Finally some success autofocusing FSQ85

I would like to share my recent tests with autofocusing a FSQ85 at native focal length.

I have posted in a couple of threads recently documenting the problem with autofocusing my FSQ85 as the best focus by HFR is different from best focus for round stars. It is a documented problem with some Tak and the standard takahashi response is to focus 2/3 of the way from centre. However this is not currently implemented by SGPro. A feature request has been previously lodged for user selectable subframe focusing as a potential solution.

Last night I had clear sky with relatively full moo so I performed more testing. When bin 2x2 previously I found that I needed a step size of at least 15-18 to get consistent V curves but I know that these step sizes are larger than the CFZ of FSQ85 (by calculation it allows 5 steps for my setup). When bin2x2 the best focus by HFR is always a few click out from best focus for round stars. When I critically analysed the images manually, the best focus images actually looked quite good with round stars at the corners. In fact the central 3 AF samples all show round stars at the corners. However when actually capturing at bin1x1 the ovoid corner stars reared their ugly head.

Last night as I last resort I tried autofocus bin1x1. In doing so I can reduce the step size to 8 which is about half what I have been using. In doing so I can see that the stars go from radially outwards thru to round to circumferential about centre, this time with best focus coinciding with round corner stars! However at bin1x1 I needed 10 second exposures even with lum filter and it was very very slow. Best results also needed to reduce the min star size to 2 which I suspect also slowed things down. I seemed to be able to get good results with a few different targets using these settings with a Lum filter so I am happy.

The same thing does not quite hold for colour filters. I suspect that has something to do with the longer duration needed to get all the corner stars (which are usually dimmer for most targets) included in the calculation. I increased exposure lengths to 20 seconds and the results were still not consistent.

Anyway I am happy at least I can get good auto focus with Lum filter and I may be able to use offsets for the other filters which is probably needed anyway otherwise it may take 15 minute per autofocus run every filter change. I still think user selectable subframe focusing (not just excluding the centre) is the way to go as I can continue bin 1:1 (which seems to be the essential ingredient for success) but have the subframe at the “worst” corners only to speed up the image download and preferentially optimise the corner star focus.

Next time I will try to use half stepping to see if I can get finer granularity in the step size to optimise the focus even further.

I should also add that having guiding on ruin the autofocus run (I use OAG). So long exposure AF exposures are no go beyond 10-20 seconds for my set up.

Thank you for sharing your success story. I have also been experimenting with auto-focus and have had my best luck with 1X1 binning. With my 3 nm H-alpha filter I have been taking 18 sec focus shots.

Hi Henry.
Thanks for sharing your experience.
I’ll try as soon as I can!

I have done more testing last night. Half stepping does not seem to help. If anything it seems to make things worse, so I have reverted back to full stepping more. I still cannot get consistent good result with colour filters using offsets. It is very close but not perfect. I think if I am doing LRGB imaging with bin1x1 on L and bin2x2 on RGB it would work, as with bin2x2 there is no perceivable elongation at the corners. However I like to be able to take RGB at bin1x1 too because that’s how I like to acquire my data on some target (without L).

It is a very frustrating problem. It would seem that the Takahashi was really designed with the larger pixel cameras in mind. If my camera has pixel size of 7-8 micron instead of 4.6 that it has, then i imagine this problem of exquisitely-focus sensitive star elongation at the corners will not have surfaced since it is not visible when bin2x2 with my camera. Some people may get a better one without this problem but some like mine or yours do.

I think as the very last resort I may be forced to try different software. Ekos can do focusing without having to skew to a different target but it will automatically pick the brightest star for focusing so it may not work still. may be there is some way I can limit the selection area to the corners. I don’t know.

Henry - I don’t see any of these issues on my FSQ85 with the reducer. I am running SGP quite happily with the following settings:
step size 25, 9 samples, 6 pixel reject, 10% crop, 1x1 binning and 6 second exposures with Lum on KAF8300 chip.

I took an image and measured it with PI. The median FWHM is 1.2 pixels and eccentricity is 0.54 over a 5 minute exposure. (The fact that the FWHM is 1.2 pixels supports the notion that 2x2 binning is not the way to go.)

I strongly suspect that you either have tilt or image curvature. The focusing is even across my entire frame. This later may be caused by the wrong sensor spacing. I don’t see how different software will help in the case of the hardware not being set up correctly.
FITS : Dropbox - File Deleted

when I ran this fits file through CCD inspector, curvature was 11%

Buzz: I am not using a reducer and the fact that I could get round stars across the field also excludes tilt or spacing issue. As you know the baby Q is a petzval and its field flatness is linked to focus. When it is outside focus the corner stars are elongated circumferential to centre and when it is inside focus they are elongated towards centre. Only when it is I focus are the corner stars round. I can actually manually assess focus by looking at the corner stars and when they are round I am in focus! The margin of error for my particular scope is very small, but it is very obvious that for others the margins of error are much wider such that when they focus using standard practice / routine they get good results, whereas for mine it does not. I think it is down to individual scope variation.

