Kevin_A On my fast scopes 1s is ok but not my f5.6 scopes.

It is a consequence of how ZWO "autofocuses." It uses a fixed exposure value (1 sec, in your example, at some fixed gain value that you have chosen).

It then defocuses to get HFD around 6 to 15 range. When you do that, the photons from a star are spread over many pixels, like 100 or more pixels. I.e, if is like exposing for 1/100 seconds. If the brightest stars are too dim at 10 millisecond exposure, then ASIAIR fails on the very first attempt to get a point on the curve.

This is why narrowband filters also need longer exposures.

ZWO should have changed exposure value as the star HFD gets larger and smaller (i.e., as the same number of photons that are spread over different number of pixels change). Again, it shows the complete uneducated approach from the folks at ZWO (yes, I know because I exchange email with them; that is exactly why I would not touch their mounts with a 10 foot pole).

The other consequence of ZWO's approach is that to keep a star unsaturated (to compute HFD) when the exposure value is kept constant, is that they have to use different stars at different EAF (thus HFD). Just imagine what that does to an OTA that does not have perfect field flatness. Remember the method I use to backfocus? Based on the fact that focus is different for different parts of the frame when the image plane is not flat. And that is what ZWO's autofocus is doing -- it is picking different stars and thus different HFD as it goes through autofocus.

Here is a secret (I don't think anyone else will tell you)-- ASIAIR autofocus is actually quite accurate when the field is flat! Just excruciatingly, and unnessarily slow. When the backfocus is not perfect (i.e. HFD is more than 10% different across the frame), ASIAIR fails miserably since can't get an accurate curve -- the points on the curve comes from different stars and they have different focus when the field is not flat. But that is also where a high quality Bahtinov mask will show different focus across the frame too.

It is just that when we use the Bahtinov mask, we focus on one fixed star! (Usually 1/3 of the way out from optical axis.) If ZWO had used different exposure values by adjusting exposure time and gain, and also use a single fixed star to make its curve, it would be just as accurate as a Bahtinov mask, if not more so (without additional tools, it is not easy to find the perfect focus of a Bahtinov spike, either). I have even had the Sigma 40 and the Samyang 135 produce better ASIAIR autofocus than simply casually judging the Bahtinov mask without using rulers, etc to measure the spikes. But the backfocus must be right. The different stars must be perfectly round.

Personally, I would pick the brightest star in the field that is between 25% to 50% away from the optical axis, and then change exposure values on it to keep the fixed star from saturating as it gets in and out of focus. We know that no star (except the Sun) will saturate a sensor at 1 milliisecond type exposures, while picking the brightest one will allow the star to have enough SNR when it is way out for focus, without having to resort to exposures that are longer than 2 or 3 seconds. But science and critical thinking is not ZWO's strong point.

So, for now, unless you have a perfectly flat field, (that is why the Askar backfocus adjuster is an essential tool), a Bahtinov mask easily beats out the ASIAIR. Again, pity the people in this hobby who depend on Easy as 1,2,3 instead of doing some critical thinking. I have always said that an experienced person will not have problems with the ASIAIR. It is the TikTokers who have problems with the ASIAIR. The experienced user knows what part of ASIAIR (and when) does not work, and know what work-arounds to use (like using a Bahtinov mask when their OTA is not flat). Note that TikTokers who depend on "55mm backfocus" will also not have correct backfocus, and therefore poor autofocus. BTW, this is why the SeeStar images are so out of focus; they advertise "APO," but don't tell you that the glass lacks a flattener. The gullible will fall for the techincal word salad everytime. I'll bet 99% of the TikTokers don't know the difference between "APO" and flatness of the image plane. Not sure 99% of ZWO developers know either.

Chen

P.S. 10:40am: the machined parts for the Nikon tilt adjustment is here.

Kevin_A I would never get a seestar or the like… it is a money grab toy!

Yep. Sure fools the gullible TikTokers. I refuse to buy one even to test.

I did order another Rokinon 135 to see how it compares on stars to my old one. If it is better I will use the new one for astro and the old for daytime photography as it is very sharp.

Amazing lens for daytime. The iris is a big disappointment for nighttime, though.

