Global Shutter CMOS Image Sensor

A global shutter CMOS pixel adds a second storage node inside every pixel so the whole array captures the same instant, trading away some of the light-sensing area that a cheaper rolling shutter pixel gets to keep entirely for its photodiode. A conventional rolling shutter sensor reads out one row at a time, exposing and digitizing row 1 microseconds before row 1080, so anything moving fast enough during that readout window — a spinning propeller, a golf swing, a car badge at a stoplight — shows up warped into the diagonal or wavy "jello" distortion rolling shutter is known for. A global shutter pixel fixes this by exposing every pixel in the array simultaneously, then transferring each photodiode's charge into an in-pixel memory node before readout begins, so the stored charge sits still while the row-by-row digitization happens on its own schedule afterward.

That extra memory node is the whole story structurally, and it is also the whole cost, because it has to sit inside the same pixel pitch as the photodiode without catching stray light itself. Sony's IMX sensors built for machine vision and Omnivision's global-shutter automotive parts shield the storage node under an opaque metal light shield, since any photons that leak into it after the simultaneous exposure ends would blur into the next frame's data — a failure mode specific to global shutter called parasitic light sensitivity (PLS), absent entirely from rolling shutter designs that have no second node to leak into. Fixing PLS costs die area for the shield and often costs fill factor for the photodiode itself, which is why global shutter sensors have historically trailed rolling shutter parts in quantum efficiency and low-light performance even though they win outright on motion fidelity — the reason automotive ADAS cameras, industrial machine-vision lines, and factory robotics arms pay the fill-factor tax for global shutter while most smartphone main cameras do not bother.

Rolling Shutter vs. Global Shutter Pixel a second storage node lets every pixel hold its charge until read-out Rolling shutter pixel photodiode straight to row readout rows exposed/read at slightly different times Global shutter pixel photodiode storage node under metal shield to row readout, on its own schedule all pixels latched at once Storage node adds area and risks parasitic light leakage — the structural cost of motion fidelity
Row Exposure Timing rolling shutter skews each row in time; global shutter does not Rolling shutter row 1 row 360 row 720 row 1080 each row exposed at a different moment Global shutter all rows latched to storage nodes simultaneously Readout still happens row-by-row afterward — only the exposure instant is shared

Correlated double sampling, borrowed from CCD sensors decades ago, is what makes the stored charge usable rather than just stable. Each pixel's memory node is read twice — once as a reset reference, once holding the actual signal charge — and the difference cancels fixed-pattern and reset noise that would otherwise swamp a small simultaneous-exposure signal, which matters more for global shutter than rolling shutter because the charge sits idle in the storage node for the better part of a frame time before it is finally digitized, giving dark current and node leakage more time to accumulate. Teledyne e2v and ON Semiconductor (now onsemi) build global-shutter CMOS lines specifically for barcode scanners, high-speed inspection cameras, and automotive surround-view systems where a warped frame is not a cosmetic flaw but a measurement error, and they quote motion-artifact-free frame rates in the hundreds of frames per second rather than the low-light sensitivity numbers that dominate rolling-shutter smartphone sensor datasheets.

Read a global shutter sensor's spec sheet by its parasitic light sensitivity rejection ratio and its fill factor, not by megapixel count, because those two numbers are the actual trade a designer made to get motion-artifact-free frames. A high PLS rejection ratio means the metal shield over the storage node is doing its job; a low fill factor means it cost real photodiode area to put that shield there, and the two numbers together tell you more about which application the sensor was built for than any resolution spec does.

Take global shutter cmos image sensor further

Ask the copilot about this term, or have our engineers assess it against your process.

Ask CFSGPT about this → Request a technical assessment → Open the CFS app