 8:44:24 PM] [DEBUG] [Camera Thread] Focuser position matches requested position (4369), continuing…
[3/15/2017 8:44:40 PM] [DEBUG] [Camera Thread] Auto focus HFR calculated at: 2.29
[3/15/2017 8:44:43 PM] [DEBUG] [Camera Thread] Focuser position matches requested position (4269), continuing…
[3/15/2017 8:44:58 PM] [DEBUG] [Camera Thread] Auto focus HFR calculated at: 1.10
[3/15/2017 8:45:01 PM] [DEBUG] [Camera Thread] Focuser position matches requested position (4169), continuing…
[3/15/2017 8:45:16 PM] [DEBUG] [Camera Thread] Auto focus HFR calculated at: 1.16
[3/15/2017 8:45:19 PM] [DEBUG] [Camera Thread] Focuser position matches requested position (4069), continuing…
[3/15/2017 8:45:35 PM] [DEBUG] [Camera Thread] Auto focus HFR calculated at: 2.06
[3/15/2017 8:45:37 PM] [DEBUG] [Camera Thread] Focuser position matches requested position (3969), continuing…
[3/15/2017 8:45:53 PM] [DEBUG] [Camera Thread] Auto focus HFR calculated at: 3.25
[3/15/2017 8:45:56 PM] [DEBUG] [Camera Thread] Focuser position matches requested position (3869), continuing…
[3/15/2017 8:46:11 PM] [DEBUG] [Camera Thread] Auto focus HFR calculated at: 4.76
==============================
=====End of 7 steps, 3 inside, 3 outside but SGp does not like it
[3/15/2017 8:46:12 PM] [DEBUG] [Camera Thread] Detected focus getting worse! Shifting auto focus range
[3/15/2017 8:46:12 PM] [DEBUG] [Focuser Move Thread] Focuser moving to 4619
[3/15/2017 8:46:25 PM] [DEBUG] [Camera Thread] - Data Points: 7
[3/15/2017 8:46:25 PM] [DEBUG] [Camera Thread] - Step Size: 100
[3/15/2017 8:46:25 PM] [DEBUG] [Camera Thread] - Current Position: 4619
[3/15/2017 8:46:25 PM] [DEBUG] [Camera Thread] - Initial Move Position: 4919
[3/15/2017 8:46:31 PM] [DEBUG] [Camera Thread] Focuser position matches requested position (4919), continuing…
[3/15/2017 8:46:46 PM] [DEBUG] [Camera Thread] Auto focus HFR calculated at: 8.25
[3/15/2017 8:46:49 PM] [DEBUG] [Camera Thread] Focuser position matches requested position (4819), continuing…
[3/15/2017 8:47:04 PM] [DEBUG] [Camera Thread] Auto focus HFR calculated at: 7.42
[3/15/2017 8:47:07 PM] [DEBUG] [Camera Thread] Focuser position matches requested position (4719), continuing…
[3/15/2017 8:47:22 PM] [DEBUG] [Camera Thread] Auto focus HFR calculated at: 7.02
[3/15/2017 8:47:25 PM] [DEBUG] [Camera Thread] Focuser position matches requested position (4619), continuing…
[3/15/2017 8:47:41 PM] [DEBUG] [Camera Thread] Auto focus HFR calculated at: 6.02
[3/15/2017 8:47:44 PM] [DEBUG] [Camera Thread] Focuser position matches requested position (4519), continuing…
[3/15/2017 8:47:59 PM] [DEBUG] [Camera Thread] Auto focus HFR calculated at: 4.68
[3/15/2017 8:48:02 PM] [DEBUG] [Camera Thread] Focuser position matches requested position (4419), continuing…
[3/15/2017 8:48:17 PM] [DEBUG] [Camera Thread] Auto focus HFR calculated at: 3.16
[3/15/2017 8:48:20 PM] [DEBUG] [Camera Thread] Focuser position matches requested position (4319), continuing…
[3/15/2017 8:48:35 PM] [DEBUG] [Camera Thread] Auto focus HFR calculated at: 1.45
[3/15/2017 8:48:38 PM] [DEBUG] [Camera Thread] Focuser position matches requested position (4219), continuing…
[3/15/2017 8:48:54 PM] [DEBUG] [Camera Thread] Auto focus HFR calculated at: 1.12
[3/15/2017 8:48:56 PM] [DEBUG] [Camera Thread] Focuser position matches requested position (4119), continuing…
[3/15/2017 8:49:12 PM] [DEBUG] [Camera Thread] Auto focus HFR calculated at: 1.34
[3/15/2017 8:49:15 PM] [DEBUG] [Camera Thread] Focuser position matches requested position (4019), continuing…
[3/15/2017 8:49:30 PM] [DEBUG] [Camera Thread] Auto focus HFR calculated at: 2.55
[3/15/2017 8:49:33 PM] [DEBUG] [Camera Thread] Focuser position matches requested position (3919), continuing…
[3/15/2017 8:49:48 PM] [DEBUG] [Camera Thread] Auto focus HFR calculated at: 3.98
[3/15/2017 8:49:51 PM] [DEBUG] [Camera Thread] Focuser position matches requested position (3819), continuing…
[3/15/2017 8:50:06 PM] [DEBUG] [Camera Thread] Auto focus HFR calculated at: 5.43
[3/15/2017 8:50:07 PM] [DEBUG] [Camera Thread] New auto focus method calculated focus at: 4169
[3/15/2017 8:50:07 PM] [DEBUG] [Camera Thread] Old auto focus method calculated focus at: 4216
[3/15/2017 8:50:07 PM] [DEBUG] [Camera Thread] Using new auto focus method for focuser position
Thanks for all that! Very thourogh thanks for your time!
I tried 9 points last night and it was better and resolved nearly first time every time. But still the 4th step as at prime focus So I had 3 outside of prime on the right, focus point #5 was at or near the same hfr as point number 4 then it ran out the ‘left’ side for 4 more steps so there was a total of 10 steps?
So the behavior is similar but now it seems to resolve on the first run. Might just be a quirk of that particular focuser / OTA.
Is the focuser on the newt on top or bottom? Sounds like it may be slipping in one direction or the other on the last move? If I were to hazard a guess I’d say that the focuser/camera is on top of the OTA? May want to use backlash compensation with OUT in that case.
8 months in and I never get reliable focus and Ive read everything I can find. The most confusing part of SGP unless you get lucky imo. I would faint if I got a curve like that