In the photography world, there are people who prefer lenses that make starbursts more prominent. You can see that as one of the points of many of the reviewers. I wonder if Samyang was aiming for that, rather than simply cheapening out. It can't be that difficult to geometrically design an iris that produces a smooth (small first derivatives) transition between the iris blades -- from Fourier Transform, that will also produce the least diffraction spikes. It seems that Sigma did try to reduce spikes with the 40mm. And for what we know, Samyang may be purposely emphasing them (they probably did not target astrophotographers).

But then, the Sigma 40mm was supposed to be a show of force by the Sigma boys. Kinda like the the Lockheed skunkworks projects. That is probably why it ended up weighing twice that of other 40mm lenses. I am one of the people willing to pay 95% more for the last 5% of perfprmance, so the Sigma 40mm is really ideal for me.

I have a Sigma 135 ART that had performed dissapointedly, but now that I know how to get rid of the bayonet-to-bayonet fitting, I may give it another try -- and this time also using the Askar backfocus adjuster (a little tight, because it is a 44mm Canon flange instead of the 46.5mm Nikon flange). I am also going to check out using its original iris. I had used a front aperture mask when I tested it a year or two ago.

i have been verifying focus with a mask recently on my faster rigs.

Just check the tilt of the lens. If the tilt is small, I think what you will find is that ASIAIR "autofocus" is actually quite good (I judged by the HFD of stars after one finishes Bahtinov, and when one finishes ASIAIR autofocus).

But the tilt has to be very small, or the different stars the ASIAIR pick will have different focus positions.

Again, it is one of these things that once you understand the underlying principles, the ASIAIR can be made to work. A quick way to check tilt without download the image to use Siril is to just use the ASIAIR star detect tools. If all the HFD numbers of the stars are within 0.5% of one another (e.g., HFD of every star is 2.5 +/- 0.02) all across the frame, I think you will find ASIAIR autofcous to be really good. ASIAIR will still pick different stars, but all the stars will be on the same flat image plane. So, basically work around the crummy algorithm by giving the ASIAIR stars that all have the same focus. Now tell me how a Easy as 1,2,3 person is supposed to know this (or even know how to tune backfocus -- the Tiktokers have a hard enough time trying to get the (completele wrong) 55mm advise from ZWO).

Chen

Kevin_A

Ready for tonight, if the clouds are thin enough.

The M3 push screws are 4mm long. The plate is 3mm, and there is a 0.5mm blind hole to center the push screw on the other plate (ZWO did not include that blind hole in their cameras), so I have 0.5mm worth of adjustment for now. Will swap with 5mm screws if needed.

Using my own M54 mounting gives me backfocus freedom to play these games. Right now, the fit is just right. Nikon flange distance is 46.5mm. The camera is 17.5mm, so there is 29mm leftover. I have three plates, 2mm (right on top of the Sigma Nikon bayonet, and these two 3mm push pull plates, and there is a 3mm plate (to convert to M54 threads) from PreciseParts (black plate right next to the push-pull screws). So, 2+3+3+3+18 (Askar backfocus adjuster) = ta da, 29mm. Now, the Askar can be adjusted down to 16mm, so I actually have something close to 2mm worth I can play with. I did add a 0.1mm shim to get a more convenient camera angle, but PreciseParts also claim their plate is 2.9mm thick instead of 3mm.

The Sigma EOS flange is only 44mm. But I also bypass its bayonet completely to recover about that amount, so a tilt plate will work for the Samyang 135 too. Sunday night project to draw the Samyang/EOS version for the web machine shop.

I'll need lots of electrical tape, though, by the time I am done :-). Right now, I am wondering if I should use Tuck tape instead of the push screws, once I have the result measured. I'll bet Tuck tape is more stable than the push screws, even with those push screw blind holes.

Chen

Kevin_A No red fringing but coma in two corners, round in 1 corner and elongated in one.

Are you sure this is not caused by tilt from the bayonet mount?

So far, I am finding that 20º or 30º of a turn of a M3 screw will change the tilt enough to affect the shape of corner stars significantly for the 40mm lens (not sure about effect on 135mm lens yet -- waiting for machine shop). M3 screw pitch is 0.5mm, so 1/15th of one turn would be 0.03 mm (much thinner than Tuck tape can correct).

