Home

Report on the FM ILT PhFPU functional tests

1. P A C S Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 15 Channel signal evolution matrix M 1 20060717_test2 Channel signal evolution matrix M 2 20060717_test2 6x10 610 ET as ae a a es ae ae Tre a IES toy ee eae 5x10 ax10 3x10 j 2x10 1x10 o o 1 A POS re 1 L rsrsrsrs Lio aiii 1 o 1000 0 3000 4000 5000 Sampling counter Figure 3 Continued Signal timeline for the 16 channels are blue matrices Channel signal evolution matrix M 5 g 6x10 T T T 20060717_test2 T Row pixel signal 2x10 Sampling counter ignol Row pixel si Sampling counter of matrices M1 and M2 in test 2 These Channel signal evolution matrix M 6 20060717 test2 6x10 5x10 4x10 3x10 2x10 1x10 o 1000 3000 Sampling counter 4000 5000 Figure 3 Continued Signal timeline for the 16 channels of matrices M5 and M6 in test 2 These are blue matrices In fact M6 is really M8 of the PhFPU as can be seen from the missing channel PAC S Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 16 Channel signal evolution matrix M 1 20060717_test3 6x10 5x10 a ool CC a e gt VE E GH E 4x10 ES EE E 3 1 a
2. Figure 7 Time evolution for the second set of biases Units on this figure are Volts Note that to produce a clean figure I have median filtered the HK values On the figure I give the actual HK values uncorrected for the GND BU level This figure shows test 1 ey bias evolution group B Key bias evolution group B_3 r m r 1 T vcc 0 0 f GG 0 0 PACS Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 26 Table 9 Bias commanding checks for the last set for biases indicates that the command is correctly executed For the BU biases I indicate both the commanded value and that value corrected by GND BU level in parenthesis Units in the table are Volts This table contains the values observed during test 1 A checkmark in the status column 24 08 2006 Bias name Commanded values Group 1 Group 3 Value Status Value Status VGL BU 3 0 3 46 3 42 27 3 46 J VDL BU 4 2 4 66 4 60 2 4 65 J VSS BU 1 0 1 46 1 45 y 1 46 y VGL 3 0 3 02 3 00 J VDL 3 0 3 04 2 3 00 J vss 0 7 0 70 J 0 70 J VSMS L 3 0 3 01 J 3 00 J VINJ 3 0 3 04 2 3 00 J D t SAp PACS MS 0652 06 PACS Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 27 4 2 2 Signal analysis We now turn to the data part of the telemetry As explained above because of the hardw
3. 12 3 17 07 2006 Signal timeline Matrices 1 and 2 test 1 13 3 17 07 2006 Signal timeline Matrices 5 and 6 test 1 14 3 17 07 2006 Signal timeline Matrices 1 and 2 test 2 15 3 17 07 2006 Signal timeline Matrices 5 and 6 test 2 15 3 17 07 2006 Signal timeline Matrix 1 test 43 o o o 16 3 17 07 2006 Signal timeline Matrix 5 test 43 0 o o 16 4 17 07 2006 VH VL sequence on matrices M1 and M2 test 1 17 4 17 07 2006 VH VL sequence on matrices M5 and M6 test 1 18 4 17 07 2006 VH VL sequence on matrices M1 and M2 test 2 19 4 17 07 2006 VH VL sequence on matrices M5 and M6 test 2 19 4 17 07 2006 VH VL sequence on matrices M1 and M5 test 3 20 5 17 07 2006 Power spectrum of the signal on M2 during the VH VL sequence Test 1 20 6 24 08 2006 Time evolution for the most important biases test 1 24 7 24 08 2006 Time evolution for the second set of biases test 1 25 8 24 08 2006 Signal timeline Matrix 1 test Hl o o 27 8 24 08 2006 Signal timeline Matrix 3 test H1l o o 28 8 24 08 2006 Signal timeline Matrices 1 and 2 test 2 29 8 24 08 2006 Signal timeline Matrices 5 and 6 test 2 29 8 24 08 2006 Signal timeli
4. 6205 z lt 7315 6200 7310 Geo EE PE A a oe R A E A ee EE E E E 2600 2800 3000 3200 3400 2600 2800 3000 3200 3400 Frame counter Frame counter Figure 4 Continued this time for matrices M1 and M2 during test 2 The non white noise component is not present here so the transitory periods are much more evident Zoom on VH VL sequence M5 test2_20060717 Zoom on VH VL sequence M6 test2 20060717 m T T T T y 7 T T T T T 4 VH 5875 8390 5870 pop dg iia bl v H o 5865 3 8385H 3 a L a gt L gt 5 5 2 L 5860 2 2 lt F X 8380 5855 8375 5850 TT Tt fiviitiviiti L 1 5845 1 1 1 L 4 1 2600 2800 3000 3200 3400 2600 2800 3000 3200 3400 Frame counter Frame counter Figure 4 Continued this time for matrices M5 and M6 during test 2 The non white noise component is not present here so the transitory periods are much more evident The data line shows quite a lot of spikes that were not present on figure 3 This is because here the signal was not median filtered P A OS Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 20 Zoom on VH VL sequence M1 test3_20060717 Zoom on VH VL sequence M5 test3_20060717 T T F id T T T T 8215 T T T VH lt lt 7665 8210 a i 8205 7660 FO pitt ith Mm 8200 Arbitrary scale
5. PAC S Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 1 Report on the FM ILT PhFPU functional tests Marc Sauvage with numerous inputs from Olivier Boulade Eric Doumayrou J r me Martignac Thomas Miller and Louis Rodriguez PACS Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 2 Document Change Record Version Date Changes 1 0 30 09 06 Document creation with tests of 24 08 2006 2 0 10 10 06 Included the tests of 17 07 2006 2 1 11 10 06 Included a discussion on correlated noise in 24 08 2006 3 0 29 11 06 Included the tests of 11 10 2006 Corrected various typos 4 0 01 12 06 Included the tests of 31 10 2006 Corrected various typos including the mis numbering of Section 5 s conclusions as section 6 D t SAp PACS MS 0652 06 PAC S ep j 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 3 Contents 1 Reference Documents 6 2 Purpose of this document 6 3 The warm functional tests of 17 07 2006 6 3 1 Test description s ss hk Sake as a a et Sew BS Le ak a ke er 6 3 2 ANALYSIS A A A ORE end oe A A Be 7 3 2 1 Bias commands execution ever rv rn knr rn kran 7 3 22 Signal Analysis vise ae A A a en GST ATA EE ED 13 3 80 Conelusions e s rc a A A a Sn ce 21 4 The warm functional tests of 24 08 2006
6. Arbitrary scale 7655 8195 8190 7650 AA AA 11 1 4 n 1 1 1 2600 2800 3000 3200 3400 2600 2800 3000 3200 3400 Frame counter Frame counter Figure 4 Continued this time for matrices M1 and M5 during test 43 Contrary to the previous panels as these are the two red matrices the data are obtained from two different groups The non white noise component is not present here so the transitory periods are much more evident Power spectrum M2 testl part2 20060717 1x107 8x10 6x1076 Volt Hz 05 4x107 2x107 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 5 10 15 20 frequency Hz Figure 5 The power spectrum of the mean signal on matrix M2 observed during the second part of the VH VL sequence VH 0 1 VL 0 1 V of test 1 P A CS Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 21 3 3 Conclusions The objective of this test is to check that all electrical lines to the bolometers are functionning This objective is fully reached here However it is worth mentioning that e One has to check that the commanding conversion tables correspond to the version of the BOLC hardware used e The relative level of the signal between the Sref_only and the Sb_only mode is different from that observed in SAp and the origin of this is unknown e We pick up an extra noise component with a characteristic frequency of 10 Hz w
7. I cannot however exclude that the file called Tm hk QM1 txt that resides in PIRE is not the correct file to use for BOLC QM1 On the other hand since I have the LTU logfile I can make a check of the analog to digital conversion First I have inspected the logfile and converted again almost all the analog values to their hexadecimal codes using the FM tables This reveals that almost all the conversions where indeed done with FM tables This is likely the reason for most of the discrepancies observed between the commanding and the HK i e the conversion tables do not correspond to the hardware This can indeed be verified using RD2 where the conversion coefficients for the commanding of BOLC QM1 are listed For all the biases showing commanding HK descrepancies of tables 7 to 9 I have converted back the digital command into an analog value This shows indeed that for all the discrepant biases the digital command used corresponds to an analog command equal to what we find in the HK Therefore we can conclude that all the discrepancies observed here are due to the use of incorrect commanding conversion tables This analysis also revealed two intriguing facts first the VH BLIND hexadecimal values are offset by 3 units from the expected values using FM conversion tables Second the VGG command setting it to 0 V as the hexadecimal argument 0001 Both the QM1 and FM conversion tables would give a negative Document SAp PACS MS 0652 06 PAC S Date 01
8. all group commanding method 11 10 2006 Bias setting values Group 1 VH_BLIND VDD VRL VH VL Value Status 1 20 1 20 1 20 1 66 1 66 1 22 1 68 1 68 1 24 1 70 1 70 1 26 1 72 1 72 1 22 1 22 1 24 1 24 1 26 1 26 2 60 3 06 3 06 1 80 1 80 0 30 0 30 0 40 0 40 0 50 0 50 0 40 0 40 0 30 0 30 1 15 1 15 0 10 0 10 0 1 0 10 0 00 0 00 0 00 0 00 ZR RSS SNS SSA SAS on all groups In other terms the commanding system appears to work as long as the right conversion tables are used This impression is confirmed when we turn to the second set of bias which is again shown as a time ordered sequence on table 12 and graphically on the right panel of figure 11 As can be seen from the table the commands are all perfectly executed Finally we examine the last set of bias Since those are commanded only once there is no need to build a timeline and table 13 simply shows the commanded and observed values This table contains one question mark which could ruin the impression given by the previous tables However inspecting the biases on the other groups shows that this is an exception Therefore the conclusion of this examination of the commanding is that on 11 10 2006 commanding was perfectly executed Document SAp PACS MS 0652 06 PAC S Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 38 Key bias evolution group B_1 test1_20061011 Key
9. the symetric 5 steps sequence Then we have a long plateau that corresponds to the VH VL sequence invisible with the full dynamical range but see later followed by the sequence of readout modes Sbolo_only Sref only and Sbolo Sref It is for this last sequence that we observe a clear difference with the reference sequence of RD1 the signal in the Sbolo_only is higher that in Sref only here while the opposite is true in RD1 The reason for this difference is not straightforward it cannot be the different VH_BLIND levels used for this run as the VH_BLIND differenciation is performed whatever the readout mode see RD1 It should therefore not affect the relative positionning of the Sbolo_only and Sref only signals It is very unlikely that it is a result of the FM QM1 commanding problem as this affects only group 1 biases yet both group 1 and group 3 signals show the same behavior PAC S Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 29 Figure 8 Continued Signal timeline for the 16 channels of matrices M1 and M2 in test 2 These are blue matrices Channel signal evolution matrix M 1 test2 Channel signal evolution matrix M 2 test2 T T T T T T T T TT 5x10 E A 5x10 a 4x10 0 3 3x10 B 3x10 gt gt E E H amp 2x10 2x10 1x10 1x10 fs o o a wera arar lr rr EE EE PE
10. 00 01 Valider enregistrement TM S 09 Set all groups bol bias 22 VDD PROT BU ON 1 P 00 16 0001 Set all groups bol bias 21 VDD PROT CL ON 1 P 00 15 0001 Set all groups bol bias 23 GND BU ON 1 P 00 17 0001 Set all groups bol bias 05 VCH to 0 00000000 Volt 0 P 00 05 0000 Set all groups bol bias 19 VGL BU to 3 00000000 Volt 2455 P 00 13 0997 Set all groups bol bias 18 VDL BU to 4 20000000 Volt 3436 P 00 12 OD6C Set all groups bol bias 17 VSS BU to 1 00000000 Volt 819 P 00 11 0333 E a A TU A A A A A Nr P BP e Ree PA o ion ct o a ct Q E w G Set all groups bol bias 15 VGG to 0 00000000 Volt 1 P 00 OF 0001 Set gain low P 08 00 00 01 Set all groups bol bias 20 VH BLIND to 0 70026000 Volt 575 P 00 14 023F Set all groups bol bias 16 VDD to 1 20000000 Volt 981 P 00 10 03D5 Attendre 5000 ms S 01 001388 Set all groups bol bias 16 VDD to 1 22000000 Volt 998 P 00 10 03E6 Attendre 5000 ms S 01 001388 Set all groups bol bias 16 VDD to 1 24000000 Volt 1014 P 00 10 03F6 Attendre 5000 ms S 01 001388 Set all groups bol bias 16 VDD to 1 26000000 Volt 1030 P 00 10 0406 Attendre 5000 ms S 01 001388 Set all groups bol bias 20 VH_BLIND to 0 72000000 Volt 591 P 00 14 024F Attendre 5000 ms E Ti ae eee A EE ee E A i ee GE sem PACS Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU function
11. 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 10 Figure 1 Time evolution for the 5 most important biases Units on this figure are Volts Note that to produce a clean figure I have median filtered the HK values On the figure I give the actual HK values uncorrected for the GND BU level This figure shows test 1 Key bias evolution group B_1 20060717_test1 Key bias evolution group B_3 20060717_test1 Volts Volts 1000 2000 3000 4000 5000 0 1000 2000 3000 4000 5000 0 Sampling counter Sampling counter code for a command of 0 V so this is in fact a protective feature of the LTU software This is why for that command we indeed get what we want 0001 is virtually indistinguishable from 0000 i e 0 V on VGG while non 0 V commands on VGG reveal the commanding problem PACS Herschel Document Date Version SAp PACS MS 0652 06 FM ILT PhFPU functional tests Table 4 Bias commanding checks for the second bias set of the test This is a time ordered table though the timing of the commands is not indicated checkmark in the status column indicates that the command is correctly executed For the BU biases I indicate both the commanded value and that value corrected by GND BU level in parenthesis Units in the table are Volts This table contains the time sequence test 1 17 07 2006 Bias setting values Group 1 Group 3 V
