Which CMOS camera to suit both long and short focal lengths?

[kens]

4 hours ago, Adam J said:

The thing with binning the ASI1600mm at 1600mm FL  is that you could get a similar result by using it at 800mm without binning it and benefit from a much lighter (potentially faster) easier to Handel scope. So something like a F5 / F6, 6 inch Newtonian. You also would not suffer from the binning losses due to the additional read noise per (virtual) pixel.

Am I missing something? As far as I am aware a 6 inch aperture is easily able to resolve down to ~1arcsec / pixel...resolution in the final image is likely limited by seeing and so in fact a faster shorter scope might even help because you could image something like a galaxy by using the v.short exposure method. 

 

There is no read noise advantage from binning a CMOS camera. That only works with CCD cameras but the CMOS architecture is different. Hardware binning in a CMOS camera happens in the digital domain (after the ADC) whereas in a CCD the charges on the binned photoreceptors are combined.

So with CMOS you can happily oversample and bin (aka downsample) in post processing.

[Adam J]

1 hour ago, coatesg said:

For a given camera, I think the big Newt still wins out (mechanical/optical considerations aside). For a given focal length, the quicker scope has higher SN ratio and that would be the larger scope (aperture wins here).

NxN Binning with the cmos camera, by theory should give ~ an N^1/2 increase in SNR again unbinned, but comparing different scopes binned Vs unbinned probably won't hold in practise.

It would be much less demanding on the mount using a shorter scope, agreed. Not sure I understand what you mean by binning losses though, unless talking about total number of pixels?

None of this means either situation won't work of course - there are plenty of examples out there to show either situation gives great results!

I am not going to get into another debate about f-ratio and image scale, but SNR certainly does not scale with aperture.

What I mean by binning losses is quite simple, to get a image scale of 1 arcsec per pixel from a 1600mm FL scope you need to bin the ASI1600mm 2x2 to give you a single virtual pixel with (read noise = 3.2e = (4x1.6e^2)^0.5) that means that you have more read noise per pixel than if you achieve the same pixel scale by using a 800mm focal length scope as in this case you will have only 1.6e noise per pixel, so you have increased the noise per arcsec^2 by a factor of two.

Beyond this at the same image scale but fixed F-ratio the photon flux will be higher in the larger scope as you noted....but we are not talking about the same F-ratio here as per the above we are talking about F8 in the case of the RC vs  a 6 inch F5.3 Newtonian (in reality it would be a F5 @ 750mm just making the math easy). So thats a 200mm mirror vs a 150mm mirror and so we are collecting 1.78x more photons per pixel with the 8" RC as with the 6" Newtonian. Yet as per the math above the 8" RC has 2x the noise. Hence the 6 inch F5.3 Newtonian has 12% better S/N in comparison to the 200mm F8 RC and not the other way around as you seem to suggest. 

Adam

 

[Allinthehead]

Adam just be aware that the Asi 1600 gives horrible  diffraction artifacts around very bright stars. It was quite severe with my ed 80 and barely noticeable with my Star 71. I've yet to try it with the tak. You might want to check if the qhy or atik versions have the same issue.

Zwo do another 4/3 sensor which may be worth looking at. My advice would be to take your time in making your decision after all astro dark won't be with us for a while yet. 

Here's an example of the issue

If you decide to go ahead with it here's the soul nebula with the 80ed and 1600 to give you an idea of the fov

[Adam J]

44 minutes ago, kens said:

There is no read noise advantage from binning a CMOS camera. That only works with CCD cameras but the CMOS architecture is different. Hardware binning in a CMOS camera happens in the digital domain (after the ADC) whereas in a CCD the charges on the binned photoreceptors are combined.

So with CMOS you can happily oversample and bin (aka downsample) in post processing.

Yes it should be done in post processing not in firmware, also in strict terms the SNR remains the same following software binning, what you are actually doing is trading resolution for perceived image quality...which is in fact the same net effect as improving SNR via true CCD binning... but strictly speaking you cant call it that as strictly it not the same thing. Also of note you dont actually get anything like 4x improvement in SNR from Binning a CCD 2x2 as there are other sources of noise (read down stream and up stream noise), in the end its something more like 2.5x to 3x. 

[tooth_dr]

10 minutes ago, Allinthehead said:

Adam just be aware that the Asi 1600 gives horrible  diffraction artifacts around very bright stars. It was quite severe with my ed 80 and barely noticeable with my Star 71. I've yet to try it with the tak. You might want to check if the qhy or atik versions have the same issue.

