Spectral Regrowth

Spectral regrowth originates from intermodulation distortion (IMD) . When a band-limited signal passes through a nonlinear power amplifier, the odd-order nonlinearities cause spectral components within the signal bandwidth to mix, generating new frequency products. These third-order intermodulation products (IM3) fall directly into the adjacent channels, while fifth-order products (IM5) extend further out. The severity of regrowth is directly proportional to the amplifier's AM-AM and AM-PM distortion characteristics and the signal's peak-to-average power ratio (PAPR) .

ACPR is the definitive regulatory metric for quantifying spectral regrowth. It measures the ratio of power leaked into an adjacent frequency channel to the power in the main transmission channel.

• ACPR Lower: Power in the lower adjacent channel relative to the main channel.
• ACPR Upper: Power in the upper adjacent channel relative to the main channel.
• Typical targets: -45 dBc to -55 dBc for 3GPP compliance. Asymmetry between upper and lower ACPR often indicates significant memory effects in the amplifier.

In an ideal memoryless nonlinearity, spectral regrowth appears as symmetric shoulders on both sides of the main channel. However, real power amplifiers exhibit memory effects—thermal, electrical, and trapping phenomena—that cause the distortion to depend on past signal values. This creates asymmetric spectral regrowth, where the upper and lower adjacent channel power levels differ significantly. Asymmetry is a critical indicator that a simple memoryless predistorter will be insufficient, requiring a memory polynomial or Volterra series based linearizer.

The bandwidth of spectral regrowth is a multiple of the original signal bandwidth. For a nonlinearity dominated by third-order distortion, the regrowth extends to three times the signal bandwidth. Fifth-order distortion expands it to five times, and so on.

• A 20 MHz LTE signal generates regrowth spanning approximately 60 MHz.
• A 100 MHz 5G NR signal can produce regrowth across 300 MHz. This bandwidth expansion dictates the linearization bandwidth required of the digital predistorter and the sampling rate of the feedback path.

While ACPR measures out-of-band distortion, EVM quantifies the in-band signal quality degradation caused by the same nonlinearity. Spectral regrowth and EVM are correlated manifestations of amplifier distortion.

• Hard compression (severe AM-AM) degrades both EVM and ACPR.
• AM-PM conversion primarily impacts EVM but also contributes to spectral asymmetry.
• Optimizing for ACPR alone can sometimes degrade EVM if the predistorter over-compensates. A robust linearization strategy must balance both metrics simultaneously.

Spectral regrowth is not a fixed characteristic; it depends heavily on the statistical distribution of the input signal.

• High PAPR signals (e.g., OFDM) drive the amplifier deeper into compression, generating more severe regrowth.
• Constant envelope signals (e.g., GMSK) produce minimal regrowth.
• The complementary cumulative distribution function (CCDF) of the signal determines how often the amplifier operates in its nonlinear region. Accurate behavioral modeling requires using test signals with statistics representative of the deployed waveform.

The primary regulatory metric for quantifying spectral regrowth. ACPR is the ratio of the integrated power in an adjacent frequency channel to the power in the assigned channel. It is a direct measure of out-of-band emissions caused by intermodulation distortion. Regulatory bodies like the FCC and 3GPP mandate strict ACPR limits to prevent interference.

• Measured in dBc (decibels relative to the carrier).
• Directly worsened by AM-AM and AM-PM distortion.
• The target metric for most Digital Pre-Distortion (DPD) systems.

The physical mechanism that causes spectral regrowth. When a band-limited signal with multiple frequency components passes through a nonlinear device, the components mix to create new, unwanted frequencies. Third-order intermodulation products (IM3) fall closest to the original carrier and are the primary source of adjacent channel interference.

• Caused by the nonlinear transfer function of the Power Amplifier.
• Even-order products fall far out of band and are often filtered.
• Odd-order products fall in-band and in adjacent channels.

The dependence of a power amplifier's current output on past input values. Memory effects cause the spectral regrowth pattern to become asymmetric, meaning the upper and lower adjacent channels have different power levels. This asymmetry cannot be corrected by memoryless linearization.

• Electrical memory is caused by bias circuit impedance.
• Thermal memory is caused by die temperature fluctuations.
• Trapping effects in GaN HEMTs cause slow, complex memory.

While ACPR measures out-of-band distortion, EVM measures the in-band signal quality degradation caused by the same nonlinearity. Spectral regrowth and EVM are linked; a highly nonlinear amplifier will exhibit both poor ACPR and high EVM. DPD systems aim to improve both simultaneously.

• Represents the deviation of actual symbols from ideal constellation points.
• Excessive AM-PM distortion is a major contributor to EVM degradation.
• Critical for high-order QAM modulation schemes (e.g., 256-QAM).

The primary active linearization technique used to combat spectral regrowth. A DPD system applies an inverse nonlinear characteristic to the signal in the digital baseband before the power amplifier. The cascaded response of the DPD and PA becomes linear, dramatically reducing out-of-band emissions.

• Relies on accurate behavioral models like the Memory Polynomial.
• Uses Indirect Learning Architecture (ILA) or Direct Learning Architecture (DLA).
• Enables the PA to operate closer to saturation, improving efficiency.

A signal characteristic that directly influences the severity of spectral regrowth. Signals with a high PAPR, such as OFDM used in 5G and Wi-Fi, force the power amplifier to operate with a large back-off to avoid clipping and nonlinear distortion. High back-off reduces efficiency, creating a fundamental trade-off.

• Crest Factor Reduction (CFR) is used to lower PAPR before the PA.
• High PAPR signals are more susceptible to generating spectral regrowth.
• The combination of CFR and DPD is standard in modern base stations.