predistortion

Predistortion is a linearization technique used in communications and radar systems to compensate for the nonlinear and memory effects of power amplifiers and other nonlinear devices. The predistorter applies the inverse of the amplifier’s transfer characteristics so that, after passing through the nonlinear stage, the overall system response is approximately linear.

In practice, nonlinear distortion in amplifiers manifests as AM/AM and AM/PM distortion and, when memory effects are present, distortion that depends on signal history. Predistortion attempts to pre-distort the signal in the opposite way, so the combination yields a flatter gain, reduced spectral regrowth, and improved adjacent-channel leakage. Predistortion can be implemented as analog circuits or in digital form.

Analog predistortion uses a circuit that approximates the inverse transfer function of the amplifier, typically operating at RF frequencies. Digital predistortion builds a model of the inverse function in baseband or digital IF, using polynomials (such as memory polynomial or generalized polynomial models), look-up tables, or more complex models, and applies predistortion to the digital signal before upconversion. The model parameters are adapted using a feedback or indirect learning architecture to minimize residual distortion, often measured by ACPR and EVM.

Predistortion is widely used in wireless transmit chains, including cellular base stations, satellite uplinks, and radar, to achieve high linearity without sacrificing efficiency. Challenges include modeling accuracy for wide bandwidths, memory effects, temperature drift, and ensuring stability and real-time adaptation. Two-tone tests and other metrics are used to evaluate performance.