What specs should I look for in a mono cams sensor?

[Mcwaffles2003]

Specs as I understand them:

Full well depth: The amount of electrons each pixel can store before becoming saturated. Higher numbers lead to greater levels of dynamic range

ADC: The degree of accuracy to which the fullness of a well is read. Higher numbers lead to greater dynamic range capping out when capable of reading per electron

Quantum efficiency: The chance when a photon hits the sensor that it will be read. Higher numbers lead to more brightness in the image and that brightness has hyperbolic scaling (a 60% sensor exposed for 8 minutes is equivalent to an 80% sensor exposed for 6 minutes or a 40% sensor exposed for 12)

Sensor size: Literal size of the sensor. Larger sensor gives a greater field of view limited by the image circle formed by the telescope.

Pixel size: Literal size of a pixel. Large pixels will gather light more quickly according to their area while smaller pixels will increase angular resolution up to dawes limit (if seeing allows)

Read noise: The average amount of noise, in extra electrons, per pixel per second. Higher noise leads to reduced contrast.

 

Please correct any inaccuracies in my understanding, or point out any nuances that I have missed.

 

[SiD the Turtle]

I'm still learning this stuff myself, but you must match your telescope to the correct pixel size to prevent over or under sampling. Here's a useful calculator:

http://astronomy.tools/calculators/ccd_suitability

[Pompey Monkey]

18 hours ago, Mcwaffles2003 said:

Specs as I understand them:

Full well depth: The amount of electrons each pixel can store before becoming saturated. Higher numbers lead to greater levels of dynamic range

ADC: The degree of accuracy to which the fullness of a well is read. Higher numbers lead to greater dynamic range capping out when capable of reading per electron

Quantum efficiency: The chance when a photon hits the sensor that it will be read. Higher numbers lead to more brightness in the image and that brightness has hyperbolic scaling (a 60% sensor exposed for 8 minutes is equivalent to an 80% sensor exposed for 6 minutes or a 40% sensor exposed for 12)

Sensor size: Literal size of the sensor. Larger sensor gives a greater field of view limited by the image circle formed by the telescope.

Pixel size: Literal size of a pixel. Large pixels will gather light more quickly according to their area while smaller pixels will increase angular resolution up to dawes limit (if seeing allows)

Read noise: The average amount of noise, in extra electrons, per pixel per second. Higher noise leads to reduced contrast.

 

Please correct any inaccuracies in my understanding, or point out any nuances that I have missed.

 

Spot-on, apart from the Read-noise. This is a constant for all exposures relating to the electronic process of converting the charges on the sensor into an output signal. This is also known as Bias.

The noise component measured in electrons per pixel, per second, is the accumulated charge on the sensor (hence "per second") for each exposure. his is known as Dark current. Dark current is a lot lower in the newer CMOS cameras compared to CCD. There are other differences, but That is for someone more knowledgeable than me to answer!  

[vlaiv]

18 hours ago, Mcwaffles2003 said:

Full well depth: The amount of electrons each pixel can store before becoming saturated. Higher numbers lead to greater levels of dynamic range

Correct.

Side note - dynamic range is quantity that is important in single exposure. With astrophotography, it is really not important as each next image in the stack increases dynamic range of whole stack. Dynamic range of stack is much much larger than any single exposure and you can always increase dynamic range of final stack by adding more exposures.

Some may say that Full depth well is important so you don't saturate your stars or bright parts of target - but again, we always have option of mixing different exposure lengths (needs to be done properly) and hence FW is not very important metric either.

18 hours ago, Mcwaffles2003 said:

ADC: The degree of accuracy to which the fullness of a well is read. Higher numbers lead to greater dynamic range capping out when capable of reading per electron

Not quite. It is number of bits needed to store read out value after gain has been applied. Designers of sensor make sure that you don't really loose much in rounding up because there is read noise and that masks what is called quantization error.

Again - not really important metric. Have 12 bit sensor? Stack 16 subs and you'll have 16 bit precision. You can and will achieve much higher precision in final stack (20+ bits), and again single sub precision is not very important. It is usually well matched to read noise and FW by designers of the chip and appropriate shooting technique will make it equally effective as any other ADC version.

For example, I have ASI1600 and it has 12 bit ADC. It has rather decent FW capacity per pixel, but the way I use it, by setting unity gain, I actually lower my FW to 4095 - maximum number that can be recorded by 12bit. Keep exposure times low and you'll have same effect as using 16bit 60K FW camera with 6e read noise.

18 hours ago, Mcwaffles2003 said:

Quantum efficiency: The chance when a photon hits the sensor that it will be read. Higher numbers lead to more brightness in the image and that brightness has hyperbolic scaling (a 60% sensor exposed for 8 minutes is equivalent to an 80% sensor exposed for 6 minutes or a 40% sensor exposed for 12)

First part is correct - probability that incoming photon will be converted into electron and subsequently contribute to total number of electrons recorded by exposure.

Second part I don't really get - scaling is linear.

Out of 100 photons 60% QE sensor will capture 60e on average, while 40% QE sensor will capture 40e on average. Want to know how much you need to increase exposure time to match - use reciprocal of QE, or ratio of two QEs. In this case 40% QE sensor will need to record for 60/40 = 6/4 = 3/2 = 50% more time than 60% QE sensor to achieve same level of signal per exposure.

