Principal:
Deputy:
Analyst:
AORKEYS:
Last Updated:
The following steps were followed in analyzing the data: 1) V2.31 of mips_sloper was used to reduce the data. We used flags -l -j CamG for the stim-calibrated dark data and -l -q -j CamG for the read noise data. The stim-calibrated darks were also run through mips_sloper with the flags -d -C CamG. 2) We used get_rdnoise_new.pro on the 70um data to calculate read noise and dark current. We set n_dce_rej=10. We also used the bad70 flag that Karl G added to the get_rdnoise program to exclude the B-side of the array and readout 4,4 on the A-side. 3) The stim-calibrated dark data were run through the IDL program mips246_stimcal.pro with flag 'skip=10' to determine the mean stim brightness and the stim repeatability. This program determines the array mean for each of the stim flashes, ignoring the first 'skip' flashes, and then determines the mean of these mean brightnesses. It also takes the brightness of each odd-numbered flash, divides that number by the average of the adjacent even-numbered flashes, and then calculates the standard deviation of these normalized values. This is not the best way of assessing stim repeatability, but it does provide a crude first estimate. It is a quantitative representation of the array-average brightness plots presented by John S below. 4) Karl Misselt performed a stim repeatability assessment, using the same methodology used in his earlier report on 70um stim repeatability. Details and plots will follow in a later version of this writeup. 5) The stim-calibrated dark data were run through the IDL program mips246_darkcal.pro. This program determines the array mean for each of the dark DCEs, then calculates the mean and standard deviation of these array means. It is possible to set the number of DCEs to ignore after each stim flash, to minimize the effect of stim latents. In practice, since the cosmic ray filtering is not perfect, I used the skip parameter as best I could to clean out the effect of bad DCEs. A median filter would have been more effective and will be implemented later, when there is more time to reanalyze the data. I will also try using the -q flag in mips_sloper, for 3sigma rather than 5sigma cosmic ray filtering. For now, the dark levels from the read noise data are probably a more accurate assessment of dark levels, even though they are not stim-calibrated. I record the time since the last thermal anneal in the tables below so that an eyeball correction can be made for the effect of cosmic rays. This post-anneal time is also useful in looking at Karl's stim repeatability results. 6) These results can be combined with insights into the effects of bias voltage on the dynamic range of the detector, the linearity of the data ramps, and the behavior of the B-side of the 70um array in deciding how best to set the detector bias.
A-Side only Both sides
AORID 70Bias 160Bias 70ReadNoise 70DarkCurrent 70ReadNoise 70DarkCurrent
(mV) (mV) (e) (e/s) (e) (e/s)
70ReadNoise 70DarkCurrent
7152640 30 15 583 +/- 208 385 +/- 156 764 +/- 651 397 +/- 171
7155200 30 20 617 +/- 233 381 +/- 132 768 +/- 600 394 +/- 154
7152896 35 20
7155456 35 25 644 +/- 217 472 +/- 163 669 +/- 252 474 +/- 167
7153152 40 25 718 +/- 270 531 +/- 202 736 +/- 289 531 +/- 204
7153408 40 30 697 +/- 257 538 +/- 183 716 +/- 286 541 +/- 185
7155712 40 35 721 +/- 266 654 +/- 228 738 +/- 292 654 +/- 228
7152384 45 35 783 +/- 293 615 +/- 210 804 +/- 323 616 +/- 211
7155968 45 15 829 +/- 340 773 +/- 261 834 +/- 347 772 +/- 261
7154688 50 15 871 +/- 351 773 +/- 268 898 +/- 399 772 +/- 268
A-Side only Both sides
70ReadNoise 70DarkCurrent 70ReadNoise 70DarkCurrent
7630848 55 920 +/- 418 702 +/- 236 982 +/- 578 711 +/- 258
7631104 60 1096 +/- 546 864 +/- 321 1171 +/- 731 872 +/- 326
7631360 65 1319 +/- 736 1029 +/- 369 1373 +/- 838 1032 +/- 373
7641088 70 1498 +/- 828 1213 +/- 420 1528 +/- 882 1218 +/- 427
For each bias voltage, we measured 278 DCEs with a stim flash on the 2nd
DCE and every 6 DCEs after that, for a total of 47 stim flashes. The IDL
routines mips246_stimcal and mips246_darkcal and Karl Misselt's stim
repeatability routine provided the following results:
70um Analysis:
AORID 70Bias 160Bias Mean Stim Stim K.M.StimRep Dark Time Since
(mV) (mV) (DN/s) StdDev (sigma) (DN/s) Anneal (h)
70 160 70 160
7152640 30 15 9106 +/- 296 .0138 8.8 10.5 105 +/- 18 0 2+
7155200 30 20 8872 +/- 78 .00863 7.0 86 +/- 20 1 0
7152896 35 20 9.7 0 1
7155456 35 25 10477 +/- 178 .0103 7.4 103 +/- 37 1 0
7153152 40 25 11404 +/- 388 .00811 9.6 78 +/-158 0 1
7153408 40 30 11664 +/- 161 .0122 --- 109 +/- 31 0 0
7155712 40 35 13102 +/- 305 .00767 6.7 140 +/- 66 1 0
7152384 45 35 13022 +/- 302 .0116 9.3 122 +/- 91 0 1
7155968 45 15 15366 +/- 408 .00770 6.3 10.4 154 +/- 21 1 0
7154688 50 15 14727 +/- 684 .0130 9.0 10.7 140 +/- 81 0+ 1+
7630848 55 11203 +/- 292 .0115 8.3 152 +/- 42 0
7631104 60 13367 +/- 733 .0111 8.2 172 +/-102 0
7631360 65 15079 +/- 897 .0106 7.7 234 +/- 52 0
7641088 70 16688 +/-1081 .0156 7.5 268 +/-110 0
A downlink occurred prior to 7154688 and 7152640, which is why the time since
the last anneal is larger than the value listed in the table. I do not know
the duration of the downlink.
John Stansberry has produced plots of brightness vs DCE number for the stim-calibrated dark data. This includes array averages and is very useful for assessing stim repeatability and responsivity drifts. These plots are presented below:
Karl Misselt produced plots of repeatability vs bias. For the 70um array, there is a dependence on settling time but no dependence on bias voltage. The one discrepent point at 50mV in the plot below was actually taken with zero settling time and so should have been plotted in red.