12. 38 e The harnesses are reconnected so as to now see matrices M5 M6 M7 and M8 i e the actual groups 3 and 4 Remember that M8 is the matrix with a missing readout channel This is test 2 of this section Telemetry files for this test have a date between 19 48 07 and 19 49 49 e The harnesses are reconnected so as to now see matrices M9 and M10 the read matrices i e groups 5 and 6 This is test 3 of this section Telemetry files for this test have a date between 20 01 48 and 20 03 30 Before moving on to the analysis I simply recall that judging from the logbook notes of that test everything appeared to proceed correctly 3 2 Analysis Because the test was performed with a private equipment it was also analyzed with a private system called PIRE which is inherited from the CAM interactive analysis and thus is in IDL An important aspect is the conversion of the HK information from decimal to analog values As the test was per formed with BOLC QM1 I have used a conversion file called Tm hk QM1 txt Using QM1 conversion factors actually required some slight modifications of PIRE to make it more flexible in its handling of the multiple BOLC versions The analysis performed is rather straightforward first we inspect the biases time sequence to check that the bias commands are correctly executed then we turn to the pixel signal to check that we observe the expected variations 3 2 1 Bias commands execution The first step her
13. 5 Group 6 0 457 0 458 0 455 0 457 0 459 0 459 As usual to simplify the inspection of the commanding I group the HK into three sets those that are used to modulate the signal during the test those that are modified so that the signal can be observed and those that are set only once The comparison between the commanding and the HK for the first set of bias is displayed in table 11 in the time order in which the commanding is made and graphically on the left panel of figure 11 I only show the values observed on group 1 as all the other groups show identical sequences once the marginal difference due to the all group commanding method are taken into account Table 11 shows that contrary to what we have observed so far all biases reach their commanded value P A CS Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 37 Table 11 Bias commanding checks for the 5 most important biases of the test This is a time ordered table though the timing of the commands is not indicated A checkmark in the status column indicates that the command is correctly executed Units in the table are Volts For the BU biases I indicate both the commanded value and that value corrected by GND BU level in parenthesis This table contains the time sequence for group 1 Identical sequences are observed for the other groups with marginal differences due to the
14. 9 followed by the sequence of readout modes Sbolo_only Sref_only and Sbolo Sref For this last sequence we observe a clear difference with the reference sequence of RD1 the signal in the Sbolo_only is higher that in Sref_only here while the opposite is true in RD1 See the text for a discussion of this discrepancy Chonnel signal evolution matrix M 1 test TIT a TT TT TT TTT 5x10 Raw pixel signal 1x10 pi pit 0 1000 2000 3000 4000 5000 Sampling counter Each of the channels displayed on figure 8 except one shows a similar pattern which is a first good Remember also that PIRE is in IDL and thus the index of M1 is 0 6 This is the absent channel of matrix M8 so this is expected and part of the success criteria for the functional test P A CS Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 28 Figure 8 Continued Signal timeline for the 16 channels of matrix M5 in test 1 This is the other red matrix Channel signal evolution matrix M 5 test Row pixel sign 1 0 1000 2000 3000 4000 5000 Sampling counter sign This pattern is almost identical to that of RD1 the first 4 downward steps correspond to the VDD sequence The following 3 upward steps correspond to the VH_BLIND scale They are followed by another upward step corresponding to the setting of VH_BLIND before the VRL scale This VRL scale is
15. Again a situation identical as that observed on 17 07 06 Thirdly there is a last set of biases that I ve grouped together because they are only set once I do not include biases that are set to 0 V Those are commanded with the 0000 hexadecimal code which is always correct This last set contains VGL BU VDL BU VSS BU VGL VDL VSS VSMS L and VINJ Table 9 compares the commanded values for these biases to the observed value This table is not time ordered as those biases are only set once during the test Again no significant difference is observed between the three tests so only the information of test 1 is given Here again there are some slightly discrepant values mostly for group 1 exactly identical to those we found during the tests of 17 07 2006 Finally the two remaining biases VCH and VSMS H that are supposed to remain at 0 V I have checked that this is indeed the case during all three tests In conclusion of this exploration of the bias commanding during this test we see that we are exactly in PAC S Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 24 Figure 6 Time evolution for the 5 most important biases Units on this figure are Volts Note that to produce a clean figure I have median filtered the HK values On the figure I give the actual HK values uncorrected for the GND BU level This figure shows test 1 Volts par 2000 3
16. E C E 1 590x10 1 462x10 L F F 1 589x10 1 461x10 1 588x10F 2600 2800 3000 3200 3400 2600 2800 3000 3200 3400 Frame counter Frome counter Figure 13 The VH VL sequence as observed on the mean signal for matrices M1 and M2 The effect of changing the VH VL biases is evident on the mean signal while it is much more difficult to indentify at a pixel level The transitory period that follows any bias change is most evident at the start of the sequence at this scale The non white noise that was apparent in previous test is gone but the strong correlations between matrices of a given group is still clear This series of figure reveals that the 10 Hz component observed in previous tests is gone This was expected since the shielding is much better when the complete FM setup is used but it is nevertheless a relief The power spectrum plots for the VH VL sequence do not reveal any interesting feature so I have not displayed them Similarly to what I had done previously I have looked at the correlation coefficients between the mean signals I again find that there is a very high level of correlation gt 0 9 between the mean signals of two matrices belonging to the same group which is not very surprising given that the signal is completely command driven except for group 3 which is principally due to the high noise level of matrix 6 Now that the 10 Hz component has disappeared the correlation coefficient bet
17. FM ILT PhFPU functional tests Figure 15 Continued Signal timeline for the 16 channels of matrices M5 and M6 Channel signal evolution matrix M 6 20061031_test4 li i Channel signal evolution matrix M 5 20061031_test4 a x x a teuis roxid wey 2 0x10 1 5x10 2 0x10 1 0x10 Sampling counter 5 0x10 1 0x10 1 5x10 5 0x10 Sampling counter Figure 15 Continued Signal timeline for the 16 channels of matrices M7 and M8 Channel signal evolution matrix M 8 20061031 test4 2 0x10 1 6x10 5 0x10 Channel signal evolution matrix M 7 20061031 test4 i ih vous txd moy 2 0x10 1 5x10 5 0x10 1 0x10 Sampling counter 1 0x10 Sampling counter 8x10 teufs jaxid moy 1 5x10 2 0x10 1 0x10 5 0x10 ee AAA 1 N a lh N N 2 L o 3 H A 75 i J a N Al O 8 lt el Aa el E E el oO ee 5 g E gt O R ai 3 a 3 E or _ ii Y E Channel signal evolution matrix M 9 20061031_test4 Figure 15 Continued Signal timeline for the 16 channels of matrices M9 and M10 FM ILT PhFPU functional tests PACS route txd moy 2 0x10 1 5x10 5 0x10 1 0x10 Sampling counter Sampling counter P A CS Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 56 The VH
18. S 01 002710 Set all groups bol bias 12 VDECX L to 0 00000000 Volt 0 P 00 OC 0000 Attendre 1000 ms S 01 0003E8 Set all groups bol bias 09 CKRLH to 2 00000000 Volt 2298 P 00 09 O8FA Attendre 1000 ms S 01 0003E8 Set seq mode Sb only P 09 01 00 02 Attendre 1000 ms S 01 0003E8 Attendre 10000 ms S 01 002710 Set seq mode Sref only P 09 01 00 01 Attendre 1000 ms S 01 0003E8 Attendre 10000 ms S 01 002710 Set seq mode Sb Sref P 09 01 00 00 Attendre 1000 ms S 01 0003E8 Attendre 10000 ms S 01 002710 Inhiber enregistrement TM S 08
19. The VH VL sequence as observed on the mean signal for matrices M1 and M2 during test 1 These are blue matrices The effect of changing these biases is evident on the mean signal while it is much more difficult to indentify at a pixel level The transitory period that follows any bias change is most evident at the start of the sequence at this scale Also evident is the presence of non white noise on the data that is clearly correlated between matrices see text for analysis The first important point to stress is that the global behavior of the VH VL sequence is in accordance with the expectations Remember that we are during this part of the test in the mode Sref_only ie that we are reading the bolometer signal or the midpoint bias level and that this signal is affected by a minus sign Therefore decreasing the midpoint level by decreasing VH or VL increases the signal level Thus this part of the test sequence is successful The second point is a little bit more worrying during test 1 we observe the presence of a supplementary noise component that is correlated between matrices of a given group as well as between groups To try and identify the origin of this noise component we have computed the power spectrum of the signal during the VH VL sequence This is in fact not a simple task as this sequence is not meant to compute power spectrum i e the plateaus are rather short about 200 readouts each thus the resulting power spectr
20. are followed by another upward step corresponding to the setting of VH_BLIND before the VRL scale This VRL scale is the symetric 5 steps sequence Then we have a long plateau that corresponds to the VH VL sequence invisible with the full dynamical range but see later followed by the sequence of readout modes Sbolo_only Sref_only and Sbolo Sref We now observe the same behaviour as in This is the absent channel of matrix M8 which is observed for the first time at the correct telemetry location since we can now operate the complete PhFPU with BOLC Therefore this is really part of the success criteria for the functional test P A CS Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 41 Figure 12 Continued Signal timeline for the 16 channels of matrices M3 and M4 Channel signal evolution matrix M 3 20061011_test1 Channel signal evolution matrix M 4 20061011_test1 82104 ot Poot Toot ot Tt T 82104 TT Poot Tot oot ot T y r Ir b s 6x10 I EA y Fi E 3 3 E 2 S amp G 4x n 4x10 z 2 2 2 z z 2 3 2x 2x10 o 11114341 MV dra o 1000 2000 3000 4000 5000 Sampling counter Sampling counter the reference sequence of RD1 the signal in the Sbolo only is lower that in S
21. et 20 03 03 03 03 03 CKRLH to 1 50000000 Volt 1724 CKRLL to 1 50000000 Volt 1724 VSMS L to 3 00000000 Volt 3447 VSMS H to 0 00000000 Volt 0 VGG to 1 300 nA VH BLIND VRL VRL VRL VRL VRL test de VRL pour chaque groupe groupe 1 15000000 Volt 941 to 1 75000000 Volt 1432 20000000 Volt 230 30000000 Volt 345 40000000 Volt 460 30000000 Volt 345 20000000 Volt 230 PACS Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 67 NHNHNHN RP RP PHN HNHNHNHN HN OP KRPOOHKHNHNHN AE DE SEE AE E E SE E SEE REE HN HN HN HN HN HS Set group 1 bol bias P 01 03 00E6 Attendre 10000 ms S 01 002710 Set group 1 bol bias P 01 03 01CC Attendre 10000 ms S 01 002710 Set group 1 bol bias P 01 03 00E6 Attendre 10000 ms S 01 002710 groupe 2 Set group 2 bol bias P 02 03 00E6 Attendre 10000 ms S 01 002710 Set group 2 bol bias P 02 03 01CC Attendre 10000 ms S 01 002710 Set group 2 bol bias P 02 03 00E6 Attendre 10000 ms S 01 002710 groupe 3 Set group 3 bol bias P 03 03 00E6 Attendre 10000 ms S 01 002710 Set group 3 bol bias P 03 03 01CC Attendre 10000 ms S 01 002710 Set group 3 bol bias P 03 03 00E6 Attendre 10000 ms S 01 002710 groupe 4 Set group 4 bol bias P 04 03 00E6 Attendre 10000 ms S 01 002710 Set group 4 bol bias P 04 03
22. group contains VH_BLIND VDD VRL VH and VL Table 7 lists the time sequence of the commands to these biases as well as the status of the command deduced from the HK Figure 6 shows graphically the time evolution of these biases For this part of the test analysis it turns out that the differences between the three tests are very small so I have only displayed the information for test 1 At this point It appears that the test is proceeding nominally apart from differences between the commanded and housekeeping values for group 1 on VDD essentially Interestingly at the 10mV accuracy we observe no difference in the bias setting between this test and that of 17 07 06 Given that the test script is exactly the same this is both expected and welcome In the second set of biases I have placed biases which are set more than once during the script although they are not directly responsible for the signal variation in the test They are adjusted so that we can see the signal variation These biases are VGG VDECX H VDECX L CKRLL and CKRLH As for the first set of bias I have placed in table 8 the time ordered sequence of bias setting commands and their execution status Figure 7 shows the graphical timeline of these biases Since most of them have the same values I have introduced artificial offsets so that the figure becomes clearer Here again P A CS Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU
23. the digital to analog conversion of the HK I have tried using the FM conversion coefficients but this did not solve the P A CS Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 9 Table 3 Bias commanding checks for the 5 most important biases of the test This is a time ordered table though the timing of the commands is not indicated The status column is decided upon the match quality between the commanded and the HK value A checkmark indicates that the command is correctly executed a question mark that some discussion is required and a double question mark that an investigation is required Units in the table are Volts For the BU biases I indicate both the commanded value and that value corrected by GND BU level in parenthesis This table contains the time sequence for test 1 17 07 2006 Bias setting values Group 1 Group 3 VH_BLIND VDD VRL VH VL Value Status Value Status 0 70 0 69 y 0 70 J 1 20 1 66 1 64 1 66 J 1 22 1 68 1 66 1 68 J 1 24 1 70 1 68 1 70 J 1 26 1 72 1 70 1 72 J 0 72 0 71 J 0 72 J 0 74 0 73 J 0 74 y 0 76 0 75 y 0 76 y 2 60 3 06 3 02 3 06 y 1 30 1 29 y 1 30 y 0 30 0 31 J 0 30 J 0 40 0 41 J 0 40 y 0 50 0 51 J 0 50 J 0 40 0 41 J 0 40 J 0 30 0 30 J 0 30 J 1 10 1 09 J 1 10 y 0 10 0 10 J 0 10 J 0 1 0 10 J 0 10 J 0 00 0 00 y 0 00 y 0 00 0 00 J 0 00 y problem
