Backfocus help needed pleeeeease!

[Lee_P]

I've officially reached the hair-tearing stage trying to get round stars to the corners of my frame, so would appreciate some help from the wise folks here!

All my kit is new: WO ZS73, 0.8x flattener / reducer, and ASI2600. The backfocus should be 56.8mm, according to: https://www.firstlightoptics.com/reducersflatteners/william-optics-x08-adjustable-reducer-flattener-for-zs73.html

I've got the image train set up as follows:

* Flattener / reducer set to 1.8mm. I used digital calipers, but it's still hard to know for sure if it's dead on.
* Extender that came with the camera, 16.5mm.
* Extender that came with the camera, 21mm.
* ASI2600 camera -- the sensor is 17.5mm back.

So, that should be the magic 56.8mm.

 

Here's a test photo. The stars at the top and bottom-left in particular look particularly bad to my eyes.

 

Here's the original FITS: Backspace.fit

I've tried tweaking the adjustable flattener to modify the distance by tiny increments, but I just can't get all corners to look good. Am I missing something obvious? Any help would be appreciated!

Thanks,

-Lee

[symmetal]

Your calculations are correct but I'm not sure why you needed calipers to set the adjustable FF as you just read the scale on the FF. Have you set the correct piece to the 1.8mm mark. Here's where the adjustment is made. It's a different flattener, but the principal is the same. The one below is set to 11.8mm using the rear section, and the smaller ring in front of it is then tightened against it.

  

You also need to add around 1/3 the thickness of any glass between the FF and the sensor to the 56.8mm to get the correct distance. Any filters and the 2mm thick protect window both need to be added on. Without any filters you need to add another 0.7mm, giving you a 57.5mm setting required. Corner stars pointing towards the centre tends to indicate the FF spacing needs to be increased but that's not always true. Try moving the FF adjustment out by a large amount, like 5mm and see what the result is. The new star shapes and type of distortion should give an indication of where your optimum distance would most likely be close to.

Alan

Edited November 19, 2020 by symmetal

[Lee_P]

Thanks for the reply! It clouded over before I could investigate further. Regarding the calipers, I used them to allow for more precise adjustment of the flattener -- not that it helped..!

I'm not currently using a filter. Is the "2mm protect window" an integral part of the camera? It sounds like 57.5mm is what I should be aiming for. 

One point to note, in my test shots sometimes one corner will seem ok, while others are bad. Could this be an indication of a sensor tilt issue? Or would backspacing explain it?

Thanks again,

-Lee

[symmetal]

Yes the protect window is the 2mm thick glass plate on the front of the camera. The manufacturers FF spacing distance is quoted for just air between the FF and the sensor so it usually has to be increased when used with most cameras, and any filters, if used. 

It's quite possible there is a bit of tilt causing your corners not to be the same. The camera has a tilt adjusting plate on the front which is handy if the screws aren't blocked by filter wheels or the like. First determine as to whether your current corner stars indicate the FF spacing needs to be increased by increasing it by a significant amount as I mentioned. If they now exhibit elongations at 90 degrees to their current direction, tangential elongation as compared to your current radial elongation then you've gone through the optimum FF distance. The stars may show little 'wings' instead of just being elongated.

If you can't find a distance where they are all equally good then it indicates tilt which is normally a bit of a pain to correct as you're not sure where in the imaging train the cause is. As the tilt adjustment pushes the camera further away at the adjustment point you want initially the FF spacing to show one corner good and the other corners showing stars towards the centre, assuming towards the centre means that edge of the camera is too close. Give one tilt adjustment screw a fairly large turn and make a note of the effect on a test image. Put that screw back and try another tilt screw, and then the third one so you know what effect each screw does and on what sides of the camera image they effect.

Before you make any tilt adjustments it's worth taking an image with the camera rotated at 90 and then 180 degrees to the current orientation to see if the image corner stars affected stay the same or whether they change corners as the camera is rotated. If they stay the same it implies the camera sensor is tilted relative to the rotation point, and if they don't it implies the focuser is tilted off axis.  A tilted sensor can be corrected with the tilt adjustment screws, while a tilted focuser will only be corrected by the tilt adjustment screws at that one camera orientation. I think that's the right way around but it's easy to get your mind in knots. 😀

A tilted focuser is most likely focuser droop under gravity so it's best to try and correct that problem at source, rather than using the tilt adjuster to correct, as a meridian flip will probably mean it droops in the other direction and your previous tilt correction adjustments will then make the result worse.

Alan

Edited November 20, 2020 by symmetal

[Lee_P]

Thanks Alan, you've given me the next steps to try. Now I just need some clear skies..!

[symmetal]

I ran your image through CCD Inspector and it reports very little tilt, and the fall off towards the corners is fairly even, reporting the FWHM better in the corners than the centre. This implies the focus has favoured the corners more than the centre. I think just increasing the FF spacing by around 1mm or so should give you better results. You invariably find the best distance is never exactly as calculated, so a little bit of fiddling is needed. At least you probably don't need to worry about tilt. 😀

The second 3D image doesn't imply that the image centre is 'closer' than the edges, it just indicates worse centre FWHM results as a more positive Z value. The brighter the colour, the worse the result. Most results show the corners rising up in the Z plane, but yours is pretty good in that respect.

