Common Drain Amplifier

A common-drain amplifier — universally called a source follower — ties the drain to a fixed supply rail, drives the input at the gate, and takes the output off the source, which makes the output chase the input almost exactly: voltage gain sits just under 1, but input impedance is enormous and output impedance is small, the complete opposite tradeoff from the common-source stage, which is why this configuration is never chosen for gain and almost always chosen for isolation.

The gain being less than 1 isn't a design flaw, it's the point — the stage is a buffer, not an amplifier. Because negative feedback is essentially built into the topology — the source node directly sets VGS, which directly sets the current, which directly sets the source voltage again — the output tracks the input almost one-to-one, with the small shortfall caused by VGS itself shifting slightly as drain current changes with signal swing; designers who need actual voltage gain never reach for a common-drain stage, they reach for it specifically when they need to move a signal from a high-impedance node to a low-impedance one without disturbing the signal's value.

Output impedance around 1/gm is what makes it a genuine load driver, unlike a bare common-source stage. A common-source amplifier's output impedance is set by its drain resistor or current-source load, which can be large; a common-drain stage's output impedance is approximately the reciprocal of its transconductance, often tens to a few hundred ohms depending on bias current, which is low enough to drive a cable, a following stage's input capacitance, or an ADC's sampling capacitor without the output voltage sagging under load — exactly the job a bare high-impedance node cannot do.

It is the standard output buffer at the end of almost every on-chip analog signal path. Op-amp output stages, voltage reference buffers, oscillator output drivers, and the final stage of countless sensor interface circuits use a source-follower (or its bipolar twin, the emitter follower) specifically because the preceding gain stage — typically a common-source or cascode stage — produces a high-impedance, easily-disturbed voltage that needs to be delivered to a real-world load without being degraded by that load's own impedance; stacking a follower after the gain stage decouples "produce the right voltage" from "drive the actual load" into two separate, independently optimizable jobs.

Its limitations are the direct mirror image of common-source's strengths, which is exactly the tradeoff that makes the pair useful together. Because voltage gain is near 1, a source follower contributes essentially nothing to a design's overall gain budget, and because the output voltage sits one VGS below the gate voltage, it also eats into the available signal swing before clipping against the supply rails — body-effect and VGS variation with current also make the "gain" slightly less than 1 and slightly nonlinear, a second-order effect that precision buffer designs have to account for with techniques like bootstrapping the body terminal or using a cascoded follower.

Read a common-drain amplifier through a *job-separation* lens rather than an *amplifier* lens: it deliberately gives up gain to gain something more useful at that point in the signal chain — the ability to deliver a voltage faithfully into a load that would otherwise collapse a high-impedance gain stage's output, which is exactly why a source follower almost never appears alone and almost always appears directly after one.