Cooled CMOS

[Petergoodhew]

I'm intrigued by cooled CMOS cameras.  Does anyone know of a good idiot's guide to them? How do they match up against cooled CCD?s Are they better in light polluted areas?  Pros and cons?

[vlaiv]

I don't think anything about sensor will help for LP, so no reason to believe either is better in LP.

Don't know any sort of idiot's guide, but can provide some "useful" info on CMOS since I've been using them.

Simplified Pro/Con list:

Pros:

- Allow for shorter exposures (especially in LP) which is good because guiding issues (probably not concern to you given 10Micron in your signature), less wasted time due to wind, cable snag, what have you ...

- Faster download times (again less imaging time wasted)

- Can be used for short exposure / planetary / EAA

- price?

Cons:

- a lot of subs usually means a lot of data to store and process. Single LRGB session / target on ASI1600, calibrated frames (float point) can take 30GB and upwards (depending on number of subs / exposure duration). More time needed to calibrate / register / stack because of large number of frames.

- Issues with bias files (I recommend calibration with darks, flats and dark flats only). Depending on model, there are some that are causing trouble with darks as well - because bias is inconsistent between power cycles. So one should check out how camera behaves prior to opting for it.

- Usually smaller pixel sizes, so not suitable for very large telescope (Max pixel size that I've seen in CMOS is <6um, but around 3.8 - 4um is the most common size).

- Does not support hardware binning (One can use software binning, and with low read noise models still end up "having" large pixels with very low read noise - compared to CMOS).

- They tend to suffer from amp glow, and flats are almost mandatory in some models - pixels each have their own amp that can have differences due to manufacture, and that needs to be handled / calibrated properly. This is really not an issue if you calibrate your frames properly.

- Have lower bit count ADC (not necessarily lower full well capacity) - again not something that is problematic in itself, but has an impact on the workflow.

 

Main thing that I think people who are used to CCD, when switching to CMOS face, is need to change workflow in acquiring and processing data. It is a bit of mindset shift that is needed. If one continues to use same methods as for CCD - they often run into trouble / sub optimal results with CMOS cameras.

When you embrace "CMOS" way of doing things - it turns out that CMOS are every bit as potent as CCDs - and some even being bigger bang for the buck (due to cmos cameras being cheaper).

[Petergoodhew]

1 hour ago, vlaiv said:

I don't think anything about sensor will help for LP, so no reason to believe either is better in LP.

Don't know any sort of idiot's guide, but can provide some "useful" info on CMOS since I've been using them.

Simplified Pro/Con list:

Pros:

- Allow for shorter exposures (especially in LP) which is good because guiding issues (probably not concern to you given 10Micron in your signature), less wasted time due to wind, cable snag, what have you ...

- Faster download times (again less imaging time wasted)

- Can be used for short exposure / planetary / EAA

- price?

Cons:

- a lot of subs usually means a lot of data to store and process. Single LRGB session / target on ASI1600, calibrated frames (float point) can take 30GB and upwards (depending on number of subs / exposure duration). More time needed to calibrate / register / stack because of large number of frames.

- Issues with bias files (I recommend calibration with darks, flats and dark flats only). Depending on model, there are some that are causing trouble with darks as well - because bias is inconsistent between power cycles. So one should check out how camera behaves prior to opting for it.

- Usually smaller pixel sizes, so not suitable for very large telescope (Max pixel size that I've seen in CMOS is <6um, but around 3.8 - 4um is the most common size).

- Does not support hardware binning (One can use software binning, and with low read noise models still end up "having" large pixels with very low read noise - compared to CMOS).

- They tend to suffer from amp glow, and flats are almost mandatory in some models - pixels each have their own amp that can have differences due to manufacture, and that needs to be handled / calibrated properly. This is really not an issue if you calibrate your frames properly.

- Have lower bit count ADC (not necessarily lower full well capacity) - again not something that is problematic in itself, but has an impact on the workflow.

 

Main thing that I think people who are used to CCD, when switching to CMOS face, is need to change workflow in acquiring and processing data. It is a bit of mindset shift that is needed. If one continues to use same methods as for CCD - they often run into trouble / sub optimal results with CMOS cameras.

When you embrace "CMOS" way of doing things - it turns out that CMOS are every bit as potent as CCDs - and some even being bigger bang for the buck (due to cmos cameras being cheaper).

Great - thanks. That helps a lot. An excellent overview.

[ollypenrice]

Pieter Vandevelde, who was here in March and is a regular of ours, has made the change. He's a tremendously talented astro-imager with an interesting and informative website.

https://pietervandevelde.smugmug.com/

Olly

[wimvb]

One advantage of cmos over ccd is the ability to change gain. You can either take long exposures at low gain, more like "traditional" ccd imaging, or short exposures at high gain. At high gain you have lower dynamic range, but since you have more subs, you can compensate for this in stacking. Shorter exposures allow you to go unguided (non-issue with your 10Micron), and it does allow you to remove subs that are not top quality. This makes imaging more flexible. If for some reason I have trouble with guiding, I just switch over from low gain to high gain and take shorter subs. This way I won't waste half a night figuring out guiding problems. (Again, a non-issue for you.)

While cmos have fewer data bits than ccd, the dynamic range needn't be less. Full well per um pixel size is actually the same or slightly higher than for ccd (SONY chips). but then again, numbers on paper is one thing, the proof is in the proverbial pudding.

While there still is some amp glow in cmos, this is much less in the newer than in older models. You still need matched darks to calibrate out any amp glow, but the residual noise is less. As Vlaiv wrote, it's better to use darks, flats and dark-flats than bias frames. I have more or less standardised my workflow on: lights + 20-30 darks + 30 flats/filter + 30 dark-flats/filter. Shooting the flats and dark-flats doesn't take more than about 10-15 minutes.