Anyway my problem is by no means isolated. Some people choose to live with it, some give up on Tak and some like myself try all sort to make it work - I am just wishing that one day there will be a software equivalent to what I am doing manually (focus using corner stars) to get good result with the baby Q, hopefully it is with SGPro since it is otherwise such a wonderful app.

Henry - I’ll have a go myself without the reducer (I have only just got this scope). If you shrink the margins down, SGP would take an aggregate FWHM of the entire image. That should be a reasonable blend of focus performance? It also occurred there would be a consistent correlation between focused central and corner stars - you could do that with a deliberate filter offset? (You would focus with a ‘focus’ filter and then use offsets to move to all other image taking filter positions.)

Hi Henry,

thank you for the update and a possible workaround!

But as you I am still hoping for an enhancement of SGPro not only to exclude some space from the borders (“crop autofocus frames by” in the autofocus options), but also from the center by a second option like “exclude from center autofocus frames by”.

I am still waiting for retesting the autofocus with my new Feathertouch focuser.

Best regards

Reinhard

I agree that it would be best if user can select a region for autofocus, especially if the autofocusing is to be done at bin1:1 which takes too long with the entire frame (longer download time, larger file to analyse).

I am wondering if SGPro can divide the entire frame into say 9 parts, and let user choose a particular part of AF. That way, those who want to use centre for focusing and choose the centre part, and those like myself who want to use the corner for focusing can do so. As the AF frame is smaller it will also speed things up. Since the division is hardwired and not randomly selectable, it should also make the coding easier (but I am no coding expert)…

I am not sure what would be easier to implement.

The option to select from 1 out of 9 predefined areas would be also a solution, but I think only a second best (as you can only focus on 1 corner/ border). The same applies (regarding the issues we have with the FSQ) for the option to give the possibility to define a custom area (e. g. by allowing to enter the top left and bottom right corner of the foucussing area).

I still think the optimal solution would be, if we could define a sort of “frame”, which is used for calculating the HFR:

(quick and dirty drawing with MS paint. The grey area is the complete image and the yellow the wanted area which should be used for calculation of the HFR)

Regarding the effort my guess is following.

For the already implemented “crop” that can be defined in the autofocus options, the image has to be - as it says - cropped, by the defined percentage, prior sending it to the routine which computes the HFR.

If there would be an additional “exclude from center” setting in the autofocus options, (which should also be a percentage number of the overal image size) I think it would be enough to past just a black rectangle of the specified size into the image center, prior sending it to the HFR routinge (this way the center is not calculated for the HFR).

So it seems not too much. But I am afraid the developer has a lot of request and fixes to do, which each for its own “is not too much” :slight_smile:

But I will not give up the hope.

Best regards

Reinhard

I have been a long time UK user of SGP and have had a FSQ85 for 3 years now, mostly used with reducer (and its three terrible grub/set screws) and with a KAF8300 sensor. I have persevered and tolerated fractionally elongated stars in a corner, sometime two. I concur with Henry’s experience of the changing orientation of the corner star shape as the AF moves extra-focal to intra-focal.

In the autumn last year after much experimentation with not using reducer, cleaning CAA rotator, imaging with CAA out of optical train, I concluded that the Tak CAA rotator induces tilt. I took this apart, cleaned it, re-greased, removed grease, reset focuser, fractionally tighened the screws on focuser, checked the seating of each filer in my QSI683, experimented with metal back spacing with reducer (found a sweet spot greater than Tak specify) but never was truly satisfied. It drove me crazy.

So after some deliberation I took the plunge and purchased the Feather Touch 3" FSQ85 focuser and adapter. This cured all of my focusing woes immediately albeit at a cost. I focus at 35 steps, bin 2, 6 sec, lum filter (plus off sets for all NB filters), focus on the whole image area, have backlash set to 50 steps. If I recall correctly my Min Pixel Size is 3.

Below is a screen shot from PixInsight of a single uncalibrated 1800s Ha sub without reducer at F5.3 including Abberation Inspector with corners for some pixel peeping. Whether I have a ‘good’ FSQ85 or whether poor CAA and focusers are more common than anticipated I’m not sure (and putting aside whether you should have to invest in a 3" FT focuser after spending on a premium refractor), the image below shows that SGP’s focus works for me and my FSQ:

Henry, you may already have investigated the stock Tak focuser and rotator and be satisfied that they are not inducing tilt; if not I would experiment with removing the rotator from your imaging train. Take some subs when pointing at the zenith straight up to remove any possibility of sag in any particular direction, rotate the ccd to see any effects/change - perhaps you’ve been through this already, so apologies if you know all of this.

I do apprecaite your frustrations however I would encourage you to hang-in there :slight_smile: as it is a super scope when you have it sorted.

HTH & CS!

Barry

Thank you Barry for sharing your expiernce. Indeed a Feathertouch improved situation a lot. Before I upgraded the focuser the elongation was asymmetrical at four corners so they never became round at the same time implying tilt. With the Feathertouch now at least with very careful focusing the stars can be all round at the same time on my Atik 460 (smallish sensor). In my opinion it is luck of the draw whether you get a good baby Q that works out of the box and whether you get a dud like mine. I probably have mentioned before but my FSQ has been checked by Takahashi Japan and was said to be within “spec”.

