| Abstract: | A correlation function for reactor noise is determined through a conditional polarity-correlation technique in which the polarity of neutron-counting fluctuations is analyzed with the start condition that the counting rate reaches a preset level. This technique, evidently, is an example of flash-start technique, developed for the purpose of improving the efficacy of reactor noise analysis. In the present case, the conditional polarity-correlation function is described mathematically assuming that the counting fluctuations of the detector conform with a Gaussian distribution. The expression thus derived reveals that this function is approximately proportional to a cross-correlation function and is a differential type of polarity-correlation function. In order to examine the validity of the obtained expression, an experiment was performed in a graphite-moderated reactor at slightly subcriticai state to measure the conditional polarity-correlation function. A novel contrivance was adopted in the experiment to eliminate the dead-time effect created during the busy time of the multi-channel time analyzer. The dependence on the preset level—i.e. flash-start condition—exhibited by the conditional polarity—correlation function showed good agreement with that expected from the mathematical expression. The prompt mode neutron decay constants were determined from the experimental data within statistical error of 3～4 %. |
