A Multi-Dimensional Calibration Based on Genetic Algorithm in a 12-Bit 750 MS/s Pipelined ADC.

Area of Science:

• Electrical Engineering
• Signal Processing
• Integrated Circuit Design

Background:

• Pipelined analog-to-digital converters (ADCs) are essential for high-speed and high-resolution data conversion.
• Accuracy in pipelined ADCs is primarily limited by errors from multiple digital-to-analog converters (MDACs), including capacitor mismatch and interstage gain errors (IGE).

Purpose of the Study:

• To develop and validate a multi-dimensional (M-D) MDAC calibration technique for a 12-bit 750 MS/s pipelined ADC.
• To compensate for static and dynamic errors in MDACs, thereby improving overall ADC performance.

Main Methods:

• Implementation of a multi-dimensional (M-D) MDAC compensation model incorporating a genetic algorithm (GA) for foreground calibration.
• Utilizing pseudo-random number (PN) injection for subsequent background calibration to address dynamic IGE.
• Development of an M-D coefficient extraction scheme based on GA to avoid local optima.

Main Results:

• The proposed calibration significantly improved performance in a prototype 12-bit 750 MS/s pipelined ADC.
• Signal-to-noise and distortion ratio (SNDR) increased from 49.9 dB to 59.6 dB, and spurious-free dynamic range (SFDR) improved from 66.7 dB to 77.5 dB at 25 °C.
• Background calibration further enhanced SNDR by 3.4 dB and SFDR by 8.8 dB at 85 °C.

Conclusions:

• The multi-dimensional MDAC calibration based on GA effectively compensates for capacitor mismatch and IGE in pipelined ADCs.
• The proposed technique enhances both static and dynamic performance, demonstrating its suitability for high-performance ADC applications.
• The method provides a robust solution for improving the accuracy of high-speed, high-resolution pipelined ADCs across varying temperatures.