What Is the Fundamental Unit of CCD/CMOS Sensors Called?

[doofus]

I was thinking about how CCD's and CMOS's almost exclusively come in multiples. To be clear, what I am referring too is how each chip is fabricated to feature many pixels. Does the fundamental unit of a CCD/CMOS have a name "commercially"? Something like a "single pixel camera"? Looking at wikipedia I see the fundamental unit is called a "MOS capacitor". Which has me thinking about phototransistors, but I was under the impression these did not have a concept of storing charge/exposure. Photodiodes being in a similar position. I think it would be very fun to have a single pixel that had a concept of an exposure time, but I'm not sure if this is a thing other people use, or if my understanding of phototransistors needs some brushing up.

Would a single pixel basically be a photodiode hooked up to several transistors?

 

[phyzguy]

Why would anyone want a "single pixel camera"? It would not be useful for building up an image. So the manufacturers build them with many pixels so they can detect an entire image. Also CCDs and CMOS imagers have a lot of overhead circuitry that surrounds the imaging region to be able to read out the image. To build all of the support circuitry for a single pixel would be very inefficient. You can always take a multi-pixel CCD or CMOS imager and just look at one of the pixels if you want.

For a CCD imager, the basic unit is indeed a specially designed MOS capacitor, where the charge is stored in a potential well in the silicon.

 

[berkeman]

CMOS's almost exclusively come in multiples.

Can you give a couple examples of what you are calling "CMOS's"? The term CMOS (complementary metal-oxide semiconductor) usually refers to a type of transistor construction. Those transistors can be used in the circuitry of an imaging device, but they are not generally used as imaging sensors on their own:

https://en.wikipedia.org/wiki/CMOS

 

[rbelli1]

The light sensing part of the CMOS image sensor is a photo-diode. There are also integrated amplifiers. You could build a singe one of them with a discrete photo-diode and a high impedance operational amplifier (op-amp).

BoB

 

[doofus]

Why would anyone want a "single pixel camera"? It would not be useful for building up an image. So the manufacturers build them with many pixels so they can detect an entire image. Also CCDs and CMOS imagers have a lot of overhead circuitry that surrounds the imaging region to be able to read out the image. To build all of the support circuitry for a single pixel would be very inefficient. You can always take a multi-pixel CCD or CMOS imager and just look at one of the pixels if you want.

For a CCD imager, the basic unit is indeed a specially designed MOS capacitor, where the charge is stored in a potential well in the silicon.

The idea on its surface is goofy. The use case isn't for taking images, more-so for a typical photodiode like application. But, the nice thing about being able to store charge from light exposure is, well, assuming thermal noise and other-wise are OK, you could measure low light levels without wiring up an amplification stage (or two).

It's true that someone could ignore the other pixels, but it has me wondering about the minimal elements and how available they are.

 

[doofus]

Can you give a couple examples of what you are calling "CMOS's"? The term CMOS (complementary metal-oxide semiconductor) usually refers to a type of transistor construction. Those transistors can be used in the circuitry of an imaging device, but they are not generally used as imaging sensors on their own:

https://en.wikipedia.org/wiki/CMOS

Sorry this is my ignorance. I am referring to CMOS imaging sensors because I am uncultured and unaware of their other use cases. I'm specifically interested in what the smallest unit is, and if its possible to build/purchase for low light level measuring. Where I can learn more about this, or circuits I could explore to understand this area more or if it's all semiconductor fab level stuff.

 

[doofus]

The light sensing part of the CMOS image sensor is a photo-diode. There are also integrated amplifiers. You could build a singe one of them with a discrete photo-diode and a high impedance operational amplifier (op-amp).

BoB

Thanks Bob, that's pretty enlightening. How do CMOS image sensor pixels "store charge" to be read out later?I'm somewhat familiar with op-amps and amplification of photodiode current in general, but I am curious how CCD and CMOS pixels have a concept of "exposure".

