Why HPAs need backoff
A satellite HPA — a TWTA or SSPA — is linear only at low drive levels. As input power rises toward saturation (Psat), the gain compresses and the output-vs-input curve bends over. Operating with Input Backoff (IBO) and Output Backoff (OBO) — dB below the saturation input/output reference — trades away some output power and EIRP in exchange for linearity, which controls gain compression and limits the intermodulation products generated when several carriers (FDMA) share one amplifier.
Models used
TWTA — Saleh model (Saleh, 1981), normalised so the output reaches exactly Psat at the saturation input reference and rolls over if over-driven, as real TWTs do:
P_out/P_sat = 4ρ / (1+ρ)², ρ = P_in / P_in,sat
SSPA — Rapp model (Rapp, 1991), a smooth, monotonic soft limiter that never turns over — a good fit for solid-state amplifiers and for OFDM/multicarrier nonlinearity studies:
P_out/P_sat = u / (1+up)1/p, u = P_in / P_in,sat
where p is the knee "smoothness" factor (larger p → sharper knee, approaching an ideal limiter). Both formulas are closed-form and invertible, so the calculator converts directly between IBO and OBO without iteration. Gain compression is the operating-point gain minus the small-signal gain G₀ — always ≤ 0 dB for a real compressive amplifier.
Multicarrier (FDMA) backoff
The backoff entered above is the total/composite value seen by the amplifier (the sum of
all carriers). For N equal-power carriers, the per-carrier backoff referenced to a single
carrier driving the amp alone is IBO_carrier = IBO_total + 10·log₁₀(N) (and likewise for OBO) —
adding carriers effectively pushes the amplifier closer to saturation for the same per-carrier power, which
is exactly why multicarrier transponders need more backoff than a single-carrier one.
Third-order intermodulation (C/I₃)
Near saturation, intermodulation products (IM3) generated by two or more carriers in the same HPA can fall back into the desired channels. In the weakly-nonlinear (well-backed-off) regime, a cubic nonlinearity makes the fundamental output grow 1 dB per dB of drive while IM3 grows 3 dB per dB of drive — so C/I₃ improves by 2 dB for every extra dB of input backoff. The slope is per dB of drive, so it applies to IBO, not OBO: the reference (C/I₃, OBO) point you supply (e.g. from a manufacturer's curve) is first carried back through the AM/AM curve above to its equivalent IBO, and the extrapolation runs from there. Once the amplifier is compressed a dB of OBO is worth much more than a dB of IBO, so working the slope directly in OBO would report an optimistic C/I₃. It is only an approximation either way: the real curve flattens out near saturation, where gain compression itself limits how bad IM3 can get.
Feed the resulting output power into the link budget's transmit power / EIRP. All computation runs locally in your browser.