cascode amplifier

A cascode amplifier stacks a common-source (or common-emitter) gain stage directly underneath a common-gate (or common-base) stage, letting the bottom device do the transconductance work while the top device shields its drain from ever seeing the full output swing — that shielding simultaneously kills the Miller effect that would otherwise choke bandwidth and multiplies the output impedance by roughly the top device's own gain, which is why the cascode shows up everywhere a circuit needs high gain, high bandwidth, and high output impedance at the same time rather than having to pick just one.

Clamping the bottom device's drain voltage is the entire mechanism, and everything else follows from it. A bare common-source stage's drain swings with the full output signal, which is exactly what multiplies its gate-drain capacitance through the Miller effect and wrecks bandwidth; stacking a common-gate device on top presents a very low impedance at its own source node (which is the bottom device's drain), so no matter how much the final output swings, the node between the two stacked devices barely moves at all — the bottom transistor effectively never sees its own output voltage change, which is why its internal capacitances stop mattering nearly as much.

The output impedance gain is multiplicative, not additive, which is what makes the cascode dramatically better rather than just somewhat better. A single common-source stage's output impedance is roughly ro, the transistor's own drain-source resistance; a cascode's output impedance is approximately gm of the top device times ro of the bottom device times ro of the top device — a product of three terms instead of one, routinely pushing output impedance up by one to two orders of magnitude, which directly translates into higher voltage gain (gain is transconductance times output impedance) without needing a larger bias current or a bigger device.

Telescopic and folded cascode topologies are just two different ways of stacking the same idea inside real op-amp designs. A telescopic cascode stacks the cascode devices directly on top of the input differential pair, which is simple and fast but burns through supply headroom quickly since every added device needs its own voltage drop; a folded cascode redirects the signal current sideways into a separate cascode branch instead of stacking straight up, trading some bandwidth for significantly more usable output swing — this exact tradeoff is why folded-cascode op-amp front ends are so common in low-voltage CMOS processes where headroom is scarce, while telescopic cascodes still dominate where raw speed matters more than swing.

The output impedance gain always comes paired with less voltage headroom, and no cascode design escapes that trade. Every stacked device needs its own VDS or VCE margin to stay in its active region, so a two-transistor cascode needs roughly twice the headroom of a single bare stage before clipping, which is a serious constraint in modern low-voltage processes where supply rails keep shrinking — this single tension (impedance and bandwidth gain versus available swing) is the reason cascode design in advanced CMOS nodes has become its own specialized subfield, with folded, gain-boosted, and regulated cascode variants all existing specifically to recover some of that lost headroom.

Read a cascode amplifier through a *voltage-shielding* lens rather than a *stacked-transistor* lens: the second device's only job is to prevent the first device's drain from ever experiencing the output signal directly, and that single act of shielding simultaneously removes the Miller bottleneck and multiplies the output impedance — two separate-sounding benefits that actually come from the exact same cause.