AM-AM Distortion

The fundamental manifestation of AM-AM distortion is the gain compression curve, which plots output power against input power. In an ideal linear amplifier, this relationship is a straight 1:1 line. In reality, as the amplifier approaches its saturation point (P1dB), the gain begins to roll off. The specific shape of this roll-off—how gradually or sharply the amplifier compresses—is determined by the semiconductor physics of the individual transistor and varies measurably between units of the same model.

AM-AM distortion is mathematically modeled using a power series or polynomial expansion, typically expressed as:

• Odd-order terms dominate the in-band distortion behavior
• The coefficients (a1, a3, a5...) represent the amplifier's transfer function
• a1 defines the linear gain
• a3 and higher-order terms capture the compression characteristics Each physical amplifier exhibits a unique set of coefficients due to manufacturing variances in doping profiles and gate oxide thickness.

The 1 dB compression point is a critical metric quantifying AM-AM distortion. It defines the output power level at which the actual gain has dropped by exactly 1 dB from the ideal small-signal gain. This point marks the transition from quasi-linear to non-linear operation. The precise P1dB value varies between individual amplifiers due to:

• Variations in transistor threshold voltage
• Differences in bias circuit component tolerances
• Subtle layout parasitic effects in the integrated circuit

The third-order intercept point (IP3) is a theoretical figure of merit extrapolated from AM-AM measurements at lower power levels. It predicts the amplifier's linearity performance. A higher IP3 indicates better linearity. The relationship between P1dB and IP3 is typically a fixed offset of approximately 9.6 dB for a memoryless third-order non-linearity, but real-world amplifiers deviate from this ideal due to higher-order AM-AM contributions and thermal memory effects.

AM-AM distortion is fundamentally distinct from AM-PM distortion, though both originate from the same non-linear power amplifier. Key differences:

• AM-AM: Amplitude non-linearity—output amplitude is a non-linear function of input amplitude
• AM-PM: Phase non-linearity—output phase shift varies with input amplitude
• AM-AM causes spectral regrowth and in-band signal distortion
• AM-PM causes constellation rotation that varies with signal envelope Both signatures are extracted simultaneously for robust device fingerprinting.

In wideband communication systems, AM-AM distortion cannot be modeled as a simple memoryless non-linearity. Memory effects cause the amplifier's current output to depend on previous input states due to:

• Thermal memory: Junction temperature changes modulate gain dynamically
• Electrical memory: Bias network impedance variations at envelope frequencies
• Trapping effects: Charge capture and release in semiconductor defects These effects create a hysteresis-like spreading of the AM-AM curve, producing a unique 2D signature for each device.

The complementary non-linearity to AM-AM, AM-PM distortion describes the unintended phase shift of a signal that varies with its instantaneous amplitude. While AM-AM compresses the signal's magnitude, AM-PM warps its phase trajectory. Together, they form a complex, device-specific non-linear transfer function that is highly effective for distinguishing otherwise identical transmitters. Analyzing the combined AM-AM and AM-PM response provides a richer fingerprint than either impairment alone.

The parent category of impairment that encompasses AM-AM distortion. It describes the deviation of an amplifier's output from a perfectly scaled version of its input. Key characteristics include:

• 1 dB Compression Point (P1dB): The output power at which the gain deviates from linearity by 1 dB, marking the onset of significant AM-AM distortion.
• Third-Order Intercept Point (IP3): A theoretical metric used to predict intermodulation distortion levels, directly linked to the polynomial coefficients that model AM-AM behavior.

A critical complication in AM-AM distortion modeling. The memory effect means the amplifier's current output depends not just on the instantaneous input amplitude, but on the history of previous signal states. This is caused by:

• Thermal time constants: Die temperature changes with signal envelope, altering transistor gain.
• Electrical time constants: Bias circuit capacitors and inductors have finite discharge times. This transforms a simple static AM-AM curve into a dynamic, history-dependent surface, significantly increasing the uniqueness of the fingerprint.

The primary spectral consequence of AM-AM distortion. When a band-limited signal passes through a non-linear amplifier, the amplitude compression generates intermodulation products that cause the signal's bandwidth to broaden, or 'regrow,' into adjacent channels. The specific shape and level of this spectral regrowth is a direct function of the amplifier's unique AM-AM transfer curve, making it a powerful, remotely observable fingerprint measured by the Adjacent Channel Leakage Ratio (ACLR).

The primary mitigation technique for AM-AM distortion. DPD intentionally pre-distorts the input signal with an inverse model of the amplifier's non-linearity. The goal is for the cascaded DPD + amplifier system to appear perfectly linear. The coefficients of a successful DPD model are a direct mathematical inverse of the device's unique AM-AM and AM-PM signatures, effectively capturing the hardware fingerprint in a digital filter. Neural network-based DPD is a growing field for complex, memory-effect-heavy amplifiers.

A comprehensive, composite metric that quantifies the aggregate impact of all hardware impairments, including AM-AM distortion. EVM measures the magnitude of the vector difference between the ideal constellation point and the actual transmitted point. While not specific to AM-AM alone, the statistical distribution and pattern of the error vectors in the I/Q plane are heavily influenced by the amplifier's compression curve, making EVM a practical, single-number indicator of the fingerprint's strength and the transmitter's signal quality.