Envelope Clipping

The simplest form of envelope limiting where any signal sample exceeding a predefined clipping threshold is instantaneously truncated to that threshold level.

• Mechanism: A non-linear memoryless operation applied directly to the I/Q samples.
• Spectral Impact: Generates severe spectral regrowth due to the sharp discontinuities in the time-domain waveform.
• EVM Penalty: Causes significant in-band distortion, degrading the Error Vector Magnitude (EVM).
• Implementation: Trivial to implement but almost never acceptable for wireless standards due to poor Adjacent Channel Leakage Ratio (ACLR).

Advanced techniques that apply smooth amplitude transitions near the saturation point to minimize spectral splatter while still reducing the Peak-to-Average Power Ratio (PAPR).

• Smooth Saturation: Uses polynomial or hyperbolic tangent functions instead of abrupt truncation.
• Spectral Containment: Dramatically reduces out-of-band emissions compared to hard clipping, easing the burden on the Digital Pre-Distortion (DPD) system.
• Trade-off: Achieves better ACLR at the cost of slightly higher in-band distortion or less aggressive PAPR reduction.
• Example: Applying a cubic soft clipper introduces a smoother compression curve, concentrating distortion energy within the channel bandwidth.

A critical design parameter defined as the ratio of the clipping threshold level to the root mean square (RMS) value of the signal envelope.

• Formula: CR = A_clip / σ, where σ is the RMS level.
• Low CR (Aggressive): Results in high PAPR reduction but severe AM-AM distortion and spectral regrowth.
• High CR (Conservative): Preserves signal integrity but provides minimal efficiency gains for the Power Amplifier (PA).
• Optimization: The CR must be jointly optimized with the Power Back-Off of the PA to balance efficiency and linearity.

The statistical properties of the error signal generated by the clipping process, which determines how distortion power is allocated between in-band and out-of-band regions.

• Clipping Noise: Defined as the difference between the original and clipped signal.
• In-Band Noise: Degrades EVM and increases the Bit Error Rate (BER).
• Out-of-Band Noise: Manifests as spectral regrowth, violating the Spectral Mask.
• Filtering: Often followed by frequency-domain filtering to suppress out-of-band components, though this can cause peak re-growth.

A repeated process that alternates between time-domain clipping and frequency-domain filtering to progressively reduce PAPR while strictly controlling spectral leakage.

• Process: Clip the signal in the time domain, transform to frequency domain via FFT, apply a rectangular filter to zero out out-of-band emissions, and transform back.
• Convergence: Typically requires multiple iterations to achieve the target PAPR without violating the Occupied Bandwidth (OBW) limits.
• Peak Re-growth: A key challenge where filtering partially restores clipped peaks, necessitating iterative correction.
• Application: Widely used in OFDM systems like 5G NR and Wi-Fi.

Peak windowing multiplies the signal by a smooth window function centered at detected peaks, offering superior spectral containment compared to raw clipping.

• Window Functions: Uses Gaussian, Kaiser, or raised-cosine windows to smooth the transition around the peak.
• Spectral Efficiency: The smooth time-domain transition results in a much sharper spectral roll-off.
• Computational Cost: Higher complexity than simple clipping due to the need for peak detection and window multiplication.
• Comparison: While clipping truncates a single sample, windowing scales a block of samples, effectively performing pulse shaping on the distortion.

A signal conditioning technique that reduces the Peak-to-Average Power Ratio (PAPR) before amplification. By limiting signal peaks, CFR enables higher average power operation without driving the amplifier into saturation. Peak windowing and noise shaping are advanced CFR methods that provide superior spectral containment compared to hard clipping, directly improving ACLR performance.

Nonlinear signal degradation caused when a power amplifier is driven beyond its saturation point, abruptly truncating waveform peaks. This hard clipping generates severe out-of-band spectral components, degrading ACLR and violating spectral mask requirements. The resulting in-band distortion also degrades Error Vector Magnitude (EVM), creating a fundamental trade-off between efficiency and signal fidelity.

A sophisticated crest factor reduction method that applies a smooth time-domain windowing function—such as Gaussian, Kaiser, or Hamming—to signal peaks exceeding a threshold. Unlike hard clipping, peak windowing produces softer transitions that dramatically reduce spectral splatter. The window duration controls the trade-off between PAPR reduction and EVM degradation.

A signal processing technique that intentionally redistributes quantization or clipping noise energy away from critical in-band frequencies to less sensitive out-of-band regions. By applying a noise transfer function (NTF) that shapes the error spectrum, noise shaping improves ACLR performance while maintaining acceptable in-band EVM. This is essential for meeting stringent spectral mask requirements.

A repeated signal conditioning process that alternately clips signal peaks and applies frequency-domain filtering to remove out-of-band distortion. Each iteration progressively reduces PAPR while controlling spectral regrowth. ICF achieves better performance than single-stage clipping but introduces latency due to multiple FFT/IFFT operations, making it suitable for offline or non-real-time applications.

The two fundamental nonlinear mechanisms that cause clipping and spectral regrowth in power amplifiers:

• AM-AM Distortion: Nonlinear amplitude-to-amplitude conversion causing gain compression
• AM-PM Distortion: Amplitude-dependent phase shift introducing spectral asymmetry Both effects are captured in behavioral models and corrected through digital predistortion (DPD).