24. we need to add wait times after the commanding of the protection biases and GND BU This will in fact become the nominal bias setting procedure starting from Draft 7 of RD1 The actual test execution is much simpler now and here the complete test script was simply executed twice Since the differences between the two run of the test are extremely small I will not treat them separately 5 2 Analysis Similarly to the previous LTU operated tests these tests were analysed with PIRE in a version that has been updated to handle correctly the different version of the commanding and HK conversion tables We made sure to specify that FM Main versions had to be used here The next sections deal with the two steps of the analysis the bias time sequence and the signal time signal In this second section we pay special attention to the noise characteristics 5 2 1 Bias commands execution As usual here we first need to record the values of GNB BU as this voltage will be added to all the BU bias values These values are listed in table 10 They have changed slightly since the previous test but this can be attributed to the major change that constitutes the replacement of BOLC QM1 by BOLC FM Table 10 The median value of GND BU measured during the LTU tests The units are Volts The values are exactly the same between the two tests therefore only the values measured during test 1 are shown 11 10 2006 Group 1 Group 2 Group 3 Group 4 Group
25. 00 6195 6190 6185 6180 Zoom on VH VL sequence M1 test3_20060824 AAA a D E ee i a A AAA EE err a n 1 1 1 1 1 1 1 1 fi 1 1 3200 3400 3600 3800 Frame counter Arbitrary scale 7315 7310 7305 7300 7295 TTS O E LE Zoom on VH VL sequence M2 test3_20060824 AA AA A gt AA T cloaca pe le q 1 1 3200 3400 3600 3800 Frame counter Figure 9 Continued The VH VL sequence as observed on the mean signal for matrices M1 and M2 during test 3 These are blue matrices belonging to the same BOLC group Arbitrary scole 8380 8370 8360 8350 8340 Zoom on VH VL sequence M5 test3_20060824 AAA gt Lg T VH A TTT TTT TT TTTTTTTTTTT porto tiiiit 1 n 1 4 1 n 1 1 1 1 n L L 3200 3400 3600 3800 Frame counter Arbitrary scale 5850 5845 5840 5835 5830 5825 5820 Zoom on VH VL sequence M6 test3_20060824 T T VH a a EL EL la a il tr 1 1 1 1 1 1 n 1 1 1 1 1 1 3200 3400 3600 3800 Frame counter Figure 9 Continued The VH VL sequence as observed on the mean signal for matrices M5 and M6 during test 3 These are blue matrices belonging to the same BOLC group P A CS Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 34 Power spectrum M2 test2 part2 20060824
26. 00 3000 3200 3400 Frame counter Frame counter Figure 4 Continued this time for matrices M5 and M6 during test 1 As can be seen the non white noise component is also correlated between groups between matrices and groups during a given test lts physical origin is probably related to the fact that the test equipment used here does not guarantee a complete shielding of the PhFPU BOLC system The QM harnesses used to connect BOLC QM1 need special adapters to connect to the FM PhFPU and these break the shielding continuity When the complete FM instrument is assembled the PhFPU and BOLC will be completely shielded I have performed the same exercice for test 2 where figure 4 seems to indicate that this noise component is absent This reveals that we in fact still pick up the 10Hz noise component but at a much fainter level the peak s height is generally a factor 3 to 10 times smaller If one wants to be optimistic this variation in the amplitude of the 10 Hz component can be seen as a confirmation that it is indeed due to noise pick up through the BOLC PhFPU connection harnesses are these are disconnected and reconnected between tests D t SAp PACS MS 0652 06 PAC S i P 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 19 Zoom on VH VL sequence M1 test2_20060717 Zoom on VH VL sequence M2 test2_20060717 T T T T A E T T i T i T T 4 6215 7325 6210 7320 gt gt
27. 00 Volt 0 P 00 02 0000 Attendre 5000 ms S 01 001388 Set all groups bol bias 12 VDECX L to 0 00000000 Volt 0 P 00 0C 0000 Set all groups bol bias 09 CKRLH to 2 00000000 Volt 2298 P 00 09 O8FA Set seq mode Sb only P 09 01 00 02 Attendre 5000 ms S 01 001388 Set seq mode Sref only P 09 01 00 O1 Attendre 5000 ms S 01 001388 Set seq mode Sb Sref P 09 01 00 00 Attendre 5000 ms S 01 001388 Inhiber enregistrement TM S 08 A A A E A A oe A A A ni ene Ss Se A a A A A et pe ge Gee ge Ce A 2 Test script of 24 08 2006 The script used was identical to that of 17 07 2006 A 3 Test script of 11 10 2006 Comparing this script to the previous ones reveals subtle differences First the VH_BLIND values are now those of RD1 as we have executed this test with the FM hardware Second we have introduced wait times after the switch on of the protection biases and of GND BU Draft 7 of RD1 elaborates on D t PACS De SAp PACS MS 0652 06 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 61 these wait times Reset bias all groups P 00 00 00 00 Set all groups bol bias 22 VDD PROT BU ON 1 P 00 16 0001 Attendre 1000 ms S 01 0003E8 Set all groups bol bias 21 VDD PROT CL ON 1 P 00 15 0001 Attendre 1000 ms S 01 0003E8 Set all groups bol bias 23 GND BU ON 1 P 00 17 0001 Attendre 1000 ms 17 00 S 01 0003E8 Set all groups bol bias 05 VCH to 0 00000000
28. 000 1000 Sampling counter I VHBLIND r aa VH_BLIND 0 1000 2000 3000 100 Sampling counter the same situation as on 17 07 2006 Therefore we know that the reason for the bias command HK discrepancies is the use of incorrect commanding tables One should remark that the purpose of repeating these tests is to make sure nothing changes between each occurence of the test s far as the biases are concerned this is indeed the case and thus the test is successful P A OS Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 25 Table 8 Bias commanding checks for the second bias set of the test This is a time ordered table though the timing of the commands is not indicated A checkmark in the status column indicates that the command is correctly executed For the BU biases I indicate both the commanded value and that value corrected by GND BU level in parenthesis Units in the table are Volts This table contains the time sequence test 1 24 08 2006 Bias setting values Group 1 Group 3 VGG VDECX H VDECX L CKRLL CKRLH Value Status Value Status 0 0 0 46 0 46 J 0 46 J 1 9 2 36 2 34 2 36 J 0 0 0 00 y 0 00 J 0 0 0 00 J 0 00 J 1 5 1 51 J 1 50 J 1 5 1 51 J 1 50 y 1 8 2 26 2 24 2 26 y 0 0 0 00 J 0 00 y 0 0 0 00 J 0 00 y 2 0 2 01 J 2 00 y 2 0 2 01 y 2 00 J 0 0 0 00 y 0 00 J 2 0 2 01 J 2 00 J
29. 01388 Set all groups bol bias 02 VL to 0 00000000 Volt 0 P 00 02 0000 Attendre 5000 ms S 01 001388 Set all groups bol bias 12 VDECX L to 0 00000000 Volt 0 P 00 OC 0000 Set all groups bol bias 09 CKRLH to 2 00000000 Volt 2298 P 00 09 O8FA Set seq mode Sb only P 09 01 00 02 Attendre 5000 ms S 01 001388 Set seq mode Sref only P 09 01 00 01 Attendre 5000 ms S 01 001388 Set seq mode Sb Sref P 09 01 00 00 Attendre 5000 ms S 01 001388 Inhiber enregistrement TM S 08 Fin Batch A 4 Test script of 31 10 2006 This test script corresponds to the 4K level test There are some bias differences with the test scripts corresponding to the 300K level Also when comparing it to the previous test script or to the test script of RD1 you will see that we have introduced artificial wait times to mimic the incompressible rate of two commands per second introduced by the CUS Batch de test 4K Driv du test 300K tel que effectu a garching le 11 10 06 avec ajout de 0 5s avec chaque TC pour simuler SC0S2000 Tests fonctionnels 4K SN ASA gt gt SS sos Valider enregistrement TM D t PACS De SAp PACS MS 0652 06 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 64 IS SEE AE AE AE AE DE AE DE DE DE OE AE AE DE AE AE AE DE AE AE AE AE DEE HN EENEN S 09 Attendre 1000 ms S 01 0003E8 Set data mode Bolo amp HK P 09 02 00 O1 Atte
30. 01CC Attendre 10000 ms S 01 002710 Set group 4 bol bias P 04 03 00E6 Attendre 10000 ms S 01 002710 groupe 5 Set group 5 bol bias P 05 03 00E6 Attendre 10000 ms S 01 002710 Set group 5 bol bias P 05 03 01CC Attendre 10000 ms 03 03 03 03 03 03 03 03 03 03 03 03 03 03 VRL VRL VRL VRL VRL VRL VRL VRL VRL VRL VRL VRL VRL VRL to to to to to to to to to to to to to to 20000000 40000000 20000000 20000000 40000000 20000000 20000000 40000000 20000000 20000000 40000000 20000000 20000000 40000000 Volt Volt Volt Volt Volt Volt Volt Volt Volt Volt Volt Volt Volt Volt 230 460 230 230 460 230 230 460 230 230 460 230 230 460 P A CS Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 68 S 01 002710 Set group 5 bol bias 03 VRL to 0 20000000 Volt 230 P 05 03 OOE6 Attendre 10000 ms S 01 002710 groupe 6 Set group 6 bol bias 03 VRL to 0 20000000 Volt 230 P 06 03 00E6 Attendre 10000 ms S 01 002710 Set group 6 bol bias 03 VRL to 0 40000000 Volt 460 P 06 03 01CC Attendre 10000 ms S 01 002710 Set group 6 bol bias 03 VRL to 0 20000000 V
31. 10 which is on group 6 showed dramatic signal drifts during both tests that led us to re open the LTU box and perform another series of LTU driven tests 6 1 Test Description The aim with this series of test was to perform an LTU test that would be as close as possible to the CUS driven test in order to eliminate all possible sources of difference in case the results observed with these two set ups would not be the same To that effect we introduced artificial delays in the LTU driven test Eventhough we performed more than one test with or without the delays I am only going to present here the results from the last LTU run where the delays are as close as possible to those introduced by the CUS if need be I will refer to this test as the CUS like test No execution or signal difference was observed between the different runs On October 31 this was the 4 fourth test executed The actual script can be found in Appendix A 6 2 Analysis This analysis is done with PIRE with the same version as that used to analyse the tests of October 11 PIRE does not evolve much 6 2 1 Bias commands execution The first step is to check the values of GND BU as it is added to all the BU HKs and needs to be taken into account when checking the correct execution of bias commands The measured values are given in table 14 Note that these values are exactly the same as those observed on 11 10 06 This is supposed to be the case but it nevertheless r
32. 1276 VGG to 1 30000000 Volt 1064 VDD to 2 60000000 Volt 2128 VGL to 3 00000000 Volt 3447 VDL to 3 00000000 Volt 3447 VSS to 1 30000000 Volt 1494 VDECX H to 0 00000000 Volt 0 VDECX L to 0 00000000 Volt 0 PACS Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 66 S 01 0003E8 Set all groups bol bias P 00 09 06BC Attendre 1000 ms S 01 0003E8 Set all groups bol bias P 00 OA O6BC Attendre 1000 ms S 01 0003E8 Set all groups bol bias P 00 OE OD77 Attendre 1000 ms S 01 0003E8 Set all groups bol bias P 00 OD 0000 Attendre 1000 ms S 01 0003E8 Set all groups bol bias P 00 OF 03AD Attendre 1000 ms S 01 0003E8 Courant de CL entre 100 Set all groups bol bias P 00 14 0598 Attendre 1000 ms S 01 0003E8 Attendre 10000 ms S 01 002710 Set all groups bol bias P 00 03 00E6 Attendre 1000 ms S 01 0003E8 Attendre 10000 ms S 01 002710 Set all groups bol bias P 00 03 0159 Attendre 1000 ms S 01 0003E8 Attendre 10000 ms S 01 002710 Set all groups bol bias P 00 03 O1CC Attendre 1000 ms S 01 0003E8 Attendre 10000 ms S 01 002710 Set all groups bol bias P 00 03 0159 Attendre 1000 ms S 01 0003E8 Attendre 10000 ms S 01 002710 Set all groups bol bias P 00 03 00E6 Attendre 1000 ms S 01 0003E8 Attendre 10000 ms S 01 002710 fin de test de VRL 09 10 14 13 15
33. 2 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 62 Attendre 5000 ms S 01 001388 fin de test du BU blocage PEL commut Set all groups bol P 00 OF 0614 Set all groups bol P 00 10 084E Set all groups bol P 00 08 0D78 Set all groups bol P 00 06 0D77 Set all groups bol P 00 07 0324 test du vrl Set all groups bol P 00 0B 0000 Set all groups bol P 00 OC 0000 Set all groups bol P 00 09 06BC Set all groups bol P 00 OA O6BC Set all groups bol P 00 OE 0D77 Set all groups bol P 00 OD 0000 Set all groups bol P 00 OF 05C2 bias bias bias bias bias bias bias bias bias bias bias bias Courant de CL entre 0 5 Set all groups bol P 00 14 05C1 Attendre 5000 ms S 01 001388 Set all groups bol P 00 03 0159 Attendre 5000 ms S 01 001388 Set all groups bol P 00 03 01CB Attendre 5000 ms S 01 001388 Set all groups bol P 00 03 023E Attendre 5000 ms S 01 001388 Set all groups bol P 00 03 01CB Attendre 5000 ms S 01 001388 Set all groups bol P 00 03 0159 Attendre 5000 ms S 01 001388 fin de test de VRL Set all groups bol P 00 09 0000 Set all groups bol P 00 OA 0000 Set all groups bol P 00 OB 08FA bias bias bias bias bias bias bias bias bias 15 16 08 06 07 11 12 09 10 14 13 15 uA 20 03 03 03 03 03 09 10 11 VGG to 1 VDD to 2 VGL to 3 VDL to
34. 21 41 MTestrdescriptiO Mt sa RN Gr aa a A AO a Be 21 422 AMAS O a hk Ps ek Sale E E ohh Att ok Gr 22 4 2 1 Bias commands execution 2 2 2 00 eee ee ee ee 22 4 0 2 Signal analysis olaa A ee eels AOA Gale 27 4 3 TCONCIUSIONS ya e a ri GR ARA a o e a al 34 5 The warm functional tests of 11 10 2006 36 5l Test Description 2054 gee a a a e eg 36 5 2 Analysis uses agree a is Be ee a 36 5 2 1 Bias commands execution e 36 5 2 2 Signal analysis es A a ES e A a ale oe r 40 5 35 CONCIUSIONS gt 2 99 TOGO RE ee eh ee AAA len kst ia a 44 6 The LTU driven 4K functional tests of 31 10 2006 48 6 1 Test Description a A ia Be 48 052 Analysis Dit E A A ee A yee Ah ee ot GE Ge Rede Ads 48 6 2 1 Bias commands execution 2 e 48 6 2 2 Signal Analysis 2 ss ek ma ae eae a ee REE a 51 6 3 Conclusions Dl g a ae Gadla ee A eg en Peed ee eh 56 A The test scripts 58 Aad Test seipt FAT OT 2000 e A ARMA AA AAA AA 58 A 2 Test script of 24 08 2006 4 saa a a OE Se ORS eS 60 Ass Test serpy Ob M W C2000 os eno ee en EN Bal 60 Aa Test script of 31 10 2006 a eS cet hae BA EG ee RA 63 PACS Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 4 List of Figures 1 17 07 2006 Time evolution for the most important biases test 1 10 2 17 07 2006 Time evolution for the second set of biases test 1
