CMOS Image Sensors - Pixel Design Basics
August 11, 2023
This time, let’s talk about the most critical part of CISQ: pixels! It can be said that almost all CIS performance indicators are ultimately determined by the quality of the pixel design. Let's put it this way: the design of the readout circuit determines the lower performance limit of an image sensor, and the pixel design determines the upper performance limit of an image sensor!
There are many types of pixels, such as:
1. The most commonly used visible light is 3T, 4T, and 5T pixels. If you want to add more functions, such as global shutter (GS, Global Shutter), high dynamic range (HDR, HighDynamic Range), etc., you need to add more transistors;
2. X-ray is used for CTIA-based pixel of Photon Counting;
3. DVS pixel for Event sensor;
4. The SPAD pixel used for 3D imaging is often used as dToF (direct time-of-flight), and it is often used as the demodulation pixel of iToF (indirect time-of-flight); 5. Another example is infrared direct injection, CTIA, etc., bolometer Use the pixel of the resistance measurement;
6. In addition, there are TDI pixel, CCD 2/3/4-phase pixel and so on.
We have dozens of types of pixels for large categories. If we list all the pixels designed by humans, there may be hundreds or thousands of types. So it is impossible for us to enter all these pixels. The vast majority of CMOS image sensors Q (>95%) in the world are based on 2D imaging of visible light, so we only focus on 3/4/5T pixels here. If you have the opportunity to work in different companies, or do full custom design, then get in touch with more interesting things
First of all, let's understand a concept: Shutter. The shutter determines how and how much light enters the pixels, which also determines the image quality. The shutter can be divided into mechanical shutter (Mechanical Shuter) and electronic shutter (Electronic Shutter), and electronic shutter can be divided into rolling shutter (Roling Shuter) and global shutter (Global Shutter).
The mechanical shutter is composed of two shutter blades. Before exposure, the first blade covers the entire sensor, and then generally speaking, the blade is removed from top to bottom, and then the second blade is moved out from top to bottom after the exposure is started. , block the entire sensor, and the exposure ends. The moving speed of these two blades is very, very fast, almost up to 35kmh. So for a 35mm full-frame camera, the blade can allow the sensor to start or end the exposure in about 1/400s, so it can be said that all mechanical shutters are "Approximate" global shutter. So how does the mechanical shutter achieve a faster exposure time? This is relatively "simple", that is, when the first blade is not fully opened, the second blade has already started to move down to block, of course, there is a limit to this , the fastest is about 1/8000s. If you want a faster speed, you need an electronic shutter.
As shown in the picture below, you can see the difference between the rolling shutter and the global shutter in the electronic shutter. For rolling shutter, the exposure time of each row of pixels is at least separated by the reading time of one row of pixels. For example, it takes 10us to read a row, and we have 1000 rows of pixels, then the exposure of the first row of pixels and the last row of pixels starts/ The end time is about 10ms apart, so that when shooting sports physics, the image will be significantly distorted; correspondingly, the global shutter means that all pixels have the same exposure start and end time