ASI585MC non-cooled vs ASI585MC Cooled

andrea tasselli:
I used and still do use uncooled sensors/cameras (i.e., DSLRs) and thermal noise has never been a really limiting factor thus far (and my "thus far" goes a very long way off...). Yes, cooled sensor are better and even more so at high temperatures but if your sensor temperatures are in the low 30s (Centigrade) and you limit exposures in the minute or so then go for it.

D. Jung:
The full well of the sensor is 50Ke.

D. Jung:
I can't say much about your first question but regarding your second question: 

Dark current is measured in e/s/px
For the 585mc, the dark current at
-20C is ~0.001e/s/px
30C is ~0.5e/s/px

Meaning that if you do a 100s exposure at -20C you collected 0.1e as dark current noise per pixel, while at 30C you are at 50e.

The full well of the sensor is 50Ke.

• The dark current itself is not the problem and is not noise if it is constant, i.e. having a variance σ^2 of 0. Then it would just be an offset which could be removable by simple subtraction.
• The dark current varies between pixels and these variations increase with temperature, such that the variance σ2 increase with temperature with some function σ^2(t).
• So if we subtract a dark frame from our light frame, we remove the offset, but the variance of the dark current embedded into the light frame remains. even worse, since the dark frame itself is noisy, we actually increase the noise of the light frame by said subtraction. in the case of identical variance σ^2 of the dark and light frame, the noise, i.e. the standard deviation would increase by sqrt(2σ^2) = 1.414. even if the variance of the dark frame was 0 we would still be left with the original variance in the light frame after subtraction!
• If we assume the dark current to follow a poison distribution the standard deviation σ is the sqrt of the amplitude. And for above dark current example would yield for -20C a σ=sqrt(0.001) and for the 30C case σ=sqrt(0.5), i.e. a noise increase of factor of 22.

D. Jung:

D. Jung:
I can't say much about your first question but regarding your second question: 

Dark current is measured in e/s/px
For the 585mc, the dark current at
-20C is ~0.001e/s/px
30C is ~0.5e/s/px

Meaning that if you do a 100s exposure at -20C you collected 0.1e as dark current noise per pixel, while at 30C you are at 50e.

The full well of the sensor is 50Ke.

I have been thinking a bit more about my answer here and I think it does not fully capture the actual problem.

• The dark current itself is not the problem and is not noise if it is constant, i.e. having a variance σ^2 of 0. Then it would just be an offset which could be removable by simple subtraction.
• The dark current varies between pixels and these variations increase with temperature, such that the variance σ2 increase with temperature with some function σ^2(t).
• So if we subtract a dark frame from our light frame, we remove the offset, but the variance of the dark current embedded into the light frame remains. even worse, since the dark frame itself is noisy, we actually increase the noise of the light frame by said subtraction. in the case of identical variance σ^2 of the dark and light frame, the noise, i.e. the standard deviation would increase by sqrt(2σ^2) = 1.414. even if the variance of the dark frame was 0 we would still be left with the original variance in the light frame after subtraction!
• If we assume the dark current to follow a poison distribution the standard deviation σ is the sqrt of the amplitude. And for above dark current example would yield for -20C a σ=sqrt(0.001) and for the 30C case σ=sqrt(0.5), i.e. a noise increase of factor of 22.

That is my feable attempt at statistical math (likely wrong) reminding me of this one below

Clayton Ostler:
Ok, I will admit I have talked about this on other forums but never felt like real logic is applied to the answers, so I I hoping for some real conversation here. 

I own an ASI585MC non-cooled camera, this is not my only camera, and yes I own better and cooled cameras. 

I have put together a super small portable widefield rig, The goal is to get the whole solution (including portable power and tripod)  in a single carry-on 

I have accomplished this but due to power and weight needs, I had to stick with the ASI585MC as the camera. I am pretty happy with the field of view, but constantly worry about "how much am I missing out on" by using a non-cooled camera. 

More info:
Yes I take calibration frames, and expect to continue to do so. 
For this setup my images are limited to no more than 60 seconds, mostly due to guiding, but also hoping to limit heat related noise on the sensor. 
The sensor in the ASI585MC seems pretty solid as far as dark/current noise, but it still has some hot pixels and without a cooler the actual temp of the sensor is somewhat variable.

Me talking to myself:
I know the general answer is that (cooled is always better), but in this case it would blow up my whole setup, adding new power requirements, pushing up my weight, and other complications that I am not sure are worth the work. 

Banter for the forum

On the newer sony sensors where there is no amp glow and low dark noise, how much difference will a cooled camera make? Really, not in theory. 
Will limiting exposure time to 60 seconds help mitigate the lack of a cooled sensor? 
Should I still try to run this camera at Unity gain / In HGC mode? or should I dial it down to lower gains hoping to compensate for the lack of a cooled sensor?

Please chime in, but be more thoughtful than (you should just get a cooled camera), Know that I have a asi533mc pro, that could easily be connected to the setup, but this would blow up my power, weight and convenience factors.