35. 3 VSS to 0 VDECX H to 0 00000000 Volt 0 VDECX L to 0 00000000 Volt 0 CKRLH to 1 50000000 Volt 1724 CKRLL to 1 50000000 Volt 1724 VSMS L to 3 00000000 Volt 3447 VSMS H to 0 00000000 Volt 0 VGG to 1 80000000 Volt 1474 et 2 uA VH_BLIND VRL to 0 VRL to 0 VRL to 0 VRL to 0 VRL to 0 CKRLH to CKRLL to VDECX H to 2 00000000 Volt 2298 90000000 Volt 1556 60000000 Volt 2126 00000000 Volt 3448 00000000 Volt 3447 70000000 Volt 804 to 1 80000000 Volt 30000000 Volt 345 40000000 Volt 459 50000000 Volt 574 40000000 Volt 459 30000000 Volt 345 0 00000000 Volt 0 0 00000000 Volt 0 P A CS Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 63 Set all groups bol bias 12 VDECX L to 2 00000000 Volt 2298 P 00 OC O8FA Set all groups bol bias 04 VINJ to 3 00000000 Volt 3447 P 00 04 OD77 Set all groups bol bias 20 VH_BLIND to 1 15000000 Volt 941 P 00 14 O3AD Set gain high P 08 00 00 00 Set all groups bol bias 01 VH to 0 10000000 Volt 115 P 00 01 0073 Attendre 5000 ms S 01 001388 Set all groups bol bias 02 VL to 0 10000000 Volt 277 P 00 02 0115 Attendre 5000 ms S 01 001388 Set all groups bol bias 01 VH to 0 00000000 Volt 0 P 00 01 0000 Attendre 5000 ms S 01 0
36. 6 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 65 S 01 0003E8 Attendre 10000 ms S 01 002710 Set all groups bol P 00 10 049A Attendre 1000 ms S 01 0003E8 Attendre 10000 ms S 01 002710 Set all groups bol P 00 10 04AB Attendre 1000 ms S 01 0003E8 Attendre 10000 ms S 01 002710 Set all groups bol P 00 14 04DC Attendre 1000 ms S 01 0003E8 Attendre 10000 ms S 01 002710 Set all groups bol P 00 14 04EC Attendre 1000 ms S 01 0003E8 Attendre 10000 ms S 01 002710 Set all groups bol P 00 14 04FC Attendre 1000 ms S 01 0003E8 Attendre 10000 ms S 01 002710 fin de test du BU blocage PEL commut Set all groups bol P 00 OF 0428 Attendre 1000 ms S 01 0003E8 Set all groups bol P 00 10 0850 Attendre 1000 ms S 01 0003E8 Set all groups bol P 00 08 0D77 Attendre 1000 ms S 01 0003E8 Set all groups bol P 00 06 0D77 Attendre 1000 ms S 01 0003E8 Set all groups bol P 00 07 05D6 Attendre 1000 ms S 01 0003E8 test du vrl Set all groups bol P 00 OB 0000 Attendre 1000 ms S 01 0003E8 Set all groups bol P 00 OC 0000 Attendre 1000 ms bias bias bias bias bias bias bias bias bias bias bias bias 16 16 20 20 20 15 16 08 06 07 11 12 VDD to 1 44000000 Volt 1178 VDD to 1 46000000 Volt 1195 VH BLIND to 1 52000000 Volt 1244 VH BLIND to 1 54000000 Volt 1260 VH BLIND to 1 56000000 Volt
37. 800 3000 3200 3400 2600 2800 3000 3200 3400 Frame counter Frame counter Figure 13 Continued The VH VL sequence as observed on the mean signal for matrices M7 and M8 P A OS Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 47 Zoom on VH VL sequence M9 20061011_test1 Zoom on VH VL sequence M10 20061011_test1 FIg q gt gt _ 5H 1 301x10 VH VH 1 666 10 lt ES lt 1 300x10 1 665x10 1 299x10 1 664x10 1 298x10 1 663x10 1 297x10 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 i 1 1 1 1 3000 3200 3400 2600 2800 3000 3200 3400 Frame counter Frome counter 1 1 1 1 1 1 2600 2800 Figure 13 Continued The VH VL sequence as observed on the mean signal for matrices M9 and M10 P A CS Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 48 6 The LTU driven 4K functional tests of 31 10 2006 In principle these tests were not supposed to happen The LTU s last use should have occured on October 11 at which point control and cable connection was to be surrendered to PACS However the repetition of the warm functional test using the PACS FM set up on October 12 and later on October 17 revealed mysterious problems group 6 was on at the start of the October 12 test and matrix
38. FPU functional tests Page 51 Table 16 Bias commanding checks for the second bias set of the test This is a time ordered table though the timing of the commands is not indicated A checkmark in the status column indicates that the command is correctly executed For the BU biases I indicate both the commanded value and that value corrected by the GND BU level in parenthesis Units in the table are Volts Only group 1 is show but all the other groups show the same behavior Note again some subtle differences in the commanded bias levels with respect to the 300 K level tests 31 10 2006 Bias setting values Group 1 VGG VDECX H VDECX L CKRLL CKRLH Value Status 0 00 0 46 0 46 1 30 1 76 1 76 0 0 0 00 0 0 0 00 1 5 1 50 1 5 1 50 1 15 1 61 1 61 0 0 0 00 0 0 0 00 2 0 2 00 2 0 2 00 0 0 0 00 2 0 2 00 EA ERES ER short spike on the way to its high level is now a minor plateau Again the purpose of these artificial delays is to make the LTU driven test as similar as possible to the CUS driven test This way visual checking of the functional test success is easier and we rule out commanding as a potential source of problems in the test execution or behavior Finally in table 17 we show the measured values of the last set of biases We find here exactly the same results as on October 1 all biases are nominally commanded except VDL BU which is sligthly discrepant on group 1 This is
39. GG VDECX H VDECX L CKRLL CKRLH Value Status Value Status 0 0 0 46 0 46 y 0 46 J 1 9 2 36 2 34 2 36 J 0 0 0 00 J 0 00 J 0 0 0 00 J 0 00 y 1 5 1 51 J 1 50 J 1 5 1 51 J 1 50 J 1 8 2 26 2 24 2 26 J 0 0 0 00 J 0 00 J 0 0 0 00 y 0 00 J 2 0 2 01 J 2 00 J 2 0 2 01 J 2 00 J 0 0 0 00 J 0 00 J 2 0 2 01 J 2 00 J Table 5 Bias commanding checks for the last set for biases indicates that the command is correctly executed For the BU biases I indicate both the commanded value and that value corrected by GND BU level in parenthesis Units in the table are Volts This table contains the values observed during test 1 17 07 2006 Bias name Commanded values Group 1 Group 3 Value Status Value Status VGL BU 3 0 3 46 3 42 27 3 46 J VDL BU 4 2 4 66 4 60 7 4 65 J VSS BU 1 0 1 46 1 45 J 1 46 J VGL 3 0 3 02 3 00 y VDL 3 0 3 04 7 3 00 J VSS 0 7 0 70 y 0 70 J VSMS L 3 0 3 01 J 3 00 J VINJ 3 0 3 04 2 3 00 J A checkmark in the status column Document SAp PACS MS 0652 06 PAC S Date 01 12 06 4 0 Herschel Version Page 12 FM ILT PhFPU functional tests Figure 2 Time evolution for the second set of biases Units on this figure are Volts Note that to produce a clean figure I have median filtered the HK values On the figure I give the actual HK values uncorrected for the GND BU level This figure shows test 1 Key bias evolution gr
40. Power spectrum M1 test3 part2 20060824 T T T T T Volt H2 05 Figure 10 On the left panel I show the power spectrum of the mean signal on matrix M2 observed during the second part of the VH VL sequence VH 0 1 VL 0 1 V of test 2 This panel is equivalent to figure 5 and immediately shows that the power in the 10 Hz component is much higher now here the peak is 3 5 times higher This also reveals a frequent situation where the peak has a lot of structure and its main component appears at a lower frequency The origin of this effect is unknown but could be related to the small number of samples we have to compute the power spectra On the right panel I show the power spectrum of the mean signal on matrix M1 during the same sequence but this time from test 3 Here the peak is strong as well and better centered at 10 Hz location MPE here KT for the tests of 17 07 2006 we can only hope that the FM shielding will be efficient Second we see sometimes the 10 Hz peak shifted to smaller frequencies as on the left panel of figure 10 broadened to 2 3 Hz It is impossible to tell whether this is an effect of the small number of samples 200 we have to compute the power spectrum This leads me to recommend that we place at the end of the VH VL sequence a longer wait time to accumulate frames that would be used to characterize and possibly identify any mysterious noise source that might affect the signal Fin
41. R arar dl rr dr rr rr br rr dr rr rr AR o 1000 2000 3000 4000 5000 5000 5000 6000 Sampling counter Sampling counter Figure 8 Continued Signal timeline for the 16 channels of matrices M5 and M6 in test 2 These are blue matrices Channel signal evolution matrix M 5 test2 T T T T Channel signal evolution matrix M 6 test2 T T T T T T T 7 5x10 5x10 ad 4 7 oe ims m a I 1 4x10 x anioi 7 E El E JE y 3 3x10 8 3x10 gt gt E E z JT z CET gt A n 2x10 1x10 1x10 o o APA AAR iii Diviiiiiis Disiiiiiis EE O APA AAR iii Disiiiiiis Disiiiiiis Iisirasass o 1000 2000 3000 4000 5000 6000 o 1000 2000 3000 4000 5000 6000 Sampling counter Sampling counter PAC S Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 30 Figure 8 Continued Signal timeline for the 16 channels of matrices M1 and M2 in test 3 These are blue matrices Channel signal evolution matrix M 1 test3 T T T Channel signal evolution matrix M 2 test3 T T 5x10 3 2 3x10 F gt E E ES E e 13 2x10 o 1000 2000 3000 4000 5000 3000 4000 5000 Sompling counter Sompling counter Figure 8 Continued Signal timeline for the 16 channels of matrices M5 and M6 in tes
42. VL sequence In previous warm tests I needed to devote a section to this sequence as it was invisible on the pixel signal and I had to use the mean signal per matrix This is no longer needed here in principle as the sequence is fully visible on figure 15 It is even more so in the mean signal that is so clean that it looks artificial Noise properties In previous tests I have used the VH VL sequence to comment on the noise properties of each matrix and correlation coefficients between the different matrices As this still has some interest I shall explore this now Rather than using the mean of the full matrix which is too clean a signal it shall use the mean of the central 4x4 pixels area Because the changes of VH and VL create such drastic changes of the signal level a fourier analysis of the noise has to be restricted to a constant VH VL set I have chosen the section where VH is 0 V and VL is 0 15V because it is far away from the start of the sequence while still creating a polarization on the bolometer bridge This is important because the start of the sequence creates a strong transient effect that is seen on all arrays Finally when doing this analysis one has to remember that the gain is set to high at the start of the sequence while it was low for the rest of the test This means with BOLC FM a gain of 5 uV ADU The noise power spectrum is very clean as figure 16 shows We have no spurious spike in the spectrum However we see tw
43. Volt 0 P 00 05 0000 Set all groups bol bias 19 VGL BU to 3 00000000 Volt 2455 P 00 13 0997 Set all groups bol bias 18 VDL BU to 4 20000000 Volt 3436 P 00 12 OD6C Set all groups bol bias 17 VSS BU to 1 00000000 Volt 819 P 00 11 0333 debut test du BU Set seq mode Sref only P 09 01 00 O1 Set data mode Bolo HK P 09 02 00 O1 Set all groups bol bias 15 VGG to 0 00000000 Volt 1 P 00 OF 0001 Set gain low P 08 00 00 01 Set all groups bol bias 20 VH BLIND to 1 20000000 Volt 982 P 00 14 03D6 Set all groups bol bias 16 VDD to 1 20000000 Volt 981 P 00 10 03D5 vout autour de 30000 pas codeur Valider enregistrement TM S 09 Attendre 5000 ms S 01 001388 Set all groups bol bias 16 VDD to 1 22000000 Volt 998 P 00 10 03E6 Attendre 5000 ms S 01 001388 Set all groups bol bias 16 VDD to 1 24000000 Volt 1014 P 00 10 03F6 Attendre 5000 ms S 01 001388 Set all groups bol bias 16 VDD to 1 26000000 Volt 1030 P 00 10 0406 Attendre 5000 ms S 01 001388 Set all groups bol bias 20 VH_BLIND to 1 22000000 Volt 998 P 00 14 03E6 Attendre 5000 ms S 01 001388 Set all groups bol bias 20 VH_BLIND to 1 24000000 Volt 1014 P 00 14 03F6 Attendre 5000 ms S 01 001388 Set all groups bol bias 20 VH_BLIND to 1 26000000 Volt 1031 P 00 14 0407 EEE EE EE EE EE ee ET EE EE EE A E ARA RA RA RA A E EE EE EE EE PACS Document SAp PACS MS 0652 06 Date 01 1
44. a a z La 2x10 em 1104 o OO dy pp aai 0 1000 2000 3000 4000 5000 Sampling counter Figure 3 Continued Signal timeline for the 16 channels of matrix M1 in test 3 This is a red matrix 6x10 Raw pixel signal 2x10 1x10 TT ee Channel signal evolution matrix M 5 20060717 test3 PE MM MM MM o 1000 2000 3000 4000 5000 Sampling counter Figure 3 Continued Signal timeline for the 16 channels of matrix M5 in test 3 This is a red matrix P A OS Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 17 The VH VL sequence When the instrument is warm it is extremely hard to see the effect of the VH VL sequence see appendix A when the full dynamical range is used It is in fact also hard to clearly identify on a pixel level VH and VL are used to adjust the value of the mid point voltage but when the system is warm this is very inefficient Thus I have instead build the average signal per matrix and this is what is displayed on figure 4 with the value of VH and VL superimposed Zoom on VH VL sequence M1 test1_20060717 Zoom on VH VL sequence M2 test1_20060717 T T T T i T T 7995 8845 7990 8840 Arbitrary scale Arbitrary scale 7985 8835 7980 8830 2600 2800 3000 3200 3400 2600 2800 3000 3200 3400 Frame counter Frame counter Figure 4
45. al tests Page 59 EET NE A gE ee e e eee SIA A A E RES P PPPRP EET EE E A A A A E S 01 001388 Set all groups bol bias 20 VH_BLIND to 0 74000000 Volt 608 P 00 14 0260 Attendre 5000 ms S 01 001388 Set all groups bol bias 20 VH_BLIND to 0 76000000 Volt 624 P 00 14 0270 Attendre 5000 ms S 01 001388 fin de test du BU blocage PEL commut Set all groups bol bias P 00 OF 0614 Set all groups bol bias P 00 10 084E Set all groups bol bias P 00 08 0D78 Set all groups bol bias P 00 06 OD77 Set all groups bol bias P 00 07 0324 08 06 07 VGG to 1 90000000 Volt 1556 VDD to 2 60000000 Volt 2126 VGL to 3 00000000 Volt 3448 VDL to 3 00000000 Volt 3447 VSS to 0 70000000 Volt 804 Set all groups bol bias P 00 OB 0000 Set all groups bol bias P 00 OC 0000 Set all groups bol bias P 00 09 06BC Set all groups bol bias P 00 OA 06BC Set all groups bol bias P 00 OE 0D77 Set all groups bol bias P 00 OD 0000 Set all groups bol bias P 00 OF 05C2 09 10 14 13 15 VDECX H to 0 00000000 Volt 0 VDECX L to 0 00000000 Volt 0 CKRLH to 1 50000000 Volt 1724 CKRLL to 1 50000000 Volt 1724 VSMS L to 3 00000000 Volt 3447 VSMS H to 0 00000000 Volt 0 VGG to 1 80000000 Volt 1474 Set all groups bol bias P 00 14 042B Attendre 5000 ms S 01 001388 Set all groups bol bias P 00 03 0159 Attendre 5000 ms S 01 001388 Set all groups b
