Improvements in data reduction in direct pulse heating calorimetry

This paper describes improvements introduced in data reduction in direct heating pulse specific heat experiments. In calculations of specific heat, it is necessary to calculate the first derivative of the recorded temperature data as a function of time. The error induced by different numerical differentiation techniques can represent a significant part of the overall measurement error. Thus, different digital filtering techniques, differentiation, and smoothing algorithms were applied and tested to examine their influence on the minimization of errors induced by noise, which is unavoidable in measured signals. A minimum square error criterion was applied in designing digital filters, with arbitrary prescribed magnitude characteristics. Attention was paid to applications when one or more structural phase transitions in the specimen material occur within the temperature range covered by the experiment. The cases where the frequency spectrum of induced noise overlaps with the spectrum of temperature transient signals originating from phase transitions were analyzed in detail. The effectiveness of the methods of extracting the final specific heat data from a noisy signal using different digital filtering techniques is demonstrated.Paper presented at the Fourth International Workshop on Subsecond Thermophysics, June 27–29, 1995, Köln, Germany.