Your problem might not be the glass, but the bayonet flange itself (plastic does not help).

Since you have no way to adjust the lens tilt, you can try experimenting with the ASI2600's tilt plate (but adding back the 6 missing push screws). You should not need to use grub screws since they are a pain to adjust; just use button heads, since they will clear the camera adapters.

For times when I need to adjust camera tilt, I actually bought tthe new ZWO tilit plate pair for the ASI2600 (90mm diameter camera) which you can adjust even when there is a filter wheel blocking the tilt adjustment screws. I also have the Baader that adjusts with edge screws, but it is humongous (68mm thread).

https://agenaastro.com/zwo-m54-tilter-adapter-for-select-zwo-cameras.html
https://agenaastro.com/baader-m68-tilter-wth-9-50-10-25mm-optical-length-and-0-1-tilt-m68-tlt-2458170.html

Tilt definitely screw things up. It has to be flat even before you can even start adjusting backfocus, otherwise the backfocus adjustment will be very confusing (one corner would be corectly backfocus adjusted, while a different corner is not and produces coma).

I would use a tool like Siril and fool with tilt until it is flat before fooling with backfocus. In fact, check the lens with your existing setting to see if one corner or side is off. It need to be more flat than what I posted earlier. I got it to this last night,before clouds came in.

But this is not good enough yet to get good corner stars (at least for me).

With the 40mm, I am seeing backfocus of the order of 0.025mm to affect the corner stars. And this is just an APS-C camera. Again, no way to use Tuck tape. Since I can't leave the Askar backfocus adjuster in a DSLR lens (if I do, I cannot add a camera angle adjuster), and Tuck tape can't give me enough precision (unless by luck), I will have to think of something.

Right now, I am thinking of washers, Tuck tape plus a third adjustment plate. The Blue Fireball washers (multiples of 0.1mm) and Tuck tape (multiples of 0.076mm for rough agjustments, and a "tilt plate" that is used as a backfocus adjuster fot 0.025 mm type adjustments (need it!).

Strictly speaking, the existing tilt plate can work as a fine backfocus adjuster (by moving the screws all by the same amount), but I rather have a different set of screws to decouple the backfocus adjustement from the tilt adjustment, since tilt is so darn important.

Time to move away from bayonets?

BTW, I have started to think about a fixture to measure camera tilt, by using one of my unused OAG to point a laser pen towards to sensor :-).

The felt lined Borg 7065 is 60mm in diameter and is perfect to allow a M54 tube to rotate in it (or go buy a Pegasus camera rotator and do it all from software [INDIGO]). I.e., point two cameras at each other (one is DUT, and the other to see the results), with an OAG in between.

Speaking of Pegasus, did you see the NYX 88 announcement? Still way heavier mount than a RST-135, though. I am still looking for a baby-RST for use with short camera lenses, i.e., a "AZ-GTI" in strain wave form -- preferably with real Harmonic Drive [tm] gears instead of Chinese copycat gears, and Renishaw encoder (instead of Chinese encoders).

Chen

Kevin_A if it were tilt I would expect different elongations but what I am seeing is very bad long fuzzy comet coma in two kitty corners and not just on one side or one corner.

I was also getting terrible fuzzy (coma filled) stars on some corners before I started to focus [sic] on getting tilt adjusted to 0.02-ish mm.

For a quick look, take the lens out of focus so you get large Airy dics (HFD perhaps over 15). Are the corner disks all circular? If not, it could be a tilt problem. If none of the corners is circuar, it is a backfocus problem. So, at least get the backfocus adjusted first so that at least one corner has a circular disk.

If the non-circular disks turn circular when you change backfocus, it is a tilt problem. I.e., as you change backfocus, the Airy disks of different corners become circular. This is just a 1 minute exercise if you have the Askar backfocus adjuster.

BTW, the 40mm Sigma is not perfect. When tilt, backfocus and focus are adjusted properly (so that the corner stars are tack sharp), you can see a small red spot deviate (x-y) from the "white" spot. Tiny amount, perhaps less than a pixel or two, but it is visible. You can see this in some of the Takahashi spot diagrams, by the way.

