Home

Analysis of Petal Longterm test data for the CMS

Contents

1. 56 T outside T_petal_frame set_cooling plant 107 ho ja 5 0 a 10 20 c lt ___ gt lt ___ gt SECLTFIRST SECLTCOLD SECLTLAST 30 T T T T T T T 0 5 10 15 20 25 30 35 40 Time hours Figure 4 8 Temperature profile of a long term test with three cold phases Functionality check OptoScanRun TimeTuneRun PedRun PeakInvOn Minimum CalProRun PeakInvOn PedRun PeakInvOff set of tests PedRun DecInvOn PedRun DecInvOff SaveRec Temperature HV Status WARM 17C 400V SECLTFIRST N 3 COLD 25C ne WARM 17C aa COLD 25C ee WARM 17C hae COLD 25C yyy SECLTCOLD N Y fa ww D u D N A WARM H7C joy SECLTLAST a N Figure 4 9 Structure of a long term test scenario 48 40 Chapter 4 Long term test 4 7 Test procedure With these tests more general sensor problems can be found In the following sections opto scan run timing run and the extended I V run will be explained Pedestal run and calibration profile run are already explained in Chap 3 A comparison of pedestal test with and without HV allows to determine if the HV reaches the sensors 4 7 1 Timing run The complete readout is time critical The timing is different for each module because the optical and electrical paths are different for each module So an optimizati
2. sjauueyd 088857 3 5 4 4 5 5 CMSnoise in dec inv off channels u o as So P 0 12 flagged by LT sjauueyd 088857 0 0 5 1 1 5 Li fe eee Fi 2 25 3 listen 3 5 4 4 5 5 CMSnoise in dec inv off Figure D 6 Noise distributions taken in deconvolution mode The noise is normalised to the average module wise First row is first warm phase second row in cold phase and third row again in warm phase APV edge strips are excluded Only petals tested at CERN are taken into account 89 Appendix D Noise distribution excluding APV edge strips excluding APV edge strips 5313 modules with 25012 apv s on 243 petals 5313 modules with 25012 apv s on 243 petals E 0 11 flagged by LT 9 9 eb lt 10 lt 10 c c 5 10 0 12 flagged by LT 10 10 SISUUEU9 096V0LE SISUUEU9 O96VOLE 10 10 10 10 10 19 1 1 0 0 5 1 1 5 2 2 5 3 3 5 4 4 5 5 0 0 5 1 1 5 2 2 5 3 3 5 4 4 5 5 CMSnoise in peak inv off CMSnoise in peak inv on excluding APV edge strips excluding APV edge strips 5313 modules with 25012 apv s on 243 petals A 5313 modules with 25012 apv s on 243 petals u 210 210 c c Fo g Os OL 5 10 AR 10 o E 0 11 flagged by LT 2 0 11 flagged by LT 2 10 3 10 2 O O a gt E 3 10 10 10 102 10 10 1 1 0 0 5 1 1 5 2 2 5 3
3. 1 1 1 1 1 1 1 1 1 1 1 1 1 50 100 150 200 250 300 time ns Figure 3 8 Calibration pulse of a saturated channel Peak height is very low For comparison see Fig 3 2 3 3 3 Short This defect type occurs when two channels are connected Fig 3 9 a The reason could be again a scratch on the sensor surface or some sort of pollution The consequence is a lower peak height as the capacitance at the amplifier is higher Fig 3 9 b If exactly two strips are connected the capacitance is doubled and therefore the height is halved Eq 3 8 otherwise it is respectively lower 3 3 4 Noisy channel This is one of the harmless problems The channel behaves normal just the noise of the strip is slightly higher It is a bit more difficult to distinguish between signal and noise because the signal to noise ratio is reduced The reasons for a noisy strip are various and were never investigated in detail For APV edge strips it is known that the increased noise can be reduced by a optimzed grounding scheme A more detailed explanation of these defect types and sources and of other defect classes can be found in 23 33 3 3 5 ARC test procedure and defect classification The minimum set of tests performed by ARC to find strip defects are the noise measurements in all four APV modes the calibration profile runs in all four APV modes and some other 29 3 3 Defect types Chapter 3 Single module test height ADC count
4. 23 a Charges created by an ionizing particle drift to their respective electrode and induce a signal that indicates the passage of a particle b Charges drifting to neighboring electrodes induce signals and the signal height is a measure for the amount of charge drifting to the respective electrode By weighting the signals the spatial resolution can be improved to values below the size width of the segments c High energetic primary electrons can create charges and thus signals in regions far away from the particle s track hereby the spatial resolution gets deteriorated High energetic primary electrons are responsible for the long tail of the Landau Distribution 17 Schematic design of one corner of a silicon strip sensor 18 Block diagram of one channel of an APV25 readout chip 85 19 Result of the deconvolution mode if one or two peaks are convoluted a Ideal CR RC function sampled in 3 125 ns intervals is used as input for the deconvolution algorithm 30 b Two peak mode pulses separated by two clock Gyles POl s eses a HR RWI HH ne E ru a OO a we ee 20 Typical APV data fame BI ck SA eh ee eRe OH ER EHH REESE 21 Photograph of an AOH a LL De EP ae EM Ree oe 22 Photograph of a fiber mapping used during the long term test A lot of fibers coming from the AOHs can be seen These are connected to six ribbons which have twelve input slots each The ribbon number used in append
5. Chapter 3 Single module test The single module test setup ARC is a specialized setup to test modules and to determine their faults with high reliability This implies not only to find general problems with a module but also to find and identify defects of single strips In the following two test procedures of the ARC setup and some defect types of the module will be introduced Those will also be used later during the LT test Detailed information about the ARC setup and the tests can be found in 23 33 3 1 Pedestal test During the pedestal test sometimes also called noise test raw data DR is measured strip wise The index ch is the channel under investigation and n the event Repetitive measure ments allow to extract the mean value P the pedestal Eq 3 1 and the RMS RM Sen the noise Eq 3 3 of the raw data of every strip Fig 3 1 The total amount of measurements is N 5000 in both test setups ARC and LT test Furthermore it is possible to extract the common mode CM which is a common shift of the signal height of all strips of one APV in event n Each APV is divided into 4 subgroups of 32 channels and the UM group 15 calculated per group Eq 3 2 Reasons for this shift could be variations in the power supply voltage which affects all strips and pick up noise To extract the RMS of every strip the common mode is subtracted In the following the equations to calculate Pen CMn group and RM Sen are listed N 1 Bm g
6. 10 15 peak time ns oO Q lt g ats ro e se x e 5 8 5 10 15 20 L L L L L L L L L L l l L l L L L L L l L 15 10 5 0 5 15 peak time ns c noisy channel peak height ADC peak height ADC Chapter 6 Analysis of LT measurements 15 10 5 0 5 10 15 peak time ns b S S open AS 1 1 1 1 L L i L l 1 1 1 1 1 1 L L 1 L 1 1 1 1 15 10 5 0 5 10 15 peak time ns d short peak height ADC 5 10 peak time ns 15 e saturated channel BE PAS open BE SSopen noisy channel WE saturated channel M short Figure 6 9 Scaled peak height versus peak time Plots are separated into the different defect types of ARC Only channels identified as defective by ARC and LT are taken into account 96 Chapter 6 Analysis of LT measurements 6 2 Calibration pulse test defect declaration The upper and lower bounds of PA S open S S open and noisy are chosen to keep areas small and hence achieve a high purity The same argument is true for the left border of PA S open and the right of noisy The position of each inner line separating two areas of defect classes is chosen based on the purity p Eq 6 6 of the corresponding samples Fig 6 10 defla denotes the amount of strips in area A1 with defect 1 i e assigned correctly and Su
7. 2 PedRun i2cdec _pd 2 I2cDump dummy dummy 2 SaveRec 1 no 2 PedRun i2cdecinv _pdi 2 I2cDump dummy dummy 2 CheckEnv dummy _d 60 I2cDump dummy dummy 2 SaveRec 1 no 2 ChangeHV 400 _hv400 2 I2cDump dummy dummy 2 PiaReset pllinit dummy 2 CheckEnv dummy _d 60 OptoScanRun i2cpeak desclbl fullscan 2 TimeTuneRun i2cpeakinv ts 6 PedRun i2cpeakinv _ppi 2 I2cDump dummy dummy 2 CalProfRun i2cpeakinv _cfpi 2 I2cDump dummy dummy 2 CheckEnv dummy _d 60 I2cDump dummy dummy 2 SaveRec 3 SECLTFIRST 19 Appendix C Scenario File 2 PedRun i2cpeak _pp 2 I2cDump dummy dummy 2 I2cDump dummy dummy 2 SaveRec 14 SECLTCOLD 2 SaveRec 3 SECLTFIRST 2 PedRun i2cdec _pd 2 PedRun i2cdec _pd 2 I2cDump dummy dummy 2 12c Dump dummy dummy 2 SaveRec 14 SECLTCOLD 2 SaveRec 3 SECLTFIRST 2 PedRun i2cdecinv _pdi 2 PedRun i2cdecinv _pdi 2 I2cDump dummy dummy 2 I2cDump dummy dummy 2 SaveRec 14 SECLTCOLD 2 SaveRec 3 SECLTFIRST 2 CheckEnv dummy dummy 2 CheckEnv dummy _d 60 T2cDump dummy dummy 60 I2cDump dummy dummy 2 SaveRec 14 SECLTCOLD 2 SaveRec 3 SECLTFIRST 2 ChangeHV 0 _hv0 2 ChangeHV 0 _hv0 2 PiaReset pllinit dummy 2 ExtIvRun i2cdecinv iv 2 I2cDump dummy dummy 20 HardReset pllinit recover_iv 30 ChangeCool 16 _t16 2 I2cDump dummy dummy 30 ChangeCool 16 _t16 2 PiaReset pllinit dummy 600 TempReached 16 tr 2 ChangeHV 0 _hv0 3600 I2cDump dummy dummy 2 I2cDump dummy dummy 2 CheckEnv dummy _d 2
8. 6 18 This plot shows the amount of strips flagged by both test systems upper dashed line and how many of these have the same classification lower dashed line In Fig 6 18 a it can be seen that 88 of all strips flagged by the ARC test are also flagged by the LT test and that in this case 76 1 have the same classification The error is determined by varying the width and height of the different areas of each defect type simultaneously by 10 The borders in 6 11 and 6 13 are varied 0 05 counts which is roughly 10 of the width of the peak at a height of 10 As the channels 1 and 2 are flagged by the ARC test disproportionately high Fig 6 18 b shows the same numbers excluding both channels Here 92 of the channels flagged by the ARC test are found by the LT test and 77 1 have the same classification It is worth mentioning that the LT test flagged 204 channels as unknown because there is no information about the height of their tick mark due to a software bug This was discovered so lately that OI 6 3 Comparison between ARC and LT test Chapter 6 Analysis of LT measurements Flags per Apv channel Flags per Apv channel 1281 modules with 6120 apv s on 54 petals 5299 modules with 24940 apv s on 242 petals ARC flagged I ARC flagged ARC flagged as short ARC flagged as short Oo O1 ES A A D defective channels defective channels CS OO TELLLLIFFFF FF FF
9. During the start up luminosity phase the innermost layer can be used at a distance of r amp 44mm After this phase it will be removed due to radiation damage A second 6 Chapter 1 Introduction 1 3 The CMS Experiment To 1 T T T T T T T i T T I T T T T T T T 1 m T T T T T T T T T T T T T T T T T T T OX OF E e u pt 1GeV Je en ae a u pt 10GeV 3 e ERDE as O oe ey e jener bog Wn o m u pt 100GeV DLL eo vO ow T ee m RS nn en oo J O 1 0 Po a ee ee Bea m ee ee En ee RER semer IE ET TIE E T annuel te D DR LES eee TERN E CRE PER ET RA 4 E i EEE a T a a ES 4 ie cre ee es re Audi O ETE eer eee CCE ae LEE ES nn i 4 ei eas IE EBERLE 2 USE De even pee ENGEN 4 i e i i i i LS ND j TI _ 4 or 7 eee 26 A o gt A P A Pe ee Des ns a ee ee a er ETES eaten ETEN ARETE PIETE T eet MS cds aM get Res gested RAE eet caress eee ete SE de i i ee Var ae l l l l l l l l l l l l l 1 9 2 0 a SE SE ES EE E u pt 1GeV pe eee ee ne er g Ll A U pt 10GeV TS NO 10 A cece pt0Gev eg Henne benennen brennen teen Bresse E BE 4 w i Figure 1 5 Resolution of several track parameters for single muons with transverse momenta of 1 10 and 100 GeV transverse momentum left transverse im
10. L_ inner discs TID Figure 2 1 A quarter of the CMS tracker The different subdetectors of the tracker are marked in different colors 16 2 1 TIB Chapter 2 The Silicon Strip Tracker 2 1 TIB The TIB has four cylindrical layers with each of them build of four half shells Fig 2 2 two for the plus and two for the minus side so in total 16 half shells The half shells are located at radii of 255mm 399mm 418mm and 498mm On the first two layers double sided modules are mounted In total TIB has 2724 modules with 1 787 904 readout channels Figure 2 3 View of TIB TID Visible are the silicon modules of the first TIB layer as well as the inner radius of the TID disks at the back 14 10 Chapter 2 The Silicon Strip Tracker 2 2 TID 2 2 TID To each end of TIB one TID is attached Three discs Fig 2 4 with an inner radius of approximately 23cm and outer radius of approximately 51cm are mounted at a distance between 80cm and 90cm from the interaction point On rings one and two double sided modules are mounted In total 816 modules with 565 248 readout channels are assembled Fig 2 3 shows a photograph of TIB and TID Figure 2 4 The innermost ring of a TID disc 34 2 3 TOB TIB and TID are located inside of TOB TOB consists of six cylindrical layers at radii between 600 and 1 080 mm Each layer is composed of substructures called rods Fig 2 5 Rods are the equivalent to the halfshells o
11. Petal Integration for the CMS Tracker End Caps CMS Note 2008 028 Integration of the End Cap TEC of the CMS Silicon Strip Tracker CMS Note to be published Doktorarbeit Analysis of Petal Longterm test data for the CMS Experiment to be published
12. Screen shot of the DAQ An opto scan run in gain2 for a ring 4 module is shown Logical one and zero for both lasers are plotted into one diagram 43 AOH on an ICB Marked are the screw which connects the AOH to the petal body and the connector with which the AOH is plugged to the ICB 48 Noise distribution taken during the cold phase The noise is normalized to the APV average The APV mode is peak inverter on In red channels are marked with a noise which deviates more than 10 from the APV average Those channels are declared as bad and hence flagged by LT All good channels are plotted in green Bl cro csa a NUE ne ah ea dy 50 Noise distribution taken during the cold phase The noise is normalized to the average The APV mode is peak inverter on APV edge strips are excluded In red channels are marked with a noise which deviates more than 10 from the APV average Those channels are declared as bad All good channels are plotted in green 51 cc ee a on aa ERK EERE nah re hr 50 Number of flags after noise tests Colors indicate the ARC test results 51 The first column shows the amount of good strips the second those which are flagged by LT and the third those flagged by ARC The dashed line shows the amount of strips flagged by both test systems 51 52 Number of defective strips per APV channel seen by the ARC and the LT test A defective strip for the LT test implies th
13. The radiation induces impurity in the lattice which results in additional energy levels between the valence and conduction band Therefore the leak increases after several years of operation The time evolution of such impurities and their impact are described by two effects The first effect is called annealing and takes place within some hours up to some weeks and reduces the leakage current and the depletion voltage The second effect is called reverse annealing and takes place on a time scale of month In contrast to annealing it has a negative effect to the leakage current and the depletion voltage Fortunately this second effect could be suppressed if the sensor is kept at low temperatures Therefore the CMS silicon tracker will be operated at temperatures below 0 C the design value for the sensors is 10 C The support frame The sensors of the TECs are glued to a U shaped support frame made of graphite or a com bination of graphite and carbon fiber Fig 2 12 This support frame provides the necessary 18 Chapter 2 The Silicon Strip Tracker 2 4 TEC stability to the sensors Further it carries the readout electronics Graphite and carbon fiber provide the requirements of high stiffness low mass and efficient heat removal from the sen sors as radiation hardness as well In addition they have approximately the same thermal coefficient of thermal expansion as silicon so that the difference could be absorbed by the glue
14. Unfortunately TIB TID TOB and TEC have different defect declaration methods and a different noise level so that the relevance of this comparison is not clear ni Total Number of Channels 3975680 m Total Number of Channels 3851440 3590 3366 l 3122 2939 3000 sed 967 UO SAMPON 0859 2000 Number of Channels Number of Channels 1000 spuueyd 98p9 AdV MONM spej9d 967 UO SANPOWN 0859 0 unflagged unflagged channels FI are channels FI an a with APV edge channels b without APV edge channels Figure 6 18 Number of channels flagged during LT and ARC tests The upper dashed line gives the number of channels flagged in both test systems the lower one the number of same flags 51 Fig 6 19 shows the number of flagged channels split into channels flagged only by the LT test only by the ARC test or by both setups From Fig 6 19 a and Fig 6 19 b it can be derived that around 40 of the channels flagged only by ARC are edge channels In contrast the number of channels flagged only by the LT test are similar in both plots Channels flagged only by LT are channels which are probably affected during assembly of the petals A misclassification of those channels by LT or ARC is possible as well The truth can eventually revealed by an optical inspection of the corresponding channels which is in general not possible anymore 6 4 Comparison between LT and sector test After the LT tes