Zwo do another 4/3 sensor which may be worth looking at. My advice would be to take your time in making your decision after all astro dark won't be with us for a while yet. 

Here's an example of the issue

If you decide to go ahead with it here's the soul nebula with the 80ed and 1600 to give you an idea of the fov

Beautiful images Richard. Can you tell me what other 4/3 sensor it is? The diffraction pattern is weird(!)

I did not realise there were Atik and QHY version. 

[Adam J]

21 minutes ago, Allinthehead said:

Adam just be aware that the Asi 1600 gives horrible  diffraction artifacts around very bright stars. It was quite severe with my ed 80 and barely noticeable with my Star 71. I've yet to try it with the tak. You might want to check if the qhy or atik versions have the same issue.

Zwo do another 4/3 sensor which may be worth looking at. My advice would be to take your time in making your decision after all astro dark won't be with us for a while yet. 

Here's an example of the issue

This is something I thought long and hard about when I chose my ASI1600mm pro and to be honest...in the end I decided that in balance the chip offered many other advantages (sensitivity and size) that were more important to me that the diffraction effect.

Also interestingly I image with Newtonian telescopes and it does not seem to present as badly as i have seen in some refactors.

I think its a combination of coatings and F-ratio that make the effect more or less apparent in addition to the properties of the star halo, hence better optics in your Star 71 gives tighter stars than the SW ED80 and so the focused star halo is spread over less micro lenses on the chip surface and the diffraction patter is weaker (like having a lower number of slits on a diffraction grating). All these factors make it hard to directly compare cameras across setups, however, unfortunately from my research its a problem inherent in the chip itself (due to it lacking anti-reflection coatings) and so QHY and ATIK both have exactly the same issues as the ASI1600mm pro. Narrower filters can also help as they in effect make the bright stars less bright and so less given to show the diffraction effect. 

In the end its personal choice...but its better to be aware if the issue going in then find out about if after purchase.

The only other 3/4 sensor ZWO do is the ASI294c and it does not come in mono only OSC.

[vlaiv]

13 minutes ago, Adam J said:

I am not going to get into another debate about f-ratio and image scale, but SNR certainly does not scale with aperture.

What I mean by binning losses is quite simple, to get a image scale of 1 arcsec per pixel from a 1600mm FL scope you need to bin the ASI1600mm 2x2 to give you a single virtual pixel with (read noise = 3.2e = (4x1.6e^2)^0.5) that means that you have more read noise per pixel than if you achieve the same pixel scale by using a 800mm focal length scope as in this case you will have only 1.6e noise per pixel, so you have increased the noise per arcsec^2 by a factor of two.

Beyond this at the same image scale but fixed F-ratio the photon flux will be higher in the larger scope as you noted....but we are not talking about the same F-ratio here as per the above we are talking about F8 in the case of the RC vs  a 6 inch F5.3 Newtonian (in reality it would be a F5 @ 750mm just making the math easy). So thats a 200mm mirror vs a 150mm mirror and so we are collecting 1.78x more photons per pixel with the 8" RC as with the 6" Newtonian. Yet as per the math above the 8" RC has 2x the noise. Hence the 6 inch F5.3 Newtonian has 12% better S/N in comparison to the 200mm F8 RC and not the other way around as you seem to suggest. 

Adam

 

It's not as simple as that. Lets make math easy and stick to x1.78 more light and same resolution (RC8" binned vs 6" native). There are many sources of noise in the image - one of them being shot noise.

Let's take for example target that emits 1e/s per pixel. Now we take 1 minute exposure. 6" will gather 60e - shot noise will be ~7.746e and add 1.6e read noise we get 7.93e noise. SNR in this case will be 60/7.93 = 7.5662.

RC8" on the other hand will gather 60 x 1.78 = 106.8e of signal. Its shot noise will be ~10.334e and combined with 3.4e noise that will give total noise of:  ~10.88e. So in this case SNR will be 9.817 or x1.3 higher than with 6".

Again this is not full math for SNR, as with software binning - you are adding up dark current and LP, both of which have their associated noise, so it SNR gain of 8" binned over 6" native will depend on many factors (exposure length included - with binned you can actually go longer without saturating pixels as each pixel gets 1/4 of signal compared to native - this helps with overall SNR - when read noise is dominant term - go for longer exposure).

Btw, on subject of those strange diffraction effects - I suspect a lot is in play, and I haven't seen it that often - I believe it has to do with type of NB filters used, their distance to sensor, light beam F/ratio. All of those contribute to some strange interference that produces those effects.