 

[rbelli1]

The CMOS sensor's diodes don't have charge storage. If charge storage is used then it will be in a capacitor after an amplifier. I don't know if that is used or some other readout mechanism is.

The exposure will start by discharging the storage capacitor or zeroing out an accumulator register or some other initialization. The input from the photo-diode will be connected to the measurement system (or some combination of diodes to measurement/accumulation systems) for an amount of time then disconnected. That is the exposure time.

For a CCD the system simply sets the CCD to a known initial condition then measures at a later time the change to charge in the cells. This is often done by reading the first line then clearing that line and then shifting the charge from all of the other lines over one place and repeating.

BoB

 

[Tom.G]

Try this site for an explanation:
https://global.canon/en/technology/s_labo/light/003/05.html

Above, and many more, found with:
http://www.google.com/search?hl=en&q=how+cmos+image+sensor+works

Have Fun!

Cheers,
Tom

 

[Borek]

For light intensity measurements you can always use the smallest resolution CMOS imaging sensor you can find (alternatively: the cheapest, no matter the resolution), and just sum up (or average) signals from all pixels.

 

[Rive]

I was thinking about how CCD's ... almost exclusively come in multiples.

The very point behind the CCDs was about reading them out through their Coupling. A single unit cannot be coupled

If you are flexible enough you can take any PV item as single pixel. Even up to a complete rooftop solar panel.

 

[doofus]

Try this site for an explanation:
https://global.canon/en/technology/s_labo/light/003/05.html

Above, and many more, found with:
http://www.google.com/search?hl=en&q=how+cmos+image+sensor+works

Have Fun!

Cheers,
Tom

Thanks Tom.G I am a fan of LMGFY as well :D. The link is cool. I love the diagrams. Here's what I still don't "know" yet. How is charge stored with an open circuit? Most photodiodes have relatively speaking very low capacitance. Often by design, because high capacitance loads are difficult to stabilize in a transimpedance sense compared to their low capacitance counterparts. Charge isn't getting stored there I don't think. Maybe it's charged on the series capacitor over the gain resistor in a typical TIA circuit.

If we put a capacitor in series with the output of an op-amp would that be enough? They demonstrate the gating with a switch, but I assume they are actually using a transistor.

I wish I could find a simple analog circuit for a single "pixel" CMOS. I think it would be really fun to breadboard and play with to understand what goes into the design for this sort of thing.

 

[doofus]

For light intensity measurements you can always use the smallest resolution CMOS imaging sensor you can find (alternatively: the cheapest, no matter the resolution), and just sum up (or average) signals from all pixels.

True! Its a little limiting in some characteristics. For example, if someone wants a high ADC count, you may be buying a lot of unused functionality that goes unused. Its still a good point and worth making.

 

[doofus]

The very point behind the CCDs was about reading them out through their Coupling. A single unit cannot be coupled

If you are flexible enough you can take any PV item as single pixel. Even up to a complete rooftop solar panel.

That would be part of the problem with my searches, to be fair though looking for "charge device" in google probably isn't going to be very helpful either :P. For some reason PhotoVoltaic's didn't cross my mind as a search term. Thanks for this

 

[rbelli1]

The CMOS sensor's diodes don't have charge storage.

After some more research I have found that there is a charge storage capacitor associated with each photo-diode. The operation of the two devices is more similar than I thought.

BoB

 

[doofus]

What I still haven't been able to figure out, and maybe its deserving of a new Doofus post is, what does the "simplest" circuit for this look like? I am not an EET, haven't even played one on TV, but what I don't understand is how I could close a switch for example and dump charge stored up from something like a photodiode to something like an ADC. IE if I leave the switch closed for 2s I see 1V across a resistor when the switch is opened, then if I close it for 4s I see roughly 2v.

Maybe even remove the amplifier, lets assume a solar cell like PD like a BPW34 or something and a bright light source, heck why not a proper pocket photon cannon operating at 5mW radiant output.