15. 100 A o nl ie Le 2 80 5 T f E 60 J B E 4 Z l j q 204 ED He 20 J 20e nn 20 gg ae 0 50 100 150 200 0 50 _ 100 150 200 Time ns Time ns a b Figure 2 16 Result of the deconvolution mode if one or two peaks are convoluted a Ideal CR RC function sampled in 3 125 ns intervals is used as input for the deconvolution algorithm 30 b Two peak mode pulses separated by two clock cycles 30 In deconvolution mode dec mode this problem can be reduced by a weighted summation of three subsequent samples of the shaped pulse The weights used by the APV can be calcu lated to wy W2 w3 1 457 1 759 0 534 including a scaling factor so that the pulse height in deconvolution mode is approximately the same as in peak mode 28 In Fig 2 16 the result of such a deconvolution can be seen Unfortunately the deconvolution logic introduces additional noise to the signal so that this mode is foreseen to be used at high event rates only LHC high luminosity phase It is expected to reach an signal to noise ratio S N of 27 33 for thin sensors and 35 42 for thick sensors in peak mode and in deconvolution mode a ratio of 19 22 for thin and 20 24 for thick sensors 14 The third operation mode three sample mode returns also a summation of the same three pipeline cells but without a weighting For this thesis the APV was used in four different modes peak mode inverter on peakinvo
16. 1400 1600 1800 delay ns Figure 4 10 Time tune run for one Laser 12Tracker Phase locked loop Al 4 7 Test procedure Chapter 4 Long term test i Moduleid 0200020013799 BN 11 31 2 redraw pphmmmmmmnmmmqpemmmmmmnmmmmmmnmnmmmmmepenmmmnmsmmmmpaumtsmnmsmnmmmqemnmmnmsmnmmmuenmnmsmmemepess m ef em mp an fon af am ere Rien of tem mp ew ere af an ren El ju exam pan on em a an on pm etes ere ap a eue ap be ere np am garam apa euren ny dl non es ares ag a ai wp lis en af an ares ag an fes my an ra m pong LL ee ee LL Le ewsngewenows tween noneonwose dam wonowenoweshaowomecn owen bac wowoen owe nd sen owonge es sise ue se 0 s LastReports elapsed tim start initialization please wait IT mM O o hitman a our F Time 30 TimeTune UNIVERSITET Figure 4 11 Screen shot of DAQ On this tab the result of a time tune for a ring 3 module is shown 4 7 2 Opto scan run Every AOH has two or three lasers These lasers can be operated in four gain settings gainO to gain3 Gain0 means minimal gain3 maximal amplification A second tuning parameter is the bias current of the laser which is regulated between 10mA and 40mA This is a fine tuning parameter for the intensity of light The opto scan run varies these parameters for the logical zero and one of each APV In Fig 4 12 and Fig 4 13 the result for one APV in one gain setting is shown The optimum is a maximal differential signal betw
17. 2 2 Os 10 4 95 w 10 w 0 10 flagged by LT 2 0 09 flagged by LT 2 10 8 i 8 10 4 10 2 2 i 10 j 10 10 10 10 1 RES nl SER nu Lau nn typ un Lun nn dun ua ta ta EEE 0 0 5 1 1 5 2 2 5 3 3 5 4 4 5 5 0 0 5 1 1 5 2 CMSnoise in dec inv off CMSnoise in dec inv on excluding APV edge strips excluding APV edge strips 5313 modules with 25012 apv s on 243 petals 5313 modules with 25012 apv s on 243 petals 9 9 210 S 108 c c 5 5 10 10 i 10 S 10 3 O O 2 10 z 10 2 10 10 10 10 1 1 0 05 1 1 5 2 25 3 35 4 45 5 0 05 1 1 5 2 25 3 35 4 45 5 CMSnoise in dec inv off CMSnoise in dec inv on Figure D 8 Noise distributions taken in deconvolution mode The noise is normalised to the average module wise First row is first warm phase second row in cold phase and third row again in warm phase APV edge strips and petals tested at CERN are excluded 87 88 Appendix D Noise distribution Appendix E Number of failed noise tests number of noise flags e 1281 modules with 6120 apv s on 54 petals 10000 9000 8000 channels 7000 6000 SIJOUUEU9 088892 5000 4000 3000 2000 1000 0 0 2 4 6 8 10 12 noise flags MY intact strip WE defective strip noisy strip Figure E 1 Number of flags after noise tests Colors indicate the ARC test results 89 90 Appendix E Number of failed noise tests Appendix F Calibration
18. 3 5 4 4 5 5 0 0 5 1 1 5 2 2 5 CMSnoise in peak inv off CMSnoise in peak inv on excluding APV edge strips excluding APV edge strips 5313 modules with 25012 apv s on 243 petals 5313 modules with 25012 apv s on 243 petals a 9 210 406 c E 10 10 7 10 2 10 3 oO ie 2 10 J 10 a 10 10 10 LL 1 1 0 0 5 1 1 5 2 2 5 3 3 5 4 4 5 5 0 0 5 1 1 5 2 2 5 3 3 5 4 4 5 5 CMSnoise in peak inv off CMSnoise in peak inv on Figure D 7 Noise distributions taken in peak mode The noise is normalised to the average module wise First row is first warm phase second row in cold phase and third row again in warm phase APV edge strips and petals tested at CERN are excluded 86 Appendix D Noise distribution excluding APV edge strips excluding APV edge strips 5313 modules with 25012 apv s on 243 petals 5313 modules with 25012 apv s on 243 petals 0 08 flagged by LT 2 a 10 2 05 c c F O 9 n 10 0 08 flagged by LT 10 SISUUEU9 O96VOLE 10 SISUUEU9 096V0LE 10 10 10 10 10 4 0 0 5 1 1 5 2 2 5 3 3 5 4 4 5 5 0 0 5 1 1 5 2 2 5 3 3 5 4 4 5 5 CMSnoise in dec inv on CMSnoise in dec inv off excluding APV edge strips excluding APV edge strips 5313 modules with 25012 apv s on 243 petals 5313 modules with 25012 apv s on 243 petals Tins 10 10
19. 35 ADC counts Fit based on the CRRC function see Chap 3 and Eq 3 8 6 2 1 Normalization of the discriminating variables In 23 29 it is shown that instead of peak time t and peak height hp the difference to the calibration group median tp hp should be used Eq 6 1 This method removes uncertainties in the signal propagation delay as this is the same for the whole calibration group The error onto the rise time time between the points where the peak has a height of 10 and 90 of its maximum is much larger and therefore not a good quantity to be measured So these At and Ah values will be discussed Atp tp t ARp hp Np 6 1 The measured Ah is a abstract function Eq 6 2 of the quality of the strip Gi and the tick height hiick the logical one see Sec 4 7 1 as each signal measurement scales with the tick height Normalizing Ah to Rick removes this dependance As the height of a tick is in the order of 100 ADC counts the outcome of this division is scaled by this factor Eq 6 3 used is the value used in the later analysis In principal Ah depends also on the sensor geometry but this variation is negligible Ah d s iick Fine fie 6 2 100 ADC Ah he used ne T dario 100 ADC 6 3 This is different for At It scales directly with the strip length of the sensors Eq 6 4 and Eq 2 2 Because the median subtracted value of an intact strip peaks around zero this effect 59 6 2 Cal
20. A quarter of the CMS tracker The different subdetectors of the tracker are marked in different colors 16 A half shell of the third layer of TIB 34 View of TIB TID Visible are the silicon modules of the first TIB layer as well as the inner radius of the TID disks at the back 14 2 2 2 2 The innermost ring of a TID disc 34 A rod ot the TOB 34 o ss eee EERE HEME HEME OH HRG SG Picture of TOB support structure 14 Side and front view of a TEC 14 Photograph of front and back side of a TEC front petal with seven rings 14 Structure of a part of petal body could be seen NOMEX within CFC The honey comb structure was partially destroyed during removal of the CFC skin Photograph of petal body with cooling pipe The cooling pipe can be seen since the outermost carbon skin is not yet glued 0 4 Petal after assembly of ICB CCU and AOH Only three out of 16 AOHs are a RANGE a Exploded view of a ring 6 TEC module b Photograph of a ring 6 TEC module mounted on a carrier plate 14 10 10 11 12 12 13 13 14 14 15 16 95 96 2 13 2 14 2 19 2 16 2 17 2 18 2 19 3 1 Se 3 9 3 4 3 0 3 0 3 1 3 8 3 9 List of Figures List of Figures Principle of the particle detection using a reversely biased diode
21. F Ts Figure 2 9 Structure of a part of petal body could be seen NOMEX within CFC The honey comb structure was partially destroyed during removal of the CFC skin Figure 2 10 Photograph of petal body with cooling pipe The cooling pipe can be seen since the outermost carbon skin is not yet glued 14 Chapter 2 The Silicon Strip Tracker 2 4 TEC AOHs ICB CCUs Figure 2 11 Petal after assembly of ICB CCU and AOH Only three out of 16 AOHs are marked 2 4 2 Silicon strip module The silicon strip module consist out of four main parts the silicon strip sensor the support frame the readout electronics on the so called front end hybrid and a flat Kapton circuit which delivers the bias voltage to the sensor backplanes and insulate the sensor from the frame Fig 2 12 The silicon strip sensor The major component of a module is the silicon sensor 20 The function principle of such a sensor can be described with help of a semiconductor diode Fig 2 13 a The diode is driven with reverse biased conditions Only a very small current the so called lea flows into this direction The width of the depleted region is increased If a charged particle transverses this region it creates electron hole pairs along its track A MIP looses about 260 eV per um while 3 6eV are needed to create an electron hole pair Thus gives about 70 electrons per um The charges of the respective sign move in the direction given by the electric
22. Noise After CM Subtraction a logarithmic scale b linear scale WE defect channel WE S Sopen unknown Figure 6 21 Scaled noise distribution of ring five to seven measured in TEC Categories are defined as in the LT test APV edge channels and bad APVs are excluded 30 31 Ring1234 cmsnoise_peddec_lengthscaled Ring1234 cmsnoise_peddec_lengthscaled 10 10 Ber 10 n n ER 10 a 1 5 Bio 2 2 zZ z 10 10 1 1 0 500 1000 1500 2000 2500 3000 3500 4000 0 500 1000 1500 2000 2500 3000 3500 4000 Noise After CM Subtraction Noise After CM Subtraction a logarithmic scale b linear scale WE defect channel WE S Sopen unknown Figure 6 22 Scaled noise distribution of ring one to four measured in TEC Categories are defined as in the LT test Channels marked in blue are mid sensor open as they are counted to S S open APV edge channels and bad APVs are again excluded 30 31 The distributions in Fig 6 20 are used to define defective channels in the TEC The cut pa 69 6 4 Comparison between LI and sector test Chapter 6 Analysis of LT measurements rameters are listed in Tab 6 3 This results in the distributions Fig 6 23 and Fig 6 24 which are defined in analogy to Sec 6 3 Table 6 3 In this table a definition of defect channels inside the TEC is given Noise was taken in deconvolution mode dec mode Total Number of Channels 727508 Total Number of Channels 1141172
23. The readout electronics The readout electronics on the front end hybrid consists of four or six Analogue Pipeline Voltages AP Vs 35 with 128 readout channels each one multiplexer chip APVMUX 36 one chip for the trigger decoding Tracker Phase Locked Loop TPLL 37 and one chip for the surveillance of environment parameters Detector Control Unit DCU 38 It monitors values like hybrid and sensor temperature supply voltage and currents at the hybrid as also leak The sensors are connected via wire bonds to the pitch adapter which adjusts the pitch of the sensors to the pitch of the readout unit APV 44 um The APV is the main read out component on the front end hybrid For operation it needs two voltages of 1 25 V and 2 5 V as well as a 40 MHz clock the bunch crossing frequency of the LHC The charge signals of the strips get amplified by low noise amplifiers Fig 2 15 The amplifier is followed by a signal MUX gain low noise charge 50 ns CR RC pre amplifier shaper EB 128 1 L MUX differential current inverter 192 analogue output amplifier pipeline cells ne signal C N input u L N gt a DI 8 ae y APSP pe Figure 2 15 Block diagram of
24. areas a and S S open and noisy b respectively between them the scaled noise APV average subtracted of those channels can be used see Fig 6 11 This is possible because saturated channels should have a low noise because they are always saturated and do not show any variation It can be seen that this variable has a discriminating power The same investigation can be done for the CERN setup see Fig 6 12 and Fig 6 13 The data differs from the test data of the other centers even if is also discriminating This deviation is as ununderstood as the deviation in the noise data In Tab 6 2 the resulting cut definitions are listed for each defect class 4scaled to strip length and same tick height mean of all tests in peak mode O7 6 2 Calibration pulse test defect declaration Chapter 6 Analysis of LT measurements ui io A 2 oo same peaknoise eyo 98h wo 2 SN D purity short and saturated A TTIT TTIIT TTT TTT TTT T N BE TEE EE a ER we 19 1 Oo 0 8 0 6 0 4 0 2 0 02 04 0 6 0 8 1 0 4 0 2 0 0 2 0 4 normalized CMSpeaknoise ave sub normalized CMSpeaknoise ave sub a other centers b other centers I saturated channel M short Figure 6 11 a Noise distribution of short and saturated channels The noise is scaled and the average is subtracted b Pur
25. field This induces charge pulses in both electrodes of the diode which can be used to identify the crossing of a particle It is to mention that the diode has to be thick enough so that the pulses are high enough to be detected The principle of the particle detection can be improved if segmented electrodes are used Those work like many different diodes and provides a more detailed information about the position of the transversing particle Fig 2 13 b and c The sensitive region of a diode is equal to the depleted region Therefore it is of interest to 3Minimum Ionizing Particle 15 2 4 TEC Chapter 2 The Silicon Strip Tracker silicon sensors NS Kapton foil near sensor pitch adapter SSS front end SSS _ frame carbon fibre ceramic u g Fi Cross piece graphite a Kapton foil silicon sensors aluminium Carrier plate pitch adapter front end hybrid frame carbon fibre hybrid supply and readout Per connection cross piece 3 graphite HV connector Figure 2 12 a Exploded view of a ring 6 TEC module b Photograph of a ring 6 TEC module mounted on a carrier plate 14 16 Chapter 2 The Silicon Strip Tracker 2 4 TEC measured N N N signal Amplifier Shaper Coupling le 2E le al re Capacitor Zu p strip particle trajectory high energetic n bulk primary electron Sign
26. group subtracted peaktime To show that the calibration group subtracted peaktime scales with the length of the modules mean and RMS of it have to be determined Results are used afterwards in a linear fit Pin 7 6817 nat 207110 Constant 2 88 ne Constant 17 3 E 7 Mean 8 61 E Mean 9 01 E Sigma 1 24 Sigma 6 5 20 O1 sjauueyD gz 15 A N O1 ES A A 24 22 20 18 16 14 12 10 24 22 20 18 16 14 12 6 CGS peak time ns 10 8 6 CGS peak time ns Figure F 1 Calibration group subtracted CGS peaktime for PA S opens in ring 1 left and ring 2 right 91 Appendix F Calibration group subtracted peaktime CGS peaktime on Ring 3 x ndf Constant 18 Mean E E Sigma 16 o E LE 147 12H 10 8 6 4 2 s 12 10 8 6 CGS peak time ns 8 64 7 14 9 10 6 0 804 sjouuey gt 1g CGS peaktime on Ring 4 x ndf 10 7 8 ri Constant 25 9 E Mean 10 4 Es Sigma 0 721 lt oO 25 20 syouueus ZZL N 10 8 6 CGS peak time ns 24 22 20 Figure F 2 Calibration group subtracted peaktime for PA S opens in ring 3 left and ring 4 right CGS peaktime on Ring 5 channels 2 ndf Constant 12 CG Mean Sigma 10 8 6 S peak time ns 15 8 17 21 7 14 6 1 42 sjouuey gt EZ CGS
27. higher noise level than the other channels 14 23 A dedicated investigation shows that a better agreement of these two numbers can be reached by splitting the data into petals tested with the CERN LT setup and petals tested at the other setups Chap 4 Fig 6 2 Obviously the amount of strips outside the 10 region is much higher for petals tested at CERN than for those tested at the other test centers There are several explanations for this the most probable is the fact that CERN used a different DAQ software than the other five PICs and tested up to six petals simultaneously which might introduce additional noise Unfortunately assembly and test of the petals was done under lAPV channels 1 2 127 and 128 49 6 1 Pedestal test defect detection Chapter 6 Analysis of LT measurements high pressure of time This lead to the fact that no investigation of the CERN setup and its high noise was done This course of action was accepted by all decision makers The petals not tested at CERN have roughly the same amount of strips outside the 10 region compared to the ARC test 0 11 by the LT test 0 13 by the ARC test A complete set of all noise distributions is included in App D The further analysis for the defect detection will exclude APV edge strips The plots will be shown separately for petals tested at CERN and petals tested at the other PICs with APV edge strips excluding APV edge strips 6594 modules with 311
28. of this method a channel defect rate of approximately 0 09 can be measured Further defects like dead components became visible after integration of the petals into the TEC and raised this number up to 0 33 defect and non recoverable channels vl Contents Introduction 1 1 The Standard Model of Particle Physics 1 2 The Large Hadron Collider 1 3 The CMS Experiment be ERK EERE sam ga nn 13 1 The Muon system ssa 2 4 aa ad aa a ee Rare 1 3 2 The Hadron Calorimeter 1 3 3 The Electromagnetic Calorimeter la EDS PAGE ne Ee we BD Al 3 1 Hi sasari 24 1 Petal The Silicon Strip Tracker 24 2 ICO strip Module 44666 Lu ows SR a Dow a 2 4 3 AOH Single module test Pedestal test 3 2 Calibration Pr nle Test 2 246684 un a 3 9 4 1 4 2 4 3 4 4 4 9 4 6 4 7 Defect types 3 3 1 Open 3 3 2 Saturated channel e Co oo en 3 3 3 Short Oe Noisy chamnel ss ue ea ww ewe Ewe hee ee ee ee 3 3 5 ARC test procedure and defect classification Long term test General setup Communication Readout 4 3 1 K MUX Cooling Slow control DAQ Test procedure QOQ IT A WO Hr me 11 11 11 13 15 21 23 23 25 29 26 29 29 29 29 33 39 39 34 39 30 37 37 39 vil Content
