Comments on "A highly linear single p-n junction temperature sensor"

The concept of using switched currents through a single p-n junction for a linear temperature measurement as proposed by Freire et al. (Apr. 1994) has been previously described by J. Williams (1991). Williams uses asynchronous set of switches (LTC 1043) to switch separate currents through a diode-connected transistor, and also to switch capacitors synchronously with the currents. This action creates an output that is proportional to the peak-to-peak voltage of the sensor. The circuit uses relatively few components. There are, however, some design errors in the the current generator that he uses. It was pointed out to Williams that trimming is necessary due to component tolerances, and a simpler and more accurate means of generating the two selected currents was proposed (May 1991).<>