S4-1.3 Multistage & High-Frequency Amplifiers

Standard analog-electronics theory — written August 2026

What this is and why it exists

One stage rarely delivers enough gain, so real amplifiers are chains. Chaining multiplies the gains — and multiplies the frequency limits into each other, which is the actual subject of this unit. It also retires the h-parameter model: at high frequency the transistor's internal capacitances dominate, and the hybrid-pi model exists to say so honestly.

The vocabulary

  • RC coupling — stages joined through a capacitor: cheap, blocks DC, loses the very low frequencies.
  • Transformer coupling — stages joined magnetically: matches impedances well, bulky and expensive, poor at low frequency.
  • Direct coupling — stages wired straight through: amplifies down to DC, but bias points interlock and drift compounds.
  • Darlington pair — two transistors with the first's emitter feeding the second's base; behaves like one transistor with beta roughly the product of the two.
  • Hybrid-pi model — the high-frequency small-signal model: transconductance, base resistance, and the capacitances the h-parameters pretend do not exist.
  • f_T — the frequency at which the transistor's short-circuit current gain falls to one; the device's intrinsic speed limit.

The mental model

Stack panes of glass. Each pane (stage) is slightly tinted — it passes the middle frequencies well and dims the edges. Two panes stacked give the product of the tints: the middle gets darker slower than the edges, so the band that passes cleanly is NARROWER than either pane alone. That is the multistage law: overall gain multiplies, overall bandwidth shrinks. A two-stage amplifier built from identical stages has less bandwidth than one of its stages — the gain-bandwidth budget is being spent.

Why do the edges dim at all? At the high end, the hybrid-pi model answers: the base-collector capacitance, Miller-multiplied by the stage gain, shunts the input signal away as frequency rises. Following that mechanism to its end gives f_T, where the device's own current gain reaches one and amplification is over regardless of circuit cleverness. Datasheet f_T is the ceiling; Miller effect decides how far below the ceiling your circuit actually operates.

The named pairings are tools with characters: CE-CE for raw gain, CE-CB (the cascode) because the CB stage shields the CE stage from Miller multiplication and keeps bandwidth, and the Darlington (CC-CC) for enormous current gain and a gentle load on the source, paying for it with two base-emitter drops and slower switching.

What you should now be able to explain or do

Pick a coupling scheme for a stated job and name its cost. Compute overall gain and estimate overall bandwidth for cascaded identical stages. Explain what the cascode buys and what mechanism it defeats. Read f_T from a datasheet and say what it does and does not promise.

Check yourself

Gain multiplies — twenty times twenty is four hundred. Bandwidth shrinks below a single stage's, because both stages' roll-offs act on the same signal.

The CB stage holds the CE stage's collector at nearly constant voltage, so the CE stage sees almost no voltage gain across its base-collector capacitance — the Miller multiplication never happens.

The frequency where short-circuit current gain falls to one. Beyond it the transistor no longer amplifies current — no circuit arrangement recovers gain the device itself has run out of.

Two base-emitter drops of headroom, higher saturation voltage, and slower response — the second transistor cannot turn off quickly because nothing pulls its base down hard.

Direct coupling — capacitors and transformers both block slow changes. The cost is interlocked bias design and drift that amplifies along with the signal.

Go deeper

Back to Multistage & High-Frequency Amplifiers: work through the checklist