29. on a ring 3 TEC module For the fit the Eq 3 8 was used The Fitting range is 10 bins around the maximum A N peak time ns TT A O1 o TTIT TTIT peak height ADC counts _ N N ol ol N ITTV TTT O1 400 500 channel number N L l l l l L L L L L L 200 300 400 500 channel number a peak time b peak height Om _t oo Figure 3 3 Distribution of peak time and peak height of the same ring 3 module A periodic pattern of the length of eight can be seen due to the calibration groups Channel 94 again stands out in these plots indicating a single strip defect compare Fig 3 1 the implementation of those defect classes will be given The cuts were found with the help of a statistical analysis of all data 23 3 3 1 Open The first main class are the opens An open is a strip which is not or only partly connected to the readout system Depending on the location and source an open can be either a pitch adapter sensor open a sensor sensor open or a mid sensor open 26 Chapter 3 Single module test 3 3 Defect types Pitch adapter sensor open PA S open The location of this defect is somewhere between the sensor and the APV So the whole strip is disconnected from the readout system In most cases the reason is
30. one channel of an APV25 readout chip 35 inverter unit which can be switched on or off A CR RC shaper produces a pulse with the output voltage AQ t 4 U t XP 2 2 with a peaking time of rT RC 50ns Qe is the collected charge and A an amplification factor determined by the preamplifier The shaped voltage pulses are then continuously sam pled every 25 ns and stored in an analogue pipeline with 192 cells per readout channel The pipeline stores the signals for more than 4s In case of no trigger signal the cell becomes overwritten within 4 8 us The following Analogue Pulse Shape Processor APSP allows to run the APV in three different operation modes and returns data with different peaking times Depending on the operation mode one or three pipeline cells are reserved for read out The modes are peak mode deconvolution mode and three sample mode In peak mode the stored signal is transmitted to the next processing step Two consecutive hits could not be disen tangled since the shaping time of 50 ns is twice as large as the time between two hits and so entails an overlap of the two shaped signals 19 2 4 TEC Chapter 2 The Silicon Strip Tracker 100 1607 _ Peak Mode _ E a Peak Mode B CR RC function 2 1407 a measured pulse 807 cb Deconvolution Mode 5 120 a eS Deconvolution 5 J es Li computed pulse S J n so computed pulse 60 7 i f bS
31. peak time ns e short f saturated channel BE PAS open BE S Sopen noisy channel WE saturated channel short Figure 6 12 Scaled peak height versus peak time for petals tested at CERN 59 6 3 Comparison between ARC and LT test Chapter 6 Analysis of LT measurements same peaknoise sjauueyd zg f 2 1 0 8 0 6 0 4 0 2 0 02 04 06 0 8 1 normalized CMSpeaknoise ave sub CERN I Saturated channel short Figure 6 13 Noise distribution of shorts and saturated channels 6 3 Comparison between ARC and LT test Using the classification scheme that has been presented above it is possible to compare ARC and LT test results without counting noisy channels to the defective channels In Fig 6 14 the number of defect strips per APV channel is shown Fig 6 15 shows the difference between the ARC test and the LT test results As noisy channels are still operational they are excluded For most channels a good agreement of the amount of flagged channels between the ARC and the LT test result is visible for all test centers A strip by strip comparison will be shown later in this section The first two APV channels show an enhanced defect rate in the ARC test results compared to the LT test results The expected shape should be flat as one would expect every channel to show the same rate of defects The LT test results indicate such a flat distribution except for channels 127 and 128 while the ARC tes
32. test results At the end the results of the different test systems will be compared 6 1 Pedestal test defect detection Every defect which can be detected by the LT test has a noticeable signature in the pedestal run of the LT test In the following two subsections it is shown that the analysis of these runs flag roughly the same amount of strips as the ARC test see Chapter 6 1 1 Furthermore it is shown that this test is reproducible see Chapter 6 1 2 This means that most defects are found in every pedestal run 6 1 1 Defect rate To determine the defect rate the distribution of the common mode subtracted noise see Sec 3 1 during one pedestal test is used The noise is normalized to the APV average Strips with a noise which vary more than 10 w r t the mean of the corresponding APV are flagged by the LT test as bad Fig 6 1 The 10 were chosen because they are roughly the same amount of deviation as allowed during the ARC test 54 ARC used absolute cuts per ring The amount of bad strips is used as a reference value and adds up to 1 44 for all 297 petals Fig 6 1 a This amount differs from the amount of 0 14 of all strips that were not accepted by the ARC test by one order of magnitude Excluding APV edge strips improves the ratio between channels flagged by the LT and ARC test to a factor of 3 0 32 by LT Fig 6 1 b and 0 13 by ARC This exclusion is motivated by the fact that these channels have a
33. the ARC and the LT test A defective strip for the LT test implies that the strip failed at least six tests 51 6 2 Calibration pulse test defect declaration After the identification of defective strips it is necessary to classify the kind of defect for each strip Some types of defective strips may still be used afterwards e g shorts and S S opens still deliver some signal while PA S open or saturated channels are dead Also noisy strips are flagged as defective though they are functioning with only an increased noise level They can be excluded from the set of defective strips All information about the strips is stored in a datbase The reconstruction alogrithms are able to use this information and hence the 92 Chapter 6 Analysis of LT measurements 6 2 Calibration pulse test defect declaration nominal reconstruction capability of the CMS detector is improved This classification is possible with the help of the calibration pulse test see Chap 3 and Fig 6 6 Every defect has a unique signature in the peak time versus peak height plane The ARC test results are used again Chap 6 2 2 to assign signatures to the defect classes It is necessary to transform peak time and peak height of different modules to a common standard see next section to be able to compare them height ADC counts O1 200 250 300 time ns Figure 6 6 Example of a pulse with a peak time of 85ns and a height of
34. 0 10 5 5 10 a 10 g 10 10 10 10 1 1 0 0 5 1 1 5 2 2 5 3 3 5 4 4 5 5 0 0 5 1 1 5 2 2 5 3 3 5 4 4 5 5 CMSnoise in peak inv on CMSnoise in peak inv off excluding APV edge strips excluding APV edge strips 6594 modules with 31132 apv s on 297 petals 6594 modules with 31132 apv s on 297 petals a 10 10 G fo E 10 M 10 o 0 32 flagged by LT a E 0 20 flagged by LT a 4 A mS 10 3 10 F 3 5 3 2 3 3 10 a 10 g 10 10 10 10 i baal aa amtaa L ANR IR Re 0 0 5 1 1 5 2 2 5 3 3 5 4 4 5 5 0 0 5 1 1 5 2 2 5 3 3 5 4 4 5 5 CMSnoise in peak inv off CMSnoise in peak inv on 6594 modules with 31132 apv s on 297 petals excluding APV edge strips excluding APV edge strips 6594 modules with 31132 apv s on 297 petals 6 E10 10 f 10 210 o o 0 16 flagged by LT S E 0 17 flagged by LT S A 10 S 3 2 2 10 D a 10 10 1 0 0 5 1 1 5 2 2 5 3 3 5 4 4 5 5 0 0 5 1 1 5 2 2 5 3 3 5 4 4 5 5 CMSnoise in peak inv off CMSnoise in peak inv on Figure D 3 Noise distributions taken in peak mode The noise is normalised to the average module wise First row is first warm phase second row in cold phase and third row again in warm phase APV edge strips are excluded 82 Appendix D Noise distribution excluding APV edge strips excluding APV edge strips 6594 modules with 31132 apv s on 297 peta
35. 25 flagged by LT 8 A A 4 a 4 S 10 o 10 0 10 u 10 a 10 10 10 10 1 1 L L LL I il l LI J 1 0 0 5 1 1 5 2 2 5 3 3 5 4 4 5 5 0 0 5 1 1 5 2 2 5 3 3 5 4 4 5 5 CMSnoise in peak inv off CMSnoise in peak inv on 6594 modules with 31132 apv s on 297 petals with APV edge strips with APV edge strips 6594 modules with 31132 apv s on 297 petals O 1 0 1 0 f 10 210 ao ao 1 09 flagged by LT g E 1 12 flagged by LT 8 10 3 10 3 3 D gt 2 3 2 3 3 10 D 10 2 10 10 10 19 1 1 0 0 5 1 1 5 2 2 5 3 3 5 4 4 5 5 0 0 5 1 1 5 2 2 5 3 3 5 4 4 5 5 CMSnoise in peak inv off CMSnoise in peak inv on Figure D 1 Noise distributions taken in peak mode The noise is normalised to the average module wise First row is first warm phase second row in cold phase and third row again in warm phase All channels are included SO Appendix D Noise distribution with APV edge strips with APV edge strips 6594 modules with 31132 apv s on 297 petals 6594 modules with 31132 apv s on 297 petals 0 43 flagged by LT 0 52 flagged by LT _ _ channels channels s o s o 104 10 SISUUEU9 627886 SISUUEU9 627886 10 10 10 10 10 10 0 0 5 1 1 5 2 2 5 3 3 5 4 4 5 5 0 0 5 1 1 5 2 2 5 3 3 5 4 4 5 5 CMSnoise in dec inv off CMSnoise in dec inv on with APV edg
36. 31 classified as defect by the LT test Noisy channels are not labeled APVs which are known to be defect 30 and APV edge channels are excluded Taking these distributions it can be seen that defective and good strips separate well for the rings 1 to 4 but not for the rings 5 to 7 Therefore the channels of rings 5 to 7 which cannot be separated are investigated in more detail One difference between rings 1 to 4 and 5 to 7 is that the latter contain only two sensor modules and only strips of such modules can have the defect S S open Strips with a S 64 Chapter 6 Analysis of LT measurements 6 4 Comparison between LT and sector test S open have a higher noise than strips with other defects see Chap 3 This explains that channels do not separate Another reason are channels flagged as unknown Their real defect is unknown part of them should be noisy and hence do not separate Due to these facts the defective channels are grouped into three classes unknown S S open and the rest is gathered as defect Fig 6 21 shows the result of this classification Indeed most channels which do not separate are S S opens and unknown channels The same plot can be done for ring one to four see Fig 6 22 60 50 40 30 Number of Strips Number of Strips 20 H l KI 600 800 1000 1200 1400 1600 1800 2000 2200 Goo 800 1000 1200 1400 1600 1800 2000 2200 Noise After CM Subtraction
37. 32 apv s on 297 petals 6594 modules with 31132 apv s on 297 petals 1 44 flagged by LT channels s Oo o channels ES ES a er 0 32 flagged by LT sjauueyJ 078E98E sjauuey9 617886 A 0 0 5 1 1 5 2 2 5 3 3 5 4 4 5 5 0 0 5 1 1 5 2 2 5 3 3 5 4 4 5 5 CMSnoise in peak inv on CMSnoise in peak inv on a with APV edge strips b without APV edge strips Figure 6 1 Noise distribution taken during the cold phase The noise is normalized to the APV average The APV mode is peak inverter on In red channels are marked with a noise which deviates more than 10 from the APV average Those channels are declared as bad and hence flagged by LT All good channels are plotted in green 51 excluding APV edge strips excluding APV edge strips 1281 modules with 6120 apv s on 54 petals 5313 modules with 25012 apv s on 243 petals 1 15 flagged by LT 0 11 flagged by LT channels o channels sjouueyo 088852 sjouueyo 096POLE i il Il I ll J II 1 3 3 5 4 4 5 5 0 0 5 1 1 5 2 2 5 3 3 5 4 4 5 5 CMSnoise in peak inv on 0 0 5 1 1 5 2 2 5 CMSnoise in peak inv on a CERN b other centers Figure 6 2 Noise distribution taken during the cold phase The noise is normalized to the average The APV mode is peak inverter on APV edge strips are excluded In red channels are marked with a
38. Analysis of Petal Longterm test data for the CMS Experiment Von der Fakult t f r Mathematik Informatik und Naturwissenschaften der RWTH Aachen University zur Erlangung des akademischen Grades eines Doktors der Naturwissenschaften genehmigte Dissertation vorgelegt von Diplom Physiker Dirk Heydhausen aus Kempen Berichter Universit tsprofessor Dr rer nat Achim Stahl Universit tsprofessor Dr rer nat Lutz Feld Tag der m ndlichen Pr fung 15 Dezember 2008 Diese Dissertation ist auf den Internetseiten der Hochschulbibliothek online verf gbar il Zusammenfassung Der Start des Large Hadron Collider LHC am Europ ischen Zentrum f r Elementarteilchen physik CERN in Genf ist f r Ende 2008 geplant Eines der Experimente am LHC ist der Vielzweckdetektor CMS Compact Muon Solenoid Ein Hauptbestandteil des CMS Detektors ist das Spursystem Dieses besteht aus dem Silizium Pixeldetektor und dem Silizium Streifen detektor Der Pixeldetektor wird dabei vom Streifendetektor umschlossen Momentan ist das Spursystem mit einer aktiven Fl che von 198 m der gr te Silizium Detektor weltweit Der Streifendetektor wiederum besteht aus vier Subdetektoren Einer davon sind die Tracker endkappen TEC mit einer aktiven Fl che von 82m Neben dieser gro en Fl che gibt die Position im Vorw rtsbereich den Endkappen eine Schl sselrolle f r Physikanalysen da viele interessante Ereignisse in diesem Bereich erwartet werden pp Be
39. Appendix C Scenario File 2 ExtIvRun default iv 20 HardReset pllinit recover_iv 2 PiaReset pllinit dummy 2 SaveRec 17 SECLTLAST 2 ChangeHV 0 _hv0 2 PiaReset pllinit dummy 2 I2cDump dummy dummy 2 CheckEnv dummy _d 60 SaveRec 17 SECLTLAST 2 Stop 0 stop T7 18 Appendix C Scenario File Appendix D Noise distribution On the following pages the common mode subtracted noise distribution of the LT setup in peak and deconvolution mode with inverter on or off is shown The plots are separated into the three phases first warm cold and last warm with and without APV edge channels and with and without petals tested at CERN as well as petals only tested at CERN 19 Appendix D Noise distribution with APV edge strips with APV edge strips 6594 modules with 31132 apv s on 297 petals 6594 modules with 31132 apv s on 297 petals 6 6 E10 210 5 5 10 10 8 1 08 flagged by LT 1 11 flagged by LT 8 gt gt 10 10 3 3 10 a 10 g 10 10 10 m 1 1 0 0 5 1 1 5 2 2 5 3 3 5 4 4 5 5 0 0 5 1 1 5 2 2 5 3 3 5 4 4 5 5 CMSnoise in peak inv on CMSnoise in peak inv off with APV edge strips with APV edge strips 6594 modules with 31132 apv s on 297 petals 6594 modules with 31132 apv s on 297 petals m 210 10 rz rz 10 7 10 1 44 flagged by LT a 1
40. In total there were eight production and test setups in Aachen Brussels CERN Louvain two in Karlsruhe and two in Strassbourg It is worth mentioning that CERN could test six petals in parallel with one setup which implies that there were 13 testing lines in total At the end of this chapter a list of the defect components and the reason why they were exchanged is given Tab 5 4 To understand this list it is necessary to understand the grading of a petal 5 2 1 Petal grading For the grading of a petal the quality of each module was used The grading of each module can be found in Tab 5 1 and Tab 5 2 The grading of the whole petal is described in Tab 5 3 Only grade A and B petals were accepted to be built into the TEC Hence petals with grade C or D had to be repaired which implies to exchange the defective modules A module is also exchanged if an optical inspection shows that bonds are bended or damaged or some other problems are found This inspection is done before and after the LT test Normally such problems imply that a module would be graded as C as a lot of channels are lost by those problems Even if a module would pass the LT test it is not desired to have bended bonds inside the tracker as it is unknown if they still have the necessary stability or if they could introduce some shorts Type of oat Normalised common mode subtracted noise 10 Normalised Pulse Height 20 Peak Time Average Subtracted Absolute
41. Le 5 i 5 E ra E E oJ z z u 5 2 5 FA Z a unflagged unflagged channels LT TEC channels LT TEC a rings 1 4 b rings 5 7 Figure 6 23 Number of channels flagged during LT and TEC sector test APV edge channels and defective APVs are excluded 51 30 Finally 85 of all LT test flags can be reproduced by the sector test data From the 214 channels flagged by LT and not flagged by TEC only 50 channels are not flagged as S S open or unknown This implies that only 3 3 of channels flagged by LT are not found by TEC unexplained In the noise region below 1250 for Ring 1 to 4 and 1420 for ring 5 to 7 see Fig 6 20 less than 6 more channels are flagged from TEC than from LT These two numbers 3 3 and 6 show the high consistency between LT and TEC test and hence the consistency between ARC and TEC Using also the ARC information for those 50 channels only 8 channels are just flagged by LT Doing the same for the 214 channels just 28 channels are not flagged These 28 channels are equal to 1 8 of all channels flagged by LT and those 8 are equal to 0 5 This shows that a channel flagged by LT is really defect in about 99 of all cases Unfortunately this statement can not be proved as it is not possible to do an optical inspection of the modules after the petal assembly Since 34 APVs out of 15104 APVs are dead 30 only TEC the defect rate here is 0 23 Taking all defect channels from ARC LT and
42. PiaReset pllinit dummy 60 I2cDump dummy dummy 2 CheckEnv dummy _d 2 ChangeHV 400 _hv400 60 SaveRec 3 SECLTFIRST 2 PiaReset pllinit dummy 2 ChangeHV 0 _hv0 2 I2cDump dummy dummy 2 I2cDump dummy dummy 2 CheckEnv dummy _d 2 PiaReset pllinit dummy 60 OptoScanRun i2cpeak desclbl fullscan 2 CheckEnv dummy _d 2 I2cDump dummy dummy 60 ChangeCool 25 _t 25 2 TimeTuneRun i2cpeak _tt 30 ChangeCool 25 _t 25 2 I2cDump dummy dummy 120 TempReached 23 _tr 2 PedRun i2cpeakinv _ppi 1800 I2cDump dummy dummy 2 I2cDump dummy dummy 2 PiaReset pllinit dummy 2 CalProfRun i2cpeakinv _cfpi 2 I2cDump dummy dummy 2 I2cDump dummy dummy 2 ChangeHV 400 _hv400 2 CheckEnv dummy _d 2 I2cDump dummy dummy 60 I2cDump dummy dummy 2 PiaReset pllinit dummy 2 SaveRec 17 SECLTLAST 2 CheckEnv dummy _d 2 PedRun i2cpeak _pp 60 I2ecDump dummy dummy 2 I2cDump dummy dummy 2 OptoScanRun i2cpeak desclbl fullscan 2 LatRun i2cpeak _p 2 I2cDump dummy dummy 2 I2cDump dummy dummy 2 TimeTuneRun i2cpeak _tt 2 SaveRec 17 SECLTLAST 2 I2cDump dummy dummy 2 PedRun i2cdec _pd 2 PedRun i2cpeakinv _ppi 2 I2cDump dummy dummy 2 I2cDump dummy dummy 2 CheckEnv dummy _d 2 CalProfRun i2cpeakinv _cfpi 60 I2cDump dummy dummy 2 I2cDump dummy dummy 2 SaveRec 17 SECLTLAST 2 CheckEnv dummy _d 2 PedRun i2cdecinv _pdi 60 I2cDump dummy dummy 2 I2cDump dummy dummy 2 SaveRec 14 SECLTCOLD 2 SaveRec 17 SECLTLAST 2 PedRun i2cpeak _pp 2 ChangeHV 0 _hv0 T6
