CMOS on long focal lenght scopes - Forums

Topic starter

Dec 12, 2021 · Edited: Dec 12, 2021

Hi all,

I open this thread in order to have some suggestions/tips/opinions about a possible change of camera in my setup.

Currently I have a RC12 GSO (diameter 304mm, focal length 2432mm, f/() and I shoot with a KAF 16200 sensor in bin2, so pixels 12um.
The sampling is 1.01”/pixel, a good sampling considering the typical seeing of 2.5” that I have in the location of my observatory.
I’m quite satisfied with this combination and the only drawback I have is that the FOV of the KAF16200 is bigger than the native correct field of RC12, so usually I do a crop of the images.

Now on my portable setup (apo refractor 76mm) I started using an ASI 294MM and I’m well impressed by the cleanliness of the image and very low level of noise, so I’m asking if the replacement of the KAF16200 with a modern CMOS also on RC could give me benefit.

The idea would be to switch on RC12 to an IMX 571 sensor (ASI 2600MM or Moravian 26000 mono): due to small 3.76um pixels, I would have to work for sure in bin2 = 7.52um pixels.
This means that if I worked at full focal length of 2432mm I would have a sampling of 0.63”/px, which probably is too much for my typical seeing.
In this case I should reduce the focal, for example with a 0.8x reducer/corrector and then I would have 1945mm of focal length, with a sampling of 0.79”/pix, more manageable.
The possible problem in this last case could be to find a suitable reducer/corrector which does not introduce reflections in all the optical train (I had bad experience in the past).

In any case, I’ve not well understood if the classical way to calculate the image scale with CCD is “significant” also for CMOS.

What do you think?
Could the above change be a real improvement vs the current CCD sensor I have?
Has someone else already done a change CCD –> CMOS on telescopes with long focal length (>2000mm) and in case which solution have you chosen/which are your impressions?

Look forward to hear your suggestions/experiences!

Clear skies
Roberto

Helpful Insightful Respectful Engaging

andrea tasselli

Frequent contributor

Dec 12, 2021

Roberto Marinoni:
Could the above change be a real improvement vs the current CCD sensor I have?

In short? A lot. Better QE (a lot), way better read noise, way better thermal noise, no defects (that I know of). The only question is what are you waiting for ;)

Well, unless absolute photometric consistency is what you want...

Roberto Marinoni

Topic starter

Dec 12, 2021

Yes Andrea, I know that the modern CMOS sensors have a lot of advantages versus the old CCD sensors, but the meaning of my questions is how to deal with long focal lenght and small pixels.

andrea tasselli

Frequent contributor

Dec 12, 2021

The usual way: you bin them. In your case between bin 2x2 and bin 3x3 depending on the camera you want (assuming ASI here). I have a sampling of 0.78"/px with a lowly 8" I might guess I'd NOT be scared of going higher up with a 12" (which will probably do when I'll have my 12" in the field). And my seeing isn't better than yours.

John

Dec 12, 2021

you can always test it out on http://astronomy.tools/calculators/field_of_view/or use the http://astronomy.tools/calculators/ccd_suitability tools.

So with the field of view ,use it in imaging mode and you can play with it.

Torsten Mueller

Dec 13, 2021

Hi Roberto,
i was facing the same issue. Looking for a proper CMOS camera for my RC 12".
I decided to go with the ASI2400MC - 6um pixels (but OSC, so no binning) and 0.8x Reducer. The one from TS (2.5") works good, also with the full frame camera.

Of course, the sampling is far from ideal, but i found it ok (way better than with my ASI1600), and its super versatile also with my other scopes (400 to 2000mm fl)

And regarding you first question: Yes, CMOS is the way to go. I was a CCD supporter for a long time, but the (non existing) noise and the blazing fast readout with CMOS is so extremly helpful, that i wouldnt step back

Cheers
Torsten

Helpful

Roberto Marinoni

Topic starter

Dec 14, 2021

John:
you can always test it out on http://astronomy.tools/calculators/field_of_view/or use the http://astronomy.tools/calculators/ccd_suitability tools.

So with the field of view ,use it in imaging mode and you can play with it.

Thank you John.
I knew Astronomy Tools but I didn't notice the "ccd suitability tools"; by playing with it I see, as expected, that in my specific case the best would be between bin2x2 and bin3x3, as by the way also Andrea suggested above.
It makes me feel "strange" to have to work in bin3x3, but this should be....probably also bin2x2 could be ok, to be tested!

Roberto Marinoni

Topic starter

Dec 14, 2021

Torsten Mueller:
Hi Roberto,
i was facing the same issue. Looking for a proper CMOS camera for my RC 12".
I decided to go with the ASI2400MC - 6um pixels (but OSC, so no binning) and 0.8x Reducer. The one from TS (2.5") works good, also with the full frame camera.