46. ally I have written earlier that the noise due to the 10 Hz component appears very correlated This can be quantified As an example I have used the second part of the VH VL sequence Working on the mean signal per matrix I find that the correlation coefficient between M2 and M1 is 0 97 that between M6 and M5 is 0 99 and that between M6 and M1 is 0 96 Given the hypothesized origin for the 10 Hz component it is not surprising to find some correlation especially on the mean signal per matrix However these correlation levels indicate that most of the noise we observe is due to the 10 Hz component I have made a small check on the pixel level signal using pixel 8 8 in the same sequence The correlation coefficients for the pixel signals are 0 93 between M2 and M1 0 98 between M6 and M5 and 0 97 between M6 and M1 Thus even on the pixel level most of the noise is due to that 10 Hz component 4 3 Conclusions The objective of this test is to check that all electrical lines to the bolometers are functionning This objective is fully reached here However it is worth mentioning that e One has to check that the commanding conversion tables correspond to the version of the BOLC P A CS Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 35 hardware used e The relative level of the signal between the Sref_only and the Sb_only mode is different from that observed in SAp and th
47. and M2 Note that each timeline is artificially offset from the previous one for clarity The first 4 downward steps correspond to the VDD sequence The following 3 upward steps correspond to the VH_BLIND scale They are followed by another upward step corresponding to the setting of VH_BLIND before the VRL scale This VRL scale is the symetric 5 steps sequence Then we have a long plateau that corresponds to the VH VL sequence invisible with this dynamical range but see figure 13 followed by the sequence of readout modes Sbolo_only Sref only and Sbolo Sref For this last sequence we observe now a behavior similar to the reference sequence of RD1 the signal in Sbolo_only is lower that in Sref_only See the text for a discussion of this return to the normal situation The data plotted here come from the first run of the test Channel signal evolution matrix M 1 20061011_test1 Channel signal evolution matrix M 2 20061011_test1 8x10 TATT PA Tot ot T T TT 7 8310 oot ri w Row pixel signal L Row pixel signal Each of the channels displayed on figure 12 except one shows a similar pattern which is a first good sign This pattern is this time identical to that of RD1 the first 4 downward steps correspond to the VDD sequence The following 3 upward steps correspond to the VH_BLIND scale They
48. are config uration one expects test 1 to have data at the location of M1 and M5 since we are looking at the red groups and tests 2 and 3 to have data at the location of M1 M2 M5 M6 This is indeed the case and so this is a relief The complete sequence the signal analysis is rather straightforward Since the signal is com pletely generated by the bias commands i e none of it is due to the sensitive part of the bolometer we only need to check one pixel per readout channel i e 16 pixels per matrix Because PIRE is a homemade IDL package it is not completely straightforward to know which index of the arrays corresponds to the channel To circumvent this I have extracted the signal on pixels i i with i from 0 to 15 I plot the timeline of each of these pixels and compare it to the reference one shown in RD1 All these timelines 160 in total are shown in figure 8 Figure 8 Signal timeline for the 16 channels of matrix M1 in test 1 This is one of the red matrix Note that each timeline is artificially offset from the previous one for clarity The first 4 downward steps correspond to the VDD sequence The following 3 upward steps correspond to the VH_BLIND scale They are followed by another upward step corresponding to the setting of VH_BLIND before the VRL scale This VRL scale is the symetric 5 steps sequence Then we have a long plateau that corresponds to the VH VL sequence invisible with this dynamical range but see figure
49. ases 37 12 11 10 2006 Bias commanding checks for the second bias set 39 13 11 10 2006 Bias commanding checks for the last bias set 39 14 31 10 2006 GND BU during the testa va se Spee a ar A 48 15 31 10 2006 Bias commanding checks for the most important biases 49 16 31 10 2006 Bias commanding checks for the second bias set 51 17 31 10 2006 Bias commanding checks for the last bias set 52 P A CS Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 6 1 Reference Documents RD1 SAp PACS MS 0616 06 PACS FM Photometer Focal Plane Unit User s Manual RD2 SAp PACS MS 0247 04 PACS CQM Photometer Focal Plane Unit User s Manual 2 Purpose of this document As a series of roughly identical functional tests of the PhFPU will be performed through the course of the FM ILT it is probably a good idea to hold in a single document all the results and analyses of these tests Part or the totality of this document will also find its place in the final PACS FM test report Table 1 identifies the different tests analyzed in this report by their date of occurence For those tests performed with the PACS warm electronics the reference of the telemetry filename is also given The temperature level at which the tests were performed is indicated The status column indicates the
50. ation at Keyser Threde The instrument was warm and though the PhFPU was in its FM version a QM version of the warm electronics was used namely BOLC QM1 This BOLC version commands only two groups see RD1 for a description of the PhFPU warm electronics Because of this the test script is repeated three times and each time the harnesses between BOLC and the PhFPU are moved so that all 6 groups of the PhFPU can be explored Because of BOLC QM1 the test script used is different from that described in RD1 The VH BLIND levels were lower to avoid saturation on BOLC QM1 and no group per group VRL scales P A CS Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 7 were performed as this made little sense when only two groups could be tested at a time Finally one has to remember that because BOLC QM1 can only control two groups one has to change the wiring between BOLC and the PhFPU to change the groups that are switched on but as far as BOLC is concerned no change has occured Therefore in all the tests we always see the same two groups switched on 1 and 3 and we always get data at the same location of the telemetry The actual execution of the test that day was the following e The first test has BOLC connected to matrices M1 M2 M3 and M4 i e the actual groups 1 and 2 This is test 1 of this section Telemetry files for this test have a date between 19 33 56 and 19 35
51. bias evolution group B_1 test1_200610 Volts Volts 3000 4000 5000 0 1000 2000 3000 4000 5000 Sampling counter 0 1000 2000 Sampling counter Figure 11 Left panel time evolution for the 5 most important biases generating the test signal Right panel time evolution for the second set of biases Units on this figure are Volts Note that to produce a clean figure I have median filtered the HK values On the figure I give the actual HK values uncorrected for the GND BU level This figure shows only group 1 data as all the other figures are identical PACS Herschel Document Date Version FM ILT PhFPU functional tests Table 12 Bias commanding checks for the second bias set of the test This is a time ordered table though the timing of the commands is not indicated A checkmark in the status column indicates that the command is correctly executed For the BU biases I indicate both the commanded value and that value corrected by GND BU level in parenthesis Units in the table are Volts Only group 1 is show but all the other groups show the same behavior VGG 11 10 2006 Bias setting values VDECX H VDECX L CKRLL CKRLH Value Group 1 Status 0 0 0 46 1 9 2 36 1 8 2 26 0 0 2 0 0 0 1 5 0 0 2 0 0 0 0 46 2 36 0 00 0 00 1 5 1 50 1 50 2 26 0 0 0 00 0 00 2 00 2 00 0 00 2 0 2 00 Sr SS SS Table 13 Bias
52. commanding checks for the last set for biases checkmark in the status column indicates that the command is correctly executed For the BU biases I indicate both the commanded value and that value corrected by GND BU level in parenthesis Units in the table are Volts This table contains the values observed on group 1 11 10 2006 Bias name Commanded values Group 1 Value Status VGL BU 3 0 3 46 3 47 J VDL BU 4 2 4 66 4 68 VSS BU 1 0 1 46 1 46 V VGL 3 0 3 00 J VDL 3 0 3 00 J VSS 0 7 0 70 y VSMS L 3 0 3 00 y VINJ 3 0 3 00 J SAp PACS MS 0652 06 P A CS Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 40 5 2 2 Signal analysis Compared to previous test occurences it will now be much simpler to identify the signals since now we have everything in its right place i e Ml is at the location of M1 M2 of M2 and so on up to M10 Therefore I no longer need to specify that they are blue or red The complete sequence Since the signal is completely generated by the bias commands I only check pixels i i on each matrix Thus for each matrix I have 16 timelines that I plot on figure 12 each displaced by a small amount from the previous one for clarity s sake The data are displayed as raw values as converting them to volts make little sense at this stage Figure 12 Signal timeline for the 16 channels of matrices M1
53. cted by the GND BU level in parenthesis Units in the table are Volts This table contains the values observed on group 1 but all the other groups show the same behavior except for VDL BU which is only slightly discrepant for group 1 31 10 2006 Bias name Commanded values Group 1 Value Status VGL BU 2 6 3 06 3 06 J VDL BU 4 2 4 66 4 68 VSS BU 1 5 1 96 1 95 J VGL 3 0 3 00 J VDL 3 0 3 00 J vss 1 3 1 30 V VSMS L 3 0 3 00 J VINJ 3 0 3 00 y level at 0 Figure 15 shows the 16 independant channels for each of the 10 matrices Since we are now looking for the first time at a test performed at 4K and given that we have introduced some differences with respect to the test described in RD1 it is worth spending some time describing the signal patterns we observe Before the test begins we have a short period of time with no data followed by a strong peak in the signal that indicates the switch on and biasing of the detectors The first 4 downward steps that follow correspond to the VDD sequence This is known to show a strong relaxation pattern that is very visible at 4 K The following 3 upward steps correspond to the VH_BLIND scale These are always much cleaner as we observe In this CUS like test these two sequences are followed by a complex event corresponding to the preparation of the VRL scale that ends up with the setting of VH_BLIND before the VRL scale This complex event is simply due to t
54. due to the all groups commanding scheme Therefore we conclude that as far as the commanding is concerned the LTU operated 4K level test of 31 10 2006 was perfectly executed Let us now turn to the signal recorded during this test 6 2 2 Signal Analysis Just a reminder for those that jump in this document directly here matrices M1 to M8 are on the blue array and matrices M9 and M10 are on the red array The complete sequence If you have RD1 in mind you remember that the readout circuit is multiplexed therefore as we are only checking the electrical continuity of the system here I only plot one pixel per channel i e pixels 7 7 on each meatrix In the figures each timeline is displaced from the previous one for clarity The signals are plotted in raw values First because it does not really make sense to convert them here and second because it allows for a check that the three readout modes Sbolo_only Sref_only and Sbolo Sref are correctly understood the first two are unsigned 16 bits integers with a 0 V level around 16000 and the last one is in signed 16 bits integers with a 0 V P A CS Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 52 Table 17 Bias commanding checks for the last set of biases A checkmark in the status column indicates that the command is correctly executed For the BU biases I indicate both the commanded value and that value corre
55. e 5 most important biases of the test This is a time ordered table though the timing of the commands is not indicated A checkmark in the status column indicates that the command is correctly executed Units in the table are Volts For the BU biases I indicate both the commanded value and that value corrected by GND BU level in parenthesis This table contains the time sequence for group 1 Identical sequences are observed for the other groups with marginal differences due to the all groups commanding method Note that since this is a 4 K level test the commanded biases are different from those used at 300 K 31 10 2006 Bias setting values Group 1 VH_BLIND VDD VRL VH VL Value Status 1 50 1 50 1 40 1 86 1 86 1 42 1 88 1 88 1 44 1 90 1 80 1 46 1 92 1 82 1 52 1 52 1 54 1 54 1 56 1 56 2 60 3 06 3 06 1 75 1 75 0 20 0 20 0 30 0 30 0 40 0 40 0 50 0 30 0 20 0 20 0 40 0 40 0 20 0 20 1 70 1 70 0 50 0 50 0 15 0 15 0 00 0 00 0 00 0 00 lt SAA Inspection of table 15 reveals that all commands are nominally executed This is true for all 6 groups of BOLC Figure 14 is familiar but nevertheless shows new features that are worth commenting as this test is in fact the reference LTU operated 4 K level test First contrary to the test of October 11 we see the setting of almost all biases This is quite noticeable on the BU biases that now start with a OV value see for