[Allinthehead]

14 minutes ago, tooth_dr said:

Beautiful images Richard. Can you tell me what other 4/3 sensor it is? The diffraction pattern is weird(!)

I did not realise there were Atik and QHY version. 

Sorry i was thinking of the 294 which is osc only i believe. Yes the diffraction is very odd but a well known issue with this camera. As Adam j said if you weigh it up it might just be an inconvenience rather than a deal breaker. You can probably get around it by taking some very short subs. It's only a problem around the brightest of stars.

[Allinthehead]

Here's the same star with the Star 71 noticeable but not as bad.

[tooth_dr]

3 minutes ago, Allinthehead said:

Here's the same star with the Star 71 noticeable but not as bad.

Thanks Richard. Your images alone make me want this camera regardless. I’m currently waiting to hear back from a supplier, and have approval from the missus, but I won’t rush into it. 

[Adam J]

49 minutes ago, vlaiv said:

It's not as simple as that. Lets make math easy and stick to x1.78 more light and same resolution (RC8" binned vs 6" native). There are many sources of noise in the image - one of them being shot noise.

Let's take for example target that emits 1e/s per pixel. Now we take 1 minute exposure. 6" will gather 60e - shot noise will be ~7.746e and add 1.6e read noise we get 7.93e noise. SNR in this case will be 60/7.93 = 7.5662.

RC8" on the other hand will gather 60 x 1.78 = 106.8e of signal. Its shot noise will be ~10.334e and combined with 3.4e noise that will give total noise of:  ~10.88e. So in this case SNR will be 9.817 or x1.3 higher than with 6".

Again this is not full math for SNR, as with software binning - you are adding up dark current and LP, both of which have their associated noise, so it SNR gain of 8" binned over 6" native will depend on many factors (exposure length included - with binned you can actually go longer without saturating pixels as each pixel gets 1/4 of signal compared to native - this helps with overall SNR - when read noise is dominant term - go for longer exposure).

Btw, on subject of those strange diffraction effects - I suspect a lot is in play, and I haven't seen it that often - I believe it has to do with type of NB filters used, their distance to sensor, light beam F/ratio. All of those contribute to some strange interference that produces those effects.

Would be good if you could share the formula you used to calculate shot noise.

However, as you say the exposure length is a big factor in determining the dominant noise factor....in fact you should normally aim for sky noise to dominate. 1e/s is also a very large signal as I understand it. Would that not result in a SNR of over 20:1 in only 4 mins or so of intergration?

 

 

[vlaiv]

1 minute ago, Adam J said:

Would be good if you could share the formula you used to calculate shot noise.

However, as you say the exposure length is a big factor in determining the dominant noise factor....in fact you should normally aim for sky noise to dominate. 1e/s is also a very large signal as I understand it. Would that not result in a SNR of over 20:1 in only 4 mins or so? Would the read noise not be a more dominant factor for a weaker signal?

 

 

Formula is really simple - Possion noise - signal^0.5, while noise addition is that of linearly independent vectors (a^2+b^2+....)^0.5

1e/s/pixel at 1"/pixel is approximately 20mag source (per arc second squared in Lum/V band). So it is not faint, but it's not bright either - my LP is close to 18 mag.

I have a spreadsheet done for just such purpose. If you let me know specs of 6" scope (CO, FL, reflectivity of mirrors), LP, target brightness in arc seconds squared (magnitude) - I can run comparison on ASI1600 vs my RC8" and post results - for single sub and stack (well stack one will be just SNR multiplied by square root of number of subs).

[coatesg]

10 hours ago, Adam J said:

Beyond this at the same image scale but fixed F-ratio the photon flux will be higher in the larger scope as you noted....but we are not talking about the same F-ratio here as per the above we are talking about F8 in the case of the RC vs  a 6 inch F5.3 Newtonian (in reality it would be a F5 @ 750mm just making the math easy). So thats a 200mm mirror vs a 150mm mirror 

 

I was referring to the 12" newt in the post above the one you replied to (sorry if that wasn't clear), so we might be at cross-purposes - so comparing say 200mm at f6 (or 150 at f8) vs 300mm at f4.

And I think we're all agreed that in non-dark skies, read noise should be dominated by sky background shot noise - it really determines the exposure needed to do that (so longer when binned).    

[tooth_dr]

1 hour ago, coatesg said:

I was referring to the 12" newt in the post above the one you replied to (sorry if that wasn't clear), so we might be at cross-purposes - so comparing say 200mm at f6 (or 150 at f8) vs 300mm at f4.

