Adjacent Channel Leakage Ratio (ACLR)

The primary physical mechanism degrading ACLR. As a PA approaches its saturation point (P1dB), the amplitude-to-amplitude (AM-AM) and amplitude-to-phase (AM-PM) conversion curves become highly nonlinear. This spectral regrowth causes the modulated signal's bandwidth to expand into adjacent channels. The specific semiconductor technology—LDMOS, GaN, or GaAs—determines the severity and memory characteristics of this nonlinearity.

Modern wideband signals like OFDM (used in 5G and Wi-Fi) exhibit a high PAPR, often exceeding 10 dB. To avoid clipping and severe ACLR degradation, the PA must operate with a significant output power back-off (OBO). A higher PAPR forces the amplifier to operate further from saturation, directly trading off power efficiency for spectral purity. Crest Factor Reduction (CFR) is applied specifically to manage this trade-off.

The PA's output is not solely a function of the instantaneous input; it depends on past signal states. Electrical memory effects arise from bias circuit impedance and harmonic terminations, while thermal memory effects are caused by dynamic die heating. These effects create an asymmetric spectral regrowth pattern, making ACLR worse on one side of the carrier. Simple memoryless linearization cannot compensate for this.

While CFR is used to improve efficiency, the process itself is a nonlinear operation that generates distortion. Hard clipping creates sharp discontinuities, causing severe spectral splatter and degrading ACLR. Advanced techniques like peak windowing and pulse injection are designed to limit peaks while confining the resulting distortion energy within the transmit channel, minimizing the impact on adjacent channel leakage.

In direct-conversion transmitters, imperfections in the IQ modulator—specifically gain imbalance and quadrature skew—create an image of the transmit signal. This unwanted image can fall directly into an adjacent channel, catastrophically degrading ACLR. This is a linear impairment that cannot be corrected by DPD alone and requires dedicated IQ imbalance compensation algorithms.

In multi-carrier or carrier aggregation scenarios, the total composite signal bandwidth increases, and intermodulation products can fall into adjacent carriers. The frequency separation between carriers and the edge of the allocated band dictates the linearization bandwidth required from the DPD system. A wider signal bandwidth demands a DPD system with a proportionally higher observation path bandwidth to capture the 3rd and 5th order distortion terms.

The engineering discipline focused on suppressing out-of-band emissions caused by power amplifier nonlinearity. When a digitally pre-distorted signal passes through a PA, spectral regrowth occurs due to intermodulation distortion, filling adjacent channels with unwanted energy. Mitigation strategies include memory polynomial DPD, CFR-aware linearization, and iterative clipping-and-filtering techniques that jointly optimize ACLR and EVM. The goal is to maintain a clean spectral mask without sacrificing amplifier efficiency.

A measure of in-band distortion that quantifies the deviation of received constellation points from their ideal reference positions. EVM and ACLR are fundamentally linked through the nonlinear transfer function of the power amplifier. Aggressive crest factor reduction improves PAPR but introduces clipping distortion that degrades EVM. DPD systems must balance these competing metrics: excessive linearization for ACLR can inadvertently increase in-band noise, while prioritizing EVM may leave adjacent channel emissions above regulatory limits.

Regulatory bodies like 3GPP and ETSI define spectral masks that specify maximum permissible emission levels as a function of frequency offset from the carrier. ACLR is the primary metric for demonstrating compliance. A typical mask requires emissions to fall below -45 dBc at the adjacent channel edge and -60 dBc for alternate channels. Nonlinear CFR algorithms must be designed with these masks as hard constraints, often using spectrally-weighted clipping or tone reservation to confine distortion energy within the mask boundaries.

The ratio of peak instantaneous power to average power in a transmit signal, typically expressed in dB. High-PAPR signals like OFDM force power amplifiers to operate with significant back-off to avoid compression, reducing efficiency from 50%+ to below 25%. This efficiency loss is the root cause of the ACLR problem: engineers push amplifiers closer to saturation to recover efficiency, which generates intermodulation products that leak into adjacent channels. CFR and DPD exist to break this trade-off.

An iterative CFR technique where signal peaks exceeding a threshold are hard-clipped to reduce PAPR, then low-pass filtered to suppress the resulting spectral splatter. The filtering stage causes peak regrowth, where previously clipped peaks reappear above the threshold, necessitating multiple iterations. Each iteration trades off PAPR reduction against ACLR degradation. Advanced implementations use frequency-domain filtering with configurable mask-aware stopbands to directly control adjacent channel leakage during the clipping process.

The intentional reduction of input drive level to operate a PA within its linear region, expressed as the difference between the amplifier's 1 dB compression point and the average operating power. Back-off is directly proportional to the signal's PAPR: a 10 dB PAPR requires roughly 10 dB of output back-off. Each dB of back-off reduces PA efficiency by approximately 2-3 percentage points. CFR reduces the required back-off by lowering PAPR, while DPD extends the linear operating range, both contributing to improved ACLR at higher efficiency.