43. TEC including dead APVs a defect rate of 66 Chapter 6 Analysis of LT measurements 6 4 Comparison between LT and sector test Total Number of Channels 727508 Total Number of Channels 1141172 E 800 E 7 in 7 T 3 g 400 E z 600 E g O W 2 400 D E 3 0 0 only LT LT TEC only TEC only LT LT TEC only TEC a rings 1 4 b rings 5 7 Figure 6 24 Number of channels flagged during LT and TEC sector test split up into channels flagged by LT and TEC or by both test systems APV edge channels and defective APVs are excluded 51 30 approx 0 33 is estimated excluding APV edge strips and noisy channels This gives hope that the design goal of less than 1 missing strips after the whole tracker insertion can be achieved 67 6 4 Comparison between LT and sector test Chapter 6 Analysis of LT measurements 68 Chapter 7 Summary The focus of the present work is the longterm test of silicon strip modules This test has been crucial for the production of the CMS tracker to guarantee the benchmark quality of the petals Moreover in case of defect components it has been possible to exchange them or even to exchange the whole petal Furthermore the LI test provides an important independent data set which allows to characterize each single module and extrapolate the final tracker performance before the start up and final assembly of all detector components The presented results
44. a precise muon system a good electro magnetic calorimeter and a high resolution tracker to identify secondary vertices and to have a good momentum measurement For the latter also a strong magnetic field is necessary Therefore it has a superconducting solenoid with a uniform magnetic field of 4T a length of 12 5m and a diameter of about 6m The magnetic flux is returned via an iron yoke of 1 8m thickness This yoke is instrumented with muon chambers This gives the detector a compact design which is the reason for its name The coordinate system of CMS is defined as x axis pointing radially toward the center of the LHC y axis pointing vertically upward and the z axis along the beam direction completing a right handed coordinate system 0 is the polar angle measured form the z axis and the azimuth angle measured from the x axis in the x y plane The pseudorapidity 7 is defined as n In tan 2 The plus side is the side with positive z values and the minus side with negative z values A sketch of the CMS detector gives 1 2 From the outside to the interaction point it is instrumented with muon system integrated into the return yoke of the magnet superconducting magnet hadron calorimeter electro magnetic calorimeter silicon strip detector silicon pixel vertex detector In total the detector has a length of 21 6 m a diameter of 15 m and a weight of about 12 500 tons 5 1 3 1 The Muon system Muons
45. a bond which is destroyed by mishandling or disconnected due to other problems of the channel Fig 3 4 a and Fig 3 4 b The missing bond is normally between pitch adapter and sensor Therefore it is called PA S open The consequences are lower noise earlier peak time and a larger peak height The lower noise can be explained by the fact that the noise of the amplifier is proportional to the connected capacitance which is in the order of 10 25 pF see Eq 2 1 The capacitance is lowest for a disconnected strip The shaping time is also proportional to the connected capacitance and hence reduced Consequently the peak height gets larger and the peak time decreases see Eq 3 8 and Fig 3 5 This is also true for the other open types They just differ in the amount of deviation So channel 94 of the ring 3 module shows the signature of an open confirmed by optical inspection with a microscope aa SN a b Figure 3 4 a Microscopic view of a missing bond at the sensor b Photograph of destroyed APV bonds Sensor Sensor open A sensor sensor open exists only for modules with two sensors rings 5 to 7 In this case strips of the near sensor are still operational Usually the reason is a destroyed bond due to mishandling Fig 3 6 Mid sensor open As the name indicates the problem lies somewhere in the middle of a sensor In Fig 3 7 for example the source is a scratch on the sensor surface A part of a s
46. ags a CERN b other centers MY intact strip WE defective strip noisy strip Figure 6 3 Number of flags after noise tests Colors indicate the ARC test results This is similar for petals tested at CERN but the correlation is not as high due to the prob lematic noise behaviour The range up to 10 000 channels the height of bin 1 can be found in App E Thus a 10 deviation in at least six noise tests is a good criterion to find defective strips without flagging too many good strips especially for the CERN setup A lot of strips which would be flagged by the CERN setup via a simple 10 cut are rejected by requiring a misbehavior in at least six tests The performance of this method can be seen in Fig 6 4 and Fig 6 5 APV edge strips are now included Fig 6 4 shows that the amount of flagged strips by the LT test drops from approx 24 000 channels to roughly 5 000 channels from 0 6 to 0 12 Most channels are flagged by both ARC and LT test In Fig 6 5 the number of defective strips per APV channel for ARC and LT test are shown A good agreement between ARC and LT test for all petals can be seen Apart from the edge channels only some APV channels show a different behavior e g channel 8 in Fig 6 5 a This can be improved by excluding noisy strips from ARC and LT test Therefore a definition of noisy in the LT test is needed This will be done in Sec 6 2 2 The petal with ID 30250400000095 was excluded
47. al Distribution measured position measured position of passage gt position position position a b c of passage Amplitude Amplitude ie Amplitude gt Figure 2 13 Principle of the particle detection using a reversely biased diode 23 a Charges created by an ionizing particle drift to their respective electrode and induce a signal that indicates the passage of a particle b Charges drifting to neighboring electrodes induce signals and the signal height is a measure for the amount of charge drifting to the respective electrode By weighting the signals the spatial resolution can be improved to values below the size width of the segments c High energetic primary electrons can create charges and thus signals in regions far away from the particle s track Thereby the spatial resolution gets deteriorated High energetic primary electrons are responsible for the long tail of the Landau Distribution deplete the whole diode In addition the noise of an amplifier is proportional to the input capacitance 24 This is mainly determined by the capacitance of the p n diode which is smallest if the diode is fully depleted An other source for noise is the leakage current 25 More information can be found in 26 and 27 Each module has one or two silicon sensors two sensors for modules in TOB and modules on ring 5 to 7 of TEC of roughly 10cmx10cm and a
48. and methods have been elaborated during the time of the LT test 2005 2007 and have influenced the final classification scheme Here the following results are sum marized The layout of the Si strip detector is introduced and possible defect sources affecting single strips are presented The LT test procedure and the importance of it is described 48 The data acquisition of the LT test is introduced For efficient data taking the MUX mapping has been developed in the context of the presented work see App B A test setup has been set into operation with the help of F Beissel and D Jahn The DAQ software and slow control has been ported to the test system in Aachen with the help of W Beaumont and Th Hermanns Roughly 15 of the whole LT test data has been taken with the LT test setup in Aachen 48 The whole data set of the LT test and TEC sector test is used for the present analysis 148 49 The LT test data has been compared with the ARC test data The knowledge gained from this comparison leads to a robust defect finding algorithm 48 The efficiency to find a defect that has already been noticed by the ARC setup is at 92 see Fig 6 18 b and Fig 6 19 b 13 of all channels flagged as defect by the LT test have not been 69 Chapter 7 Summary noticed to be defect by the ARC test Reasons therefore could be the assembly of the modules onto the petals as also a bad identification of defects in ARC or LT A f
49. and the signal latency must be tuned Init Control Steering Settings Monitor ModStat StrucStat descriptor default term3 LtStruct config settings xml Test status None is Running A Figure 4 7 Screen shot of the DAQ software On this tab the manual steering of the test is shown 47 4 7 Test procedure The standard long term test scenario can be split into three parts Starting in warm conditions going to cold and back to warm conditions The results of the test performed during each part are stored in a ROOT file Each part is saved into an own directory inside the ROOT file secltfirst secltcold and secltlast This single cooling cycle and the tests are mandatory for a LT test Two further cooling cycles and tests are added but not analyzed see Fig 4 8 During the transition the temperature of the cooling liquid is changed between 17 C and 25 C The fridge is off or on respectively At each temperature an opto scan run timing run four pedestal runs one per APV mode and one calibration profile run are performed Fig 4 9 These tests are used to find strip errors Additionally in the first and last part an extended I V test and in the first part a pedestal test without HV are performed l An Object Oriented Data analysis framework Copyright by Rene Brun amp Fons Rademakers 99 4 7 Test procedure Chapter 4 Long term test nn Tin en run
50. ar it is the first test which allows to test the cooperation of all components like module ICB and AOH As the whole test program needs a lot of time this test is called long term test LT test The data of the test allows a grading of the petal By that a decision is made if the petal can be used for the TEC To speed up the assembly and the testing of the petals the components i e petal bodies modules etc were distributed to six Petal Integration Centers PICs which allowed to assemble and test the petals in parallel A summary of major problems found by this test as also which problems were found with the setup in Aachen will be given in Chap 5 4 1 General setup The main parts of the LT setup are a cooling plant a refrigerator a rack with the necessary electronics and two PC s for slow control and DAQ Fig 4 1 and 4 2 The functionality of the different parts will be described in the next subsections 4 2 Communication The communication during the longterm test with the petal works via a token ring like pro tocol A FEC card inside the DAQ PC translates the commands of the PC into the token ring protocol with LVTTL signals An FEC adapter card designed and build in Karlsruhe Front End Controller 40 2Low Voltage Transistor Transistor Logic 39 4 3 Readout Chapter 4 Long term test Figure 4 1 The long term test setup at Aachen Left the cooling plant for active cooling of a petal middle the fridge for
51. are expected to provide clean signatures for many interesting physics processes e g the Higgs decay H ZZ ptu ptpu For this reason the task of the muon system is a good muon identification charge determination and in combination with the tracker a precise 3 1 3 The CMS Experiment Chapter 1 Introduction Superconducting Solenoid Silicon Tracker Pixel Detector Very forward Yo Calorimeter Preshower Hadronic Calorimeter Electromagnetic Calorimeter N amp UN Muon ANS Detectors Compact Muon Solenoid Figure 1 2 Overview of the CMS Detector 7 muon momentum measurement Also a fast trigger decision is needed The muon chambers are the outermost part of the detector because muons are the only charged particles which are able to traverse all detector parts without significant energy loss In Fig 1 3 the three different technologies used in the detector and their position are plotted In the barrel region the muon detection is based on drift tubes DTs In the end caps cathode strip chambers CSCs are used because they can be operated even in a strong magnetic field and at high particle rate In addition both regions are equipped with resistive plate chambers RPCs Due to their fast response i
52. ast 8 hits in the strip tracker to reconstruct a track Fig 1 4 The achieved accuracy can be seen in Fig 1 5 N points 0 025 05 075 1 125 15 175 2 225 25 Figure 1 4 Number of measurement points in the strip tracker as a function of pseudo rapidity n Filled circles show the total number back to back modules count as one while open squares show the number of stereo layers layers which deliver a three dimensional measurement 14 The Silicon Strip Tracker The silicon strip detector can be split up again into four subdetectors Fig 1 6 The innermost part is the tracker inner barrel TIB which is enclosed by the tracker inner discs TID Both are inside of the outer barrel TOB To each side of TOB one tracker end cap TEC is mounted The tracker will be discussed in detail in the next chapter as parts of it are matter of this thesis The Vertex Detector The Vertex Detector 15 is a silicon pixel detector and the innermost part of the CMS detector It consist of three barrel layers TPB and two end caps TPE at each side It must be able to identify secondary vertices from b quark or T lepton decays Therefore it is necessary to have a high hit resolution In the barrel region the pixels have an area of 100 x 150 um They are delivering two dimensional points which allow a resolution of 15 um in both coordinates Fig 1 7 is a sketch of the pixel detectors with three layers in the barrel region and two discs on each side
53. at the strip failed at least six tests 51 52 Example of a pulse with a peak time of 8 ns and a height of 35 ADC counts Fit based on the CRRC function see Chap 3 and Eq 3 8 93 Line fit through the calibration group subtracted CGS peak time per ring IE PS ORCS oe kat an nn a a a haha lan 54 Peak height versus peak time Only channels identified as defective by ARC and LT are taken into account Categories are defiened as in the ARC test 55 97 List of Figures 98 6 9 6 12 6 13 6 14 6 16 6 17 6 20 6 21 6 22 6 23 6 24 B 1 Scaled peak height versus peak time Plots are separated into the different defect types of ARC Only channels identified as defective by ARC and LT are taken IMO OCCU lt e kw He eR A Oe nn a ki a a a ea Purity as given by Eq 6 6 PA S open and S S open define the two areas a and S S open and noisy b respectively a Noise distribution of short and saturated channels The noise is scaled and the average is subtracted b Purity as given by Eq 6 6 Short and saturated channels define the two areas Scaled peak height versus peak time for petals tested at CERN Noise distribution of shorts and saturated channels Number of defective strips per APV channel for ARC and LT Noisy channels ere excluded 51 2614 0 0 0 u Rw Kee ROR Oh Re D
54. counts A wen exp a 3 8 o o where hp gives the height of the peak tp the position and the width which is equal to the shaping time 7 RC 50ns t is equal to the time on the r axis and ADCcounts to the height on the y axis Using this equation for a fit the position t peak time and the maximum h peak height can be determined The distribution of the peak heights and peak times for a single module can be found in Fig 3 3 Problems with a strip can be observed if the peak height or peak time differs from reference value The amount of the deviation depends on the type of defect of the strip So a classification of strip defects is possible 3 3 Defect types There are different types and sources of defects Some affect the whole module some a group of 128 channels and others only individual strips In the following single strip defects will be discussed These can be divided into different defect classes The general signature is given in the following At the end of this chapter a short overview of the ARC test procedure and 2LATENCY and CSEL are APV registers for fine tuning LATENCY delays the write signal with respect to the trigger in steps of 25 ns CSEL can adjust the time of calibration injection in eight steps of 3 125 ns 25 3 3 Defect types Chapter 3 Single module test peak height height ADC counts peak time 250 300 time ns Figure 3 2 Typical profile of one channel Here measured
55. d LT test Flags per Apv channel Flags per Apv channel 5 odules wi 1281 modules with 6120 apvs on 54 petals LT Test JARC Test N O1 defect channels N defect channels er N v a LILI 2 gt rm D 1 2 O D H u D ct un cr jauueyo OZEZELE Isuueu9 098882 PS O1 gt oO l In i 1 E N in i i ii IL All h i un me Hh a LS il af dl n Mh pr 4 ao 60 80 100 120 ME a ki 60 80 100 720 APV channel APV channel a CERN b other centers Figure 6 14 Number of defective strips per APV channel for ARC and LT Noisy channels are excluded 51 Flags per Apv channel Flags per Apv channel odules wi 1281 modules with 6120 apv s on 54 petals _ A N A euueu9 09 8 o ISUUBU9 OZEZELE N difference of defective channels N gt TT RER difference of defective channels N o o L L L L L L 1 1 1 1 1 L L L L l 1 L L 1 L 1 1 L L L L L L 20 40 60 80 100 120 20 40 60 80 100 120 APV channel APV channel a CERN b other centers Figure 6 15 Difference of defective strips per APV channel between ARC and LT ARC LT Noisy channels are excluded 51 Another way to compare the ARC test with the LT test is to investigate how many strips are flagged by the two tests Fig