56. e M2 test2 20060824 IT FF een E A Da 8840 T TT iid YA lt 8830 8820 8810 O O O T 8800 AAA 3400 3600 3800 4000 4200 4400 Frame counter Figure 9 Continued The VH VL sequence as observed on the mean signal for matrices M1 and M2 during test 2 These are blue matrices belonging to the same BOLC group The presence of non white noise on the data is clearer now with a strong correlation between matrices of the same group Zoom on VH VL sequence M5 test2_20060824 r q r 7800 T T 7790 y R o e Arbitrary scale 7770 7760 Prr 7750 4 4 1 L 1 1 3400 3600 1 p 4 1 1 P ni P 1 1 1 1 3800 4000 4200 4400 Frame counter Arbitrary scale 6520 6510 6500 6490 6480 6470 6460 FETT TTTTTT A LN UD LDL DELAS BI RALES 6450 Zoom on VH VL sequence M6 test2_20060824 r AE FETE APA FEET 3400 3600 3800 4000 4200 4400 Frame counter Figure 9 Continued The VH VL sequence as observed on the mean signal for matrices M5 and M6 during test 2 These are blue matrices belonging to the same BOLC group The non white noise is indeed correlated between the two active groups of this test PACS Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 33 Arbitrary scale 6205 62
57. e commanded only once during the test I do not show here the biases that are never commended and hence set at OV by the reset all bias command at the start of the test procedure or commanded to be at OV during the script In fact only two biases are in that category VCH and VSMS H I have however checked that they are indeed at 0 V during the whole test The first set of biases contains VH BLIND VDD VRL VH and VL Remember that VDD is a BU bias hence the comparison between the HK and commanded value has to take into account GND BU The time sequence of commands to these biases as well as the corresponding HK values is given in table 3 as well as displayed graphically on figure 1 There is not significant differences in the housekeeping data for this set of biases between the three tests therefore only the data for test 1 are displayed As can be seen from table 3 and figure 1 the test appears to proceed quite correctly Everything is nominal on group 3 while VDD does not reach the commanded values on group 1 sometimes by 40 mV We will come back to that later I have considered that differences between the commanded and HK value of less than 20mV are no cause for alarm The second set of biases contains VGG VDECX H VDECX L CKRLL CKRLH Remember that VGG is a BU bias Table 4 gives the timeline of the commands to these biases as well as the value read in the HK This timeline is graphically displayed on figure 2 This time the discr
58. e delay before the start of the actual test scripts therefore tests 1 2 and 3 differ slightly in their initial sequence 4 2 Analysis Similarly to the 17 07 06 tests these were analyzed with PIRE with the conversion file called Tm_hk_QM1 txt to transform the HK into analog values The analysis performed is rather straightforward first we inspect the biases time sequence to check that the bias commands are correctly executed then we turn to the pixel signal to check that we observe the expected variations 4 2 1 Bias commands execution When performing this part of the analysis remember that for all the BU biases one has to add the value of GND BU to all the commanded values before comparing them to the HK values In the current test GND BU is around 0 46 V see table 6 for the actual median value of GND BU As can be seen from a comparison with table 2 changes of the order of 1mV have occured This is not significant Table 6 The median value of GND BU measured during the three tests The units are Volts 24 08 2006 Test 1 Test 2 Test 3 Group 1 Group 3 Group 1 Group 3 Group 1 Group 3 0 463 0 463 0 463 0 462 0 462 0 462 For the sake of clarity I have grouped the biases in three sets The first set contains the biases that are extensively used in the test either because we regularly need to tune the electronics with them or because we rely on these biases in the test This
59. e is to check the value of GND BU as this value will be present in all the BU HKs and thus we need to add it to all the BU commands to check whether the command matches the HK Remember that this bias is either on or off i e one does not command its value it has to be read in the HK themselves and there is one value per group In the current test GND BU is around 0 46 V see table 2 for the actual median value of GND BU Note that the rms of GND BU during a test is typically around 1 2mV There are a large number of biases that can be set in each group but for the purpose of this test we can restrict our inspection to the biases that see their state change during the test This give a total of 18 biases that I have grouped into 3 sets The first set contains the biases that at some point in I As of 03 10 06 these have now been integrated in the reference version of PIRE P A CS Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 8 Table 2 The value of GND BU measured during the three tests The units are Volts 17 07 2006 Test 1 Test 1 Test 1 Group 1 Group 3 Group 1 Group 3 Group 1 Group 3 0 463 0 462 0 463 0 462 0 463 0 463 the test generate the signal changes the second set contains the biases that need to be modified so that the signal changes are visible and the last set contains similar biases but that ar
60. e origin of this is unknown This relative position is the same as that observed in KT on 17 07 2006 e We pick up an extra noise component with a characteristic frequency of 10 Hz which is likely due to the general lab power supply 50 Hz frequency and the lack of complete shielding between BOLC QM1 and the FM PhFPU This shielding will be complete when BOLC FM is used nevertheless the levels observed at MPE for this component are much higher than those observed in KT e The current procedure is not adequate to quantify the noise on the signal it was not meant to be Since we do observe extra noise sources it would be good to introduce a waiting plateau after the VH VL sequence long enough to accumulate more than 200 samples on the signal P A CS Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 36 5 The warm functional tests of 11 10 2006 5 1 Test Description These tests were performed right at the start of the actual FM ILT They used the FM version of BOLC and the SAp Local Test Unit As such they form the reference against which the tests performed the following day with the PACS warm electronics and complete FM ILT setup can be compared This time we can control the complete instrument i e the 6 groups and the 10 matrices at once Since we have the FM hardware we are almost back to using the test script described in RD1 Almost only as we found out that
61. eassuring Table 14 The median value of GND BU measured during the LTU test The units are Volts Only those values observed during the CUS like test are shown 31 10 2006 Group 1 Group 2 Group 3 Group 4 Group 5 Group 6 0 457 0 458 0 455 0 457 0 459 0 459 Let us now turn to the three sets of housekeepings these three sets are defined as 1 the set of biases that are used to generate the signal 2 the set of biases that need to be commanded so that we see the signal and that are commmanded more than once and 3 the set of biases that are only commanded once The comparison between the commanded values and the HK for the first set is shown as a time ordered sequence in table 15 Note that since this is a 4K level test some of the commanded biases are different from what we have seen before Also worth mentionning are the extra VRL steps These P A CS Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 49 come from the fact that we repeat a part of the VRL scale on a group per group basis rather than on the usual all groups basis The bias sequence of this first set is also shown graphically on figure 14 Here again I will only show in detail what is observed on group 1 No major difference is observed between groups apart from that due to the all groups commanding scheme Table 15 Bias commanding checks for th
62. epancies are minor and affect only VGG a BU bias on group 1 again Interestingly VGG is not incorrect when it is set to 0 V we will come back to that later The final set of biases contains those that are set only once i e VGL BU VDL BU VSS BU VGL VDL VSS VSMS L VINJ For those biases I list in table 5 their commanded values and the measured HK values For this set it is rather straightforward to indentify the BU biases Note that this is not a time ordered table As with the first two sets of biases the differences in the HK between the three tests are not significant Here again there are some discrepant values mostly for group 1 In conclusion of this exploration of the bias commanding during the test we see that almost all of the biases reach their correct commanded value However out of the total of 82 commanded biases we have 5 major discrepancies by 40 mV or more and 7 minor ones by 20 mV or more All these discrepancies occur on group 1 The first explanation that comes to mind is that this is related to the use of the all groups com manding method in the test script As this uses average conversion factors this will necessarily lead to errors at the individual group level However these errors are too small to explain what we see here Another possibility is that something is wrong either in the analog to digital conversion of the com mands or in the digital to analog conversion of the housekeepings Regarding
63. er Figure 9 The VH VL sequence as observed on the mean signal for matrices M1 and M5 during test 1 These are red matrices and contrary to the following figures they do not belong to the same BOLC group The effect of changing the VH VL biases is evident on the mean signal while it is much more difficult to indentify at a pixel level The transitory period that follows any bias change is most evident at the start of the sequence at this scale Also evident is the presence of non white noise on the data expecially on M5 that is correlated between groups this will become clearer on the next panels see text for analysis As with the test of 17 07 2006 the first important point to stress is that the global behavior of the signal during the VH VL sequence is as expected Remember that VH and VL are used to adjust the level of the midpoint voltage which is in fact the bolometer signal when the instrument is cold As explained in RD1 in readout mode Sref_only which is the mode where we read only the bolometer the signal from the bolometer has a minus sign therefore decreasing the midpoint level either through a VH or a a VL decrease will increase the signal level But again as with the test performed on 17 07 2006 we see that we have an extra non white noise component that is correlated between matrices of the same group and groups of the same test I have again performed computations of the power spectra of these mean s
64. functional tests Page 23 Table 7 Bias commanding checks for the 5 most important biases of the test This is a time ordered table though the timing of the commands is not indicated A checkmark in the status column indicates that the command is correctly executed a question mark that some discussion is required and a double question mark that an investigation is required Units in the table are Volts For the BU biases I indicate both the commanded value and that value corrected by GND BU level in parenthesis This table contains the time sequence for test 1 24 08 2006 Bias setting values Group 1 Group 3 VH_BLIND VDD VRL VH VL Value Status Value Status 0 70 0 69 J 0 70 J 1 20 1 66 1 64 1 66 J 1 22 1 68 1 66 1 68 y 1 24 1 70 1 68 y 1 70 J 1 26 1 72 1 70 de 1 72 J 0 72 0 71 J 0 72 J 0 74 0 73 J 0 74 J 0 76 0 75 y 0 76 y 2 60 3 06 3 02 7 3 06 y 1 30 1 29 y 1 30 J 0 30 0 31 J 0 30 J 0 40 0 41 J 0 40 J 0 50 0 51 J 0 50 J 0 40 0 41 J 0 40 J 0 30 0 30 J 0 30 J 1 10 1 09 y 1 10 J 0 10 0 10 J 0 10 y 0 1 0 10 J 0 10 y 0 00 0 00 y 0 00 y 0 00 0 00 J 0 00 J the differences observed between the three tests are extremely small and for clarity I have only shown the information related to test 1 For this second set of bias we observe again that the test appears to proceed quite nominally except for some slightly discrepant values on VGG for group 1
65. he introduction of artificial delays between each bias setting At this location of the test we have 13 biases to set so it takes some time In previous test this event was condensed in a single spike The VRL scale is the symetric 5 steps sequence Then we have a long plateau during which the group per group VRL scale occurs This plateau ends with another complex event that signals the preparation of the VH VL sequence 6 biases to set This 4 steps VH VL sequence follows now clearly visible since we are at 4K and the test ends by the sequence of readout modes Sbolo_only Sref_only and Sbolo Sref Taking into account the differences between the test script used here and that of RD1 we can state that the signal behaves exactly as expected Therefore regarding the signal behavior the functional test of 31 10 06 is a success With respect to the CUS driven test that occured in the meantime it is worth mentioning that we do not see here the very strong drifts that affected matrix 10 at 300 K We have no good explanation for this P A CS Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 53 Figure 15 Signal timeline for the 16 channels of matrices M1 and M2 Note that each timeline is artificially offset from the previous one for clarity The first 4 downward steps correspond to the VDD sequence The following 3 upward steps correspond to the VH_BLIND scale In this CUS
66. hich is likely due to the general lab power supply 50 Hz frequency and the lack of complete shielding between BOLC QM1 and the FM PhFPU This shielding will be complete when BOLC FM is used e The current procedure is not adequate to quantify the noise on the signal it was not meant to be Since we do observe extra noise sources it would be good to introduce a waiting plateau after the VH VL sequence long enough to accumulate more than 200 samples on the signal 4 The warm functional tests of 24 08 2006 4 1 Test description These tests were performed at the end of the integration of the PhFPU at MPE The instrument was obviously warm and since the FM warm electronics of PACS was not complete we have used the SAp Local Test Unit LTU to perform the test Another major difference in the setup is the use of the QM1 version of BOLC rather than the FM version This BOLC model can only control two groups therefore a series of tests had to be performed to explore the behavior of all the groups of the PhFPU see RD1 for a description of the PhFPU warm electronics Also because of BOLC QM1 peculiarities some differences exist between the test script as described in RD1 and the actual test script used here The VH_BLIND levels were lower to avoid saturation on BOLC QM1 and no group per group VRL scales were performed as this made little sense when only two groups could be tested at a time Finally one has to remember that because BOLC QM1 can o
