Abstract: A sub-ranging analog to digital converter utilizes open loop differential gain amplifiers and analog switches to implement a pipeline. Each stage of the converter contains two fine range transfer amplifiers, sampling switches and hold capacitors, a low resolution sub-range analog to digital converter and a resistive ladder. The sampling switches behave as a digital to analog converter. Each stage then converts the held analog value to a digital code, which is used to operate the transfer switches to select the proper sub-range result for the next stage. The transfer switches are analog switches that perform the function of both the sampling and the sub-range transfer. The interstage amplifiers are simple open loop differential amplifiers with a rather imprecise absolute gain. Because the reference and the signal are both amplified by this imprecise gain, both the reference and the signal are amplified by the same amount.

Abstract: The present invention provides a sub-ranging analog-to-digital (A/D) converter with improved speed and power consumption characteristics relative to known sub-ranging converters. The sub-ranging A/D converter utilizes information relating to the values of the bits determined in one stage to define the range of operation for a subsequent stage. In one embodiment, the subsequent stage utilizes three-input comparators in determining the value of a bit. Two of the inputs are used to receive signals representative of the upper and lower limits of the range of operation that has been determined by the prior stage and the other input is used to receive the analog signal. The three-input comparator operates to produce an output signal that is indicative of the relationship of the analog signal to a threshold level within the defined range of operation determined by the prior stage.

Abstract: A modified quasi-Gray encoding technique for use in parallel analog-to-digital converters that significantly reduces errors resulting from multiple simultaneous inputs. The encoding technique converts a one-in-(2.sup.n -1) digital code into an n-bit binary word that is the same as quasi-Gray code in all but its least significant bit position, which alternates in the same manner as standard binary code. For many multiple simultaneous inputs, the modified quasi-Gray code substantially reduces errors when compared with quasi-Gray code. For example, the modified quasi-Gray code reduces the maximum error from 3 to 2 for two simultaneous inputs separated by two bit positions (n=8). In a typical parallel analog-to-digital converter employing the modified quasi-Gray code, the one-in-(2.sup.n -1) digital code is converted into modified quasi-Gray code using a read-only memory.