56. d also be one for the power supplies They won t be able to deliver the necessary current for this module and so the module has to be cut from the power system without being exchanged Therefore sensors with a leakage current above 10 uA per sensor at 400 V are rejected A normal current at this voltage is below 1 uA Such a high current gt 10 uA was found for approximately ten modules out of 6600 modules These modules were disassembled and replaced by other modules 43 4 7 Test procedure Chapter 4 Long term test 44 Chapter 5 Problems found with long term test This chapter gives a summary of problems found during the LT tests with the Aachen setup There are three locations where problems can occure the test stand setup defect components of the petal and problems of the petal design itself 5 1 LT test setup Before testing each petal the setup itself had to be set into operation and tested The hard ware part was done with the help of the mechanican and electronics engineer The software part had to be adapted to the local conditions in Aachen This was done without bigger problems as the software was very modular and therefore only few changes were needed The commisioning of the setup with help of a test petal was not that easy It was found that the communication of the software with the petal was very instable In most cases the token ring could not be established otherwise it was lost after few minu
57. dow Help reading instruments E S DT STOP alive signal g al hz Software interlock first level SET temperature cooling IN T Es and the highest G dew point ponts alle dry ar IN int reo aai sa upper side set fridge read fridge set cooling plant EZE E 2 read cooling plant oo time between set fridge dt minutes and set cooling plant TCP IP SORT PEST PL PA BEN EHE BEE DREH EL ES EI BEER MEER 3o 4 a 9 10 12 sc u Figure 4 5 Screen shot of the slow control software On this tab the monitoring of the temperature is shown 43 OFI QUISITIO ACQUISITION connection _ QUIT Slow Control LV power supply Software Interlock Status Cooling plant alarm instruments shut down Ill Figure 4 6 Screen shot of the slow control software On this tab the interlock status is shown 43 38 Chapter 4 Long term test 4 7 Test procedure temperature and voltages Fig 4 7 The complete long term test of one petal needs two to three days depending on the petal and scenario To give the user the possibility to change the order of tests or to add some additional tests a scenario file can be loaded The standard scenario can be found in App C The main parts of the scenario are pedestal and cal prof tests in different APV modes and at different temperatures To perform these tests in a proper way the parameters of the AOHs need to be optimized
58. e strips with APV edge strips 6594 modules with 31132 apv s on 297 petals 6594 modules with 31132 apv s on 297 petals 10 10 So G 10 a 10 es 0 47 flagged by LT 0 59 flagged by LT FA P A 104 4 gt 2 10 S H 10 D 10 n 10 10 10 10 L AM LAURE es L ee 0 0 5 1 1 5 2 2 5 3 3 5 4 4 5 5 0 0 5 1 1 5 2 2 5 3 3 5 4 4 5 5 CMSnoise in dec inv off CMSnoise in dec inv on 6594 modules with 31132 apv s on 297 petals with APV edge strips with APV edge strips 6594 modules with 31132 apv s on 297 petals 6 6 10 c10 gt 5 10 a 10 0 41 flagged by LT g 0 51 flagged by LT g 4 4 10 S 10 S 9 9 3 3 10 a 10 w 10 10 10 10 1 1 0 0 5 1 1 5 2 2 5 3 3 5 4 4 5 5 0 0 5 1 1 5 2 2 5 3 3 5 4 4 5 5 CMSnoise in dec inv on CMSnoise in dec inv off Figure D 2 Noise distributions taken in deconvolution mode The noise is normalised to the average module wise First row is first warm phase second row in cold phase and third row again in warm phase All channels are included Sl Appendix D Noise distribution excluding APV edge strips excluding APV edge strips 6594 modules with 31132 apv s on 297 petals 6594 modules with 31132 apv s on 297 petals a 6 6 E10 ate O O 10 u 10 z 0 15 flagged by LT amp fe 0 16 flagged by LT 18 1
59. ed by CERN in collaboration with ITHE Brussels Optical Electrical Converter a chip developed by CERN in collaboration with IIHE Brussels which converts analog optical in to analog electrical signals 8Front End Driver 34 Chapter 4 Long term test 4 3 Readout 1 PC vy gt Control signals FEC Petal electrical Data optical Data gt Slow Control controlling supply units gt Communication DAQ Slow Control Figure 4 2 Schematic view of a longterm test station help of an 9 Bit ADC Unfortunately the FED has only eight input channels and not 68 as needed To solve this problem a switch K MUX allowing to switch between eight times ten channels is used A schematic view of communication and read out can be found in Fig 4 3 4 3 1 K MUX The K MUX allows to change the readout between different modules It consists of 80 input channels that are arranged in 8 units of each 10 input channels column Each unit allows to link one of the 10 input channels to a single output channel The readout happens row wise For sake of efficiency and to avoid systematics by reordering the cables it is desired to keep one unique mapping for front and back petals Fig 4 4 shows a photo of a fully cabled K MUX The only constraint to the mapping is to always read out complete modules It can be shown that there is no common mapping which matches
60. een logical one and zero for one bias and gain setting without saturation within the ADC of the FED Per default gain2 is used in all tests as this should work for all lasers ADC counts S 80 90 100 10 20 30 40 50 60 70 80 90 100 bias mA bias mA Figure 4 12 Opto scan run for one APV in gain3 Left logical zero right logical one 42 Chapter 4 Long term test 4 7 Test procedure a ar oma a Van Moduleid _ fo200020010026 EN 1 41 gt redraw ST daa i cane GE SB east Goh adaa u De Ok DER D enr GUN ee A A aa ed aaay OCA mn nun nu nn nu nn nn bu nu nun nun nun un nn du nn nn nn nn ne UF TI u jph I ee eee ee ee eee ha ihc ok al a ae ae al aaa eS ho a en ga jb tas admuaadasaceasehbaad aaasdanedgewaedcad dea ede needa e ede acdaldedaaedeatigwaenaaablhbacamnenaaeamacaaidian elapsed tim rm ern net A Time 30 TimeTune NERDEN Figure 4 13 Screen shot of the DAQ An opto scan run in gain2 for a ring 4 module is shown Logical one and zero for both lasers are plotted into one diagram 4 7 3 Extended I V run In the extended I V run a current voltage curve bias current against depletion voltage per module is taken Measurements are done up to 450 V in steps of 50 V A high current indicates that the module has a problem even if the detector performance is not disturbed by this As the leakage current of each module will increase during LHC running see Sec 2 4 2 this problem coul
61. entral Communication Unit AOH Analog Opto Hybrid APV Analogue Pipeline Voltage 93 Appendix F Calibration group subtracted peaktime ARC APV Readout Control CalProf Calibration Profile ADC Analoge Digital Converter PA S open Pitch Adapter Sensor open S S open Sensor Sensor open PIC Petal Integration Center LT Long Term FEC Front End Controller OFED Optical Front End Driver OEC Optical Electrical Converter K MUX Karlsruhe Multiplexer PLL Phase Lock Loop 94 List of Figures 12 1 3 1 4 1 5 1 6 Lf 2 1 2 2 2 3 2 4 2 0 2 6 AL 2 8 27 The LHC ring at CERN with its four experiments 2 Overview of the CMS Detector 7 quarter of the muon system The different technologies are labeled and Number of measurement points in the strip tracker as a function of pseudo rapidity n Filled circles show the total number back to back modules count as one while open squares show the number of stereo layers layers which deliver a three dimensional measurement 14 Resolution of several track parameters for single muons with transverse mo menta of 1 10 and 100 GeV transverse momentum left transverse impact parameter right and longitudinal impact parameter below 14 The CMS tracker The different regions of the tracker are marked in different ee The pizel detector 12 sse sau saw nu Lern DEERE DR EH end
62. ework for the analysis of sector test data RWTH Aachen 2008 Personal communication with Wim Beaumont M Axer Development of a Test System for the Quality Assurance of Silicon Microstrip Detector Modules for the Inner Tracking System of the CMS Experiment PhD thesis RWTH Aachen 2003 G Sguazzoni The construction of the CMS Silicon Strip Tracker arXiv0801 2468v1 2008 Bibliography Bibliography 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 M J French Design and results from the APV25 a deep sub micron CMSOS front end chip for the CMS tracker Nuclear Instruments and Methods A 466 pages 359 365 2001 P Murray APVMUX user guide version 1 0 2000 http www te rl ac uk med projects High_Energy_Physics CMS APVMUXPLL pdf UserGuide pdf A Marchioro et al CMS Tracker PLL Reference Manual 2000 http cmstrackercontrol web cern ch cmstrackercontrol documents PauloMoreira PLL25 20User 20Manua2 1 pdf A Marchioro et al DCU2 User Guide 2001 http cmstrackercontrol web cern ch CMSTrackerControl documents Magazzu DCU2_ User_Manual 20v2 12 pdf personal communication with ARC group in Aachen C Ljuslin and C Paillard Front End Control unit for Embedded Slow Control CERN 2003 http cmstrackercontrol web cern ch cmstrackercontrol documents FECCCU FECSpecs pdf M Ageron CMS Trigger Sequenzer Card User Manual Lyon 2001 ftp ly
63. f TIB The first two layers are layers with double sided modules so each rod has depending on the layer six or twelve modules TOB has in total 5 208 modules with 3 096 576 readout channels Fig 2 6 shows a picture of TOB 2 4 TEC Next to each end of TOB one TEC is placed Each TEC has nine discs and each disc has up to seven rings of modules discs 1 to 3 7 rings discs 4 to 6 6 rings discs 7 and 8 5 rings and disc 9 4 rings counting from the interaction point The TECs have a substructure called petals On every disc 16 petals are mounted 8 facing the interaction point front petals 11 2 4 TEC Chapter 2 The Silicon Strip Tracker Figure 2 5 A rod of the TOB 34 Figure 2 6 Picture of TOB support structure 14 and 8 on the far side back petals The front and back petals are slightly overlapping to ensure a complete coverage with silicon On the rings one two and five double sided modules are mounted Both TECs together contain 10 288 sensors on 6 400 modules with 3 866 624 readout channels More details on the tracker can be found in 14 12 Chapter 2 The Silicon Strip Tracker L lt A h l 7 a i i ae m Ri as _ aly ia Bala A A f AA M 4 4 IN t Figure 2 7 Side and front view of a TEC 14 2 4 1 Petal One petal has modules belonging to up to seven rings depending on the disc the petal is mounted to In total there are eight different type
64. from the analysis due to a corrupt dataset 3Fluctuating numbers due to slightly varying statistics ol 6 2 Calibration pulse test defect declaration Chapter 6 Analysis of LT measurements Zn Total Number of Channels 3936768 n Total Number of Channels 3954176 a 6000 a 23656 2 5462 20000 z 5 4000 5 K z gt UO z z 5 nn J J 8 10000 5 8 ae lt E2000 2 5 Ze Z 0 0 unflagged unflagged channels LT ARC channels LT ARC a flagged with one failed test b flagged with at least six failed tests Figure 6 4 The first column shows the amount of good strips the second those which are flagged by LT and the third those flagged by ARC The dashed line shows the amount of strips flagged by both test systems 51 Flags per Apv channel Flags per Apv channel 1281 modules with 6120 apv s on 54 petals 5320 modules with 25040 apv s on 243 petals LT Test LT Test JARC Test JARC Test er N O1 D defective channels defective channels O1 _ O1 SIeuueu9 09 82 sj9uuey9 0ZLSOZE i 30 1007 H ft Li lu Ill A 1044 l Em 1 ali Fin E ue LR a Pah i pi ut 1 lent neal oral MAA 20 40 60 80 100 120 20 40 60 80 100 120 APV channel APV channel a CERN b other centers Figure 6 5 Number of defective strips per APV channel seen by
65. from the silicon thickness as the contribution of the interstrip capacitance cancles the contribution of the backplane ca pacitance 25 The capacitance for modules on the rings 5 to 7 of TEC and for modules of TOB is largest as the strip length for these modules is up to 20cm Hence the noise for those modules is largest To obtain a good signal to noise ratio also in this region the signal has to be increased This is done by using thick sensors as the signal height is proportional to the sensor thickness The capacitance and therefore also the noise is not influenced by this change see Eq 2 1 The back side of the sensor consists of a uniformly metalized n layer This layer is connected to positive voltage of up to 500 V The n layer provides the necessary ohmic contact between the bulk and the aluminum layer The sensitive area is surrounded by a p bias ring and several pt guard rings The bias ring is at ground potential The guard rings are at floating potential in order to gradually reduce the electric field between the bias ring and the n layer at the sensor edges At the end of each strip are two bond pads to connect the strips to the readout electronic or in the case of two sensors the far sensor the sensor which has more distance to the readout chips is connected via wire bonds to the near sensor Each strip has a test pad which is directly connected to the p implant During LHC operation each sensor suffers from radiation damage
66. he energy of high energy showers additional scintillator layers are placed just outside the magnet Together they reach eleven absorption lengths The forward direction is situated within a harsh radiation field Therefore instead of brass iron is used as absorber This leads to narrower and shorter hadronic showers which fit the requirements in the forward region 1 3 3 The Electromagnetic Calorimeter The electromagnetic calorimeter ECAL 12 should give a precise measurement of the direc tion and energy of photons and electrons An interesting process which could be detected by the ECAL is H yy A clear signature is also given for a lot of other physical processes with leptons with a large transverse momentum e g semi leptonic t quark decays 13 The ECAL consists of more than 80 000 lead tungstate PbWO crystals In the barrel region EB they have a front face of about 22 x 22 mm which matches well the Moli re radius and a length of 23cm This allows a good separation of adjacent showers In the endcap region EE the front face is about 28 6 x 28 6 mm and the length 22 cm ECAL Barrel SECAL Endcap 1 3 The CMS Experiment Chapter 1 Introduction 1 3 4 The Tracker Within the ECAL the tracker is situated It can be divided into two components the silicon strip detector and the silicon pixel detector Both are needed to determine tracks of charged particles and their vertices Up to n 2 5 there are at le
67. hreiben Lutz Feld der bereit ist die Arbeit als Kokorrektor zu lesen der Arbeitsgruppe mit der Ideen erabeitet und besprochen wurden meinen Kollegen aus dem I Phys Inst die gute Anregungen zur Doktorarbeit hatten und mit denen man auf blank liegenden Nerven pfiff den Kollegen die im Tanzsaal oder in der Kaflepause f r regen Gedankenaustausch sorgten den Emigranten denen es an kreativen Ideen nie mangelte und die mir den Einstieg in die Arbeit sehr erleichtert haben Besonders dem Anstreicher fuer die Poesie dem Computersupport und den anderen die f r die n tige Sicherheit sorgten dem berflieger f rs Korrekturlesen und f r ein offenes Ohr allen Korrekturlesern f r eben selbiges meiner kleinen frechen Hexe die mich in der Zeit ertragen und unterst tzt hat meinen Eltern f r die selbiges gilt allen die ich vergessen habe 105 Danksagung 106 Lebenslauf Pers nliche Daten Name Geburtsdatum ort Dirk Heydhausen Diplom Physiker 20 03 1978 n Kempen Familienstand ledig Staatsangeh rigkeit deutsch Studium seit 01 2005 Promotion am III Physikalischen Institut B RWTH Aachen Analysis of Petal Longterm test data for the CMS Experiment 10 2008 Abgabe der Doktorarbeit 11 2006 03 2007 Aufenthalt am CERN in Genf 1998 2004 Physikstudium Diplom an der RWTH Aachen 2003 2004 Diplomarbeit am II Physikalischen Institut A RWTH Aachen 2001 Vordiplom Z
68. ibration pulse test defect declaration Chapter 6 Analysis of LT measurements is visible only for defective channels as they differ from zero The PA S open strips provide a good signature for the dependency on the strip length In Fig 6 7 the correlation between the At of PA S open and the strip length is shown The distribution of At per ring can be found in App F Due to this behavior and the fact that a strip length is of the order of 10cm 14 At is scaled to this length Eq 6 5 At f strip X striplength 6 4 10 cm At tp_use 10 f stri 6 9 p used striplength Cm F Gstrin QE tpused is the value used in the later analysis In the following the terms peak time and peak height indicate the scaled values tp used and Ap used Mean RMS from CGS peak time for PA S open per ring offset 2 0 1 3 slope 0 8 0 095 CGS peak time 8 10 12 14 16 18 20 Striplength cm Figure 6 7 Line fit through the calibration group subtracted CGS peak time per ring for PA S opens 6 2 2 Classification of defect types As described above the scaled peak height and peak time of all defects allow to classify the type of defect Fig 6 8 shows the LT calibration pulse test results of peak height versus peak time as a scatter plot for channels identified as defective by ARC and LT six failed noise tests The classification of a defective channel is given by the ARC test
