Cascode amplifier

A cascode amplifier is a common-source (or common-emitter) stage feeding directly into a common-gate (or common-base) stage. Stacking the two transistors this way gets you the high gain of a common-source/emitter input and the wide bandwidth of a common-gate/base output at the same time — something neither stage achieves alone.

The conflict it resolves

A single Common-source amplifier (or Common-emitter amplifier) gives high gain $|A_v|=g_m(R_D\parallel r_o)$, but it is bandwidth-limited by the Miller effect: its gate–drain (or base–collector) feedback capacitance is multiplied by $(1+|A_v|)$ and dominates the high-frequency pole. A Common-gate amplifier / Common-base amplifier has no Miller effect (its input and output do not share a swinging feedback capacitor) and so is fast — but its very low input resistance makes it useless as a voltage-input stage on its own.

The cascode combines them so each does what it is good at:

• Bottom transistor — CS/CE. Driven by the input. It is the transconductor: input voltage → signal current $i=g_m{v}_{in}$.
• Top transistor — CG/CB. Its source/emitter sits on the bottom transistor’s drain/collector; its gate/base is held at a fixed AC-ground bias. It passes the signal current straight through to its own high-impedance drain/collector.

Why this is the best of both

The reason the cascode beats a single stage on both axes:

• Bandwidth. The bottom transistor’s drain/collector now sees the low input resistance of the top common-gate/base device ($\approx 1/g_m$ or $r_e$), not a large resistance. With almost no voltage swing at that node, the bottom transistor’s feedback capacitance is barely multiplied — the Miller effect is suppressed. The wide-band CG/CB output then carries the current up to a high impedance with no Miller penalty of its own.
• Gain / output resistance. Looking into the top transistor’s drain/collector you see a much higher output resistance than a single stage (the lower device degenerates the upper one, multiplying its $r_o$ by roughly $g_m{r}_o$). Driving that huge output resistance with the full transconductance $g_m$ of the input device gives a very high voltage gain, $\sim (g_m{r}_o{)}^2$ order of magnitude.

So the cascode delivers high gain and high bandwidth simultaneously. That is why it is the standard gain stage inside operational-amplifier inputs and in RF amplifiers, and why it appears wherever a single common-source/emitter stage’s gain–bandwidth trade-off is unacceptable. The MOSFET version is the CS–CG cascode; the BJT version is the CE–CB cascode; mixed (e.g. CS–CG with one of each) cascodes are also common.

[Background from general knowledge, not the source PDF]