ChronotechAn atlas of how we made the world

Semiconductors · Vol. V|CMOS image sensor|1990s

CMOS image sensor

1990s

A camera on a chip: each pixel a photodiode plus transistors, made in the same CMOS lines as logic. Charge-coupled devices were better at first; CMOS won on cost, power, and integration. When a phone carries two or three of them, photography stops being a separate object. The sensor is why the present is drowning in pictures.

Fossum and colleagues (active pixel); many firms · United States, then global

CMOS image sensor — Original plate drawn for this atlas. Not a photograph of a museum object or a copyrighted portrait.

How the mechanism is arranged

Each pixel is a well of charge. Colour filters split the light; the chip reads the wells as a picture.

How it works

Light knocks charge free in a photodiode.
On-pixel transistors amplify and address that charge.
A processor turns the array into a color image.

Why it mattered

Collapsed camera, computer, and network into one slab.
Made computer vision a default, not a lab.
Raised new questions about surveillance and memory.

Who turned the wheel

c. 965–1040Ibn al-Haytham

What it unlocked

For further study

Fossum, IEEE papers on APS. Active-pixel sensor.
Camera industry histories. CCD to CMOS shift.

Check

What manufacturing fact helped CMOS sensors beat CCDs on price?