67. ignals to characterize this noise source Each sequence gives rise to 4 spectra as I have to compute one per value of the VH VL pair The exploration of the noise properties indicates again that we are picking up the general 50 Hz modulation from the lab s power supply see figure 10 This is not very surprising as the hardware set up is the same i e we still suffer from the lack of complete shielding between BOLC QM1 and the PhFPU There are two noticeable differences with the tests of 17 07 2006 First the noise level associated with the 10Hz component is significantly higher here I have compared the peak values observed for each sequence of each matrix during each test and found that on 24 08 2006 they were 3 to 8 times larger than on 17 07 2006 As the only identified difference lies in the test If you are lost now it is either a sign that your training in the ways of Saclay s Fuzzy Logic that you have not read RD1 each being terrible for your karma is not complete or PACS Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 32 Zoom on VH VL sequence M1 test2_20060824 _ 1 5 7980 7975 7970 y a a Arbitrary scale 7960 7955 TET ta a a E TT be tb ta 7950 1 1 f 1 1 1 1 f 1 fi 1 1 1 1 f 1 3400 3600 3800 4000 4200 4400 Frame counter Arbitrary scole Zoom on VH VL sequenc
68. instance VDD This is due to three reasons 1 the command to start downlinking telemetry has been moved to the start of the script see Appendix A 2 because of this shift we can see the effect of the delays introduced at the setting of the protection biases and of GND BU and that now form the standard procedure to bias the detectors and 3 artificial delays are introduced after each command to simulate the fact that with the flight commanding equipment we cannot send more than two commands per second Continuing on the VDD example after the telemetry is requested it takes 4s to switch on GND BU and another P A OS Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 50 Key bias evolution group B_1 20061031_test4Key bias evolution group B_1 20061031_te I VH_BLIND VGG 0 0 Volts Volts 0 5 0x10 1 0x10 1 5x10 2 0x10 0 5 0x10 1 0x10 1 5x10 2 0x10 Sampling counter Sampling counter Figure 14 Left panel time evolution for the 5 most important biases generating the test signal Right panel time evolution for the second set of biases Units on this figure are Volts On the figure I give the actual HK values uncorrected for the GND BU level This and the fact that we have now introduced delays between each commands explains why for instance VDD shows an intermediate level around 0 46 V which does not appear in table 15 This only signals the switchi
69. is 0 3This is the absent channel of matrix M8 so this is expected and part of the success criteria for the functional test P A CS Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 14 Channel signal evolution matrix M 6 20060717_test1 T T T T 30 2000 3 Sampling counter Sompling counter Figure 3 Continued Signal timeline for the 16 channels of matrices M5 and M6 in test 1 These are blue matrices invisible when the full dynamical range is used see later followed by the sequence of readout modes Sbolo_only Sref_only and Sbolo Sref It is for this last sequence that we observe a clear difference with the reference sequence of RD1 the signal in the Sbolo_only is higher that in Sref only here while the opposite is true in RD1 The reason for this difference is not straightforward it cannot be the different VH_BLIND levels used for this run as the VH_BLIND differenciation is performed whatever the readout mode see RD1 It should therefore not affect the relative positionning of the Sbolo_only and Sref_only signals It is very unlikely that it is a result of the FM QM1 commanding problem as this affects only group 1 biases yet both group 1 and group 3 signals show the same behavior
70. l signal evolution matrix M 20061011 test1 8241 04 AI TT TT TT TT TT TTTTTTTTT TT 7 Bao oot mr FA mr Se gee E 4 E I 3 a 6x10 I 4 6x10 r Ir J f E 4x10 4 400 L r E 4 E 5 4 3 tt 2x10 2x10 o o ass O A PM FF o 1000 2000 3000 4000 5000 o 1000 2000 3000 4000 5000 Sampling counter Sampling counter Row pixel signal PAC S Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 43 Figure 12 Continued Signal timeline for the 16 channels of matrices M9 and M10 Channel signal evolution matrix M 9 20061011_test1 Channel signal evolution matrix M 10 20061011_test1 8x10 T 4x10 Row pixel signal 2x10 2x10 o 1000 2000 Sompling counter 4000 5000 Sompling counter P A CS Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 44 The VH VL sequence Since the instrument is still warm I again examine here the behavior of the mean signal per matrix during the VH VL sequence Figure 13 shows this sequence for each matrix with the VH and VL sequences superimposed for clarity Zoom on VH VL sequence M1 20061011_test1 Zoom on VH VL sequence M2 20061011_test1 T T T T T T T T 1 464x10 E E 1 591x10 1 463x10
71. like test they are followed by a complex event corresponding to the preparation of the VRL scale that ends up with the setting of VH_BLIND before the VRL scale This VRL scale is the symetric 5 steps sequence Then we have a long plateau during which the group per group VRL scale occurs This plateau ends with a rather complex event that signals the preparation of the VH VL sequence This 4 steps VH VL sequence follows now clearly visible since we are at 4K and the test ends by the sequence of readout modes Sbolo_only Sref_only and Sbolo Sref We observe a behavior similar to the reference sequence of RD1 the signal in Sbolo_only is lower that in Sref_only Channel signal evolution matrix M 1 20061031_test4 Channel signal evolution matrix M 2 20061031_test4 T TA qq AA rr q AAA E Bx10 i T T i Raw pixel signal s 1 s r e 4 1 0x10 1 5x10 2 0x10 Sampling counter Figure 15 Continued Signal timeline for the 16 channels of matrices M3 and M4 Channel signal evolution matrix M 3 20061031_test4 Bx10 q _ Channel signal evolution matrix M 4 20061031 test4 TA _ 5 1 M I y Raw pixel signal tc Pe z 8 Ld t pa E PG z y 5 E ao poo e e 1 0x10 1 5x10 2 0x10 Sampling counter SAp PACS M5 0652 06 Document Date PACS Herschel 01 12 06 4 0 Page 54 Version
72. ndre 1000 ms S 01 0003E8 Reset bias all groups P 00 00 00 00 Set all groups bol bias 22 VDD PROT BU ON 1 P 00 16 0001 Attendre 1000 ms S 01 0003E8 Set all groups bol bias 21 VDD PROT CL ON 1 P 00 15 0001 Attendre 1000 ms S 01 0003E8 Set all groups bol bias 23 GND BU ON 1 P 00 17 0001 Attendre 1000 ms S 01 0003E8 Set all groups bol bias 05 VCH to 0 00000000 Volt 0 P 00 05 0000 Attendre 1000 ms S 01 0003E8 Set all groups bol bias 19 VGL BU to 2 60000000 Volt 2125 P 00 13 084D Attendre 1000 ms S 01 0003E8 Set all groups bol bias 18 VDL BU to 4 20000000 Volt 3435 P 00 12 OD6B Attendre 1000 ms S 01 0003E8 Set all groups bol bias 17 VSS BU to 1 50000000 Volt 1227 P 00 11 04CB Attendre 1000 ms S 01 0003E8 debut test du BU Set seq mode Sref only P 09 01 00 01 Attendre 1000 ms S 01 0003E8 Set all groups bol bias 15 VGG to 0 00000000 Volt 1 P 00 OF 0001 Attendre 1000 ms S 01 0003E8 Set gain low P 08 00 00 01 Attendre 1000 ms S 01 0003E8 Set all groups bol bias 20 VH_BLIND to 1 50000000 Volt 1227 P 00 14 04CB Attendre 1000 ms S 01 0003E8 Set all groups bol bias 16 VDD to 1 40050000 Volt 1146 P 00 10 047A Attendre 1000 ms S 01 0003E8 vaut autour de 30000 pas codeur Attendre 10000 ms S 01 002710 Set all groups bol bias 16 VDD to 1 42000000 Volt 1162 P 00 10 048A Attendre 1000 ms PACS Document SAp PACS MS 0652 0
73. ne Matrices 1 and 2 test 3 30 8 24 08 2006 Signal timeline Matrices 5 and 6 test 3 o o o o o 30 9 24 08 2006 VH VL sequence on matrices M1 and M5 test 1 31 9 24 08 2006 VH VL sequence on matrices M1 and M2 test 2 32 9 24 08 2006 VH VL sequence on matrices M5 and M6 test 2 32 9 24 08 2006 VH VL sequence on matrices M1 and M2 test 43 33 9 24 08 2006 VH VL sequence on matrices M5 and M6 test 43 33 10 24 08 2006 Power spectrum of the signal during the VH VL sequence 34 11 11 10 2006 Time evolution for the two most important set of biases 38 12 11 10 2006 Signal timeline Matrices 1 and 2 aoaaa e 40 12 11 10 2006 Signal timeline matrices 3 and 4 41 12 11 10 2006 Signal timeline matrices 5 and 6 42 12 11 10 2006 Signal timeline matrices 7 and 8 42 12 11 10 2006 Signal timeline matrices 9 and 10 43 13 11 10 2006 VH VL sequence on matrices Ml and M2 44 13 11 10 2006 VH VL sequence on matrices M3 and M4 45 13 11 10 2006 VH VL sequence on matrices M5 and M6 46 13 11 10 2006 VH VL sequence on matrices M7 and M8 0 46 D t SAp PACS MS 0652 06 PAC S j 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional
74. ng on of GND BU that occurs some time before the commanding of VDD This figure shows only group 1 data as all the other figures are identical 9s to set VDD to its first commanded value Previously all the action before the setting of GNB BU was happening before the downlinking of the telemetry and all commands from the GND BU setting to the biasing of VDD proceeded almost instantaneously Note that the artificial delays could also be seen with keen eyes and thinner lines as now all of the bias settings occur at different times Second we see a new step on VRL This corresponds to a repetition of a part of the VRL scale on a group per group basis This is why so much time elapses between that repetition on group 1 and the VH VL sequence we need time to repeat the VRL scale on each of the remaining 5 groups successively Finally the values used for the VH VL sequence are different here The second set of bias is explored now In table 16 we give the time line of bias commands while this timeline is also shown graphically on the right panel of figure 14 Inspection of the housekeepings for the second set of biases table 16 reveals again that the com manding is nominal The right panel of figure 14 is also familiar and as the left panel shows the effect of the artificial commanding delays see for instance the evolution of VGG what used to be a very P A CS Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT Ph
75. nly control two groups one has to change the wiring between BOLC and the PhFPU to change the groups that are switched on but as far as BOLC is concerned no change has occured Therefore in all the tests we always see the same two groups switched on 1 and 3 and we always get data at the same location of the telemetry The actual test execution was as follows e A first run was performed with BOLC connected to groups 5 and 6 of the PhFPU i e the two red matrices M9 and M10 In this test stops were placed at crucial points to allow visual checking of the test progress mostly the execution of bias commands This test was performed mostly to validate the test script in the present electronics configuration and is not analyzed here e This first test was repeated without the stops In this section this is going to be test 1 e The wiring between BOLC and the PhFPU was changed so that BOLC was connected to groups 1 and 2 i e matrices M1 M2 M3 and M4 and the test script was repeated In this section this is going to be test 2 PAC S Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 22 e The wiring between BOLC and the PhFPU was changed again so that BOLC was connected to groups 3 and 4 i e matrices M5 M6 M7 and M8 In this section this is going to be test 3 Telemetry recording was started each time after the sequencer loading script but with a variabl
76. o things First we still have a rather strong component at low frequency This is due to the transient effect on the signal Second the noise level is different from what we had at 300 K It is now around 1 2 1075 V Hz 2 This is quite higher than what we observed at 300K But this is not a real problem at 300 K we were not picking up noise from the bolometer bridge but rather from the electronics At 4K the bolometer bridge starts to be reactive as revealed by the fact that we see dead pixels at that temperature Hence we are starting to pick up its contribution to the noise Since we are far from the operating conditions the fact that it is quite high is no cause for alarm Finally on the noise issue matrix 6 is no longer noisier that the other matrices Regarding the correlation between signals observed on different matrices I first measure that all signals are strongly correlated correlation coefficients of 0 65 and greater whatever the pair of matrices used However this is strictly due to the strong transient that affect all matrices If I subtract from each signal its median filtered version using a large median window e g 21 readouts I see that the resulting signals are completely uncorrelated even for two matrices of the same group 6 3 Conclusions Inspection of this test show that is was performed successfully be it from the commanding side or from the signal side We can therefore consider that it constitutes the reference LTU dri
77. ol bias P 00 03 01CB Attendre 5000 ms S 01 001388 Set all groups bol bias P 00 03 023E Attendre 5000 ms S 01 001388 Set all groups bol bias P 00 03 01CB Attendre 5000 ms S 01 001388 03 03 03 03 VH BLIND to 1 30000000 Volt 1067 VRL to 0 30000000 Volt 345 VRL to 0 40000000 Volt 459 VRL to 0 50000000 Volt 574 VRL to 0 40000000 Volt 459 P A CS Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 60 Set all groups bol bias 03 VRL to 0 30000000 Volt 345 P 00 03 0159 Attendre 5000 ms S 01 001388 fin de test de VRL Set all groups bol bias 09 CKRLH to 0 00000000 Volt 0 P 00 09 0000 Set all groups bol bias 10 CKRLL to 0 00000000 Volt 0 P 00 OA 0000 Set all groups bol bias 11 VDECX H to 2 00000000 Volt 2298 P 00 OB O8FA Set all groups bol bias 12 VDECX L to 2 00000000 Volt 2298 P 00 OC O8FA Set all groups bol bias 04 VINJ to 3 00000000 Volt 3447 P 00 04 OD77 Set all groups bol bias 20 VH_BLIND to 1 10000000 Volt 903 P 00 14 0387 Set all groups bol bias 01 VH to 0 10000000 Volt 115 P 00 01 0073 Attendre 5000 ms S 01 001388 Set all groups bol bias 02 VL to 0 10000000 Volt 277 P 00 02 0115 Attendre 5000 ms S 01 001388 Set all groups bol bias 01 VH to 0 00000000 Volt 0 P 00 01 0000 Attendre 5000 ms S 01 001388 Set all groups bol bias 02 VL to 0 000000