Of course, the sampling is far from ideal, but i found it ok (way better than with my ASI1600), and its super versatile also with my other scopes (400 to 2000mm fl)

And regarding you first question: Yes, CMOS is the way to go. I was a CCD supporter for a long time, but the (non existing) noise and the blazing fast readout with CMOS is so extremly helpful, that i wouldnt step back

Cheers
Torsten

Thank you Toersten.
I had not good experience with TS 2.5" reducer due to reflections issues; I tested two models of the same reducer with the same result, so probably it is an issue related to my specific optical train.
Interesting your statement about the fact that you are far from ideal sampling but anyway you find it ok: this was exactly the point I issued, that is how much we have to rely on the formula of ideal sampling when we talk about CMOS. (ok, physic is physic but probably the result of the pure calculation is only indicative and the final word should be looking at the result of the final image).

Olaf Fritsche

Dec 14, 2021

Roberto Marinoni:
John:
you can always test it out on http://astronomy.tools/calculators/field_of_view/or use the http://astronomy.tools/calculators/ccd_suitability tools.

So with the field of view ,use it in imaging mode and you can play with it.

Thank you John.
I knew Astronomy Tools but I didn't notice the "ccd suitability tools"; by playing with it I see, as expected, that in my specific case the best would be between bin2x2 and bin3x3, as by the way also Andrea suggested above.
It makes me feel "strange" to have to work in bin3x3, but this should be....probably also bin2x2 could be ok, to be tested!

I am far from an expert, but I have read that there is no real binning in CMOS. The pixel data is only processed by the software. Therefore, I always work with binning 1x1 and combine the pixels if necessary later when processing the images.

Well Written Concise

andrea tasselli

Frequent contributor

Dec 14, 2021

Olaf Fritsche:
Roberto Marinoni:
John:
you can always test it out on http://astronomy.tools/calculators/field_of_view/or use the http://astronomy.tools/calculators/ccd_suitability tools.

So with the field of view ,use it in imaging mode and you can play with it.

Thank you John.
I knew Astronomy Tools but I didn't notice the "ccd suitability tools"; by playing with it I see, as expected, that in my specific case the best would be between bin2x2 and bin3x3, as by the way also Andrea suggested above.
It makes me feel "strange" to have to work in bin3x3, but this should be....probably also bin2x2 could be ok, to be tested!

I am far from an expert, but I have read that there is no real binning in CMOS. The pixel data is only processed by the software. Therefore, I always work with binning 1x1 and combine the pixels if necessary later when processing the images.

It depends on the CMOS sensor you're using. Some cameras perform the "binning" in hardware but it isn't the same as binning a CCD sensor.

Well Written

Torsten Mueller

Dec 14, 2021

andrea tasselli:
Olaf Fritsche:
Roberto Marinoni:
John:
you can always test it out on http://astronomy.tools/calculators/field_of_view/or use the http://astronomy.tools/calculators/ccd_suitability tools.

So with the field of view ,use it in imaging mode and you can play with it.

Thank you John.
I knew Astronomy Tools but I didn't notice the "ccd suitability tools"; by playing with it I see, as expected, that in my specific case the best would be between bin2x2 and bin3x3, as by the way also Andrea suggested above.
It makes me feel "strange" to have to work in bin3x3, but this should be....probably also bin2x2 could be ok, to be tested!

I am far from an expert, but I have read that there is no real binning in CMOS. The pixel data is only processed by the software. Therefore, I always work with binning 1x1 and combine the pixels if necessary later when processing the images.

It depends on the CMOS sensor you're using. Some cameras perform the "binning" in hardware but it isn't the same as binning a CCD sensor.

Yes, CMOS can (by design) only bin "later" in the processing. The consumer types in the software, the more scientific maybe also in the FPGA.

Torsten Mueller

Dec 14, 2021

Roberto Marinoni:
Torsten Mueller:
Hi Roberto,
i was facing the same issue. Looking for a proper CMOS camera for my RC 12".
I decided to go with the ASI2400MC - 6um pixels (but OSC, so no binning) and 0.8x Reducer. The one from TS (2.5") works good, also with the full frame camera.

Of course, the sampling is far from ideal, but i found it ok (way better than with my ASI1600), and its super versatile also with my other scopes (400 to 2000mm fl)

And regarding you first question: Yes, CMOS is the way to go. I was a CCD supporter for a long time, but the (non existing) noise and the blazing fast readout with CMOS is so extremly helpful, that i wouldnt step back

Cheers
Torsten

Thank you Toersten.
I had not good experience with TS 2.5" reducer due to reflections issues; I tested two models of the same reducer with the same result, so probably it is an issue related to my specific optical train.
Interesting your statement about the fact that you are far from ideal sampling but anyway you find it ok: this was exactly the point I issued, that is how much we have to rely on the formula of ideal sampling when we talk about CMOS. (ok, physic is physic but probably the result of the pure calculation is only indicative and the final word should be looking at the result of the final image).

