The application of sensor design and modelling techniques from earlier work in differential heat conduction calorimeters to DSC (Differential Scanning Calorimetry) has produced a prototype DSC with a 10 fold increase in effective sensitivity. The prototype was placed in service and evaluated in late 1985. After several months, noise problems developed that were traced to the sensors. Since these sensors are designed and manufactured for use as heat pumps (Peltier mode of operation) and not sensors (Seebeck mode of operation), the manufacturer could not supply information on noise performance. We then initiated a program to evaluate the noise and failure modes of sensors from several different manufacturers. One noise source was traced to mechanical failures in the thermocouple elements caused by large stresses developed across the thermopile when the calorimeter temperature was ramped from 30 deg C to 80 deg C. These stresses were created by the """"""""sandwich"""""""" type mounting of the sensors together with a mismatch in the coefficients of thermal expansion. A new mounting design allows the sensors to """"""""float"""""""" thus minimizing stresses as the temperature changes. The screening procedure and new mounting scheme reduced the noise to a level one fourth that of a commercial DSC. In a complete analysis of noise sources in the DSC however, we found that the noise contribution from the sensors is still approximately five times larger than any other noise source in the system. Thus, a significant reduction in the system noise is still possible if the sensor noise can be reduced further.
