DC Offset

The primary physical mechanism causing DC offset in zero-IF receivers is local oscillator (LO) self-mixing. This occurs when LO leakage from the mixer port radiates to the antenna, reflects off the environment, and is downconverted back to baseband by the same LO. Since the reflected signal is at the identical LO frequency, the mixing product is a zero-frequency DC component. Secondary causes include transistor mismatch in the mixer core and second-order non-linearity in the low-noise amplifier (LNA).

DC offset poses an existential threat to receiver performance because it appears at the center of the baseband spectrum (0 Hz). Key consequences include:

• ADC Saturation: The constant voltage consumes dynamic range, causing the analog-to-digital converter to clip prematurely.
• Amplifier Saturation: Subsequent variable-gain amplifiers (VGAs) saturate, rendering the signal path non-linear.
• EVM Degradation: For OFDM systems, the DC subcarrier is typically nulled, but residual offset leaks into adjacent subcarriers, degrading Error Vector Magnitude.
• Demodulation Failure: In single-carrier systems, a rotating DC offset after carrier recovery creates a circular constellation shift.

DC offset is not always a fixed value. It is categorized into two distinct types:

• Static Offset: A constant voltage caused by fixed transistor mismatches. This is predictable and can be calibrated out during manufacturing or a one-time power-up sequence.
• Dynamic (Time-Variant) Offset: This is the more challenging impairment. It fluctuates rapidly due to changes in the LO leakage path as the antenna impedance varies (e.g., a hand moving near a mobile phone). This requires a real-time tracking loop, often implemented as a DC offset cancellation (DCOC) circuit in the analog domain or a digital tracking filter.

While analog AC-coupling can remove DC offset, it also removes spectral content near DC, which is unacceptable for wideband signals. Digital correction methods preserve the spectrum:

• Averaging and Subtraction: Estimate the mean of the received IQ samples over a long window and subtract it. This fails for dynamic offset.
• High-Pass Filter (HPF): A digital HPF with a very low cutoff frequency (e.g., 1 kHz) can track slow drift without destroying the signal. IIR filters are preferred for their low computational cost.
• Adaptive Tracking Loop: A feedback loop that integrates the residual DC error and subtracts a correction term, effectively acting as a narrowband notch filter at 0 Hz.

DC offset and IQ imbalance are distinct but often correlated impairments in direct-conversion receivers. DC offset is an additive impairment (a constant vector added to the origin), while IQ imbalance is a multiplicative impairment (gain and phase mismatch between I and Q branches). Critically, a strong DC offset can bias the estimation algorithms used for IQ correction, leading to inaccurate compensation coefficients. Robust receivers must perform joint estimation or correct DC offset first before addressing the IQ constellation distortion.

DC offset is directly observable in the IQ constellation diagram as a shift of the entire point cloud away from the origin. In the frequency domain, it appears as a distinct spike at 0 Hz in the power spectral density (PSD) plot. The magnitude is typically measured in dBFS (decibels relative to full scale) or as a percentage of the ADC's dynamic range. A well-designed receiver should maintain a DC offset below -40 dBFS to prevent significant performance degradation.