69. ifference of defective strips per APV channel between ARC and LT ARC LT Noisy channels are excluded 51 Only channels flagged just by ARC are counted Noisy channels are excluded Peak height Pulse Peak versus channel number The channels 482 486 495 and 503 have a strongly increased peak height These channels are also suspi cious in other tests Channels 1 and 2 have a small peak height which differs only slightly from the distribution which has a slope between 0 5 and 1 0 ADC count per channel But as the deviation from the median is taken into account these channels are flagged as shorts Using other tests for the same strips does not indicate defects Number of channels flagged during LT and ARC tests The upper dashed line gives the number of channels flagged in both test systems the lower one the number of same flags 51 Number of channels flagged during LT and ARC test split into channels flagged only by LT and ARC or by both test systems 51 Scaled noise distribution measured in TEC Channels flagged by the LT test are marked in red APV edge channels and bad APVs are excluded 30 31 Scaled noise distribution of ring five to seven measured in TEC Categories are defined as in the LT test APV edge channels and bad APVs are excluded Mile go ober bee ea ee eee eee eee kee ee eee ee ee eS Scaled noise distribution of ring o
70. inv off CMSnoise in peak inv on Figure D 5 Noise distributions taken in peak mode The noise is normalised to the average module wise First row is first warm phase second row in cold phase and third row again in warm phase APV edge strips are excluded Only petals tested at CERN are taken into account 84 Appendix D Noise distribution excluding APV edge strips 1281 modules with 6120 apv s on 54 petals excluding APV edge strips 1281 modules with 6120 apv s on 54 petals channels s 3 1 0 18 flagged by LT sjauUeYd 088852 ee Co ree 2 25 3 35 4 45 5 CMSnoise in dec inv on excluding APV edge strips 1281 modules with 6120 apv s on 54 petals channels s s A ol w 10 E 0 28 flagged by LT sjouuey gt 088852 1 5 2 2 5 3 3 5 4 4 5 5 CMSnoise in dec inv on excluding APV edge strips 1281 modules with 6120 apv s on 54 petals channels 5 1 u P 10 10 E 0 14 flagged by LT sjouuey gt 088852 Lo BL er 2 25 3 35 4 45 5 1 5 CMSnoise in dec inv on channels s o 104 excluding APV edge strips 1281 modules with 6120 apv s on 54 petals 0 15 flagged by LT Pre sjauueyd 088894 li 2 5 3 CMSnoise in dec inv off channels u o excluding APV edge strips 1281 modules with 6120 apv s on 54 petals 0 19 flagged by LT
71. ip defect The values described above are all derived by offline calculations Instead of this method an event by event calculation is preferred as there is no need to keep all raw data and the information is available instantaneously Unfortunately some of those on line calculations are only approximations to the values mentioned above The Pedestal F nm for channel ch in event m is calculated with following equation Pim Penm 1 m 1 RD cham 3 4 m For m N this equation is identical to Eq 3 1 For the noise calculation a stable pedestal with only low statistical fluctuations is needed Therefore the noise computation starts after the first l events with l 1000 for the ARC test and later l 200 for the LT test The raw noise RNG m for channel ch in event m is given with following term RN m1 m 1 RD eam N RN im E 3 5 and the common mode C M in event m is calculated by y 2 CMn 198 2 Dern m 3 6 The common mode corrected noise CM Nen m for channel ch in event m is finally calculated by 24 Chapter 3 Single module test 3 2 Calibration Profile test CM Neamt m l 1 Tr RDen m UMm zu Pia CMN4 chim m l 3 7 The situation gets even more complicated as the calculation of the common mode is again for groups of 32 channels In addition noisy and dead channels have to be excluded from the computation Therefore an iterative algorithm is necessary The exact algor
72. irst row is first warm phase second row in cold phase and third row again in warm phase APV edge strips and petals tested at CERN are excluded 87 E 1 Number of flags after noise tests Colors indicate the ARC test results 89 F 1 Calibration group subtracted CGS peaktime for PA S opens in ring 1 left WO nat 91 F 2 Calibration group subtracted peaktime for PA S opens in ring 3 left and ring ad mn ne ey a ae 92 F 3 Calibration group subtracted peaktime for PA S opens in ring 5 left and ring Pen che Oe EAE POE Oe He ee Rees 92 F 4 a Calibration group subtracted peaktime for PA S opens in ring7 b Linear fit through the calibration group subtracted peaktime per ring 92 99 List of Figures List of Figures 100 Bibliography 1 2 S Eidelman et al Physics Letters B 592 1 Particle Data Group 2004 CMS Collaboration UMS Posters for Point 5 http cmsinfo cern ch Welcome htm1l CMSdocuments Point5Posters CMSp5posters_ index html ALICE Technical Proposal CERN LHCC 95 71 CERN 1995 ATLAS Technical Proposal CERN LHCC 94 43 LHCC P2 CERN 1994 CMS Collaboration The Compact Muon Solenoid Technical Proposal CERN LHCC 94 38 LHCC P1 CERN 1994 LHCb Technical Proposal CERN LHCC 98 4 CERN 1998 CMS Collaboration Detector Drawings http cmsinfo cern ch outreach CMSdocuments DetectorDrawings DetectorDrawings html CMS Collaborati
73. ithms and their performance can be found in 21 3 2 Calibration Profile test During the Calibration Profile CalProf test the APV emits a charge adjustable between 0 fC and 25 5 fC to each single strip The APV does this simultaneously for every eighth strip called calibration group Each calibration group differs slightly in the amount of charge sent to the strips In a second step the APV gets a trigger to read out the sensor The time between emitting the charge and readout can be controlled with help of LATENCY and CSEL in 80 steps of 3 125 ns 23 Thus the time evolution of the APV response could be measured which is a method to get information about the connected capacitance Subtracting common mode CM and pedestal P from the measured height RD gives the signal height S RM P CM To minimize statistical fluctuations onto this height each measurement is repeated 400 times Further a new pedestal is taken after every LATENCY change The exact implementation can be found in 33 The implementation for the LT test is a bit different As an approximation for pedestal and common mode the median med of the neighbouring and next to neighbouring channels are taken S RM med 32 A standard pulse which is read back can be found in Fig 3 2 It is to mention that the pedestal is subtracted A behavior like a CR RC circuit is expected see Eq 2 2 This form of this peak could be generalized with the following equation ADC
74. itting devices and were chosen due to their high linearity Each laser diode is steered by a laser driver which receives the data stream from the APVMUX and provides a bias current to the laser diode The laser diodes can be driven in four gain modes With help of those gain modes and a bias offset the optimal working point of the laser diode can be determined During the longterm test always gain two was chosen More details on the AOHs can be found in 14 The whole assembly of ICB CCU and AOH and also modules was performed in cooperation of several institutes across Europe After the assembly of all components a long intensive test was necessary to check the state of each component the long term test Chapter 4 Furthermore this test is the first test on the complete substructure of the TEC including communication readout and cooling The test routine contains the co operation of all component as well as several stress tests to reach knowledge if a petal could withstand several LHC shut down cycles Test of a whole petal directly after the assembly of all modules Described in Chap 4 21 2 4 TEC Chapter 2 The Silicon Strip Tracker Figure 2 18 Photograph of an AOH m Figure 2 19 Photograph of a fiber mapping used during the long term test A lot of fibers coming from the AOHs can be seen These are connected to six ribbons which have twelve input slots each The ribbon number used in appendix A 1 is shown in yellow 22
75. ity as given by Eq 6 6 Short and saturated channels define the two areas OTHERS peak time ns peak height ADC on ee gt unes Gamal sr eus 0 lt 6 lt 02 Pas eme lt a o Ss gt a lt a gt 2 lt a myama gt lt o gt 2 lt a mom TT NE CERN peak time ns peak height ADC ao TS lt 0 sn cm sn ur crane gt 10 lt 0 gt 2 lt 10 lt Paso gt se ase e en rss open gt kassa lt 5 Eos ma gt lt 6 gt lt a unknown everything else Table 6 2 Upper and lower bound of the classification for each defect class and testing center S S open overrule PA S open definition D Chapter 6 Analysis of LT measurements 6 2 Calibration pulse test defect declaration peak height ADC peak height ADC sjauueyd 229 10 10 a p Fe a 2 20 20 l l l l l l 1 l l l i l l 2 l l 1 l 1 l l l l l l ji l l 1 l l l 1 l ji l l 1 l l l l i l 15 10 0 5 10 15 15 10 5 0 5 10 15 peak time ns peak time ns a CERN b PA S open Q Q lt lt D D 3 3 xX x Q Q 10 5 0 5 10 15 10 15 peak time ns peak time ns c S S open d noisy channel peak height ADC peak height ADC 10 15 20 15 10 5 0 5 10 15 10 15 peak time ns
76. ivildienst 1997 1998 Haus Broich Anrath Schulische Ausbildung 1994 1997 LFS Muhlhausen Abitur 1988 1994 Joh Kepler Realschule S chteln 1984 1988 Katholische Grundschule Grefrath Weitere T tigkeiten w hrend des Studiums 2001 2008 Anstellung als Ubungsgruppenleiter f r Mathematik I IV und Physik I III Betreuung von Versuchen im Fortgeschrittenenpraktikum Physik Vorbereitung von Versuchen f r die Vorlesung Physik I 2006 Teilnahme an der LHC School of Physics theoretical tools and experimental challenges in Lecce Italien 2005 Teilnahme an der Joint Dutch Belgian German Graduate School in Texel Niederlande 2002 2003 studentischer Vertreter in Berufungskomissionen 1999 2000 Leitung eines Tutoriums f r Physik Erstsemester Au eruniversit re T tigkeiten und Hobbys freiwilliger Helfer beim Weltjugendtag 2006 Organisation von Wallfahrten und Jugendfahrten Betreuung von Messdienergruppen Klettern Wandern Fussball Musizieren Computerkenntnisse Linux Windows 95 98 2000 NT XP DOS Office Excel Word Power Point OpenOffice TEX Pascal C C Fortran ROOT MatLab LabVIEW Gimp Sprachkenntnisse Englisch Grundkenntnisse in Franz sisch und Hebr isch Ver ffentlichungen Diplomarbeit Untersuchung der Kopplung von phononischen und orbitalen Anregung mittels Raman Streuung in La xSr MnO 2004 The CMS experiment at the CERN LHC 2008 JINST 3508004
77. ix A 1 is shown We chee Rr 22 Pedestal and common mode subtracted CMS noise of a ring 3 module Ob viously the noise of the APV edge channels 1 128 256 384 is higher The noise of channel 94 indicates a strip defect 0 0 2 004 24 Typical profile of one channel Here measured on a ring 3 TEC module For the fit the Eq 3 8 was used The Fitting range is 10 bins around the maximum 26 Distribution of peak time and peak height of the same ring 3 module A periodic pattern of the length of eight can be seen due to the calibration groups Channel 94 again stands out in these plots indicating a single strip defect compare Fig 3 1 eee ee beet eed 26 a Microscopic view of a missing bond at the sensor b Photograph of de stroyed APV bonds sss ssc snerta kaden e k e da a d aof aen 27 Calibration profile of a PA S open a and a faultless channel b The PA S open has compared to the faultless channel an earlier peak time and a raised pak AREA ra i A a eg 28 Photograph of touched and destroyed bonds between two sensors 28 Microscopic view of a scratch on the sensor surface 29 28 Calibration pulse of a saturated channel Peak height is very low For compar WOU CGS PIG ie a ed ROP EEE REED SEGRE ESS OD 29 a Microscopic view of two connected strips 23 b Calibration pulse of a short strip The peak height is very low More than two strips are connected For comparison
78. jeuueys o9eesZ o jouueyo 0ZEZ61E D RES RERE 0 Muli tee oll al 80 120 20 40 60 80 100 120 APV channel APV channel a CERN b other centers Figure 6 16 Only channels flagged just by ARC are counted Noisy channels are excluded Pulse Height vs Channel 85 80 75 70 65 Pulse Peak ADC Counts 60 55 50 45 100 200 300 400 500 Channel Figure 6 17 Peak height Pulse Peak versus channel number The channels 482 486 495 and 503 have a strongly increased peak height These channels are also suspicious in other tests Channels 1 and 2 have a small peak height which differs only slightly from the distribution which has a slope between 0 5 and 1 0 ADC count per channel But as the deviation from the median is taken into account these channels are flagged as shorts Using other tests for the same strips does not indicate defects 62 Chapter 6 Analysis of LT measurements 6 4 Comparison between LT and sector test reprocessing the petals was not an option This prevents a correct classification see prev section The corrected amount of strips flagged identically by the ARC and LT test is around 82 1 A scenario in which all strips flagged by the ARC and LT test are bad results in a defect rate of 0 09 excluding APV edge strips and noisy channels For TIB and TID together this number adds up to 0 07 for TOB to 0 13 53 52 So the 0 09 of TEC LT are in a comparable range
79. llider 2Conseil Europeen pour la Recherche Nucl aire today European Organization for Nuclear Research Large Electron Positron collider 1 2 The Large Hadron Collider Chapter 1 Introduction leptons Table 1 1 The three particle generations of quarks and leptons and their quantum numbers is the electric charge given in elementary charge 73 is the third component of the weak isospin and Y the hypercharge Figure 1 1 The LHC ring at CERN with its four experiments 2 Chapter 1 Introduction 1 3 The CMS Experiment It will provide proton proton collisions with a center of mass energy of s 14TeV and at startup for the first year an instantaneous luminosity of L 10 cm7 s7 and later a luminosity of L 10 em s The beams will circulate in two separate pipes Dipole magnets with a field of 8 4 T force them onto their orbit At four points the pipes intersect and the beams will collide every 25ns which equates a bunch crossing frequency of 40 MHz At each of these four interaction points one of the following experiments is placed s Fig 1 1 ALICE A Large Ion Collider Experiment 3 ATLAS A Toroidal LHC ApparatuS 4 CMS Compact Muon Solenoid 5 LHC b Large Hadron Collider beauty Experiment 6 1 3 The CMS Experiment CMS was designed as a typical multipurpose detector It should be able to detect nearly all produced particles originating from the proton proton collisions It is equipped with
80. ls 6594 modules with 31132 apv s on 297 petals 0 10 flagged by LT 0 10 flagged by LT channels channels s o s o 10 10 SISUUEU9 OV8E98E SISUUEU9 OV8E98E 10 10 10 10 10 10 0 0 5 1 1 5 2 2 5 3 3 5 4 4 5 5 0 0 5 1 1 5 2 2 5 3 3 5 4 4 5 5 CMSnoise in dec inv off CMSnoise in dec inv on excluding APV edge strips excluding APV edge strips 6594 modules with 31132 apv s on 297 petals 6594 modules with 31132 apv s on 297 petals 10 210 G o E 10 5 10 g 0 13 flagged by LT a 0 11 flagged by LT a 10 3 10 2 10 u 10 a 10 10 10 10 1 Misch 1 A Dem mn On Er Pre 0 0 5 1 1 5 2 2 5 3 3 5 4 4 5 5 0 0 5 1 1 5 2 2 5 3 3 5 4 4 5 5 CMSnoise in dec inv off CMSnoise in dec inv on 6594 modules with 31132 apv s on 297 petals excluding APV edge strips excluding APV edge strips 6594 modules with 31132 apv s on 297 petals 6 6 E10 il 5 _ 5 _ 10 a 10 g 0 10 flagged by LT S 0 09 flagged by LT S 10 S 10 S gt 5 5 3 2 3 2 10 o 10 10 10 10 10 1 1 2 5 3 3 5 4 4 5 5 3 3 5 4 4 5 5 CMSnoise in dec inv off CMSnoise in dec inv on Figure D 4 Noise distributions taken in deconvolution mode The noise is normalised to the average module wise First row is first warm phase second row in cold pha
81. ma the total amount of strips in Al and analog for 2 deflai def2a2 6 6 Suma Sumag iG The cut is chosen to give an almost constant purity under slight variation Using the cut values shown in Tab 6 1 a purity for these three regions in the order of 90 is achieved For some of the strips classified differently in ARC and LT test it is known that the ARC test has declared them as S S open but visual investigation showed that the real defects are PA S open 39 ARC is a very precise test system with low noise but there are still some misidentifications and classifications Unfortunately the real amount of those misidentifications and classifications is unknown Some other differently classified strips can be explained by the fact that the DAQ software had some problems in the start up phase of the LT test software bugs which lead to a wrong peak time and peak height Obviously shorts and saturated channels cannot be distinguished that easily To distinguish u 2 1 0 9 u 9 2 u D 0 8 2 09 DT n C O1 8 5 F 5 0 7 pa H fz E 5 0 8 5 Q 3 p C 0 6 o 2 E lt 2 0 7 a f H 0 6 a 0 4 E 0 3 0 5 N T Oo 16 14 12 10 8 6 8 6 4 2 0 2 4 2 peak time ns peak time ns a PA S open and SS open b SS open and noisy Figure 6 10 Purity as given by Eq 6 6 PA S open and S S open define the two
82. me phase of the LHC clock 20 Chapter 2 The Silicon Strip Tracker 2 4 TEC APV _APVDataframe Frame 12 bit ee are header 250 ADC counts N O O vI O 100 50 a bit pipeline level of digital 1 address tick mark analogue data N mark oe level of digital O 20 20 40 60 80 100 120 140 160 180 Figure 2 17 Typical APV data frame 23 A Kapton circuit attached to the backplane of the sensor is used to connect the sensors to HV The return line goes via bonds to the pitch adapter More detailed information about the modules can be found in 14 23 30 2 4 3 AOH The AOH Fig 2 18 transforms analogue electrical signals coming from the APVMUX into analogue optical signals These signals are sent in case of the longterm test via optical fibers to a patch panel where the fibers are connected to six optical multi ribbon cable Fig 2 19 Each ribbon has twelve fibers Due to the length of the fibers and the wish to always have a complete module on one ribbon there are constraints to the mapping of AOH to ribbon The mapping used in the longterm test setup can be found in appendix A 1 Each AOH is responsible for one module and contains one laser transmitter per data stream coming from the APVMUX This means that an AOH that is responsible for a single sided module has two lasers and three lasers for one side of an double sided module The lasers are commercially available multi quantum well InGaAsP edge em
