<< Effect of Power on/off/on cycle on Image Quality >> Technical Report Hajime Kawakami DOCUMENT:XMM-OM/MSSL/TC/0046.01 1 September 1997 1. Introduction The XMM-OM instruments will power off during perigee to avoid radiation damage. Image intensifier is one of the most sensitive component for the high energy particles. There are 2 possibilities on the operation of the image intensifier. The 1st one is shutting down all HVs, while the 2nd is shutting down photocathode voltage but keeping MCPs and anode voltages at 70% of the nominals. The 1st solution is of course simpler, but may causes degradation of image quality. Since the engineering time will be allocated after the perigee, inaccurate calibration parameters may be employed for an orbit if recovery time of the intensifier is not fast enough. To assess the recovery time of an image intensifier, the image intensifier was completely off for more than 15 hours, and photon counting images were taken before and after the power on This extreme test gives the answer which operation should be employed with an idea about margin. 2. Set-up of experiment DEP TYPE-PP0370B (SN B9710003) intensifier was kept on at nominal voltages for a month and was powered off for 15.6 hours. A F-F was projected on the detector with the count rate of 19,000 c/s (full detector area) to assess the modulation pattern in centroiding image. A CCD camera format of 256x256 was employed, which provided a frame rate of 100Hz. The centroiding was carried out by software, which employs parabola algorithm and cross-hair data sampling to emulate flight electronics. The processing speed was 12 frames/sec. This software centroiding is capable of determining characteristic curve, which describes the relation of calculated position to real position, during building up the centroiding image. The characteristic curves were derived by analysing event profiles within 512(H)x1536(V) subpixels area, and centroiding images were built up from 512(H)x512(V) subpixel area out of the 512(H)x1536(V). Two measurements were carried out before the power-off (i.e. Obs #1 and #2), and 11 measurements were followed after the power-on (i.e. Obs #3-#13). The boundary values were derived from the characteristic curve of Obs #1 and were used for all imagings (i.e. FR#2-#13). There is no centroiding image corresponding to FR#1, as there was no boundary value for the 1st measurement. Table 1 summarizes 13 sets of boundary values from the 13 measurements for both of X- and Y- directions. The room temperature was kept 23 Celsius throughout this experiment. The temperature of the intensifier was not actively controlled, but the temperature measured at its mechanical attachment showed constantly 23 Celsius. 3. One dimensional results Boundary values for centroiding will be up-dated in the beginning of every orbit and will be used for all imagings in the orbit. If the intensifier will not recover before the calibration time, wrong boundary values will be derived, hence the quality of all images in the orbit will be degraded. Figure 1 describes how boundary values are determined from characteristic curve, which is derived by analysing many events at different positions within a CCD pixel. The characteristic curve is easily affected by a little changed of event profile, and is especially sensitive with the parabola algorithm if event size is very small. Unfortunately, the event size of the DEP-PP0370B intensifier is really small. The curve in figure 1 shows extreme non-linearity, which is caused by the very small event size. Table 1 of Obs#1 shows that central 6 subpixels enjoy from -.17832 to +.15935 only in calculated values. While, 2 boundary pixels cover the residual range of 0.66233. This causes numerical instability and extreme sensitivity of image quality to a change in detector system. Two characteristic curves from the 2 measurements, Obs#1 and #13, are drawn in figure 1, but both curves coincide very well fortunately. This implies the intensifier was fairly stable through whole experiment. To see the more details, individual X-characteristic curves are subtracted by that of Obs#13 and are drawn with magnified scale (see figure 2). Amplitudes themselves are small but those of Obs#1 and #2 are larger than the others. The standard