Zero-IF Architecture

The local oscillator (LO) frequency is set exactly equal to the desired carrier frequency, mixing the baseband I/Q signals directly to RF in one step. This eliminates the need for intermediate frequency (IF) stages, image-reject filters, and multiple mixers found in superheterodyne architectures.

• Single mixer stage per channel reduces component count and board area
• No IF filters required, enabling monolithic integration on a single silicon die
• Baseband signal spectrum is centered at DC before upconversion
• Simplifies frequency planning by removing IF selection constraints

In zero-IF transmitters, the image signal overlaps exactly with the desired signal because the LO is at the carrier frequency. Unlike superheterodyne receivers that use filtering to reject images, zero-IF relies entirely on quadrature accuracy to suppress the unwanted sideband.

• Any gain imbalance or phase imbalance between I and Q paths creates a mirror image that falls directly on top of the transmitted spectrum
• Image rejection is achieved through complex signal processing rather than analog filtering
• Typical uncorrected image rejection is only 25-35 dB, requiring digital compensation to reach 60+ dB
• The image is a conjugate copy of the desired signal, scaled by the mismatch coefficient

Because the LO operates at the exact transmit frequency, any DC offset in the baseband I or Q paths is upconverted directly to the carrier frequency, appearing as an unmodulated tone at the center of the output spectrum. This LO leakage or carrier feedthrough degrades Error Vector Magnitude (EVM) and can violate spectral emission masks.

• Sources include transistor mismatch in the mixer, LO self-mixing, and PCB trace coupling
• DC offset as small as a few millivolts can produce significant carrier leakage
• Compensation requires DC offset cancellation loops or digital pre-correction
• Particularly problematic in OFDM systems where the carrier tone creates interference on the DC subcarrier

While simple I/Q imbalance models assume frequency-independent gain and phase errors, real zero-IF transmitters exhibit frequency-dependent I/Q mismatch across the signal bandwidth. This is caused by mismatched anti-aliasing filters, unequal trace lengths, and component tolerances in the I and Q baseband paths.

• Requires complex FIR filter structures rather than simple scalar correction
• I/Q skew (timing mismatch between channels) creates linear phase distortion versus frequency
• Wideband signals in 5G NR and Wi-Fi 7 are particularly susceptible
• Correction filters must adapt to temperature and voltage variations during operation

The elimination of IF stages, image-reject filters, and external SAW filters makes zero-IF the dominant architecture for high-volume consumer devices and software-defined radios. Modern CMOS transceivers integrate the entire RF front-end, baseband processing, and digital compensation on a single chip.

• Bill of materials (BOM) reduction of 30-50% compared to superheterodyne designs
• Enables multi-band, multi-mode operation through digital reconfiguration
• Found in virtually all smartphones, IoT modems, and small cell base stations
• The cost of digital I/Q compensation logic is negligible in modern nanometer CMOS processes

Zero-IF receivers are particularly vulnerable to second-order intermodulation distortion (IM2) because low-frequency mixing products fall directly into the baseband. In transmitters, even-order nonlinearities in the mixer and baseband amplifier create distortion components near DC that are upconverted around the carrier.

• Differential circuit design is essential to suppress even-order products through common-mode rejection
• IP2 (second-order intercept point) becomes a critical specification for zero-IF designs
• Envelope signal components from amplitude-modulated waveforms can self-mix and create in-band distortion
• Requires careful layout symmetry and balanced impedance matching on I and Q paths