78. olt 230 P 06 03 00E6 Attendre 10000 ms S 01 002710 fin du test groupe par groupe Set all groups bol bias 09 CKRLH to 0 00000000 Volt 0 P 00 09 0000 Attendre 1000 ms S 01 0003E8 Set all groups bol bias 10 CKRLL to 0 00000000 Volt 0 P 00 OA 0000 Attendre 1000 ms S 01 0003E8 Set all groups bol bias 11 VDECX H to 2 00000000 Volt 2298 P 00 OB O8FA Attendre 1000 ms S 01 0003E8 Set all groups bol bias 12 VDECX L to 2 00000000 Volt 2298 P 00 OC O8FA Attendre 1000 ms S 01 0003E8 Set all groups bol bias 04 VINJ to 3 00000000 Volt 3447 P 00 04 OD77 Attendre 1000 ms S 01 0003E8 Set all groups bol bias 20 VH_BLIND to 1 70000000 Volt 1391 P 00 14 O56F Attendre 1000 ms S 01 0003E8 Set gain high P 08 00 00 00 Attendre 1000 ms S 01 0003E8 Set all groups bol bias 01 VH to 0 50000000 Volt 575 P 00 01 023F Attendre 1000 ms S 01 0003E8 Attendre 10000 ms S 01 002710 Set all groups bol bias 02 VL to 0 15000000 Volt 416 P 00 02 0140 Attendre 1000 ms S 01 0003E8 Attendre 10000 ms S 01 002710 Set all groups bol bias 01 VH to 0 00000000 Volt 0 P 00 01 0000 Attendre 1000 ms S 01 0003E8 Attendre 10000 ms D t PACS De SAp PACS MS 0652 06 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 69 S 01 002710 Set all groups bol bias 02 VL to 0 00000000 Volt 0 P 00 02 0000 Attendre 1000 ms S 01 0003E8 Attendre 10000 ms
79. ositioning is driven by the BOLC model though we have no explanation for this e The extra noise component is gone This was expected since the shielding of the complete FM configuration is much better that the shielding for the QM1 FM configuration used previously As a result the correlation or absence thereof between different signals is much easier to understand P A OS Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 46 Zoom on VH VL sequence M5 20061011_test1 Zoom on VH VL sequence M6 20061011 test1 q A A TF 1 255x10 F 1 421x10 lt 1 254x10 1 420x10 1 253x10 1 419x10 1 252x10 1 418x10 1 251x10 1 1 1 1 1 n 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2600 2800 3000 3200 3400 2600 2800 3000 3200 3400 Frame counter Frame counter Figure 13 Continued The VH VL sequence as observed on the mean signal for matrices M5 and M6 In both tests performed that day the noise level on matrix 6 was significantly higher than that observed on the other matrices Zoom on VH VL sequence M7 20061011_test1 Zoom on VH VL sequence M8 20061011_test1 T T T T T T 1 333x10 1 726x10 lt 1 332x10 1 725x10 1 331x10 1 724x10 1 330x10 1 723x10 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2600 2
80. oup B_1 20060717_test1 Key bias evolution group B_3 20060717_test1 Volts Volts 0 1000 2000 3000 4000 5000 0 1000 2000 3000 4000 5000 Sampling counter Sampling counter P A CS Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 13 3 2 2 Signal analysis We now turn to the data part of the telemetry As explained in the test description we expect to have in test 1 and 2 data at the telemetry location of M1 M2 M5 and M6 while for test 3 we expect to have data only at the telemetry location of M1 and M5 This is indeed the case and thus it is a relief The complete sequence The analysis of the signal is straightforward Since the signal is com pletely generated by the bias commands i e none is due to the sensitive part of the bolometer we only need to check one pixel per readout channel i e 16 pixels per matrix Because PIRE is a homemade IDL package it is not completely straightforward to know which index of the arrays corresponds to the channel To circumvent this I have extracted the signal on pixels i i with i from 0 to 15 I plot the timeline of each of these pixels and compare it to the reference one shown in RD1 All these timelines 160 in total are shown in figure 3 Channel signal evolution matrix M 1 20060717_test1 T T T T 30 2000 3 Sampling counter Sompling co
81. processing status for the test data and OK is the label of successful functional tests Table 1 References for the functional test In the status column I give some indication of the test results OK is for successful tests Commanding implies some inconsistencies between the commanded biases and the housekeepings Signal means the signal behavior is not what we expect and Noise means the noise behavior is not what we expect In all these cases more details can be found in the respective sections of this report Test Date Electronics set up Temperature filename Status 17 07 2006 LTU BOLC QM1 Warm private telemetry format Commanding Signal Noise 24 08 2006 LTU BOLC QM1 Warm private telemetry format Commanding Signal Noise 11 10 2006 LTU BOLC FM Warm private telemetry format OK 31 10 2006 LTU BOLC FM 4K private telemetry format OK The rest of this document is a test by test report on each functional test As the purpose of the test is to first check that the instrument is functional and second to check that it has remained unchanged since the last occurence of the test it is impossible to avoid a certain feeling of d ja vu all over again as our english speaking friends would add for some obscure reason while reading the report 3 The warm functional tests of 17 07 2006 3 1 Test description These tests were performed at the end of the integr
82. ref_only you will need to read the PhFPU user s manual RD1 to understand why this is compatible with the fact the the signal in Sbolo Sref mode is still positive This is welcome However we can only conclude that the relative position of these two modes in terms of signal level is driven by the warm electronics and for this we have no good explanation D t SAp PACS MS 0652 06 PACS Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 42 Row pixel signal Figure 12 Continued Signal timeline for the 16 channels of matrices M5 and M6 Row pixel signal channel signal evolution matrix M 20061011 test1 Channel signal evolution matrix M e 20061011 test1 8241 0 TT TT TT TT TT TT TETT TT TT TT TT TT TT T 8201 0 TT 5 TT TT TT TT TT TT T 1 4 GF pg S a 7 ae E 6x10 h I 4 6x10 T 4 J T 4 A 7 E axiot I Fano 2 4 z 4 2 2x10 E 4 2x10 o 4 o ass AA AO o 1000 2000 3000 4000 5000 o 1000 2000 3000 4000 5000 Sampling counter Sampling counter Figure 12 Continued Signal timeline for the 16 channels of matrices M7 and M8 Channel signal evolution matrix M E 20061011 test1 Channe
83. t 3 These are blue matrices The constant channel on the M6 panel allows the unambiguous identification of this matrix with matrix M8 of the PhFPU see RD1 Channel signal evolution matrix M 6 test3 T T x 3 3x10 2 2 a 2 El El 2x10 13 13 1x10 o PURO pass lisis o 1000 2000 3000 4000 5000 Sampling counter Sompling counter P A OS Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 31 The VH VL sequence When the instrument is warm it is extremely hard to see the effect of the VH VL sequence see appendix A when the full dynamical range is used It is in fact also hard to clearly identify on a pixel level VH and VL are used to adjust the value of the mid point voltage but when the system is warm this is very inefficient Thus I have instead build the average signal per matrix and this is what is displayed on figure 9 with the value of VH and VL superimposed Zoom on VH VL sequence M1 test1_20060824 Zoom on VH VL sequence M5 test1 20060824 w _ or w an gt 7660 8185F 8180 7650 8175 7640 8170 Arbitrary scale Arbitrary scale 7630 8165 8160 7620 8155 morr p orr ar aT 7610 2600 2800 3000 3200 3400 2600 2800 3000 3200 3400 Frame counter Frame count
84. tests Page 5 13 11 10 2006 VH VL sequence on matrices M9 and M10 47 14 31 10 2006 Time evolution for the two most important set of biases 50 15 31 10 2006 Signal timeline Matrices 1 and 2 53 15 31 10 2006 Signal timeline matrices 3 and 4 53 15 31 10 2006 Signal timeline matrices 5 and 6 54 15 31 10 2006 Signal timeline matrices 7 and 8 54 15 31 10 2006 Signal timeline matrices 9 and 10 o 55 16 31 10 2006 Power spectrum of the signal during the VH VL sequence 57 List of Tables 1 References for the functional tests e e 6 2 17107 2006 GND BU during the tests bal AAA cd Beis 8 3 17 07 2006 Bias commanding checks for the most important biases test 1 9 4 17 07 2006 Bias commanding checks for the second bias set test 1 11 5 17 07 2006 Bias commanding checks for the last bias set test 1 11 6 24 08 2006 GND BU during the tests o e eee eee 22 7 24 08 2006 Bias commanding checks for the most important biases test 1 23 8 24 08 2006 Bias commanding checks for the second bias set test 1 25 9 24 08 2006 Bias commanding checks for the last bias set test 1 26 10 11 10 2006 GND BU during the testa ii e SEE gras 36 11 11 10 2006 Bias commanding checks for the most important bi
85. um is itself not well sampled Therefore I had to separate the VH VL sequence into four parts corresponding to the 4 different settings of the VH VL pair For each sequence I have performed of power spectrum calculation Figure 5 shows on of these power spectra here obtained from the second part of the VH VL sequence for VH VL at 0 1 0 1 on matrix M2 during test 1 This figure nicely shows a rather typical situation for test 1 the power spectra show a rather strong component at 10 Hz For some sequences the peak is smaller and wider but this is likely due to the small number of samples 200 used to compute the power spectrum This is a situation that has occured before at SAp and corresponds to the pick up of the general 50 Hz frequency of the lab s power supply though various channels that are not or sometimes cannot be completely shielded This origin can help understand why this noise component is quite correlated Those who are lost now should check back into RD1 P A OS Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 18 Zoom on VH VL sequence M5 test1_20060717 Zoom on VH VL sequence M6 test1_20060717 m 7 7 7 2 Tr 7815 7810 6520 6515 7805 OL VL Arbitrary scale Arbitrory scale 7800 6510 rrer 6505 7795 TT 6500 ATTTT 7790 tr 2600 2800 3000 3200 3400 2600 28
86. unter Figure 3 Signal timeline for the 16 channels of matrices M1 and M2 in test 1 These are two blue matrices Note that each timeline is artificially offset from the previous one for clarity The first 4 downward steps correspond to the VDD sequence The following 3 upward steps correspond to the VH_BLIND scale They are followed by another upward step corresponding to the setting of VH_BLIND before the VRL scale This VRL scale is the symetric 5 steps sequence Then we have a long plateau that corresponds to the VH VL sequence invisible with this dynamical range but see figure 4 followed by the sequence of readout modes Sbolo_only Sref_only and Sbolo Sref For this last sequence we observe a clear difference with the reference sequence of RD1 the signal in the Sbolo_only is higher that in Sref_only here while the opposite is true in RD1 See the text for a discussion of this discrepancy Each of the channels displayed on figure 3 except one shows a similar pattern which is a first good sign This pattern is almost identical to that of RD1 the first 4 downward steps correspond to the VDD sequence The following 3 upward steps correspond to the VH_BLIND scale They are followed by another upward step corresponding to the setting of VH_BLIND before the VRL scale This VRL scale is the symetric 5 steps sequence Then we have a long plateau that corresponds to the VH VL sequence Remember also that PIRE is in IDL and thus the index of M1
87. ven 4K level test and that the LTU can now be safely put back in its box The explanation for the strong drifts observed with the CUS driven 300 K level test has however not been found D t SAp PACS MS 0652 06 PAC S ep j 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 57 Power spectrum M6 test4 20061031 T T T T T T T T T 1077 T T T T T 10 NAAA 107 Volt Hz 05 107 107 0 5 10 15 20 frequency Hz Figure 16 The power spectrum of the mean signal on the central 4x4 pixels area of matrix M6 observed during the third part of the VH VL sequence VH 0 0 VL 0 15 V of test 4 We see a strong low frequency component corresponding to the long term transient seen in the signal Otherwise the noise level is quite homogenous although higher than observed during the 300K tests P A CS Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 58 A The test scripts Only the test part of the actual executed script is shown i e switch on and switch off sequences are ommitted A 1 Test script of 17 07 2006 Note that the conversion tables use to go from analog to digital both the decimal and the hexadecimal codes are the FM tables This is a mistake for the test in question that used BOLC QM1 Reset bias all groups P 00 00 00 00 Set seq mode Sref only P 09 01 00 O1 Set data mode Bolo amp HK P 09 02
88. ween signals of matrices belonging to different group is compatible with the absence of correlation Therefore at this stage we can probably say that in a sequence where the signal is purely generated by the electronics there is a strong correlation between signal observed on circuits that have a large number of components in common such as within a group but this correlation disappears when one compares signals generated through different circuits 5 3 Conclusions For this test all the red lights recorded previously have now turned to green though we can provide little explanations for some of these favorable changes e Now that the conversion tables are compatible with the test equipment we find that the agree ment between the commanding and the HK values is nominal e The relative level of the signal between the Sref_only and the Sb_only modes is back to what P A OS Document SAp PACS MS 0652 06 Date 01 12 06 Herschel Version 4 0 FM ILT PhFPU functional tests Page 45 Zoom on VH VL sequence M3 20061011_test1 T T T T 1 273x10 1 461x10 1 272x10 F 1 460x10 1 271x10 E 1 459x10 F 1 270x10 F 1 458x10 1 269x10 2600 2800 3000 3200 3400 2600 2800 3000 3200 3400 Frome counter Frome counter Figure 13 Continued The VH VL sequence as observed on the mean signal for matrices M3 and M4 we observed in SAp This can only mean that this relative p

Report on the FM ILT PhFPU functional tests

Contents

Download Pdf Manuals

Related Search

Related Contents