Direct Conversion Transmitter

Unlike a superheterodyne architecture, a direct conversion transmitter modulates the baseband I/Q signals directly onto the final RF carrier. The local oscillator (LO) frequency is set exactly equal to the desired carrier frequency. This eliminates the need for IF filtering and reduces component count, enabling highly integrated, low-cost radio implementations. The baseband signal is mixed up to RF in one step, making the architecture inherently wideband and flexible for multi-standard operation.

The architecture is critically sensitive to mismatches in the I and Q signal paths. Gain imbalance causes the constellation to stretch along one axis, while phase imbalance (deviation from the ideal 90-degree LO offset) causes inter-symbol interference and constellation rotation. Because the desired signal and its image occupy the same spectrum, these impairments cannot be filtered out and must be corrected through digital I/Q mismatch compensation.

A defining impairment of the zero-IF transmitter is LO leakage, which manifests as an unwanted continuous wave tone at the exact carrier frequency. This is primarily caused by DC offset at the modulator input, often due to transistor mismatch or LO self-mixing. The leaked tone degrades the Error Vector Magnitude (EVM) and can violate spectral emission masks, requiring active DC offset cancellation loops.

In an ideal quadrature modulator, the image sideband is perfectly canceled. In practice, I/Q imbalance creates a mirror image of the desired signal. The Image Rejection Ratio (IRR) quantifies the suppression of this unwanted image. Without external filtering, achieving high IRR requires precise digital pre-distortion. A typical uncorrected modulator might achieve only 25-35 dB of IRR, whereas modern standards demand 50 dB or more, necessitating adaptive I/Q equalization.

I/Q mismatch is categorized into two types:

• Frequency-Independent Imbalance: Static gain and phase errors constant across the entire signal bandwidth. Correctable by a simple complex scalar multiplication.
• Frequency-Dependent Imbalance: Errors that vary with frequency, caused by mismatched anti-aliasing filters, trace lengths, or DAC responses. This requires a complex I/Q mismatch filter, typically a widely-linear FIR structure, to equalize the response across the band.

The non-ideal behavior of a direct conversion transmitter is mathematically modeled as a widely-linear system. The impaired output is a linear combination of the ideal baseband signal and its complex conjugate. This is represented by a 2x2 I/Q mismatch matrix that maps the ideal I/Q vector to the impaired output. Digital compensation applies the inverse of this matrix, effectively restoring circularity to the constellation and suppressing the image sideband.