Why does AM-AM conversion happen?

An ideal amplifier has a perfectly flat gain (e.g., exactly 10 dB) regardless of the input power. However, real transistors are limited by their DC power supply. If you keep increasing the input RF power, the output RF voltage eventually hits the 'rail' (the maximum voltage of the DC supply). It physically cannot go any higher. Therefore, as you approach that limit, the gain slowly starts dropping from 10 dB down to 9 dB, then 8 dB. This drooping gain curve is the AM-AM conversion.

How does AM-AM ruin a 64-QAM signal?

In 64-QAM, data is encoded by transmitting dots at specific amplitudes (distances from the center of the constellation graph). The outer dots are high-power; the inner dots are low-power. Because AM-AM compression only affects the high-power signals, the inner dots of the 64-QAM grid are amplified perfectly, but the outer dots are compressed inwards. The perfect square grid warps into a rounded, collapsed shape, causing the receiver to misinterpret the data (high EVM).

How is AM-AM conversion fixed?

The simplest fix is 'back-off'—turning the input power down so the signal never reaches the nonlinear saturation zone, but this ruins thermal efficiency. The modern fix is Digital Predistortion (DPD). The digital baseband processor measures the exact AM-AM curve of the amplifier. If the amplifier naturally compresses a peak by 2 dB, the DPD algorithm artificially boosts that specific peak by 2 dB *before* sending it to the amplifier. The pre-boost and the amplifier compression perfectly cancel out, resulting in a flawless, linear output.

System Performance

AM-AM Conversion

A telecom engineer tests a new base station amplifier using a 256-QAM OFDM signal. When they test the signal at 1 Watt, the constellation diagram on the analyzer is a perfect, flawless square grid. The amplifier is operating in its linear region. However, to maximize range, the engineer turns the power up to 20 Watts. The amplifier begins to enter saturation. At this high power, the gain of the amplifier is no longer constant; it starts dropping (compressing) during the highest voltage peaks of the OFDM signal. This nonlinear amplitude distortion is called AM-AM Conversion. Because the gain is dropping only on the peaks, the outer points of the 256-QAM constellation are violently crushed inwards toward the center, while the inner points remain mostly intact. The square grid morphs into a blurred circle. The receiver's error correction fails, and the data link drops entirely. To fix it, the engineer must activate a Digital Predistortion (DPD) algorithm to mathematically un-crush the peaks.

Category: System Performance

Cause: Amplifier gain compression near saturation

Primary Mitigation: Digital Predistortion (DPD) or Power Back-off

The AM-AM Compression Curve

Input Power Level	Amplifier Gain	AM-AM Distortion	Signal Integrity
Low (Linear Region)	Constant (e.g., 15.0 dB)	Zero (Output tracks input perfectly)	Excellent (Low EVM)
High (P1dB Point)	Drooping (e.g., 14.0 dB)	Moderate (Peaks are crushed by 1 dB)	Degraded (Outer QAM points warp)
Maximum (Saturation)	Collapsing (e.g., 8.0 dB)	Severe (Output power flatlines)	Destroyed (Massive clipping)

AM-AM Measurement:
AM-AM conversion is typically plotted on a graph where the X-axis is the Input Power (in dBm) and the Y-axis is the Output Power (in dBm). For a perfectly linear amplifier, the plot is a perfectly straight line with a slope of 1. As the amplifier saturates, the curve bends downwards. The severity of the AM-AM distortion is the vertical distance between the ideal straight line and the actual bending curve.

The Taylor Series Expansion:
Mathematically, an amplifier's output voltage (V_out) is modeled as a polynomial based on the input voltage (V_in):
V_out = a₁V_in + a₂V_in² + a₃V_in³ ...
The 'a₁' term is the linear gain. The 'a₃' term is the primary cause of AM-AM conversion. Because it is negative and cubic, it aggressively subtracts voltage from the output as the input gets large, causing the compression.

Common Questions

Frequently Asked Questions

What is AM-PM conversion?

AM-PM conversion is the sister effect to AM-AM. As the amplifier saturates, not only does the amplitude get crushed (AM-AM), but the internal parasitic capacitances of the transistor change value due to the high voltage. This changing capacitance causes the phase of the signal to shift. So, an increase in Amplitude Modulation (AM) causes an unwanted Phase Modulation (PM). On a QAM constellation, AM-AM crushes the dots inward, while AM-PM causes the dots to swirl in a circular vortex.

Does AM-AM conversion cause out-of-band emissions?

Yes, absolutely. When an amplifier crushes the top of a smooth sine wave (clipping), the wave becomes jagged. Mathematically, a jagged wave is composed of high-frequency harmonics and intermodulation products. This newly created distortion power bleeds into the adjacent frequency channels, ruining the Adjacent Channel Leakage Ratio (ACLR). This is why the FCC enforces strict linearity limits on cell towers.

Do constant-envelope signals suffer from AM-AM?

No. If a signal has a constant envelope (like FM radio or GSM), the input amplitude never changes. Because the input amplitude is locked, the amplifier just sits at a single, fixed point on the AM-AM curve. The gain might be compressed, but because it is *consistently* compressed by the exact same amount at all times, no distortion occurs. This allows constant-envelope signals to be amplified at absolute maximum saturation (100% efficiency).

Related Terms

← Allan Variance Anode (Tube) →

System Performance

AM-AM Compression Simulator

Input your amplifier's 3rd-order intercept point (IP3) and P1dB specifications. Upload your QAM constellation map. Instantly visualize how the mathematical AM-AM and AM-PM curves warp the constellation points and calculate the resulting EVM degradation.

Simulate Constellation Warping