My suspicion is that the babyQ was designed in the era of film camera or large pixel CCD where the large pixels / grain size hide the problem (the star elongation is not visible when I bin 2x2) but becomes apparent when one starts using smaller pixel cameras. I am actually quite tempted to get one of the older CCD with larger pixels and be using the darn thing rather than fiddle with it.

Just for you, I used my first clear night in weeks to reconfigure the Tak85 without a reducer. I kept the same AF settings, but reduced the margin to 1%. This was first light with this configuration.

I just let SGP do is stuff and it worked a treat.

1.4pix median FWHM
CCD inspector has this as perfectly flat - a good sign that it is in focus. Very slight tilt which shows up in bottom two corners a little. I cannot fault the focus though.

ONE THING - do not have the focus lock even remotely engaged, it induce drag and causes the tube to skew (you see it with image shift when you change focus)

I use a different Q (FSQ106) but @buzz is right – you do not want the focus lock even remotely engaged at all. I use a different focuser on my FSQ (Robofocus 3.1) but ensuring you have the lock completely disengaged is important.

Hi buzz and rockstarbill
Thank you. Do you use a Feathertouch or a stock focuser? I use a Feathertouch, as the stock has bad tilt. I do have a bit of image shift when the focuser is moving. May be I will try to loosen the focuser lock.

I have the original focuser on the '85. The scope is new and it has the heavier duty one. I do have FT on my WO refractors. They run on two sets of three teflon pads - and the top adjuster sets the friction/play. I basically set that so the motor does not stall. These refractors (FLT132 and FLT98) are much longer FL, so if there was any image shift, it would have noticed by now.

Henry - I have been imaging with my rig over the last week and I have noticed that if there are not sufficient stars being measured over the whole image area, the results can show elongated stars at the corner. I now focus with luminance only and use offsets for the other filters (I used a GoldFocus mask to measure the offsets). I doubled my exposure time to 10 seconds and the results are more consistent, with good V curves without strange bumps.

@HenryNZ I use the stock 4" micro focuser on the FSQ106.

Hi Buzz and RockstarBill

Thanks for all the feedback and tips. Weather has been bad and I have not done much, but did manage to sneak in some additional tests last night.

I think by now it is pretty clear that my babyQ has field curvature which is at the upper limit of the specifications. The integral flattener will only produce flat field if focus is carefully achieved. Your baby-Q is likely better than mine so you have been able to get good results with standard AF parameters. I know it is not tilt as round stars can be achieved - it is just fiddly to do so.

At the beginning, I have been assuming that bad field flattening was due to an apparent difference between the best focus for HFR and the best focus for round stars. I have been trying to determine what this difference was so that I could focus for best HFR and then apply an offset to get round stars. However, I have not had any luck with it. Firstly, the focus point for best HFR was often different for different runs; and secondly, the offset that was needed to get round stars was also different for different runs.

Then I thought, may be I could minimise this difference by including as many corner stars as possible in the AF routine for HFR calculation. My hypothesis was that when the corner stars were round, they would also have the lowest HFR and by including as many corner stars as possible in the calculation, I would also optimise the corner stars. To this end, I have changed to bin 1x1 and reduced the min star size to 2. I thought I hit the jackpot as this strategy appeared to produce good results with L-filter, albeit very slowly. Unfortunately I did not have much luck with the other color filters.

But then I started thinking, why was the AF focus so inconsistent between runs? Clearly the inconsistency was the reason why simply applying a constant offset to the AF result did not work. Perhaps there was something about the autofocus process that could be improved?

So two nights ago I set up to test whether I could get more consistent AF results by just using the centre of the FOV for focusing. I activated the “crop” function so that the elongated corner stars were excluded from the analysis. That’s when I found something quite interesting:

Without changing the step size, I found that the previous V curve became a very very shallow U. When the corner stars were excluded and using the old step size, the HFR at max outside focus was like 1.5, it went through focus at 0.8-1 and back to about 1.5 at max inside focus. The centre 3-5 data points all had the HFR around 0.8-1.0. It became very clear that the AF focus was very susceptible to error (a slight change in the HFR results in a big change in the calculated focus point). To make a better V-curve with well defined slopes, I had to very substantially increase the step size from 18 to 55 and the min star size from 2 to 10!!

It would appear that previously the majority of the HFR change through the focus range was in fact due to the corner stars, and the centre stars did not change much through the focus range at all. It also means that the AF focus result previously was likely not very accurate.

I have since changed back to bin2x2, with 40% cropping and increasing the step size from 18 to 55 and min size to 10. By doing this, it appears that I am able to get good corners with my LRGB filters although more tests are required.

It may just be spacing. If the sensor is to close, it is under corrected. A mm can make the difference. If it is too far away, it can over correct. Try Baader Delron shims