Oh wow, thats not good the hear. I have compared the reducer only to the CCDT67 and i like the big one more.

Well... the sampling is 0.6"/px for my setup (6um pixel with 2000mm fl)
But for me, i love the performance CMOS so much, i would not trade it for the perfect pixel size (which may be around 9 um) with a "old" CCD.

If i could afford it, i would buy a scientific CMOS (GSENSE4040). But the price is... astronomical.
Cheers
Torsten

Bruce Donzanti

Dec 14, 2021

The numbers do suggest that you should do software binning or get a CMOS camera with a larger pixel size than the 3.76 used in the ASI2600 and ASI6200 cameras, but frankly, I have been using the ASI6200 recently on my C11" EdgeHD @ f/10 (2800 mm FL) with very good success (for exampleNGC 2174: The Monkey Head Nebula ( Bruce Donzanti ) - AstroBin and NGC 7635 (Bubble Nebula) in SHO ( Bruce Donzanti ) - AstroBin ). My seeing is average and I am in a Bortle 7 sky, but I mostly do narrowband. I guess it all depends on your acceptance as to what you think is a quality image, but I think: 1) the 3.76 um/px CMOS cameras work way better at long focal lengths than what they are given credit for and 2) I have found doing software binning doesn't seem to improve things that much. So, maybe I just have a lower threshold as to what I think is a nice image.

However, I would like to hear if anyone has gotten dramatic improvements in their images following software binning from a scope with a long focal length. Perhaps I can greatly improve on my current work.

Helpful Engaging Supportive

andrea tasselli

Frequent contributor

Dec 14, 2021 · Edited: Dec 15, 2021

An easy test to make is to take an image at full resolution and downsample it by a factor of 2x (3x). If the image doesn't improve in contrast and/or perceived noise when fully stretched then there is no point in binning. Personally I only binning when the subject is too faint to afford otherwise.

Helpful Concise

Torben van Hees

Dec 15, 2021

My sampling is at 0.39“/px with an ASI2600 and a 10“ RC. It works perfectly fine. Due to the low read noise, exposures do not have to be excessively long even with the small pixels. The good dynamic range makes the 10“ feasible without blowing out too many stars in LRGB. On very good nights, I‘ll leave the whole resolution, on worse nights I can just bin/reample in post. Two things to note: The small pixels will relentlessly reveal any collimation or spacing issue and the inherent tilt of many ZWO camera makes them difficult to use.

Helpful Insightful Respectful Concise

Torben van Hees

Dec 15, 2021

Bruce Donzanti:
The numbers do suggest that you should do software binning or get a CMOS camera with a larger pixel size than the 3.76 used in the ASI2600 and ASI6200 cameras, but frankly, I have been using the ASI6200 recently on my C11" EdgeHD @ f/10 (2800 mm FL) with very good success (for exampleNGC 2174: The Monkey Head Nebula ( Bruce Donzanti ) - AstroBin and NGC 7635 (Bubble Nebula) in SHO ( Bruce Donzanti ) - AstroBin ). My seeing is average and I am in a Bortle 7 sky, but I mostly do narrowband. I guess it all depends on your acceptance as to what you think is a quality image, but I think: 1) the 3.76 um/px CMOS cameras work way better at long focal lengths than what they are given credit for and 2) I have found doing software binning doesn't seem to improve things that much. So, maybe I just have a lower threshold as to what I think is a nice image.

However, I would like to hear if anyone has gotten dramatic improvements in their images following software binning from a scope with a long focal length. Perhaps I can greatly improve on my current work.

I‘ve seen quite some improvements with less/smoother noise in the background. If you do noise reduction in post anyway, it will make the difference harder to detect, but it allows to stretch more aggressively or tone done on the other noise reduction algorithms. On the other hand, I feel many noise reduction algorithms do a better job than downsampling with preserving detail. My test is actually the reverse of Andrea‘s: I downsample and upsample and compare to the original: If no detail is lost, it‘s worth to downsample for better SNR.

Björn Arnold

Dec 15, 2021

Hi all,

Just a quick add to the CMOS binning. Andrea has mentioned it and one should take a closer look at the specific sensor and camera implementation. For example, the ASI294 uses the IMX492 sensor which has a high conversion gain that only kicks in if gain is set to 120 and binning to 2x2 (to avoid confusion, physical resolution yields a 44MP image with 2,315um pixels which the current driver for the Mono shows as 1x1 binning). There are indeed some sensors that do some hardware binning, although not equal to CCD binning, it is improving the read noise on the physical level.

Cheers,
Björn

Helpful Insightful Respectful