Caelus Technologies Limited

Xiaoyong He

Chi Fung Lok

Zhi Jun Li

An N-channel interleaved Analog-to-Digital Converter (ADC) has a variable delay added to each ADC's input sampling clock. The variable delays are each programmed by a Successive-Approximation-Register (SAR) during calibration to minimize timing skews between channels. An auto-correlator generates a sign of a correlation error for a pair of ADC digital outputs. SAR bits are tested with the correlation sign bit determining when to add or subtract SAR bits. First all pairs are calibrated in a first level of a binary tree of mux-correlators. Then skews between remote pairs and groups are calibrated in upper levels of the binary tree using auto-correlators with inputs muxed from groups of ADC outputs input to the binary tree of mux-correlators. The binary tree of mux-correlators can include bypasses for odd and non-binary values of N. Sampling clock and component timing skews are reduced to one LSB among both adjacent channels and remote channels.

Calibration of timing skews in a multi-channel interleaved analog- to-digital converter (ADC) by auto-correlation of muxed-together channels in binary output tree

A first calibration measures capacitor array mis-match and updates a Look-Up Table (LUT) with calibrated weights that are copied to both a positive LUT and a negative LUT, and then adjusted for non-linearity errors by a second calibration using a Least Mean-Square (LMS) method. The binary code in the Successive-Approximation Register (SAR) is complemented to generate a complement code with a sign bit. When the sign bit is positive, entries for complement code bits=1 are read from the positive LUT and summed, a first offset added, and the sum normalized to get a corrected code. When the sign bit is negative, entries for complement code bits=0 are read from the negative LUT and summed, a second offset added, and the sum normalized to get the corrected code. A Multi-Variable Stochastic Gradient Descent method generates polynomial coefficients that further correct the corrected code.

Calibration of residual errors using least-mean-squares (LMS) and stochastic-gradient methods for an analog-to-digital converter (ADC) with a pre-calibrated lookup table

A self-calibrating Analog-to-Digital Converter (ADC) performs radix error calibration using a Successive-Approximation Register (SAR) to drive test voltages onto lower-significant capacitors. The final SAR code is corrected by performing LSB averaging on LSB averaging capacitors and then accumulated, and the measurement repeated many times to obtain a digital average measurement. An ideal radix or ratio of the measured capacitor's capacitance to a unit capacitance of an LSB capacitor is subtracted from the digital average measurement to obtain a measured error that is stored in a Look-Up Table (LUT) with the ideal radix. Radix error calibration is repeated for other capacitors to populate the LUT. During normal ADC conversion, the SAR code obtained from converting the analog input is applied to addresses the LUT, and all ideal radixes and measured errors for 1 bits in the SAR code are added together to generate an error-corrected digital value.

Calibration of radix errors using Least-Significant-Bit (LSB) averaging in a Successive-Approximation Register Analog-Digital Converter (SAR-ADC) during a fully self-calibrating routine

Growing community of inventors

Hong Kong, China

Xiaoyong He