AM-PM Distortion

AM-PM conversion arises primarily from voltage-dependent parasitic capacitances in transistor junctions. As the input drive level changes, the input and output capacitances of the active device vary, altering the phase of the transmitted signal. In GaN and GaAs FETs, the gate-to-source capacitance (Cgs) and gate-to-drain capacitance (Cgd) exhibit strong nonlinear dependence on terminal voltages. This effect is compounded in Doherty amplifiers where load modulation dynamically shifts the impedance presented to the main and peaking devices.

AM-PM distortion causes constellation-dependent phase rotation that is correlated with instantaneous signal amplitude. Higher-order modulation schemes are disproportionately affected:

• QPSK: Moderate degradation, primarily at constellation corners
• 16-QAM: Inner and outer constellation points experience different phase shifts
• 64-QAM and 256-QAM: Severe degradation, as closely spaced points become indistinguishable
• OFDM: High PAPR signals experience time-varying phase distortion across symbols

The result is increased Error Vector Magnitude (EVM) and degraded bit error rate performance.

AM-AM and AM-PM distortion are coupled phenomena that occur simultaneously in real amplifiers. While AM-AM describes amplitude compression or expansion, AM-PM captures phase rotation. Key relationships:

• Both intensify near the 1 dB compression point and saturation
• In Class AB amplifiers, AM-PM typically increases monotonically with drive level
• In Doherty architectures, AM-PM exhibits complex behavior due to load modulation from the peaking amplifier
• Memory effects cause frequency-dependent variations in both AM-AM and AM-PM characteristics
• Joint compensation requires complex-valued predistortion addressing both impairments simultaneously

AM-PM distortion is quantified by measuring the phase difference between input and output as a function of instantaneous input power. Standard characterization methods include:

• Vector Network Analyzer (VNA) measurements with power sweeps to extract AM-PM transfer curves
• Real-time vector signal analysis using modulated test signals to capture dynamic behavior
• Two-tone intermodulation tests revealing phase asymmetry in IM3 products
• Complex envelope extraction comparing ideal and measured baseband waveforms

The resulting AM-PM characteristic curve plots output phase shift versus input power and serves as the basis for predistorter design.

Digital predistortion (DPD) compensates AM-PM distortion by applying an inverse phase rotation in the digital baseband before the power amplifier. The predistorter introduces a phase advance that exactly cancels the amplifier's phase lag at each instantaneous power level. Effective compensation requires:

• Complex gain models that capture both magnitude and phase nonlinearity
• Memory polynomial structures with complex coefficients to address frequency-dependent AM-PM
• Look-up tables (LUTs) indexed by instantaneous amplitude storing pre-computed phase corrections
• Adaptive coefficient updates to track changes due to temperature, aging, and channel frequency

AM-PM distortion becomes increasingly challenging at wider bandwidths and higher frequencies:

• Wideband signals (100+ MHz): Frequency-dependent AM-PM requires models with memory depth to capture dispersion across the band
• mmWave (28 GHz, 39 GHz): Phase noise and AM-PM interact, requiring joint estimation
• Massive MIMO arrays: Each antenna path exhibits unique AM-PM characteristics demanding per-branch DPD
• Carrier aggregation: Cross-modulation between carriers creates additional AM-PM products
• GaN technology: While offering higher efficiency, GaN devices exhibit stronger AM-PM nonlinearity requiring more sophisticated compensation

The companion nonlinear impairment to AM-PM, Amplitude-to-Amplitude distortion describes how input signal amplitude variations cause nonlinear changes in output amplitude. While AM-PM affects phase, AM-AM causes gain compression or expansion. Both arise from the same nonlinear transfer characteristic and must be jointly compensated in digital predistortion systems. Key characteristics:

• Measured via single-tone power sweep to plot Pin vs. Pout
• 1 dB compression point (P1dB) marks the onset of significant nonlinearity
• Saturation region exhibits severe AM-AM and AM-PM distortion simultaneously

The dependence of a power amplifier's current output on past input values, causing frequency-dependent distortion that complicates AM-PM behavior. Memory effects mean the phase shift at any instant depends not just on the current amplitude but on the envelope history. Sources include:

• Thermal memory: Die temperature changes with signal power, altering transistor parameters over milliseconds
• Electrical memory: Bias circuit impedance variations at envelope frequencies, affecting modulation bandwidth
• Trapping effects: Charge capture/release in semiconductor defects, particularly in GaN HEMT devices

A lowpass equivalent signal representation that captures both amplitude and phase modulation while omitting the high-frequency carrier. Essential for AM-PM analysis because phase distortion appears directly in the complex envelope. The baseband equivalent maps the bandpass signal to:

• In-phase (I) component: Real part of the complex envelope
• Quadrature (Q) component: Imaginary part of the complex envelope
• AM-PM distortion manifests as a rotation of the complex vector dependent on instantaneous magnitude

A critical in-band signal quality metric that directly reflects AM-PM distortion impact. EVM measures the magnitude of the vector difference between the ideal reference constellation point and the actual transmitted symbol. AM-PM distortion causes:

• Phase rotation of constellation points, especially at higher amplitudes
• Increased EVM in outer constellation rings of QAM schemes
• Degraded bit error rate (BER) performance
• EVM is expressed as a percentage or in dB relative to the reference signal power

A simplified Volterra series model that efficiently captures both nonlinearity and memory effects, including AM-PM distortion with memory. The memory polynomial includes only diagonal terms, dramatically reducing coefficients compared to full Volterra while maintaining accuracy. Structure:

• Combines polynomial nonlinearity with FIR filter taps at each order
• Each term: x(n-m) * |x(n-m)|^(k-1) for delay m and order k
• Odd-order terms dominate for bandpass nonlinearities
• Widely used as the baseline predistorter architecture in FPGA implementations

The appearance of unwanted frequency components in adjacent channels caused by intermodulation distortion when a band-limited signal passes through a nonlinear amplifier. AM-PM distortion is a primary contributor because phase nonlinearity generates asymmetric spectral regrowth. Key regulatory metric:

• Adjacent Channel Power Ratio (ACPR) quantifies leakage into neighboring channels
• AM-PM typically produces more upper-sideband regrowth than AM-AM alone
• Must meet 3GPP/ETSI spectral mask requirements for commercial deployment