83. n peak mode inverter off peakinvoff deconvo lution mode inverter on decinvon and deconvolution mode inverter off decinvoff After the APSP the data can be read out through an analogue multiplexer that combines all 128 readout channels to one single channel A typical data frame released by the APV can be found in Fig 2 17 It is composed of a digital header with 12 bit length The analogue data follows with a length of 128 bit corresponding to the 128 strips The data frame is terminated by a tick mark a digital 1 If no data is present every 35 clock cycles a tick is sent to keep synchronization with the DAQ system In order to test the connected strips and the corresponding APV channel the APV can release a certain amount of charge to the strips The feedback gives information about the character istics of the corresponding channel The APVMUX chip is the next step in the readout chain of the hybrid It is a multiplexer chip which in order to minimize the number of readout channels combines the output of two APVs to one single channel This results in two readout channels per module for single sided modules and three for one side of a double sided module In total a petal can have up to 68 readout channels here front petal for disc 1 3 The TPLL chip allows to shift the signal of the clock in steps of 1 04ns up to 25ns With help of this chip it is possible to synchronize the APVs of different modules to the sa
84. ne to four measured in TEC Categories are defined as in the LT test Channels marked in blue are mid sensor open as they are counted to S S open APV edge channels and bad APVs are again Be Od ok ek ee He BE ae ER SE RES DEKE GO Number of channels flagged during LT and TEC sector test APV edge channels and defective APVs are excluded 51 30 Number of channels flagged during LT and TEC sector test split up into channels flagged by LT and TEC or by both test systems APV edge channels and defective APVs are excluded 51 30 Mapping of long ribbon cables to K MUX The mapping is given for a back petal In row 0 column 7 the entry is 5 12 This means that channel 12 of longribbon 5 has to be plugged in slot 7 row 0 To switch from back to front petal the blue have to be exchanged against the green ones This can be done by changing just one whole ribbon which means one connection List of Figures O7 61 List of Figures List of Figures D 1 Noise distributions taken in peak mode The noise is normalised to the average module wise First row is first warm phase second row in cold phase and third row again in warm phase All channels are included 80 D 2 Noise distributions taken in deconvolution mode The noise is normalised to the average module wise First row is first warm phase second row in cold phase and third row again in warm pha
85. noise which deviates more than 10 from the APV average Those channels are declared as bad All good channels are plotted in green 51 50 Chapter 6 Analysis of LT measurements 6 1 Pedestal test defect detection 6 1 2 Reproducibility As shown above every noise test flags around 0 1 of all strips except for APV edge strips and the CERN setup The total amount contains no information about the reproducibility of a single test i e if the same strips are flagged in all pedestal runs In total twelve noise tests are performed If the flagging was totally random approx 1 1 0 999 12 0 1 of all strips would be flagged at least once and the test would not be very reliable Therefore it is interesting to analyze how often a strip is flagged in all noise tests This can be seen in Fig 6 3 The color code is given by the ARC test results Concentrating on the petals not tested at CERN and using the information from the ARC test it can be said that a defective strip is flagged at least six times or never Intact strips are normally flagged never or only a few times lt 6 number of noise flags number of noise flags 0 A 1281 modules with 6120 apv s on 54 petals 5292 modules with 24912 apv s on 242 petals S N O1 500 channels channel 2000 400 1500 sIeuueus 088857 SIeuueu9 096260 300 1000 200 100 o 0 2 4 6 8 10 12 0 2 4 6 8 10 12 noise flags noise fl
86. ns Table 5 1 Valid ranges for module test parameter values 48 Modified in Chap 6 CB IR lt n lt I lt 3yA sensor 1 lt n lt 2 gt 3uA sensor Cote LL eS ho Table 5 2 Module grading criteria as a function of the number of bad channels n and the sensor leakage current leak 148 5 2 2 List of exchanged components and their defects In the following a short summary of the components exchanged in Aachen is given Tab 5 4 The reason why the modules were exchanged can be found in Tab 5 5 the reasons for the 46 Chapter 5 Problems found with long term test 5 3 Petal design Z Nesom 3A gt Neon YA OB 05 lt N lt I lt 50 of B modules no C module lt Nees A 0 5 lt N lt 1 lt 50 of B modules no C module SEN sos t IUA 1 lt N lt 1 5 1 C module with lt 2 5 of bad channels DE D Any other combination Table 5 3 Petal grading criteria as a function of the total number N of bad channels in a petal the module quality and the total sensor leakage current Ijeax 48 AOHs in Tab 5 6 The reasons for the exchange of modules are touched bonds I2C problems i e inter chip communication problems during the LT test modules graded as C by LT or damaged during the rework of the petal clearance problem 22 the insertion into TEC at the Tracker Integration Center TIC or the preparation for the LT test The categorie other includes scratches on the sensor surface or b
87. o smaller and smaller pieces This leads to the topic of particle physics In the last century physicists developed the Standard Model of Particle Physics SM describ ing effects and particles which were measured Nevertheless there is one particle predicted by the SM which could not be found until now the Higgs boson Furthermore there are questions the SM cannot answer To solve these problems to measure the Higgs boson and to find new physics a new accelerator was built the Large Hadron Collider LHC 1 1 The Standard Model of Particle Physics According to the SM the material in the universe is made up of fermions The interactions between the fermions the gravitational electromagnetic weak and strong interaction are mediated by bosons the Graviton G photon y weak gauge bosons Z Wt W and eight gluons g All elementary fermions have spin 5 while bosons have an integer spin Both are cgiven in units of In Table 1 1 an overview of the fundamental fermions is given These can be divided into two categories the quarks and the leptons In addition these are grouped into three generations which differ only in mass To get a more detailed overview of the SM see 1 1 2 The Large Hadron Collider The LHC which is under construction at CERN in Geneva will take into operation end of 2008 Its accelerator ring has a circumference of about 27km and was integrated into the old LEP tunnel Fig 1 1 Large Hadron Co
88. oftp in2p3 fr cms Tsc tscO1 ps Philips Company The I C bus specification Version 2 1 2000 http www semiconductors philips com acrobat_download literature 9398 39340011 pdf O Militaru et al Slow Control for the Petal long term test 2005 http www fynu ucl ac be users o militaru SlowControl petaltest html F Beissel Cooli Cold Box Control Serial Interface Version 2 0 Aachen 2003 F Beissel DEPP Aachen 2001 http www physik rwth aachen de fileadmin user_upload www_physik Institute Inst_3B Forschung CMS Detektorentwicklung ARC DEPP_2 pdf http web physik rwth aachen de alinn Petalstatus_AachenIII_web htm W Beaumont Long term software Universiteit Antwerpen 2004 http www hep ua ac be cms testing software ltstruct ltswdocstruct4 php T Bergauer et al Petal Integration for the CMS Tracker End Caps CMS Note to be published R Bremer et al Integration of the End Cap TEC of the CMS Silicon Strip Tracker CMS Note to be published V Zhukov LtStruct test software User Manual Karlsruhe 2005 Layout is adjusted to plots from G Kaussen and M Zoeller 103 Bibliography Bibliography 152 M D Alfonso et al Validation tests of the CMS TIB TID structures CMS NOTE CERN 2008 53 personal communication with G Sguazzoni and D Abbaneo 54 personal communication with V Zhukov 104 Danksagung Dank sagen m chte ich Achim Stahl der es mir erm glicht hat die Doktorarbeit zu sc
89. on CMS Posters http cmsinfo cern ch Welcome html CMSdocuments CMSposters CMSposters_index html CMS Collaboration The Muon Project Technical Design Report CERN LHCC 97 32 CMS TDR 3 CERN 1997 E James Y Maravin N Neumeister Muon Identification in CMS CMS NOTE 2006 010 CERN 2006 CMS Collaboration The Hadron Calorimeter Project Technical Design Report CERN LHCC 97 31 CMS TDR 2 CERN 1997 CMS Collaboration The Electromagnetic Calorimeter Project Technical Design Report CERN LHCC 97 33 CMS TDR 4 CERN 1997 S Kasselmann Top Quark Mass Measurement in the Lepton Jets Channel using Full Simula tion of the CMS Detector PhD thesis RWTH Aachen 2007 CMS Collaboration The CMS experiment at the CERN LHC 2008 JINST 3 S08004 2008 CMS Collaboration The Tracker Project Technical Design Report CERN LHCC 98 6 CMS TDR 5 CERN 1998 101 Bibliography Bibliography 16 17 18 19 20 21 22 23 24 25 126 27 28 29 30 34 102 D Abbaneo The tracker layout http abbaneo home cern ch abbaneo cms layout whole html A Marchioro Communication and Control Unit ASIC for Embedded Slow Control 2002 http cmstrackercontrol web cern ch cmstrackercontrol documents Sandro CCU25Specs 20v2 1 pdf J Troska et al Prototype Analogue Optohybrids for the CMS Outer Barrel and Endcap Tracker Proceedings of the 6th Workshop on electronics for LHC Experimen
90. on Strip Tracker As described in the last chapter the silicon strip tracker consists of four subdetectors TIB TID TOB and TEC They are mounted into the tracker support tube TST The TST has a length of 5 5m and a diameter of 2 2m On the inner side of the tube a thermal shield is mounted The shield is necessary to guarantee the temperature difference of 30 C between tracker and ECAL Fig 2 1 shows a longitudinal cut through a quarter of the silicon strip tracker It shows the position of sensitive areas and gives information about the module type The strips of a module are orientated along the direction of z for TIB and TOB and in direction of r for TID and TEC The strip pitch is in the range of 80 um to 205 um On some positions a second module with a stereo angle of 100 mrad is mounted This provides a measurement of the second coordinate z in the barrel region and r on the disks These modules are called stereo modules the others are called normal modules A normal module and stereo module in combination is a double sided module Single sided modules are normal modules without a stereo module They have 512 readout channels double sided modules 768 channels each 01 02 03 04 05 06 07 14 15 m III IIIILET T a a 0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400 2600 2800 zimm single sided modules _ outer barrel TOB __ endcap TEC double sided modules L inner barrel TIB
91. on is done at the modules and not at the FED the timing run During the timing or time tune run each APV sends every 3200 ns a logical one tick This signal can be delayed within the TPLL in 24 steps of 1 04ns The FED reads every 25ns In Fig 4 10 and Fig 4 11 the result of such a run can be seen As the tick of one APV has a duration of 25ns and two ticks of two APVs are multiplexed together the duration of the tick in the plot is 50ns A good value for timing is in the center of the first 25ns of the plotted tick as this is the most stable position 32 It must be pointed out that each laser has its own timing run but the chosen delay is used per module In addition this test allows to measure the height of the tick hia and therefore to check if the optical connection is in a good state i e if enough light is transmitted from the lasers In case of a bad transmission tick height is less than 100 ADC counts the optical connections had to be checked and cleaned from dust If no tick above 100 ADC counts could be reached the AOH had to be exchanged Reasons for such a low tick are often a kink in the fragile fibers ADC counts 200 150 i fe ps A m h mi F r A p pl p p m ph pr n a ru AL r a L AA m m ol el ai oi p T u LoL Lf Li g fe ry uf Tu uf uf uf Lr Li Le Li u 1 rr r fr wur Li w Li UT TE TE Ts u war uf 100 20 D a 0 200 400 600 800 1000 1200
92. ow the benchmark rate of 1 and so a very promising result as the performance of the tracker and therefore physics analyses are not harmed by those defects 22 This lets us eagerly await the first physics run at LHC to see new physics TO Appendix A Fiber Mapping ribbon ch 1 2 3 4 5 6 7 8 9 10 ll 1 430 432 440 442 610 612 620 622 630 632 640 2 420 422 410 412 330 332 320 322 310 312 3 240 241 242 230 231 232 220 221 222 210 211 4 140 141 142 130 131 132 120 121 122 110 111 5 T50 752 740 742 540 541 542 530 531 532 730 6 120 722 820 921 922 510 511 512 710 712 Table A 1 Mapping of AOH Fibers to long ribbon cables given for a frontpetal 12 642 212 112 132 The First position in this table is 430 This means AOH on ring 4 position 3 fiber 0 has to be plugged on longribbon cable 1 channel 1 71 12 Appendix Fiber Mapping Appendix B K Mux Mapping KMUA Gack pelali tor a tron pelal change shartribbon 1 against 5 ehigroup To RE RE HAE sa MEET a a a an RE NE EY BEER CL EE 660 69 6 6 d 61 EEE Figure B 1 Mapping of long ribbon cables to K MUX The mapping is given for a back petal In row 0 column 7 the entry is 5 12 This means that channel 12 of longribbon 5 has to be plugged in slot 7 row 0 To switch from back to front petal the blue have to be exchanged against the green ones This can be done by changing just one whole ribbon which means one connection 13 74 Ap
93. pact parameter right and longitudinal impact parameter below 14 layer is mounted at a radius of 73mm In the phase of highest luminosity a third layer will be mounted at 102mm During the high luminosity phase it is expected that the innermost layer stays operational for at least two years and could be exchanged afterwards The end caps have a radius from 60 mm to 150 mm and are mounted at a distance of 34 5 cm and 46 5 cm to the vertex Thus they cover a region up to 7 2 4 Introduction Chapter 1 1 3 The CMS Experiment SE EEE SISTERS IT TEILEN III TI TRS JSS SIMO If ERR MW TI a IT ON RE MEANS TRS NN WERDEN K F SS DI MR RRRINNQN RIRE ARRAS MN N MY N SLR NA YY ERPS TRS ME SEHEN 7 7 IV RES WY HSL TPR SRA UE ORS AAAS My Y AN ESOS IN WL ANY PIN y Ha VINI N A AN 1 ll y N MUG 10 NA YY N 40 Myr W f 00 N MMM NG M ANN WY IIIA KY LRR LY OM RMS CRETE P ER EST LUURE J N IA nv x ys oy j FU X Ke RS SAW A A A 00001 NN HH ERIK EL E RSS ua SKK 7777 Inner Barrel TIB Vi Outer Barrel TOB 7772 nner Disks TID Vii Endcap TEC a H N 77 2 Pixel Barrel TPB 7777 Pixel Endcap TPE The different regions of the tracker are marked in different Figure 1 7 The pixel detector 15 Figure 1 6 The CMS tracker colours 8 Chapter 2 The Silic
94. passive cooling The petal is placed inside the fridge On the right side the rack with all electronics and the PCs for data recording are placed converts the LVTTL into LVDS In addition to this conversion the adapter card was designed to filter noise from the power supply used for the control ring on the petal Via a shielded twisted pair cable the CCUs on the petal get these commands translate them into the PC protocol and route them to the different modules and AOHs The response of the components returns on the same path to the computer see also Fig 4 3 4 3 Readout In case of a positive trigger decision the FEC sends the readout signal to the modules Trigger and clock are generated inside the TSC a PCI computer card steered by the DAQ software The modules send their data to the AOHs where it is converted into analog optical signals The next step in the readout chain is the OFED housing the OECs Here the analog optical signal is transformed back into an analog electrical signal In total there are six OEC one for each multi ribbon cable In principle it would now be possible to go directly to the FED a PCI card within the computer This card converts the analog signals to digital signals with Low Voltage Differential Signal I C stands for Inter Integrated Circuit and is a multi master serial computer bus invented by Philips 42 Trigger Sequenzer Card 41 Optical Front End Driver a VME crate card develop
95. peaktime on Ring 6 x ndf 10 8 11 n Constant 17 6 20 Mean 15 6 z Sigma 1 08 S 18 sjauuey gt gzL 10 8 6 CGS peak time ns Figure F 3 Calibration group subtracted peaktime for PA S opens in ring 5 left and ring 6 right CGS peaktime on Ring 7 ao O1 amp channels 25 20 2 ndf Constant 19 7 12 23 6 Mean Sigma 10 8 6 17 5 1 02 sjauuey gt 894 CGS peak time ns CGS peak time 1 Mean RMS from CGS peak time for PA S open per ring 20 15 offset 2 0 1 3 slope 0 8 0 095 104 8 10 12 14 16 Striplength cm b 18 20 Figure F 4 a Calibration group subtracted peaktime for PA S opens in ring7 b Linear fit through the calibration group subtracted peaktime per ring 92 Glossary SM Standard Model LHC Large Hadron Collider CERN Conseil Europe n pour la Recherche Nucl aire LEP Large Electron Positron collider ALICE A Large Ion Collider Experiment ATLAS A Toroidal LHC ApparatuS CMS Compact Muon Solenoid LHC b Large Hadron Collider beauty Experiment HCAL Hadron Calorimeter ECAL Electromagnetic Calorimeter TIB Tracker Inner Barrel TID Tracker Inner Disc TOB Tracker Outer Barrel TEC Tracker End Cap TPB Tracker Pixel Barrel TPE Tracker Pixel End cap TST Tracker Support Tube CFC Carbon Fiber Composit ICB Inter Connect Board CCU C