CCD Inspector should be used more as a guide than an absolute analysis as it seems to rely on its determined FWHM values, and errors like coma and other star 'smearing' effects don't seem to affact the calculated results as much as they ought to.

Alan

 

Edited November 20, 2020 by symmetal

[Lee_P]

Thanks Alan, that's very kind of you of check. Being about 1mm out would fit with your calculation that I should be at 57.5mm rather than 56.8mm. I'm glad to hear that I don't need to mess about with the tilt! I've made the required adjustment on the flattener, just need some clear skies to check. Obviously the forecast is for complete cloud cover for the foreseeable future...

[Lee_P]

@symmetalI caught a brief gap in the clouds to test the 57.5mm backfocus. I took three photos as suggested.

0 degrees:

 

90 degrees:

 

180 degrees:

 

Those corner stars are much better than before. I tried a few more pictures after adjusting the flattener tiny amounts in each direction, but to be honest I reckon the shots above are as good as I got. Do you think I should play about some more, or am I being a pixel peeper? It's hard to tell 😅  I've got a light pollution filter, and an L-eXtreme on order, and am dreading putting them in and having to readjust backfocus again!

[symmetal]

Hi Lee_P,

All three images are pretty much the same, regarding corner star shapes, and better than your previous post so increasing the FF dist has helped.

Top-left is pretty much spot on. Top-right and Bottom-left show slight radial elongation (towards the centre), while bottom-right is showing a tendancy towards tangential elongation, (rotating about the centre), although it's a bit of a mixture of the two.

It's worth increasing the FF a little more, about 0.5mm say to see if the top-right and bottom-left improve, without the top left getting noticeably worse. The bottom right is in a bit of a halfway house at the moment so may improve with the increased 0.5mm or get tangentially slightly worse elongation. Worth giving it a try.

The CCD Inspector plots show slight differences but small variations like this can easily happen between successive frames in an imaging run. For a more consistant analysis you generally need to take several images (6 to 8 or so) at each position and stack them. You had varying thin cloud on your images which can affect the measured results too. Did you refocus between rotations and did you manually focus or use autofocus for the images. The central hump you had on your first posted image has gone which is good as it implies your central stars are now a bit sharper.

There is an indication of very slight tilt from left to right, for which you may want to try using the camera tilt adjustment (as it's not affected by camera rotation) but it's very small and you may make it worse. Before doing that though, it's best to check an image from the other side of the meridian to check that it's not just very slight focuser droop possibly causing this.

You may find after stacking many images in normal imaging that the slight corner aberrations are masked out by the slight star 'spreading' due to rotations and shifts during stacking, so your setting as it currently is may be fine. 😀 I used up several rare good imaging nights trying to improve the corner stars on my ASI6200 (full frame sensor) but accepted that it wasn't going to get any better, and real stacked images were actually very good, especially when binned 2x2 😀

Alan

Edit. Rechecking CCDI analysis results, the 0 deg had 650 stars analysed, while the other two only had 350 or so stars used for analysis which could explain the slight variation. The clouds probably obscured more stars on these. I would take the 0 deg image as the best one to use here as an indication.

 

Edited November 23, 2020 by symmetal

[Lee_P]

Thanks Alan, I'd be lost without you! To answer your question, I manually focussed before the first picture, then moved the flattener's in-built rotator to get to the other two angles. I didn't refocus between shots. I've got an autofocusser ordered, but it's taking a while because of the current stock issues. The next time the clouds part I reckon I'll try tweaking the distance a little more, and then just go for an imaging run and see what the final stacked result looks like. 

I dread to think what kind of precision is needed to get good corner stars on your 6200!

Thanks again,

-Lee

Edited November 23, 2020 by Lee_P

[symmetal]

I mentioned focusing as a small variation in focusing can affect corner star shape as well as focus. Manually focusing on the 'thirds' can give improved corner star shapes without affecting the centre to any degree. With autofocus you're dependent on what method your software uses as to whether it goes for best average focus over the frame, or weights it towards the centre. At one stage I found that after autofocusing, if I told the focuser to move out by 5 steps, the corner stars were improved without noticibly making the centre worse. This was a problem in automated runs as it wouldn't find optimum focus. Using a Gerd-Neumann tilt adjuster on the 6200 seemed to solve this problem. With a sensor that size a tilt adjuster is most likely needed. They are more expensive than standard push/pull adjusters as found on the cameras but easier to set up.

Alan 

[Lee_P]

I have a conclusion to this thread   After many more nights tweaking backfocus settings, I contacted FLO sent them some sample images. They sent me a replacement flattener to try, but no joy. I returned the WO ZS73, and I believe FLO may be sending it to Es Reid for tests. I bought an Askar FRA400, and everything with it works with minimal hassle. Thanks for the help offered in this thread, and FLO's excellent customer service should be noted too