2026-09-01
The Clapp oscillator is a Colpitts variant that solves one nagging problem: transistor parasitic capacitances (Cbe, Cce, Miller capacitance) sit in parallel with the tank capacitors and drift with temperature, bias, and supply voltage. In a standard Colpitts, those parasitics directly shift the resonant frequency. James Kilton Clapp's 1948 fix was elegant: add a third capacitor C3 in series with the tank inductor, and make it much smaller than C1 and C2.
The tank now resonates at:
f0 = 1 / (2π√(L·Ceff)), where 1/Ceff = 1/C1 + 1/C2 + 1/C3
Because series capacitance is dominated by the smallest term, C3 sets the frequency almost entirely. C1 and C2 — the ones swamped by transistor parasitics — become nearly irrelevant to the frequency, though they still set the feedback ratio C1/C2 for oscillation startup (Barkhausen).
Design example — 10 MHz Clapp with a JFET:
The ratio C1/C2 = 1 gives unity feedback attenuation, which a JFET with gm·RD > 1 easily overcomes. Notice how huge C1 and C2 are — you want them massive so parasitics disappear into the noise.
Rule of thumb: Make C1 and C2 at least 10× larger than the largest expected transistor parasitic capacitance, and make C3 at least 10× smaller than C1||C2. This decouples frequency from active-device drift.
Real-world use: Clapp topology dominates in HF and low-VHF signal generators, amateur radio VFOs (Variable Frequency Oscillators — replacing C3 with a variable cap or varactor gives clean tuning), and lab-grade RF sources up to ~200 MHz where crystals are impractical but Colpitts drift is unacceptable. The classic Heathkit and Collins receivers used Clapp VFOs for exactly this reason.
The tradeoff: C3 must be a high-Q, low-TC part — an NP0/C0G ceramic or silvered-mica cap. Use an X7R here and you'll get thermal drift worse than a plain Colpitts. Also, tuning range shrinks compared to Colpitts because C3 alone dominates — for wideband VFOs, a Colpitts with variable C1/C2 tunes further.