Abstract: The conversion is effected by adding equal and opposite reference voltages to the analog signal to provide two first output voltage levels. The voltage difference between these levels is converted into a first given number of pulses. The polarity of the analog signal is then reversed during the second half of the cycle so as to be subtracted from the reference voltages for providing two second output voltage levels. The voltage difference between these levels is converted into a second given number of pulses. A function of the first and second number of pulses provides a numerical count directly proportional to the analog signal with all zero drift eliminated. A second conversion can be made to provide a digital read-out which is free from both zero drift and span drift.