Answer to Question #7958 Submitted to "Ask the Experts"
The following question was answered by an expert in the appropriate field:
I am using a Ludlum 4400 Scaler with uptake probe, peaking with a 137Cs rod source. My previous quarterly checks of FWHM (full width at half max) and chi-square were within limits with the high voltage set at 256.For that quarter: FWHM = 9.4 percent, chi-square = 22.8
This quarter, with the high voltage set at 258, the chi-square 20 values were in range but high at 26.80 and the FWHM was high at 11.5 with the peak count at 670. Prior peak counts were at 640. I have repeated the instrument test with no change in the high voltage.
Is this a significant change?
I assume that your reference to "chi-square 20" implies that you made 20 independent measurements to perform the test. If you have been using the commonly applied acceptance criterion involving a two-tailed test with a total 5 percent probability of obtaining a chi-square result that falls outside the acceptance region when the data are actually acceptable, the acceptable range of chi-square values would be between 8.91 and 32.7. In such case the value of 26.8 that you obtained would be acceptable, implying that the counting system was not exhibiting any more or less variance than expected.
An acceptable value of the chi-square statistic, however, does not preclude the possibility that the operation of the counting system has degraded. A reproducible increase in the value of chi-square from an earlier test to a recent test may be cause for concern that some part of the system is not performing as it should and that this is contributing to the increased variance in the data. The fact that you have observed a notable degradation in the energy resolution of the system (increased FWHM) is consistent with the greater chi-square value and could well be an indication that the detector, or possibly one or more electronic components, is beginning to fail.
You should continue to perform your tests of the counting system in order to evaluate whether system degradation is progressing. I would recommend performing the chi-square test and the resolution measurements on a more frequent basis than quarterly, weekly if feasible, at least until you demonstrate that the system is functioning in a consistent and acceptable fashion. Naturally, you should also be performing simple daily operational checks with appropriate sources to ensure day-to-day counting reproducibility.Good luck.
George Chabot, PhD, CHP