How to focus manually a lens for which it was not designed.

[fivemack]

One of my other hobbies is astrophotography.

I have an optically-decent autofocus 70-200 f/2.8 lens, which I expect ought to be able to take pretty good pictures of star clusters.

The problem is that autofocus doesn't work on point sources; the manual focus ring is very loose and the difference between fifty-pixel-diameter blobs one side of focus and fifty-pixel-diameter blobs the other side of focus is no more than five degrees of motion. I can use live-view and zoom in as far as it goes, but that is not one-screen-pixel-per-sensor-pixel, still less the ten screen pixels per sensor pixel that I would like.

Is there some sort of clever contraption that could be made (I'm thinking of some kind of worm-and-wheel) to turn the focus ring in hundredths-of-a-degree increments, and if so is it an item of commerce? Is there some kind of optical contraption which could make accurate-infinity-focus discernible on something other than a star? There is no focus-lock mechanism that I can see.

I am reasonably confident that the odd-shaped stars are a result of optical aberrations rather than of a wobbly mounting, because they are very consistent in shape over several exposures of the same field.

[chalsall]

Originally Posted by fivemack

One of my other hobbies is astrophotography.

Is there some sort of clever contraption that could be made (I'm thinking of some kind of worm-and-wheel) to turn the focus ring in hundredths-of-a-degree increments, and if so is it an item of commerce? Is there some kind of optical contraption which could make accurate-infinity-focus discernible on something other than a star? There is no focus-lock mechanism that I can see.

Calm down little chicken.

Problems can be solved.

[kladner]

Originally Posted by fivemack

One of my other hobbies is astrophotography.

I have an optically-decent autofocus 70-200 f/2.8 lens, which I expect ought to be able to take pretty good pictures of star clusters.

The problem is that autofocus doesn't work on point sources; the manual focus ring is very loose and the difference between fifty-pixel-diameter blobs one side of focus and fifty-pixel-diameter blobs the other side of focus is no more than five degrees of motion.

.....

The stars are not at infinity focus? I take it this is not a DSLR, and Live View is your only option. I can imagine a pinch roller with a gear train to reduce the motion of the ring, but I don't know what kind of off-the-shelf parts are out there. Can you do anything to induce a gentle amount of drag on the ring so that it does not jump around so fast?

Check these out-
https://www.bhphotovideo.com/c/buy/L...20Gear%20Rings
https://www.bhphotovideo.com/c/buy/F...0Drive%20Gears
https://en.wikipedia.org/wiki/Coma_(optics)

[Uncwilly]

Try making a reducing gear set-up with pulleys first. You might be able to make a 2 stage system with readily available items.

[a1call]

Part of the lack of focus in any existing collector based optics is the scattering of light at the edge of the collector lens or mirror. This limits how sharp an image can be focused. The problem is reduced, the larger the collector gets, but is not eliminated and there are practical precision and weight limitations on how large a collector can be fabricated,
* Is there a general solution?
** Yes, but I haven't patented it yet.

[xilman]

Originally Posted by fivemack

I am reasonably confident that the odd-shaped stars are a result of optical aberrations rather than of a wobbly mounting, because they are very consistent in shape over several exposures of the same field.

Your image shows classic coma. The point spread function (PSF in the trade) is comet shaped, hence the name. There might also be some spherical aberration but that's hard to tell given the amount of coma displayed. The simple out-of-focus PSF is a flat disk.

Fix the coma first, if possible. It may not be possible with your lens. How you fix it on your particular kit is left as an exercise, but it's generally caused by one or more optical surfaces being tilted with respect to the optical axis.

[xilman]

Originally Posted by xilman

Fix the coma first, if possible.

If you can't correct the optics you could try correcting the image.

Determine the PSF around the field of view by measuring the shape of star images, which should be bright enough to show the intensity distribution but not bright enough to saturate the detectors. Given that information it is possible to deconvolve your subsequent images, though there are many pitfalls arising from modelling (or not) the noise in the image. If you are lucky the PSF will be location independent and you can use relatively cheap FFT-based algorithms.

This post and its predecessor should be enough to get you started but I can point you at any number of sources of further information.

[a1call]

Perhaps a dumb question, but just to eliminate the possibilities, are you using a clock drive?
https://en.m.wikipedia.org/wiki/Clock_drive

[xilman]

Originally Posted by a1call

Perhaps a dumb question, but just to eliminate the possibilities, are you using a clock drive?
https://en.m.wikipedia.org/wiki/Clock_drive

It doesn't really matter right now. The principal aberration is coma, not trailing, the PSF of which is a line segment. Most anywhere a reasonable distance from the celestial poles it can be treated as a straight line for simplicity, rather than an arc of a circle whose curvature varies over the field.

[fivemack]

Originally Posted by kladner

The stars are not at infinity focus? I take it this is not a DSLR

This is a DSLR (Nikon D90). The stars are by definition at infinity focus, but the point at which the focus ring on the lens stops rotating is a couple of degrees beyond the infinity focus point - I would not be surprised were the focus ring just input to an angle-encoder, with a motor inside the lens moving the glass around.

In answer to other questions: I am using a clock drive (an iOptron SkyTracker v2, which I'm very happy with). The image scale with this setup is about six arc-seconds per pixel, and that's a 30-second exposure so the sky would have moved 450 arc-seconds were it not for the clock drive.

I am a bit worried about the coma: the cutout I posted was from essentially the centre of the field, and the impression I had was that coma was an unavoidable aberration at the edge of the field from a wide-field lens, and a sign of a very unhappy lens if you had it in the middle of the image.

PSF measurement and deconvolution seems like a fun rabbit-hole to go down, and I've got reasonable experience of it from my time in X-ray crystallography; currently all I'm doing processing-wise is fitting and subtracting the background noise, and stacking images once I have images that are reasonably usable.

[xilman]

Originally Posted by fivemack

I am a bit worried about the coma: the cutout I posted was from essentially the centre of the field, and the impression I had was that coma was an unavoidable aberration at the edge of the field from a wide-field lens, and a sign of a very unhappy lens if you had it in the middle of the image.

PSF measurement and deconvolution seems like a fun rabbit-hole to go down, and I've got reasonable experience of it from my time in X-ray crystallography; currently all I'm doing processing-wise is fitting and subtracting the background noise, and stacking images once I have images that are reasonably usable.

Could you make the full image available please? I'd like to take a look at it to see what else can be gleaned. Posting here, DropBox or email would all work for me.

I've also tried to identify the bright star but can't quite place it as there aren't enough field stars to make it obvious. Albireo perhaps?

In case folk haven't guessed by now, I've had quite a bit of experience in both amateur astronomy and image processing. Never got the hang of X-ray crystallography and related inverse phaseless FFT reconstruction problems during my time as a student so you're one up on me there.