A voltage-to-time conversion circuit compares a ramp-wave voltage, which steps up at a certain gradient, with each of a reference voltage, an input voltage, and a reference voltage, and produces a PB pulsating signal representing the times which the voltages require for having a predetermined relationship to the ramp-wave voltage. An encoder circuit converts the times into coded data items according to the ratios of the times to a common unit time. A normalization circuit determines a conversion characteristic curve on the basis of the coded data items, into which the times required by the reference voltages are converted, and A/D-converted values predefined for the reference voltages, and fits the coded data, into which the time required by the input voltage is converted, to the characteristic curve. Thus, the A/D-converted value of the input voltage Vin is calculated.