TX IQ mismatch pre-compensation

A wireless transmitter circuit based on a direct conversion architecture, comprising:

• a TX (transmit) signal chain coupled to receive digital TX baseband data x(n), and including
• a digital TX IQ_mismatch pre-compensator to compensate for in-phase (I) and quadrature (Q) mismatch, the TX IQ_mismatch pre-compensator coupled to receive the digital TX baseband data x(n) and provide digital pre-compensated TX baseband data y(n) by performing IQ pre-compensation filtering based on IQ pre-comp filters w1 and w2;
• an analog TX RF (radio frequency) unit to generate an upconverted TX RF signal based on the pre-compensated TX baseband data y(n);
• a feedback receiver circuit (FBRX), coupled to receive the TX RF signal, and to generate corresponding FBRX baseband data z(n) based on downconverting the TX RF signal;
• and
• an adaptation module coupled to the TX IQ_mismatch pre-compensator to perform direct learning adaptation of the IQ pre-comp filters w1 and w2, modeled respectively as an x(n)_direct transfer function w1 receiving x(n) as input, and a complex conjugate x(n)_image transfer function w2 receiving a complex conjugate of x(n) as input,
• the adaptation module coupled to receive the TX baseband data x(n), and the corresponding FBRX baseband data z(n);
• the adaptation module including:
• a TX delay estimator configured to modify x(n) corresponding to a delay through the TX RF unit, producing a delay x(n) signal;
• a FBRX delay estimator configured to modify z(n) corresponding to a delay through the FBRX unit, producing a delay z(n) signal;
• a TX/FBRX error signal generator configured to generate an adaptation error signal corresponding to a difference between the delay x(n) and delay z(n) signals;
• and
• a DL (direct learning) adapter to perform direct learning adaptation to adjust the IQ pre-comp filters w1 and w2 to minimize the adaptation error signal.

A method for use in direct conversion radio frequency transmission, comprising:

• receiving in a transmit signal chain digital TX baseband data x(n), and generating corresponding TX RF (radio frequency) signals, including IQ (in-phase I and quadrature Q) modulation and upconversion, where IQ modulation and upconversion results in the generation of IQ mismatch signals;
• generating, from the TX baseband data x(n), pre-compensated TX baseband data y(n) to compensate for the IQ mismatch signals by performing IQ pre-compensation filtering based on IQ pre-comp filters w1 and w2;
• receiving in a feedback receiver (FBRX) the TX RF signal, and generating corresponding FBRX baseband data z(n) based on downconverting the TX RF signal;
• and
• operating on the TX baseband data x(n), and the corresponding FBRX baseband data z(n) to perform a direct learning adaptation of the IQ pre-comp filters w1 and w2, including modeling the filters w1 and w2 respectively as an x(n)_direct transfer function w1 receiving x(n) as input, and a complex conjugate x(n)_image transfer function w2 receiving a complex conjugate of x(n) as input, including:
• modifying x(n) corresponding to a delay in generating the TX RF signals, producing a delay x(n) signal;
• modifying z(n) corresponding to a delay in generating the FBRX baseband data z(n), producing a delay z(n) signal;
• generating an adaptation error signal corresponding to a difference between the delay x(n) and delay z(n) signals;
• and
• performing direct learning adaptation to adjust the IQ pre-comp filters w1 and w2 to minimize the adaptation error signal.