deviations are tabulated in table 2. This suggests the event profile of the intensifier changed between the power-off. The variation of Y-characteristic curve is very small and no significant difference between before and after the power-off as shown in table 2. If characteristic curve changes but the boundary values are kept same, (i.e. wrong boundary values), subpixels have different sizes. This causes undulation in F-F image, namely CCD pixel based modulation pattern. Figure 3 in bottom panel shows the time sequence of the modulation patterns from FR#2-#13. The 8x8 cell corresponds to each subpixel. As the modulation appears even with a small change of event profile, it is the best parameter to assess image quality. The modulation along X-subpixels is shown in figure 4 and tabulated in table 3. Outputs in image were added together along y-direction. The modulations from FR#2 to FR#13 were 3-5%. The modulation in FR#2 is not fantastic in spite of adjacent observation time from the up-date of boundary values. To see the effect of settling time on the image quality after the power-on, the modulation patterns in all frames were normalized by that of FR#13. Figure 5 and Table 4 tabulate results. The improvement in modulation is noticeable for frames after the power-on, ~2%. This suggests to allow calibration after the power-on without waiting warming up of the intensifier. The different characteristics for the FR#2, before the power on, is again obvious. Figure 6 and table 5 are modulations along Y-subpixels. As expected from the stability of Y-characteristic curve, the output of Y-pixels are quite flat, 2-3%. Figure 7 and Table 6 are results normalized by that of FR#13. There is not remarkable improvement, but the flatness is still excellent. 4. Two dimensional results Two dimensional modulation pattern will be recorded in the beginning of an orbit for precise calibration of scientific data. If the intensifier will not recover before the engineering time, wrong data will be stored and the precise calibration will fail. Table 7 summarizes the statistics among the 8x8 modulation cells. There are 12-17% modulation in maximum, which are caused by asymmetric event profile and too small event size of the intensifier. The orientation of CCD to the image intensifier will be accurately tuned for the flight detector for minimizing the effect of asymmetry in event profile. But, the orientation was not tuned at all for this test. These modulation patterns were normalized by that of FR#13 to suppress 2-dimensional undulation and to see more detail, which is shown in the top panel of figure 3. Table 8 summarizes the result. This normalization improves flatness from FR#3 to FR#11, for instance from 15% to 9% in FR#9. The characteristics look same from FR#3-#13, therefore this suggests again to start calibration without waiting the warm up of the intensifier. The normalization did not work well for the FR#2. The change of characteristics before and after the power-off is obvious. 5. Discussion The characteristics look similar after the power-on until 3 hours. It is clear that it is not worth waiting the warm up of the intensifier for calibration. But, it is not guarantied that the calibration parameters are valid throughout an orbit, because the results might be due to very slow change of the intensifier. The followings must be clarified (1) Do characteristics change with very long time scale ? (2) Is the power-off essential for the change of characteristics or long time difference more important ? The total photons acquired are not sufficient to investigate 2 dimensional characteristics, because each cell has only 900 counts in FR#2, whose photon shot noise is 3.3%. And, only single image was taken before the power-off. It is ideal to take several frames before power-off and many after the power-on to continue more than 24 hours to investigate long time stability. Each of frame should contain more than 1 million photons (15,625 counts per each 8x8 modulation cell). The input light intensity should be kept low, e.g. 10,000 c/s (full detector area). The Q-model hardware can process 100 frames/sec with the CCD camera format of 256x256. It takes 30min to acquire 1 million photons in 512x512 area, but this is fast enough to investigate long