Power Amplifier Non-Linearity

The amplitude-to-amplitude transfer characteristic defines how output power compresses relative to input drive. As the amplifier approaches saturation, the gain curve flattens, producing a unique compression profile. The 1 dB compression point (P1dB) and the shape of the transition from linear to saturated operation vary between individual amplifiers due to semiconductor doping variations and transistor geometry differences. This characteristic curve is measured by sweeping input power and recording output power, yielding a device-specific polynomial coefficient set.

The amplitude-to-phase conversion effect introduces an unintended phase shift that varies with instantaneous signal envelope power. Unlike AM-AM distortion, this mechanism causes constellation rotation that changes dynamically with modulation amplitude. The phase shift originates from the voltage-dependent capacitance of the transistor's drain-to-gate junction, which alters the amplifier's input impedance as signal levels change. Each amplifier exhibits a distinct AM-PM curve due to parasitic capacitance variations and bias network tolerances.

The history-dependent behavior of a power amplifier means its current output depends not only on the instantaneous input but also on previous signal states. This arises from two primary mechanisms: electrical memory caused by bias network impedance variations at envelope frequencies, and thermal memory from transistor junction temperature fluctuations that track recent power dissipation. The resulting distortion pattern creates a hysteresis-like trajectory in the AM-AM and AM-PM curves, forming a unique multidimensional signature tied to the amplifier's physical layout and thermal impedance.

When a modulated signal passes through a non-linear amplifier, intermodulation between spectral components generates new frequency content outside the intended channel. This adjacent channel leakage manifests as a broadening of the transmitted spectrum, with the specific shape and amplitude of the regrown sidebands determined by the amplifier's non-linearity coefficients. Third-order intermodulation products are typically dominant, creating shoulders that fall at offsets equal to the signal bandwidth. The asymmetry between upper and lower sidebands often reveals the presence of memory effects.

Non-linear amplification generates integer multiples of the carrier frequency that radiate as spurious emissions. The second harmonic (2f₀) and third harmonic (3f₀) are typically the strongest, with their relative amplitudes determined by the symmetry of the amplifier's transfer function. Even-order harmonics indicate asymmetric clipping behavior, while odd-order harmonics reveal symmetric saturation. Output matching network design partially filters these products, but the residual harmonic levels and their exact frequency spread constitute a hardware-specific fingerprint tied to transistor physics and matching component tolerances.

When multiple carrier frequencies pass through a non-linear amplifier simultaneously, sum and difference products appear at predictable frequency offsets. For two tones at f₁ and f₂, the third-order intermodulation products at 2f₁-f₂ and 2f₂-f₁ fall close to the original carriers and are difficult to filter. The third-order intercept point (IP3) quantifies this behavior, but the precise amplitude and phase of each intermodulation product varies per device. This multi-tone response provides a richer fingerprinting feature space than single-carrier measurements alone.

The amplitude-to-amplitude transfer characteristic describing how input signal magnitude maps to output magnitude. As the amplifier approaches gain compression, the relationship deviates from linear, producing a characteristic compression curve unique to each device. This curve is typically modeled using Saleh or Rapp models and is the primary source of in-band distortion and spectral regrowth.

The amplitude-to-phase conversion effect where input amplitude variations induce unintended phase shifts at the output. Unlike AM-AM distortion, AM-PM affects the phase trajectory of the modulated signal, rotating constellation points in a power-dependent manner. This distortion is particularly damaging to phase-modulated waveforms like QPSK and QAM, and its specific curve serves as a highly discriminative fingerprint feature.

The dependence of the amplifier's current output on previous input states, caused by:

• Thermal time constants: Die temperature changes with signal envelope
• Electrical time constants: Bias network capacitors and inductors store energy
• Trapping effects: Charge capture in semiconductor defects

This creates a history-dependent distortion pattern that cannot be corrected by memoryless pre-distortion and is highly individual to each amplifier's physical construction.

The broadening of a transmitted signal's bandwidth beyond its allocated channel, caused by intermodulation products generated in the non-linear amplifier. The specific spectral shoulder shape—the roll-off pattern of power into adjacent channels—reflects the amplifier's non-linearity order and coefficients. This out-of-band emission is both a regulatory compliance concern and a rich source of identifying features for RF fingerprinting systems.

Unwanted frequency products generated when multiple signal components mix in the non-linear amplifier transfer function. Key characteristics:

• Third-order products (IM3) fall closest to the carrier and are most problematic
• IM3 intercept point (IP3) is a figure of merit for linearity
• The relative amplitudes of IM products form a unique spectral signature determined by the specific polynomial coefficients of the amplifier's non-linearity

Integer multiples of the fundamental carrier frequency generated by the amplifier's non-linear transfer function. The second harmonic (2f₀) and third harmonic (3f₀) are typically strongest, with their precise amplitudes determined by the amplifier's Taylor series expansion coefficients. While harmonics are often filtered before transmission, their residual presence and the specific ratio between harmonic orders provide a hardware-specific identifier exploitable for emitter classification.