#743

Episode 743
A description of the 3T pixel used in CMOS imagers
Be a Patron:
/ imsaiguy

[@herbertsusmann986]

The first digital "camera" I ever had was a 3rd party board for my Apple 2. It consisted of a dynamic RAM chip with a quartz window so light could get at the chip. The RAM chip was on the end of a 2 foot long ribbon cable and a standard camera lens was used to focus light onto the RAM chip. The Apple 2 would write all 1's to the RAM chip and then wait a few 10's of ms or so. It would then read out the entire RAM chip. The memory cells that had enough light striking them would read back 0's. The light bled off the charge on the RAM memory cells. The ones with not so much light would still read back 1's. Thus you got an image that consisted of each pixel either being white or black. Extremely bad resolution but it did work! I think I even still have that board around here somewhere.

[@t.w.experiments2122]

That’s cool! I was learning about old eprom chips and I was wondering if something like that was possible!

[@BigWolf130]

Thanks for this video not alot of people talk about this part of how pixels sensors work. It's all about light filters, and not what happens to the light once it gets to the sensor. All people ever say is that it works. This is definitely the more interesting part of pixel sensors.

[@BJcanal270]

I was just looking for this info, That's why I stick your channel!! Thank you!!

[@ec11368]

Thank you very much for going to the basics and explain the principle.

[@hahasimpsons]

Loved it! Super simple and easy to finish your sentences, as you break it all down super simple, for those of us still using crayons. lol

[@BilalHeuser1]

This is very interesting and explained really well. I'm not sure if a scanner works in the same camera does, because scanning a page or photo with a scanner takes longer than a camera. Do the photo sensors in scanner work much in the same way?

[@IMSAIGuy]

scanners are usually a line scanner so the imager is only one row of pixels and the paper is moved (scanned) over the line to build up the picture.

[@aduedc]

I would replace photo diode, with photo FET. Then you probably do not need reset or source follower, the photo FET itself could be used as source follower. So could have two transistor ( photo FET and the other select FET) and no diode, no reset , and no source follower. Check out photo FET, they use it for high frequency microwave amplifiers and oscillators, embedded with its own transmission line. Also, you are probably using direct band-gap semiconductors, the process engineer instead of grinding, could do filp-chip, or process both side of the die. one side photo element with color filter and the other side part of the signal processor chip.

[@IMSAIGuy]

the schematic is more of a block diagram. the actual device is more of what you say the 'diode' is really the gate of the follower. you need the reset for the function of an electronic shutter. otherwise the the photo area will just saturate.

[@Kashif_Javaid]

What happened to the capacitor that was in parallel with photodiode in your second schematic?

[@schwachmatxyz]

I appreciate this fundamental description. The kind of image sensor you are describing is an APS (active pixel sensor), i assume. In real I think the reset circuitry works inversly, but for your amazing presentation it would make no difference in understanding to not complicate it more as neccessary. Reading photodiodes is highly linear if you convert the photocurrent in short circuit (10 or more decades, more than 32 bit). How linear can this APSs be? And in some cataloges, especialy for spectrometers, there are CCD sensors mounted. From datasheets of CCDs I know that their dynamic is only about 7 or 8-bits. Why are CCDs nevertheless used?

[@IMSAIGuy]

The photons create holes (not electrons) so the reset sets the photodiode to Vcc. CCDs can have a much lower noise floor so are used in measurement applications.

[@schwachmatxyz]

@@IMSAIGuy Ok, thanks for the quick reply. So it is my misunderstanding of the term "RESET". The not illuminated pixel is read as nearly Vdd. The higher the Illumiation of the pixel, the lower is the voltage read.

[@aryaprincess2479]

Your hands are all over the images that we were supposed to see.

[@manueljenkin95]

This was a very good presentation. Thank you very much. One suggestion though, I wouldn’t call anything of biological systems (especially not the sense systems) to be junk. (4:20) All of them have a purpose, that constitutes a holistic system.

[@gregsullivan7408]

I absolutely LOVED the way he referred to it as "junk" - it's just a bit of tongue-in-cheek humour. 😉

[@MaxParadiz]

Thank you a lot for the detailed explanation! One question.... Why is this a CMOS sensor and not an NMOS sensor? Where do the PMOS transistors come in? I see that M1, M2, and M3 are all NMOS.

[@IMSAIGuy]

from wiki: The two main types of digital image sensors are the charge-coupled device (CCD) and the active-pixel sensor (CMOS sensor), fabricated in complementary MOS (CMOS) or N-type MOS (NMOS or Live MOS) technologies. Both CCD and CMOS sensors are based on the MOS technology,[4] with MOS capacitors being the building blocks of a CCD,[5] and MOSFET amplifiers being the building blocks of a CMOS sensor.[6][7]

[@MaxParadiz]

@@IMSAIGuy I see, thank you again!!

[@lorenzoemmepi3302]

great video! super clear

[@jacobliu3316]

Great! How do you think of the novel one-transistor active sensor? many technical papers introduce the 1t-APS, while is there any 1T-APS used in a real camera?

[@IMSAIGuy]

somebody needed to get published.

[@dukebozikowski3801]

Amazing explanation!

[@IMSAIGuy]

Glad you liked it

[@dukebozikowski3801]

Could you do a video on image capture after the sensor? How it does this for video?

[@IMSAIGuy]

www2.cs.sfu.ca/~mark/ftp/SignalProcessing05/Color_image_processing_pipeline05.pdf

[@videolabguy]

You wouldn't happen to have the datasheet for the RCA SID504 would you?

[@IMSAIGuy]

no

[@IMSAIGuy]

BTW CCDs operate in an entirely different way

[@snoo333]

thank you

[@VandalIO]

Why can’t we use class AB amplifier instead of a single fet

[@IMSAIGuy]

all of this is for every pixel. needs to be as small as possible since you have 4 million or more of them.

[@tamasdebreczeni5335]

❤

[@jdmccorful]

Interesting, now I have some understandable knowledge of pixels' operation. Thankyou!

[@MrBanzoid]

Great explanation, thanks.

[@sammin5764]

🌟🌟🌟🌟🌟

[@riccardo2948]

Beautiful video, but I don't think that the story of the octopus is true. Actually they have eyes similar to the human eyes. Thanks for sharing

[@IMSAIGuy]

en.wikipedia.org/wiki/Cephalopod_eye

[@islamicworld2277]

Really bad explanation man, wasted my time on this.

[@suhaimiseliman8593]

I 🤔 I SHOULD CHOOSE A MATERIAL THAT CAN CONVERT FREQUENCY TO VOLTAGE ON IT'S SURFACE. REFERRING TO COLOR SPECTRUM ELECTROMAGNETIC WAVE FREQUENCY RANGE. HOPEFULLY THE SPECTRUM IS CORRECT & COMPLETE..🫣😎🥳😁