How does a CMOS sensor work?

[TonyW]

Some time ago there was a discussion where somebody, probably Ted (expatUSA), gave us a link which described well how a sensel in a CMOS sensor worked. I seem to remember that the description was of a capacitor with a light sensitive NP junction separating the cathode and anode. I can't find the thread where this was mentioned nor can I find a good description after many Google searches.

Can someone help me?

[dje]

Hi Tony

I've loaded a couple of articles onto Dropbox here. (I don't have direct links to the sources).

These may be of some use in getting some details of CMOS sensors. The articles probably are more in-depth than you are looking for but you can pick out the fundamentals, particularly in the first parts of the articles.

Dave

[TonyW]

Thanks Dave,

Those articles were very helpful, although they challenged my knowledge of semi-conductors. I will need to spend more time to fully absorb them.

They also raised a few other questions. For example, it occurred to me that the quantum efficiency if the sensor might depend on the colour of the light, because the difference between the photon energy and energy gap of the semi-conductor would vary. If I am right, this must be taken care of during the processing of the signal.

[dje]

Thanks Dave,

Those articles were very helpful, although they challenged my knowledge of semi-conductors. I will need to spend more time to fully absorb them.

They also raised a few other questions. For example, it occurred to me that the quantum efficiency if the sensor might depend on the colour of the light, because the difference between the photon energy and energy gap of the semi-conductor would vary. If I am right, this must be taken care of during the processing of the signal.

Tony my knowledge of semiconductor theory was left behind at the University gate when I graduated about 50 years ago!

The response of a sensor pixel does vary with wavelength and this, along with the colour filers, contributes to the overall spectral response of the three colour channels (R G and B). This is taken care of in processing with white balancing and camera colour profiling.

Dave

[xpatUSA]

Some time ago there was a discussion where somebody, probably Ted (expatUSA), gave us a link which described well how a sensel in a CMOS sensor worked. I seem to remember that the description was of a capacitor with a light sensitive NP junction separating the cathode and anode. I can't find the thread where this was mentioned nor can I find a good description after many Google searches.

Can someone help me?

Sorry, I'm late to the party.

Here's a circuit:


Pardon the appearance it's a SVG file.

Full description and better appearance here:

https://en.wikipedia.org/wiki/Active-pixel_sensor#Pixel

The capacitance is inside the photodiode. A little known fact is that the reset switch Mrst fully charges up the capacitor, which then discharges during exposure. So much for all those comfortable dumbed-down analogies of buckets filling up with ping-pong balls ...

[xpatUSA]

Thanks Dave,

Those articles were very helpful, although they challenged my knowledge of semi-conductors. I will need to spend more time to fully absorb them.

They also raised a few other questions. For example, it occurred to me that the quantum efficiency if the sensor might depend on the colour of the light, because the difference between the photon energy and energy gap of the semi-conductor would vary. If I am right, this must be taken care of during the processing of the signal.

The quantum efficiency (QE) is dependent on the wavelength of the incident radiation. "Colour of the light" is inappropriate in the context of QE because visible light only occurs over a restricted bandwidth.

With any camera, the QE of the camera and lens assembly is completely dependent on any filtration existing over the said sensor such as a Bayer CFA. Additionally, the lens and it's coatings block most of the UV coming into the body. The actual sensor QE is of course unaffected by the foregoing. Silicon itself runs out of QE steam as photons over 1150nm roar through the material virtually unimpeded.

Here's the QE of a bare Foveon sensor:



Three curves because the Foveon has three stacked photodiodes per pixel.

http://kronometric.org/phot/sensor/f...ubel_FINAL.pdf

See Page 3.

[TonyW]

Sorry, I'm late to the party.

...

The capacitance is inside the photodiode. A little known fact is that the reset switch Mrst fully charges up the capacitor, which then discharges during exposure. So much for all those comfortable dumbed-down analogies of buckets filling up with ping-pong balls ...

Thanks Ted. That is what I wanted to confirm. My interest was re-ignited when I heard someone using those crude analogies and making false deductions from them.