AM-AM Distortion

As the instantaneous input envelope amplitude increases, the amplifier's incremental gain deviates from its linear small-signal value. This gain compression causes the output amplitude to saturate rather than increase proportionally.

• Linear region: Output amplitude scales linearly with input (constant gain)
• 1 dB compression point (P1dB): Output power where gain drops by exactly 1 dB from linear value
• Saturation region: Output power plateaus regardless of input increase

The compression characteristic is typically modeled using Rapp, Saleh, or polynomial models that map instantaneous input power to instantaneous output power.

AM-AM nonlinearity directly causes spectral regrowth by distorting the envelope of modulated signals. When a band-limited signal passes through a compressive nonlinearity, the amplitude clipping generates out-of-band frequency components that spill into adjacent channels.

• The sharper the compression knee, the more severe the spectral regrowth
• Third-order nonlinearities produce the most problematic adjacent channel interference
• Spectral regrowth bandwidth is typically 3-5x the original signal bandwidth for severe compression

This directly degrades Adjacent Channel Leakage Ratio (ACLR) and violates regulatory spectral masks.

AM-AM and AM-PM distortion are the two fundamental nonlinear mechanisms in power amplifiers, often occurring simultaneously but requiring separate characterization.

• AM-AM: Amplitude-to-amplitude conversion — affects output magnitude only
• AM-PM: Amplitude-to-phase conversion — introduces input-dependent phase shift
• AM-AM dominates near saturation and primarily causes spectral regrowth
• AM-PM contributes to spectral asymmetry in adjacent channel emissions

Digital predistortion systems must compensate for both effects independently using complex baseband models that correct magnitude and phase simultaneously.

Pure AM-AM distortion is memoryless — the output at any instant depends only on the instantaneous input envelope. However, real amplifiers exhibit memory effects where past inputs influence current output.

• Memoryless AM-AM: Modeled by instantaneous transfer functions (Rapp, Saleh)
• Quasi-memoryless: AM-AM with frequency-independent AM-PM
• Memory effects: Thermal trapping, bias circuit modulation, and charge storage cause frequency-dependent behavior

Memoryless AM-AM models are insufficient for wideband signals where electrical memory causes the compression characteristic to vary with signal bandwidth and modulation rate.

Several mathematical frameworks capture AM-AM distortion with varying accuracy and complexity:

• Saleh model: Two-parameter formula widely used for traveling wave tube amplifiers
• Rapp model: Captures solid-state power amplifier compression with a smoothness factor
• Polynomial models: Odd-order Taylor series (3rd, 5th, 7th order) for mild nonlinearities
• Cann model: Extended Rapp with additional parameters for improved saturation fitting
• Look-up tables (LUTs): Directly map quantized input amplitudes to predistorted outputs

The choice depends on the compression characteristic sharpness and required predistortion accuracy.

AM-AM distortion directly degrades Error Vector Magnitude (EVM) by compressing constellation points non-uniformly based on their amplitude.

• Outer constellation points experience more compression than inner points
• QAM and APSK modulations are particularly sensitive due to amplitude-dependent symbol positions
• Higher-order modulations (64-QAM, 256-QAM) require tighter AM-AM linearity
• EVM degradation from AM-AM is deterministic and correctable via digital predistortion

For 5G NR signals with OFDM waveforms, the high PAPR means instantaneous amplitudes frequently drive the amplifier into compression, making AM-AM correction essential for meeting EVM requirements.

The complementary nonlinear mechanism where the phase shift introduced by a power amplifier varies with the instantaneous input signal envelope. While AM-AM distorts magnitude, AM-PM distorts phase, and the two effects combine to create complex, asymmetric spectral regrowth patterns. Memory effects often manifest more strongly in AM-PM characteristics, making joint compensation essential for wideband signals.

The output power level at which a power amplifier's gain deviates from its linear small-signal value by 1 dB. This metric defines the practical onset of significant AM-AM distortion. Operating beyond P1dB causes progressive gain compression, where further input power increases yield diminishing output power increments, directly generating spectral regrowth and in-band distortion.

A phenomenon where a power amplifier's current output depends on past input states due to thermal dynamics, electrical biasing circuit time constants, and semiconductor trapping effects. Memory effects cause the AM-AM characteristic to become frequency-dependent, meaning the distortion pattern changes with signal bandwidth and modulation rate. Static AM-AM correction alone cannot compensate for memory-induced distortion.

The deliberate reduction of a power amplifier's average operating power below its saturation or compression point to improve linearity. By operating in the linear region of the AM-AM transfer curve, spectral regrowth is minimized. However, this trades power efficiency for signal fidelity—a critical design tension in battery-powered and infrastructure equipment where every decibel of back-off reduces DC-to-RF conversion efficiency.

Nonlinear signal products generated at sum and difference frequencies when multiple signals pass through a nonlinear device. AM-AM distortion is the root cause of IMD, as the nonlinear transfer function creates mixing products. Third-order intermodulation products (IMD3) are particularly problematic because they fall within or immediately adjacent to the original signal bandwidth, directly contributing to spectral regrowth and ACLR degradation.

The direct manifestation of AM-AM distortion where amplifier gain decreases as input power increases. In the linear region, gain is constant; in compression, the output fails to track input proportionally. Gain compression is typically quantified by the deviation from ideal linear gain and is the primary target of digital predistortion algorithms that apply an inverse nonlinearity to restore linear amplification.