1. Relative and (maybe) Absolute RHESSI Detector Efficiency: 2002-2008 J.McTiernan 2-dec-2009

2. Test for decrease in detector efficiency: Motivated by calculation of B. Dennis for some flares which found that detector efficiency relative to D1 decreased with time. Here we do the same calculation for approximately 8900 flares, from feb-2002 through dec-2007. Also we will check for changes in D1 response in comparison with GOES Slide #1

3. The calculation: For each of a sample of flares, find T and EM for D1. Next fit data for each of the other detectors, while keeping T fixed at the D1 value. EM is then a measure of the relative sensitivity for each detector. OSPEX setup code supplied by B. Dennis Slide #2

4. Flare Sample: The sample of flares includes each flare with no attenuators, no data gaps, no SAA, no particles, no decimation, no high latitudes. Also 12 seconds before and after flare start and end must be “clean” for background subtraction. The time interval used for each flare was the same one used to find position, typically 1 to 2 minutes at peak. Slide #3

5. Results: There are spectra for 8923 flares. Plotted 60 day averages of EM / EM(D1) for detectors 3, 4, 5, 8 and 9. It’s pretty clear that by 2007 the other detectors have lost sensitivity relative to D1. Slide #4

6. Error bars are easier to see if you plot the detectors separately. Note also that the rel. sensitivity recovers post anneal. (2008 and 2009 are currently being processed.) Slide #5

7. What about absolute efficiency? To test for changes in absolute sensitivity, compare D1 with GOES. Why GOES? Well, it’s what we have, and we hope that it doesn’t lose sensitivity. Wow, it looks like D1 gains sensitivity relative to GOES data. Slide #6

8. Some notes: This seemed kind of surprising, so the calculations were repeated, using count rate > 6keV rather than EM. The results are the same. There are different GOES detectors used (10,11,12), which may change results, so the calculations were repeated using only GOES10 data. Only 7822 flares, but the results were the same. Dispersion in results for EM/GOES ratio was very large: about 3 orders of magnitude. Slide #7

9. Variation of ratio with flare size: It turns out that the ratio of D1 EM with GOES is correlated with flare size, and that’s where most of the dispersion in the ratio. Note that GOES data is not background subtracted. Slide #8

10. Variation of ratio with flare size: Here we compare the first 100 flares to the 100 flares right before anneal: No evidence of loss of sensitivity relative to GOES. Slide #9

11. Conclusions: Detectors 3, 5, 6, 8, and 9 lose sensitivity relative to detector 1, especially in 2007. No obvious loss of sensitivity in D1, relative to GOES. It seems odd that the other detectors can lose sensitivity, and not D1. Maybe GOES lost sensitivity too. Any suggestions? Slide #10