In general - higher QE the better, however, do keep in mind that QE is not a single number, although single number is often quoted - and that single number is peak QE. One camera can have higher peak QE than other but lower QE in significant wavelength - like Ha. Most QE curves are fairly similar so overall performance will scale with peak QE, but there can be differences for narrow band.

18 hours ago, Mcwaffles2003 said:

Sensor size: Literal size of the sensor. Larger sensor gives a greater field of view limited by the image circle formed by the telescope.

Indeed - very important metric, larger the sensor - means faster sensor / faster imaging. This can be somewhat strange when you hear it first time, but think of it this way:

Take 14mm diagonal sensor and 28mm diagonal sensor.

Pair first one with 80mm F/5 scope and second with 160mm F/5 scope. Each will cover the same amount of sky - will have identical FOV (provided that aspect ratio of sensors is the same), but first one will gather light with 80mm of aperture and second one with 160mm of aperture.

Which one will collect more light in the same amount of time?

(You can make sure to have same sampling rate of the two sensors by using appropriate pixel sizes and/or binning - so resolution can be fixed and then we only have aperture competing on same FOV and sampling rate).

Larger sensors will be more expensive of course - no free lunch

18 hours ago, Mcwaffles2003 said:

Pixel size: Literal size of a pixel. Large pixels will gather light more quickly according to their area while smaller pixels will increase angular resolution up to dawes limit (if seeing allows)

Important metric, but not as important as it might seem at first. Smaller pixels are better because you have more flexibility in binning them to achieve your target resolution.

Tip - optimum sampling rate for long exposure is close to average FWHM/1.6

If you have 3.2" FWHM stars in your image - you need to sample at 2"/px, but for 1"/px you need your FWHM to be around 1.6". Do under sample if that suits the purpose of going wide field, but don't over sample as it serves no purpose at all. If you over sample - just bin until you get close to optimum sampling rate - you can bin both in hardware or software.

There is difference only in how the read noise is treated. With hardware binning - nothing happens, but with software binning - read noise is effectively increased by bin factor. For example if you have 1"/px system with 2e read noise and you decide to bin to get to 2"/px, then you'll have 4e of read noise per binned pixel.

(keep this in mind as read noise is related to exposure length).

Rough guide - 80mm or less - 2"/px or more

~100mm - 1.6"/px to 2"px

~150mm - 1.4"/px to 1.6"/xp

~200mm - 1.2"/px - 1.4"/px (you can go with 1"/px if you have exceptional mount and exceptional seeing)

(moral of the story - FWHM of your image will depend on how good your optics is, aperture of your scope, seeing conditions and mount performance and guide precision - small apertures simply can't create high resolution images because of this).

18 hours ago, Mcwaffles2003 said:

Read noise: The average amount of noise, in extra electrons, per pixel per second. Higher noise leads to reduced contrast.

Nope. This one is quite off.

There are 4 principal sources of noise:

1. Target / Shot noise

2. Light pollution noise (same as above but light is provided not by target but by sky)

3. Thermal noise (also called dark current noise) - similar to those above, but electrons (not photons) are provided by temperature and electronics (behaves very similar to above ones - just different physical process produces it)

4. Read noise - this is noise injected in final value by act of reading pixel value

First three grow with time. Their value is equal to square root of corresponding physical quantity that grows linearly with time. Either target photon count, or sky photon count or dark current electron count.

Read noise is only one that does not grow with time and is fixed in value and happens once per exposure. Each exposure gets same "dose" of read noise.

This is very important for stacking. There is no difference in single long sub or many short subs that are added together (stacked) and have same total duration as long sub - if you only consider first three noise sources. They behave in additive manner regardless if you add seconds or add exposures - as long as total time is the same.

Read noise is different and this is factor that makes distinction between few long subs and many short subs (totaling to same total exposure time). Larger read noise - bigger the difference. In fact, there is rule of how this read noise contribution behaves. As long as read noise is comparatively large to any of other 3 noise sources - difference is noticeable. As soon as read noise becomes significantly smaller (about x3-5 smaller - depending on how much impact you can tolerate) difference becomes negligible. All other noises grow with exposure time and at some point, one of them will become sufficiently larger than read noise - that means you can stop your exposure there and there is no more benefit in going further.

This other noise source is almost always LP noise (with cooled cameras and when chasing faint signal - both thermal and target noise tend to be very small, but light pollution is ever growing problem) and if you search topic here, you will find recommendations on how to best determine exposure length based on sky brightness and read noise.

In the end, read noise is not that important either as it's impact can also be minimized with particular style of imaging (remember CCDs - those have higher read noise, but people also tend to use long exposures with them like 10 or 20 minutes). To some extent - choose camera based on this metric if you don't have the best mount or best guiding as low read noise means that you can keep exposure length down.

HTH

[Tommohawk]

To be honest you can learn just as much from looking at other people's equipment and images produced as you can from trawling through all of the specs.

You will notice many folk prefer certain combination s of scope/camera.

Obviously it helps if you know what you're doing too!

Edited September 9, 2020 by Tommohawk