Why matched transistor pairs are important (and expensive)

Integrated circuits often implement complex black-box like behavior, and contain thousands to millions of transistors. Even seemingly simple circuits like operational amplifiers contain dozens of transistors at a minimum. So, you might be surprised to find an IC that has a few transistors, and that’s it- nothing else. One of these ICs that you can find here at M5 on our parts wall is the ALD1105; so why does it cost nearly 10 dollars per IC?

The ALD1105, made by Advanced Linear Devices, is an IC that contains two matched pairs of N-channel and P-channel MOSFETs. The kicker here, is that both pairs have nearly identical electrical characteristics. Datasheets often define ranges for characteristics such as threshold voltage, drain current, and transconductance for transistors. For high-precision applications, it is up to you to measure these for each transistor you buy. But, the ALD1105 provides two pairs of near perfectly matched transistors.

Semiconductor characteristics, especially PN/NP junctions, differ wildly based on manufacturing processes, temperature, and the individual dies they are made on. These matched transistors are picked out from the same manufacturing die, often extremely close to each other to ensure the same thermal characteristics. Hence why the ICs cost an arm and a leg to buy (relatively).

So why bother? Surely, you could test hundreds of transistors until you find a good enough match, but what are these used for? Matched transistors are used in many analog building blocks, often those that require differential operation. Matched transistor pairs, both MOS and BJTs can be found in basic circuits such as current mirrors, differential amplifiers, audio amplifiers, and other high-precision applications that require accurate temperature drift.

Semiconductors are weird, but luckily, manufacturing takes away the pain of rummaging through hundreds of transistors to find two that behave the same.