I/Q Mismatch

The amplitude mismatch component where the I and Q branches exhibit different gain factors. This causes the ideal square constellation to stretch into a rectangle, compressing symbols along one axis while expanding them along the other.

• Quantified as the ratio ( g = G_I / G_Q ) or in decibels
• Typical uncorrected values range from 0.1 dB to 1 dB in integrated modulators
• Results in EVM degradation that is constant across all subcarriers for frequency-independent cases
• In OFDM systems, gain imbalance alone creates a scaled image of the signal on the opposite side of the carrier

The deviation from the ideal 90-degree phase offset between the I and Q local oscillator signals. This non-orthogonality causes energy from the I channel to project onto the Q channel and vice versa, creating inter-symbol interference.

• Expressed in degrees, with typical values between 0.5° and 5° before calibration
• Causes constellation points to shear into a parallelogram shape
• Phase error ( \phi ) introduces a cross-talk term proportional to ( \sin(\phi) )
• Combined with gain imbalance, produces the characteristic image sideband at the mirror frequency

An unwanted constant voltage added to the baseband I or Q signal, primarily caused by local oscillator self-mixing or component mismatch in the mixer. This manifests as an unmodulated carrier tone at the center of the transmitted spectrum.

• Creates a spurious spectral line at ( f_c ), degrading Error Vector Magnitude (EVM)
• Typical offset values range from microvolts to millivolts at the modulator input
• LO leakage power is often specified relative to the total channel power, with targets below -30 dBc
• In direct conversion transmitters, this is the dominant source of carrier feedthrough

A wideband impairment where gain and phase errors vary across the signal bandwidth, caused by mismatched anti-aliasing filters, unequal trace lengths, or component tolerances in the I and Q paths.

• Requires a complex FIR filter for correction rather than a simple scalar multiplication
• Dominant in wideband signals such as 5G NR with 100 MHz or greater bandwidth
• I/Q skew, a linear phase distortion, is a primary contributor
• Modeled as a widely-linear system with frequency-selective coefficients ( H_{IQ}(\omega) )

The most visible consequence of I/Q mismatch is the creation of an unwanted image signal at the mirror frequency. For a desired signal at ( f_c + f_m ), the image appears at ( f_c - f_m ), directly interfering with adjacent channels.

• Image Rejection Ratio (IRR) quantifies suppression, with uncorrected values typically 25-40 dB
• High-order modulation schemes like 256-QAM require IRR exceeding 45 dB
• The image is a complex conjugate of the original signal scaled by the mismatch coefficient
• In multi-carrier systems, the image of one carrier can fall directly on top of another

I/Q mismatch transforms the modulator from a linear time-invariant system into a widely-linear system, where the output depends on both the input signal and its complex conjugate.

• Mathematically represented as ( y(t) = \mu x(t) + \nu x^*(t) ), where ( \mu ) is the desired response and ( \nu ) is the image-producing coefficient
• The I/Q Mismatch Matrix is a 2x2 transformation mapping ideal I/Q to impaired I/Q
• This property is exploited by blind estimation algorithms that detect non-circularity in the received constellation
• Compensation requires an inverse widely-linear filter to restore proper signal structure

The amplitude mismatch component of I/Q imbalance, defined as the ratio or difference in gain between the I and Q branches. This causes the ideal square constellation to stretch into a rectangle along one axis, directly degrading Error Vector Magnitude (EVM).

• Frequency-Independent: A constant gain offset across the band, correctable by a scalar.
• Frequency-Dependent: Gain ripple caused by mismatched low-pass filters or trace responses.
• Measurement: Often expressed in dB, representing the power ratio between the I and Q paths.

The deviation from the ideal 90-degree phase offset between the I and Q local oscillator signals. This non-orthogonality causes inter-symbol interference (ISI) and rotates the constellation, making adjacent symbols harder to distinguish.

• Impact: Generates an image signal that overlaps the desired channel, causing self-interference.
• Sensitivity: Direct conversion architectures are highly sensitive to even fractions of a degree of quadrature error.
• Correction: Requires a complex rotation matrix in the digital baseband to restore orthogonality.

A relative timing delay between the sampling clocks or data paths of the I and Q channels. This is a form of frequency-dependent imbalance that introduces a linear phase distortion across the signal bandwidth.

• Causes: Mismatched PCB trace lengths, clock buffer propagation delays, or ADC aperture uncertainty.
• Effect: Destroys sideband suppression, particularly at the band edges of wideband signals.
• Compensation: Requires a fractional-delay filter or a complex FIR equalizer to realign the data streams in time.

The primary metric for quantifying a system's ability to suppress the unwanted image signal generated by I/Q mismatch. It is the power ratio between the desired signal and its image, expressed in decibels.

• Calculation: IRR = 10 * log10(P_desired / P_image)
• Target: High-performance systems often require > 50 dBc IRR to meet spectral mask requirements.
• Limiting Factor: IRR is directly limited by the accuracy of the I/Q Mismatch Compensation algorithm and the resolution of correction coefficients.

An unwanted constant voltage added to the baseband I or Q signal, typically caused by local oscillator self-mixing or component mismatch. This manifests as LO Leakage—a spurious tone at the carrier frequency.

• Mechanism: DC offset at the modulator input upconverts directly to the LO frequency.
• Impact: Wastes transmit power and can violate spectral emission masks at the center frequency.
• Correction: A simple additive constant is subtracted from the baseband signal before the DAC.

The mathematical framework for correcting I/Q mismatch. An impaired signal is modeled as a widely-linear transformation: y(t) = K1 * x(t) + K2 * conj(x(t)), where K1 is the direct path gain and K2 is the I/Q Mismatch Coefficient.

• Correction: Applies the inverse matrix to the digital baseband data.
• Adaptive Equalizer: A complex FIR filter that tracks time-varying K1 and K2 coefficients.
• Blind Estimation: Extracts K1 and K2 from signal statistics (e.g., circularity) without a training sequence.