When slightly defocused, this looks like coma, but it is just a short red tail. Not 100% APO from red to blue.

I have decided to just use the tilt screws to create a +/- 0.05mm offset for backfocus, by the way (use the Blue Fireball metal shims to get down to 0.1mm). I has considered using a 54mm stop plate, but am afraid the stop plate will itself introduce 0.02mm type errors.

However, I will create a larger tilt plate, so that the screws will clear the 90mm body of the camera. I.e. place the tilt screws 90/2mm + 1.5mm from the center of the tilt plate. A 95mm or 96mm plate should work. Then stick in 4 (non-ball head) long 2mm L hex wrenches into the pull screws, and strop the four to the body of the camera. The ball head hex wrenches do not provide very precise angles, clearing the camera body (by half the diameter of the wrench -- that 1.5mm) will allow the use of non-ball headed wrenches.

I may have to buy an additional tilt plate as I want to keep the 2600 as is as it is fine on all my other scopes.

If you do, get the new one, which is a pair of plates. This lets you adjust from the back of the camera (like what I am planning for my next set of tilt plates). You still end up with 5mm backfocus consumption, same as the original.

https://www.zwoastro.com/product/m54-tilter-2/

Like I said, I think they send the crap to Canada as my last Rokinon buy took 3 copies to get a decent one so I am not surprised by another sub par copy.

You may just have had 2 copies with a bad tilt. Not bad glass. If so, you should be able to correct by just correcting tilt. I am not kidding when I say that the backfocus is sensitive to even 0.025mm correction.

Now that I understand the problem better, I m going to one day play with the Sigma 135/1.4 ART again. I think now that it was just a tilt problem.

I don't think it is a Canadian problem. Heck, my Samyang and Rokinon came from some Amazon third party that sells candy, even though the web page says ships and sold by Amazon. The Sigma came from a NY electronics store that sells compuer games.

Chen

Kevin_A

Oh, BTW, I can confirm that the red glitch that I saw when I used a Hoya UI/IR cut and a didymium filter is not caused by the Hoya, but by this particular didymium filter:

https://www.amazon.com/gp/product/B07NP774RH/

The didymium filter from Maven (cheaper than above) produced no red globs and glitches with the Hoya:

https://mavenfilters.com/product/night-sky-filter/

I then tried a slightly cheaper Haida UV/IR cut filter than the Hoya with the Maven, and also no red blob.

So, I am back to using 82mm UV/IR cut and didymium filters now.

I did find a B&W 82mm filter that cuts both UV and IR. I may get it if I play more with wide angle lenses. B&W are my go-to filters for terrestial photography. Using a Chinese Haida filter on a Canon lens is like putting an SVBONY UV/IR cut filter on an Takahashi. In both cases, there is no point screwing up thousands bucks worth of optics by saving a few bucks on the filter.

Chen

Kevin_A

Another by the way, the Sigma 40mm has very little light falloff over an APS-C frame at the f/2.4 that I am using.

Looking at a flat capture (cooled to +15ºC), I see Max ADU of 37600 and a min ADU of 31700 with mean ADU at 35400.

I have turned gain to -2.5 dB to get more dynamic range. Heck, I can get almost one more bit of dynamic range by using a longer exposure than 0.2 sec.

Flat dark has Min ADU of 400-ish, and max ADU of around 1000.

Chen

Kevin_A

Hmm, I think Siril's HFD/FWHM is confused by background gradient. It is showing tilt when visual tilt is quite small. I need to look more into this.

This is what I now have with the aberration inspector with the Sigma 40 at f/2.4 (rotated 180 degrees). Image is 17 subframes of 180 second exposures at 10 dB gain; scaled down for JPEG to fit forum, no sharpening in post processing, so what you see are the real stars. Slight contrast change using macos Preview:

Sadr is at center. Almost the entire Northen Cross part of Cygnus is in frame. Albeiro is just barely outside the bottom edge. Camera angle not perfectly north-south since I can't include the camera angle adjuster.

I might just get the old Rokinon adjusted to the point it is ok

Yeah, I am going to leave the camera lenses for perhaps a week too, to get the large tilt plate made before I do more on the Sigma 40 and Samyang 135. Otherwise, I Iwill end up doing it over again.

Chen