term variation. There is already a software to record sequential images if the pixel format is 512x512. The parabola algorithm with cross hair sampling is not accurate if event profile is asymmetric and small. The DEP TYPE-PP0370B intensifier has asymmetric and very small (~15um) event profile. The modulation more than 10%(max) in 8x8 cell is due to this problem. Relatively large modulation along X-axis in FR#2, in spite of proximity measurement time from the calibration, may also be caused by the event profile. This un-ideal event profile might have amplified the change of centroiding image. Having a small air gap, which enlarges the event size and reduces the asymmetry, might have given better result. The other possibility of instability in modulation pattern is digitizing effect. The calculated positions distribute discretely after the digitizing. Namely, many of digitized position values represent none of real position, while several special digitized positions represent wide range of real positions. It is unlucky if one of the 7 boundary values coincide with the special positions. In such case, there is no perfect boundary value to flatten the output of sublixels. And an extremely small fluctuation of event profile can cause the change of the modulation pattern. The coincidence of the special position values and boundary values was investigated for FR#2, and it was found not to be the case. The results suggest to keep HVs 70% of the nominals during perigee until the more accurate experiment will guarantee the power off. Acknowledgement. The author wish to express grateful thanks to DEP for loan of the new intensifier. This work was carried out as part of the XMM-OM program for the blue detector operation. Table 1. Boundary values derived from individual measurement. The 1st row is for X-subpixels and 2nd row for Y-subpixels in each measurement. ------------------------------------------------------------------------ Pix-1 Pix-2 Pix-3 Pix-4 Pix-5 Pix-6 Pix-7 Pix-8 ------------------------------------------------------------------------ Obs# 1 Total Photons = 157682 18H 33M 29S 18H 39M 36S 1997/08/05 -.50000 -.17832 -.07323 -.02174 .01052 .03862 .07718 .15935 .50000 -.50000 -.17959 -.08250 -.03366 -.00065 .03050 .07914 .17980 .50000 Obs# 2 Total Photons = 157387 18H 42M 53S 18H 49M 00S 1997/08/05 -.50000 -.17512 -.07258 -.02141 .01086 .03895 .07646 .15751 .50000 -.50000 -.17972 -.08211 -.03353 -.00092 .03004 .07774 .17865 .50000 >>>>>>>>>>>>>> 15.6 Hours Power-off <<<<<<<<<<<<<<<< Obs# 3 Total Photons = 320653 10H 33M 43S 10H 45M 54S 1997/08/06 -.50000 -.18020 -.07547 -.02355 .00926 .03761 .07602 .15822 .50000 -.50000 -.17856 -.08197 -.03358 -.00069 .03093 .07937 .18062 .50000 Obs# 4 Total Photons = 638993 10H 33M 43S 11H 01M 06S 1997/08/06 -.50000 -.17970 -.07503 -.02347 .00902 .03715 .07527 .15665 .50000 -.50000 -.17787 -.08177 -.03348 -.00066 .03089 .07918 .18060 .50000 Obs# 5 Total Photons = 320071 11H 03M 59S 11H 16M 09S 1997/08/06 -.50000 -.17830 -.07468 -.02368 .00829 .03604 .07448 .15684 .50000 -.50000 -.17792 -.08212 -.03364 -.00094 .03031 .07795 .17894 .50000 Obs# 6 Total Photons = 321127 11H 17M 53S 11H 30M 03S 1997/08/06 -.50000 -.18157 -.07661 -.02467 .00759 .03541 .07365 .15536 .50000 -.50000 -.17737 -.08083 -.03316 -.00067 .03064 .07882 .17925 .50000 Obs# 7 Total Photons = 322246 11H 35M 04S 11H 47M 15S 1997/08/06 -.50000 -.18134 -.07621 -.02482 .00787 .03592 .07436 .15741 .50000 -.50000 -.17879 -.08192 -.03350 -.00071 .03096 .07940 .18100 .50000 Obs# 8 Total Photons = 322465 11H 53M 22S 12H 05M 33S 1997/08/06 -.50000 -.18184 -.07700 -.02531 .00729 .03554 .07389 .15617 .50000 -.50000 -.17965 -.08238 -.03362 -.00065 .03083 .07950 .18221 .50000 Obs# 9 Total Photons = 323183 12H 07M 27S 12H 19M 39S 1997/08/06 -.50000 -.18185 -.07641 -.02451 .00775 .03589 .07408 .15624 .50000 -.50000 -.17912 -.08248 -.03351 -.00057 .03156 .07963 .18163 .50000 Obs#10 Total Photons = 322737 12H 20M 49S 12H 33M 04S 1997/08/06 -.50000 -.18159 -.07649 -.02473 .00769 .03605 .07454 .15749 .50000 -.50000 -.17956 -.08273 -.03430 -.00106 .03090 .07989 .18103 .50000 Obs#11 Total Photons = 964848 12H 37M 17S 13H 13M 49S 1997/08/06 -.50000 -.18137 -.07685 -.02510 .00754 .03573 .07383 .15629 .50000 -.50000 -.17953 -.08255 -.03380 -.00077 .03113 .07957 .18064 .50000 Obs#12 Total Photons = 954068 13H 15M 40S 13H 52M 09S 1997/08/06 -.50000 -.18153 -.07645 -.02456 .00800 .03611 .07424 .15619 .50000 -.50000 -.17995 -.08254 -.03397 -.00079 .03099 .07920 .18050 .50000 Obs#13 Total Photons = 948618 13H 53M 06S 14H 29M 31S 1997/08/06 -.50000 -.18090 -.07592 -.02429 .00837 .03659 .07463 .15619 .50000 -.50000 -.17930 -.08239 -.03401 -.00071 .03122 .07962 .18107 .50000 ------------------------------------------------------------------------ Table 2. Deviation of characteristic curves from that of Obs #13. Values are standard deviation. The unit is CCD pixel ------------------------------------------------- Obs# 1 X-LUT= .00350 Y-LUT= .00067 Obs# 2 X-LUT= .00386 Y-LUT= .00144 Obs# 3 X-LUT= .00135 Y-LUT= .00076 Obs# 4 X-LUT= .00088 Y-LUT= .00077 Obs# 5 X-LUT= .00107 Y-LUT= .00146 Obs# 6 X-LUT= .00154 Y-LUT= .00145 Obs# 7 X-LUT= .00094 Y-LUT= .00049 Obs# 8 X-LUT= .00146 Y-LUT= .00060 Obs# 9 X-LUT= .00103 Y-LUT= .00052 Obs#10 X-LUT= .00103 Y-LUT= .00071 Obs#11 X-LUT= .00137 Y-LUT= .00036 Obs#12 X-LUT= .00073 Y-LUT= .00050 Obs#13 X-LUT= .00000 Y-LUT= .00000 ------------------------------------------------- Table 3. Modulation along X-direction Raw Data ---------------------------------------------------- Frame No. Standard Peak Accumulated Deviation Deviation Photons ---------------------------------------------------- 2 .0284 .0517 57250 3 .0134 -.0246 117566 4 .0173 -.0273 233464 5 .0283 -.0318 117271 6 .0287 -.0414 117711 8 .0305 -.0400 117955 9 .0295 -.0444 118371 10 .0283 -.0488 118316 11 .0323 -.0468 353498 12 .0226 -.0318 347912 13 .0217 -.0383 345250 ---------------------------------------------------- Table 4. Modulation along X-direction Normalized by FR#13 ---------------------------------------------------- Frame No. Standard Peak Accumulate Deviation Deviation Photons ---------------------------------------------------- 2 .0307 .0613 57250 3 .0112 -.0149 117566 4 .0102 -.0151 233464 5 .0169 -.0279 117271 6 .0107 .0186 117711 7 .0141 -.0205 117955 8 .0130 .0198 118371 9 .0134 -.0262 118316 10 .0134 -.0243 353498 11 .0058 -.0090 347912 12 .0000 .0000 345250 ---------------------------------------------------- Table 5. Modulation along Y-direction Raw Data ---------------------------------------------------- Frame No. Standard Peak Accumulate Deviation Deviation Photons ---------------------------------------------------- 2 .0105 -.0206 57250 3 .0094 .0178 117566 4 .0095 0175 233464 5 .0147 -.0245 117271 6 .0143 .0319 117711 8 .0113 -.0200 117955 9 .0106 .0199 118371 10 .0095 -.0144 118316 11 .0089 -.0179 353498 12 .0072 .0144 347912 13 .0103 -.0177 345250 ---------------------------------------------------- Table 6. Modulation along Y-direction Normalized by FR#13 ---------------------------------------------------- Frame No. Standard Peak Accumulated Deviation Deviation Photons ---------------------------------------------------- 2 .0123 .0204 57250 3 .0109 .0139 117566 4 .0121 -.0160 233464 5 .0188 .0283 117271 6 .0199 .0392 117711 7 .0122 -.0175 117955 8 .0089 -.0160 118371 9 .0098 -.0212 118316 10 .0092 -.0153 353498 11 .0075 -.0151 347912 12 .0 000 .0000 345250 ---------------------------------------------------- Table 7. Two dimensional modulation patten in 8x8 cell Raw Data ---------------------------------------------------- Frame No. Standard Peak Accumulated Deviation Deviation Photons ---------------------------------------------------- 2 .0517 .1224 57250 3 .0538 -.1437 117566 4 .0518 -.1263 233464 5 .0555 .1510 117271 6 .0551 .1304 117711 7 .0558 .1367 117955 8 .0578 .1435 118371 9 .0544 .1743 118316 10 .0548 .1513 353498 11 .0477 .1361 347912 12 .0491 .1586 345250 ---------------------------------------------------- Table 8. Two dimensional modulation patten in 8x8 cell Normalized by FR#13 ---------------------------------------------------- Frame No. Standard Peak Accumulated Deviation Deviation Photons ---------------------------------------------------- 2 .0437 .1122 57250 3 .0281 .0756 117566 4 .0254 .0618 233464 5 .0356 .0902 117271 6 .0342 .0787 117711 7 .0350 .0818 117955 8 .0300 .0687 118371 9 .0314 -.0970 118316 10 .0244 -.0536 353498 11 .0207 .0496 347912 12 .0000 .0000 345250 ----------------------------------------------------