Bad autofocus position in Taka TSA-102

Sequence Generator Auto Focus
1

Hi all,

I have a problem with Autofocus.
My equipment is:

  • Takahashi TSA-102 + Reducer/flattener
  • QHY9 Mono camera
  • OAG + Filterwheel
  • Seletek Armadillo focuser

My problem is that I always get elongated stars at corners.
The V curve is fine, but the real focus point is always 30-40 motor steps outer from the Focus Routine result.

Previously I had a TS 130 refractor with the same imaging equipment and I never had that issue.
The focus was always correct.

The takahashi TSA-102 has a very thin focus plane, so the V curve has to be very accurate.
I defined the steps size , number of steps, minimum star size, etc as stated at the help file of SGP.

Once I got the autofocus automatic point, if I manually move the focuser 30-40 steps out side, the elongated radial stars disappear and the stars at corners are round.

Investigating a little, if I take after focusing the saved files, and open it at SGP and calculate de HFR, the theoretical focused image has a higher HFR than the previous one (40 steps higher)!!

I tried:

  • different steps (9, 13, 15)
  • also different distance between steps (30, 40, 50)
  • also changing Bin 2x2 to Bin 1x1
  • and also changing the minimum star size (4, 6, 10, etc)

Always the focused image has elongated radial stars at corners.
The resolution of the focused central part of the image is poor compared with the image I got if moving manually the focuser 30-40 steps.

I though it was a problem of Backfocus distance from the reducer to the CCD plane, but after trying several rings (5mm, 10mm, 15 mm, 1mm, 0,2mm, 0,3mm etc) the result is always the same.

I can attach some AF saved files and two images, one with automatic focus, and the other moving the focuser out side.

https://drive.google.com/open?id=0BzglGy9bmr6fa3JYcUJzN1hoVEk

For example, if you see tha AFPack_376, the focus point is about interval 7 (See V image). But the interval 6 has a lower HFR if you open it in SGP. But in the V curve, the Interval 6 has a higher HFR!!

Also the log file for the session.

Any help?
Thanks a lot
Fernando
sg_logfile_20170309202338.txt (1019.0 KB)

2

@Errabundo

I don’t think I can help with the current iteration of AF. I looked through all your data (thank you for being so thorough in your request fro help). And it seems like you need provision to have AF focus on the outside band of the image (everything but the center… inverse crop). We have this on our backlog and between it and multi-camera support it is next up for implementation (both of these are after the 2.6 stabilization effort).

3

I am interested to know that I am not the only one who experienced this problem. I have a Takahashi FSQ85 which is exactly the same. I have posted my experience in a recent thread and also under feature request. It would seem that for many Takahashi’s the best focus by HFR as determined by SGPro is not the same as best focus for good corner stars. Takahashi’s standard response is one should focus 2/3 of the way from centre, but this is not possible using SGPro at the moment. I really do hope that Ken can implement the outside band / inverse crop AF focus routine, this would be very welcome by myself and a few other Takahashi users out there. (I should note that not all Takahashi models have this problem; and not all examples of a particular model have this problem - it seems to have something to do with individual OTA variability - some OTAs need this some don’t - but my FSQ85 certainly needs this).

4

Thx for answering to my question, Ken and Henry.
I took more images changing the backfocus distance. In my case (not Henry’s) I use a reducer/flattener, and it seems the Backfocus distance is very critical.
But I realized that focus is more critical indeed!.
I have seen that moving out the focus 30 steps (with my imaging equipment) after the Autofocus routine, in most cases I reach an acceptable focus. The whole image has round stars. I tried in Lum and Ha.
Unfortunately the weather changed and I’m in the middle of a stormy weather.

In the meantime you develop a version capable of focusing at the outside band, I don’t know if is possible an script telling SGP to move X steps out the focuser after the Autofocus routine.

Anyway, it seems that the actual autofocus routine makes a calculation of HFR different from the HFR calculation SGP makes manually.

The focus point found at the V curve routine, in theory, has the lower HFR of the serie.
But if you open the images saved at the AF folders, and calculate the HFR with SGP (the “star” icon), the Autofocus solution has an HFR bigger than others near it!.

Another thing to investigate is:
I take an image in bin 1x1 at the Autofocus automatic solution position (only central part of the image is at focus)
I take another image moving focuser 30 steps outside (the whole image is focused).
Well: If you compare ONLY the central part of the image, the manually focused image has better detail at the central stars, and a global lower HFR. (I updated the Google Drive folder with results).

That should means that in case of cropping, for instance, 70% of image for Autofocusing, the V solution should be near the manual solution, doesn’t it?

I can not try that in some days due to rain, but I’ll check asap and’ll post the results.

Thank again,
Fernando

5

You should be able to use focus offsets to compensate for this issue. You would just have to figure out the offset needed for R, G, B after focusing with L.

But if you want L to be in focus - you would need a way to provide an offset for the filter you are focusing with (L). So it would find focus with L - and then move to an offset from that point.

This might be easier than specifying some section of the image for focus - especially if you don’t have many stars.

Frank

6

Hello Frank,
That’s what I mean with a Script or something like that.
My filterwheel has 7 filters, and I use to focus with L.

In that way it’s easy to setup the offset for the other 6, but there is no way of applying an offset to the filter you use to focus (or I don’t know how).

At least, when doing Narrow band, I can keep the autofocus with offsets relatives to L.

Thx
Fernando

7

that’s an interesting idea I think I will give it a go. I have a 8 slot filter wheel so I have on slot free. I can get another lum or clear filter for focusing. How reproducible is the filter offset though?

8

I have just tried what Frank suggested but it does not work because the offset required is inconsistent.

9

That is interesting, my FSQ85 works just fine most of the time. Irrespective of the autofocus software application, I have seen several TAK scopes’ HFR V-curves with a hump in the middle. My mate’s FSQ106 does it a lot and I occasionally see it with the 85. I think that is why the slope method works best, rather than the minima version. Logically, I would have expected SGP to be better behaved, since it is looking at multiple star HFR readings, rather than the more typical single star, used by TSX / MDL / FM etc.

10

Yes Buzz, it is very frustrating. Takahashi Japan response is that it is within spec. Some FSQ85’s are better than others, obviously; and it is really down to luck of the draw. I do occasionally see a middle hump in the bottom of the V-curve. At the moment it is a nice parabola but no matter what the best focus for round star is always a few clicks in from best focus by minimum HFR, but the difference is not consistent enough to have this hard wired in.

If I can’t get this to work I think the FSQ85 is not a keeper. Automated focusing is so important for ones sanity I simply can’t afford to baby sit the scope when it is imaging, especially when it is also very sensitive to temperature change.