Fine tuning autofocus

HI all,

First thanks so much to all who have answered my rookie questions and put me on the right path. Truly appreciated.

Another question: What is the procedure to fine tune your autofocus. I have it working properly to the point that it is achieving focus and shooting with no issue. But, it seems to accept an HFR value than is higher than I would normally use. For example, it might settle on a 3.2 HFR when I can manually dial it to 2.6. I feel like there is a tweak I am missing.

Thanks, Bill

Which version of SGP are you using? And also which Auto Focus Metric? If you are on V4 and ASTAP then remember that the metrix is HFD not HFR. And different options do not always give exactly the same measurement for osensibly the same thing.

Thanks for your response, Mike.

The version is 4.0.0.651

I am not sure exactly how to answer you other question except to say that the parameter shown for measurement by SGP is HFR

Thanks, Bill

Hi Bill,

As you are using SGP v4 you have a new option to use ASTAP for detecting and measuring stars during autofocus. I think it is by far the best option but you can experiment for yourself - choose a night when the moon is near full and no good for imaging. As noted above on the AF focus curve this option shows HFD rather than HFR.

If you have the time and endurance to read it all, much of the background leading to the provision of this option is contained in this thread:

Does SGP talk to Planewave’s PWI AutoFocus?

You may already be using ASTAP for platesolving (it is very quick) but if not you will need to download and install from here.

http://www.hnsky.org/astap.htm

Once you are up and running with SGP AF there is not a great deal you can do to fine-tune it apart from:

a) Vary the step-size and/or number of steps. You need to select values that will get you reasonably out of focus (3-5 times your best obtainable HFR) while still giving you (I suggest > 5 star detections) at the extremities. With an L filte this is generally not an issue but can be more of a challenge using NB filters. For what its worth, with my f/6 refractor I use 7 x 60 steps with exposures varying from 5 - 30 seconds at 2x2 binning.

b) Atthough SGP currently uses a parabolic curve fit the actual curve should be hyperbolic so if the outermost data points with a typical 7 step AF run are not in moreorless a straight line you may have uncorrected backlash, a poor choice of step-size or poor observing conditions on the night.

c) If you start an AF run with your scope close to best focus for the conditions then you should get a nearly symetrical AF curve. An issue with the current parabolic curve fit is that it can give a poor result if the AF curve is greatly extended on one side or the other. You can guard against this to a degree by setting a more demanding AF quality fit increasing the custom variable ‘quadratic_fit_min_quality’ to say 0.98.

d) Once you are reasonably confident in your AF set-up then if you think there maybe benefits then you can start looking at filter off-sets and temperature compensation.

And, going back to your original question, while focus accuracy is a key determinant of HFR, other factors such as tracking accuracy, seeing and temperature stability can also will have an impact.

Thanks for all the help!! I will look into the ASTAP

Bill

Hi Bill,

I think you will find that using ASTAP for focus metrics will give you more consistent results than the older HFR method, especially when using denser or NB filters. In my opinion this is down to ASTAP star detection being more sensitive particularly when measuring out-of-focus stars at the extremities of the AF curve. If these values are markedly wrong it can have have a big influence on the AF curve and thus the calculated best focus position.

Having looked back to your original post, I must add that if you have a manual focus method that can regularly achieve a significant HFR reduction following a good AF run then I would be most interested to hear how you do it!

Best of luck

Mike

Thank you Mike. I just installed v4. Hoping to try it this month. Weather hasn’t been great.

KG

@Kurious_George ,
Hi KG,

I share your frustrations with the weather. The best I’ve managed is a couple of nights for about 5 hours total since December. The only two really clear nights I’ve seen have been at full moon!

As I somewhat hijacked your Planewave thread I do hope you get good results with v4 ASTAP. It works well for me but I appreciate we are using very different equipment. I’ve not seen any serious analysis from other users but the few in passing comments I’ve noticed have been favourable.

Personally I would still like to see the parabolic curve fit replaced by hyperbolic as aesthetically it looks the more pleasing but in terms of the actual calculated focus position (with my scope at least) the difference is a generally only a couple of steps, at least when the completed curve is close to symmetrical around the calculated position. I’ve observed that when the AF curve is obviously asymmetric that the variance between the calculated hyperbolic and parabolic fit position starts to diverge more markedly. I’ve guarded against this by increasing the result quality threshold to 98% so as to trigger a rerun if the result is possibly suspect.

I hope when the issues the devs have had in V4 with switches and the transition to 64-bit and ASCOM 6.5 are resolved they might get round to this change but I am reconciled to this not being a high priority.

Clear skies

Mike