 

[rbelli1]

Three Transistor CMOS Pixel
The charge storage capacitor here is the gate capacitance of Msf.
Four Transistor CMOS Pixel
The charge storage capacitor here is Cfd. That could be integrated into transistor SF too. This allows for the use of a global shutter but costs one extra transistor per pixel.

The reset signal sets the pixel to a known state. Then the diode charges the capacitor for a set amount of time. For the 3T version the charge is just read and the cycle repeats for the next area (often a row or column) of the sensor. For the 4T version the exposure time is determined by how long TG is closed. Then each section may be read with no additional change from the light hitting the sensor. Both of these allow direct measurement of the charge and no transfer must be done.

A sensor could be created so as to minimize the capacitance and it would take instantaneous measurements. That would make the fastest camera but would have terrible performance with less light. The capacitor makes the sensor work more like film. Also instantaneous measurement would be difficult with a global shutter.

IE if I leave the switch closed for 2s I see 1V, then if I close it for 4s I see roughly 2v.

Your 2s and 4s times are how long from releasing reset to reading the value in the 3T sensor and how long from closing to opening TG in the 4T sensor. This would be like discharging your solar battery at night. Then you charge all day. The state of charge the next night tells you how much sun for that day.

BoB

 

[doofus]

This helped me a ton thank you. For some reason I couldn't see what the best way to do this was. So I guess for a single pixel, in the 4T version we wouldn't need that SEL transistor because we aren't selecting rows at all. Making the simplest unit that stores charge a 3T with a capacitor.

Super fun I'll probably try making one of these sometime over the holiday. I'll report back when I get a chance, but you answered my question. Thank you so much.

 

[rbelli1]

Depending on the output current of your photo-transistor and the input impedance of your amplifier and ADC you may want the select transistor.

BoB

 

[DaveE]

what does the "simplest" circuit for this look like?

Here's how I did something like this in a pulsed laser to measure the pulse energy. Yes, the sketch is simple but not much simpler than the actual circuit, not counting the digital timing stuff. The only analog amp was built into the ADC front end. Key point though, we knew when the laser would fire.

Sorry, I may be off base, I haven't read every post yet.

 

[doofus]

Not off base at all. Most of the post is me failing to ask for examples like this because I lack the vocabulary and clarity of thought.

Super cool! I take it your ADC had a transistor that only read off the measurement when trigger went high? I like this because it shows that you only need a transfer gate transistor if you need an off-state while the circuit is live. For my goofy interests I can probably remove that transistor as well and end up at a 2 transistor model.

1 to reset and 1 to amplify/read out.

 

[rbelli1]

DaveE's circuit uses the diode in photo-conductive mode rather than photo-voltaic mode.
Photo-diode Tutorial

BoB

 

[DaveE]

I take it your ADC had a transistor that only read off the measurement when trigger went high?

They are called sampling ADCs and are VERY common. Many ADCs that don't appear to be sampling to the user will still do so internally since they can't deal with an analog voltage that changes during the conversion process.

 

[doofus]

Thank you both for the links. I'm learning a lot. Now that I think about it, I never considered how a useful ADC worked beyond the toy model ones. Interesting

 

[berkeman]

Since you are wanting to learn about these types of circuits, here are a few topics and links for your reading:

For a single photodiode, you often use a simple (photo)current-to-voltage circuit:

https://en.wikipedia.org/wiki/Transimpedance_amplifier

In order to digitize the output voltage of the photodiode current-to-voltage converter circuit, you would first* go through a sample-and-hold circuit:

https://en.wikipedia.org/wiki/Sample_and_hold

And after the S/H circuit, you would use some form of ADC circuit to digitize the voltage for further processing by a microcontroller (uC) or other processor:

https://en.wikipedia.org/wiki/Analog-to-digital_converter

*In real systems, an anti-alias filter is used to ensure that the data that is digitized is limited to the baseband signals of interest: https://www.analog.com/en/resources/technical-articles/guide-to-antialiasing-filter-basics.html

 

[DaveE]

For a single pixel design, really a photodiode, you can't beat the Analog Devices web pages. They have tons of great information about all kinds of photodiode applications.