And I think we're all agreed that in non-dark skies, read noise should be dominated by sky background shot noise - it really determines the exposure needed to do that (so longer when binned).    

I am in a reasonably dark site, does this influence my choice of CCD or CMOS?

[coatesg]

I wouldn't let it be the final say - both will work fine (and just being in a dark site puts you at a huge advantage!).

In theory, the CMOS cam with lower read noise will require shorter exposures to dominate it than ccd in all situations - in truly dark sites (or narrowband) those exposures might be quite long which taxes other aspects like tracking/guiding performance, flexture, etc. Either way, you'll need roughly equivalent total exposures.

[tooth_dr]

25 minutes ago, coatesg said:

I wouldn't let it be the final say - both will work fine (and just being in a dark site puts you at a huge advantage!).

In theory, the CMOS cam with lower read noise will require shorter exposures to dominate it than ccd in all situations - in truly dark sites (or narrowband) those exposures might be quite long which taxes other aspects like tracking/guiding performance, flexture, etc. Either way, you'll need roughly equivalent total exposures.

Ok thank-you.  I've been guiding ok with my dual rig ED80s for 900s using my DSLRs, so I'm reasonably confident in that area.

[Adam J]

 

15 minutes ago, tooth_dr said:

Ok thank-you.  I've been guiding ok with my dual rig ED80s for 900s using my DSLRs, so I'm reasonably confident in that area.

You will be very happy with the ASI1600mm pro, I got mine in April and I love it. Its not just the low read noise, its the fast read time that makes framing and focusing very easy. Not that I am saying CCD is harder to use...its just not as easy. You will have to take more time with calibration frames than with a Sony CCD but when you get calibration right the results are great. 

[tooth_dr]

Was there a 10% off ZWO stuff a wee while ago or did I imagine that?

What else is there to consider, new or used, that would be an alternative?

I have £1200 to play with.  

[Adam J]

5 hours ago, tooth_dr said:

Was there a 10% off ZWO stuff a wee while ago or did I imagine that?

What else is there to consider, new or used, that would be an alternative?

I have £1200 to play with.  

Yes, I got mine for £1090 and a 8 position filter wheel for about £245. Only way I could afford it. A friend gave me his old budget lrgb set. Is the £1200 budget for filter wheel and filters too? If so it's not really doable in that budget unless you forgot filters and just shoot lum to start off. Your scopes are too long in focal length for the IMX183m based cameras due to the very small pixels. Anything with a 8300 CCD chip will work...ASI1600mm-c do appear on the second hand market but not seen a pro model yet. The QHY163m is an equivalent but needs 36mm filters really. You won't find an ATIK horizon easily second hand. You sometimes see ATIK428ex cameras come up in the 1000pound range... But can you accept the field of view? The ATIK460ex is out of budget. 

[coatesg]

You might just get away with 1.25" filters with the 163M if you get the US wheel with the close coupling kit (though it'll depend what f ratio you work at as to how much it'll vignette). Bit tight though... And if you ever use a quick camera lens on it, you may struggle to get the flats to properly correct.

[tooth_dr]

5 hours ago, Adam J said:

Yes, I got mine for £1090 and a 8 position filter wheel for about £245. Only way I could afford it. A friend gave me his old budget lrgb set. Is the £1200 budget for filter wheel and filters too? If so it's not really doable in that budget unless you forgot filters and just shoot lum to start off. Your scopes are too long in focal length for the IMX183m based cameras due to the very small pixels. Anything with a 8300 CCD chip will work...ASI1600mm-c do appear on the second hand market but not seen a pro model yet. The QHY163m is an equivalent but needs 36mm filters really. You won't find an ATIK horizon easily second hand. You sometimes see ATIK428ex cameras come up in the 1000pound range... But can you accept the field of view? The ATIK460ex is out of budget. 

Great info Adam. That budget is for camera only. I have a set of RGB filters, but was planning on just shooting L to begin with anyway.

Is the Atik460ex worth the extra outlay?

 

[Adam J]

1 hour ago, tooth_dr said:

Great info Adam. That budget is for camera only. I have a set of RGB filters, but was planning on just shooting L to begin with anyway.

Is the Atik460ex worth the extra outlay?

 

It's a great camera, easy to get a good result with it as it's very clean. So no dithering or darks etc is possible from what I hear. But the 1600mm is lower noise, lets you do shorter exposures it's also twice the size. So if you can live with the diffraction pattern on very bright stars and put the effort in with calibration it's probably the better choice.