96. pendix B K Mux Mapping Appendix C Scenario File Each line of the scenario file represents one command for the DAQ software and each com mand can be split into four parts The first part is the information when the command should be executed positive numers are absolute times in second negative values represent a relative time to wait e g 1 means to be executed at the second 1 of the longterm test and 600 means wait 600 seconds after finishing the last test The second part of the line is the main command which should be executed by the DAQ software In addition to this command two parameters can be transferred These are the third and fourth part of the command In most cases this are just dummy text A complete documentation of the commands can be found in 50 and 48 1 Start longterm test 2 SetDt 600000 _noslctrl 2 ChangeHV 0 _hv0 2 I2cDump dummy dummy 2 PiaReset pllinit dummy 2 ChangeCool 16 _t16 2 ChangeCool 16 _t16 60 TempReached 16 _tr 600 I2ecDump dummy dummy 2 CheckEnv dummy _d 60 OptoScanRun i2cpeak fast 2 I2cDump dummy dummy 2 CheckEnv dummy _d 60 I2cDump dummy dummy 2 TimeTuneRun i2cpeak _ts 2 PedRun i2cpeakinv _ppi 2 ConnectivityRun i2cpeakinv dummy 2 PedRun i2cpeakinv _ppi 2 I2cDump dummy dummy 2 CalRun i2cpeakinv _cpi 2 I2cDump dummy dummy 2 CheckEnv dummy _d 60 I2cDump dummy dummy 2 SaveRec 1 no 2 PedRun i2cpeak _pp 2 I2cDump dummy dummy 2 SaveRec 1 no
97. results see Chap 3 This distribution allows to distinguish between the different defect types In Fig 6 9 the same plot is split up into the five classification types 94 Chapter 6 Analysis of LT measurements 6 2 Calibration pulse test defect declaration peak height ADC SJOUUELP 9993 saturated and short 15 10 5 0 5 10 15 peak time ns other centers BE PAS open BE S Sopen noisy channel I saturated channel short Figure 6 8 Peak height versus peak time Only channels identified as defective by ARC and LT are taken into account Categories are defiened as in the ARC test er mostime us pe hen ADC Table 6 1 Definition of the four regions PA S open S S open noisy channel and satu rated channel and short All channels flagged as PA S open or likely PA S open in the ARC test are marked in red upper left channels with S S open likely S S open and mid sensor open flag are marked in blue upper right In yellow those channels are marked which are noisy middle left Shorts are marked in light blue middle right and saturated channels in purple lower one In the peak height peak time plane the different types split up into four areas Tab 6 1 The plots and results for the CERN center are shown at the end of this section I 6 2 Calibration pulse test defect declaration peak height ADC
98. roken carbon fibre frames The defects that occur most are touched bonds After a training and learning phase this defect type could be avoided For the AOHs the defect classes are no signal most probably a broken fibre I2C problems damage found during petal insertion TIC and unknown For the unknown one it is only known that this AOH was exchanged The full list and description can be found in 46 Table 5 4 Amount of exchanged components in Aachen 46 touched bonds I2C TIC preperation for LT A 5 Table 5 5 Reasons for module exchange 46 gt ES E BC TIC E gt Table 5 6 Reasons for AOH exchange 46 5 3 Petal design During the LT test phase it was discovered that in a few cases the communication with some AOHs was lost during the cold phase but working again during the warm phase This 47 5 3 Petal design Chapter 5 Problems found with long term test behaviour was reproducible and occured for each AOH at it s individual temperature A deeper investigation showed that removing the screw with which the AOH is mounted to the petal body solves this problem The connection of the AOH to the petal can be seen in Fig 5 1 As removing the screw solves this problem it is assumed that the real problem is mechanical stress between the ICB and the AOH inside the connector coe Figure 5 1 AOH on an ICB Marked are the screw which connects the AOH to the petal body and the connector with which
99. rt As an example the cut values for Ring 1 can be found in Tab 3 2 31 3 3 Defect types Chapter 3 Single module test Ring 1 very very very low low high very high low cut cut cut LC_3 LC_1 HC1 noise ADC dec inv on dec inv off peak inv on peak inv off peak height dec inv on dec inv off peak inv on peak inv off peak time ns dec inv on dec inv off peak inv on peak inv off Table 3 2 Cut values for the fault identification on TEC ring 1 modules 23 32 Chapter 4 Long term test After the assembly of the modules to the petals a test was performed This test was designed to inspect the petal before installing it into the TEC and to ensure that all components are in a good state this implies that an optical inspection shows no mechanical problems and that the petal is electrical fully functional Special attention was paid to the simulation of LHC running conditions This includes the simultaneous readout of all modules and an ambient temperature of 20 C as this temperature was supposed to ensure a silicon temperature below 10 C the temperature at which the sensors should be operated to keep the impact of radiation damage small Several cooling cycles should give information if the petal could withstand ten years LHC in which it will be cooled down often The test was further designed to find single strip defects and to ensure that this amount is below 1 of all strips In particul
100. rt of the CMS detector is the tracking system that is composed of a silicon pixel detector forming the innermost part surrounded by silicon strip sensors Currently it is the largest silicon detector in the world with an active area of 198 m The strip tracker itself consists of four subdetectors One of these are the tracker end caps TEC with an active area of 82 m Besides this large aperture their position in the forward region plays a key role for physics analysis due to the fact that many of the interesting events are expected to be boosted in the forward region pp collider This area splits up into 10 288 sensors with 3 988 765 channels in total In several steps the modules constructed and tested before being mounted onto the final substructures petals An important longterm test has been performed which qualifies the petals to be installed into the detector The focus of the present work is in the longterm test The test procedure is described A method for identification and classification of defect channels is presented This method has been developed based on the test results of a previous test ARC test which has examined each module before the assembly onto the petals A cross check has been performed to compare the results with data from a subsequent test sector test that is performed after the petals have been integrated into the TEC A good agreement shows the consistency of the presented results With the help
101. s 50 100 150 200 250 300 time ns b Figure 3 9 a Microscopic view of two connected strips 23 b Calibration pulse of a short strip Ihe peak height is very low More than two strips are connected For comparison see Fig 3 2 tests not explained in this thesis as not necessary to understand the final strip flagging For each test and module geometry cuts are defined Depending on the test result a classification of the strip is performed This classification is done by the fault finding algorithm presented in 23 From the eight tests the following quantities were used for this classification common mode subtracted noise relative height of the calibration pulse to the APV median and absolute deviation of the calibration pulse peak time to the median of the calibration group Every quantity is cutted into different regions e g the peaktime is splitted into five regions LC3 lt LC2 lt LC_1 lt range of normal channels lt HC_1 3 9 where LC_x denotes low cuts and HC_x denotes high cuts Each measurement gets a corre sponding flag The results of the quantities in the different APV modes are combined in a logical OR see Tab 3 1 Example for the flag combination d dec peak peak combined inv on inv off inv on inv ofl flag yes below LO no no no yes yes ec no no between LC_3 and LC_2 y n yes between LC 2 and TCI O O O n Table 3 1 Example of the combination of flag
102. s hi CIO SAT sd Ken ere 4 7 3 Extended l V run 5 Problems found with long term test 5 1 LT test setup 5 2 Defective components D21 P tal grading sos nn a a a been 5 2 2 List of exchanged components and their defects ge I eee M ba rn HE eher 6 Analysis of LT measurements 6 1 Pedestal test defect detection 6 1 1 Defect rate 6 12 Reproducibility 4642 sa eee e ka weeds 6 2 Calibration pulse test defect declaration 6 2 1 Normalization of the discriminating variables 6 2 2 Classification of defect types 6 3 Comparison between ARC and LT test 2 222202 6 4 Comparison between LT and sector test 22 2 2 2 N Summary Fiber Mapping K Mux Mapping Scenario File Noise distribution Number of failed noise tests JT 0 Q U p Calibration group subtracted peaktime Glossary Bibliography vill Contents 69 71 73 75 79 89 91 93 101 Chapter 1 Introduction Dass ich erkenne was die Welt im Innersten zusammen h lt J W v Goethe 1808 This phrase written by Johann Wolfgang von Goethe describes the human desire to under stand the underlying mechanism of nature Since centuries mankind is trying to investigate matter Therefore they splits it int
103. s for the fault finding algorithm Though only one cut can be set for a single APV mode several flags can be set in the combined flag 23 30 Chapter 3 Single module test 3 3 Defect types These combined flags are transfered to the analysis procedure of the algorithm The analysis is performed in several steps A short summary of the analysis is given here e The channel is set to unknown faulty if it is flagged in only one test e If the noise of a channel APV edge channel passes the HC_1 cut HC_2 cut the channel is set to noisy Module edge channels will not be marked e A noise below LC_l combined with a peak height below 10 ADC counts indicates a saturated channel e A noise below LC_1 in addition to a LC_1 flag of the peaktime indicates an open between two Sensors e If the LC_2 flag of the peaktime is set and one of the low noise flags the channel is marked as an open between pitch adapter and sensor e If one of the high noise flags is set and one of the low peak time flags the channel is marked as mid sensor open e If the LC_1 and LC_2 flag of the peaktime are set and in addition the low noise flag it indicates an open The position of this open is determined to be a PA S and S S open which couldn t be realized hence the flag open with conflicting location results is set e A low or high noise in one of two neighboring channels in addition to a LC_1 flag of the pulse height in both channels leads to the flag sho
104. s of petals The modules are mounted on both sides of a petal to guarantee a complete coverage with silicon Fig 2 8 The precision with which the modules are positioned on the petal is in the order of 20 um 14 wa oe Li gz p LE RR ps d Figure 2 8 Photograph of front and back side of a TEC front petal with seven rings 14 The body of the petal is made of 10mm NOMEX in a honeycomb structure within 0 4mm Carbon Fiber Composite CFC skins Fig 2 9 A cooling pipe is integrated within the body Fig 2 10 The whole system is optimized for a good heat flow with low material budget On the petal body the interconnect board ICB 19 is mounted It sends communication signals via the CCUs and also voltages ground Low Voltage with 1 25V and 2 5 V and Flame resistant meta aramid material developed by DuPont http www dupont com Communication and Control Unit 17 13 2 4 TEC Chapter 2 The Silicon Strip Tracker High Voltage with up to 500 V to the different modules Readout signals of the modules are transported to the analogue opto hybrids AOH 18 also mounted onto the ICB and from there via optical fibers to the read out system Fig 2 11 The read out link is optical to minimize cross talk More information can be found in chapter 4 and also in 14 22 L F Fr 1 bi ry Li TY I st L F T i F Y 4 LTI II Y I Ci LE J wi un Tr Ay Se x 23 F IT5E ER Er rt i
105. schleuniger Die Endkap pen setzen sich wiederum aus 10 288 Sensoren mit insgesamt 3 988 765 Kan len zusammen Die Module wurden in zahlreichen Arbeitsschritten aufgebaut und mehrfach getestet bis sie schlie lich auf den finalen Substrukturen den sogenannten Petals integriert und einem inten siven Langzeittest unterzogen wurden welcher diese f r den Einbau in den Detektor quali fizierte Das Hauptaugenmerk dieser Arbeit liegt dabei auf dem Langzeittest Die dabei verwen dete Testprozedur wird beschrieben Ferner wird eine Methode zur Fehlererkennung und deklaration vorgestellt Diese wurde mit Hilfe der Ergebnisse fr herer Tests ARC Test die an jedem Modul vor der Integration durchgef hrt wurden entwickelt Ein Vergleich mit den Ergebnissen eines Tests nach dem Einbau der Petals in die TEC wurde als Gegenprobe durchgef hrt Sektor Test Eine hohe Ubereinstimmung zeigt die Konsistenz der gezeigten Ergebnisse Mit Hilfe der Methode konnte eine Kanalfehlerrate von etwa 0 9 bestimmt werden Weitere Defekte wie tote Komponenten welche nach der Integration der Petals in die TEC gefunden wurden erh ht die Zahl der nicht verwendbaren Kan le auf 3 3 o 111 1V Abstract The Large Hadron Collider LHC at the European Organization for Nuclear Research CERN in Geneva will start end of 2008 One of the experiments at the LHC is the multipurpose detec tor CMS Compact Muon Solenoid A key pa
106. se All channels are included 81 D 3 Noise distributions taken in peak mode The noise is normalised to the average module wise First row is first warm phase second row in cold phase and third row again in warm phase APV edge strips are excluded 82 D 4 Noise distributions taken in deconvolution mode The noise is normalised to the average module wise First row is first warm phase second row in cold phase and third row again in warm phase APV edge strips are excluded 83 D 5 Noise distributions taken in peak mode The noise is normalised to the average module wise First row is first warm phase second row in cold phase and third row again in warm phase APV edge strips are excluded Only petals tested at CERN are taken into account 84 D 6 Noise distributions taken in deconvolution mode The noise is normalised to the average module wise First row is first warm phase second row in cold phase and third row again in warm phase APV edge strips are excluded Only petals tested at CERN are taken into account 85 D 7 Noise distributions taken in peak mode The noise is normalised to the average module wise First row is first warm phase second row in cold phase and third row again in warm phase APV edge strips and petals tested at CERN ato excluded ee en 86 D 8 Noise distributions taken in deconvolution mode The noise is normalised to the average module wise F
107. se and third row again in warm phase APV edge strips are excluded 83 Appendix D Noise distribution excluding APV edge strips excluding APV edge strips 1281 modules with 6120 apv s on 54 petals 1281 modules with 6120 apv s on 54 petals 2 2 2 2 E 5 210 810 bs si 10 0 30 flagged by LT amp 104 u 0 37 flagged by LT amp z 4 4 10 A 10 3 Oo o 10 10 10 L LA aM EEE 1 E 1 0 0 5 1 1 5 2 2 5 3 3 5 4 4 5 5 0 0 5 1 1 5 2 2 5 3 3 5 4 4 5 5 CMSnoise in peak inv off CMSnoise in peak inv on excluding APV edge strips excluding APV edge strips 1281 modules with 6120 apv s on 54 petals 1281 modules with 6120 apv s on 54 petals 1 15 flagged by LT u 0 55 flagged by LT channels u 3 1 channels s o 10 y So P sjauueyd 088857 10 sjouuey gt 088852 10 10 10 10 10 2 2 5 3 3 5 4 4 5 5 0 0 5 1 1 5 2 2 5 3 3 5 4 4 5 5 CMSnoise in peak inv off CMSnoise in peak inv on excluding APV edge strips excluding APV edge strips 1281 modules with 6120 apv s on 54 petals 1281 modules with 6120 apv s on 54 petals 2 2 2 2 C 5 10 810 O S 4 10 0 35 flagged by LT amp 10 u 0 38 flagged by LT S 7 g 5 S 3 3 3 5 10 2 10 2 7 7 10 10 10 10 l eto L Robe EN 0 0 5 1 1 5 2 2 5 3 3 5 4 4 5 5 0 0 5 1 1 5 2 5 3 3 5 4 4 5 5 CMSnoise in peak
108. see Fig 3 2 30 List of Figures List of Figures 4 1 4 2 4 3 4 4 4 9 4 6 4 7 4 8 4 9 4 10 4 11 4 12 4 13 9 1 6 1 6 2 6 3 6 4 6 5 6 6 6 7 6 8 The long term test setup at Aachen Left the cooling plant for active cooling of a petal middle the fridge for passive cooling The petal is placed inside the fridge On the right side the rack with all electronics and the PCs for data recording are placed eh ee HR eee eee ee oe 34 Schematic view of a longterm test station 39 Schematic view of the communication and readout of a petal 36 Photograph of a full cabled K MUX 37 Screen shot of the slow control software On this tab the monitoring of the temperature is shown 43 38 Screen shot of the slow control software On this tab the interlock status is RO ee ee A ee ee 38 Screen shot of the DAQ software On this tab the manual steering of the test ONO Ts bp ne RETR EOD Ow EE ew Eee EEE 39 Temperature profile of a long term test with three cold phases 40 Structure of a long term test scenario 48 40 Time tune run for one Laser 41 Screen shot of DAQ On this tab the result of a time tune for a ring 3 module PP 0 E E EERE EEEN EEE EE ee 42 Opto scan run for one APV in gain3 Left logical zero right logical one 42
109. t DRenn 3 1 n 1 1 C Magro 39 i gt Din ER 3 2 chegroup gt DRenn Cure Po N RM San 3 3 The RMS depends on the strip length and the quality of the readout system Due to the fact that the readout system and strip length is the same for all strips of a module the 1 APV Readout Control 23 3 1 Pedestal test Chapter 3 Single module test noise varies only slightly from strip to strip especially in the range of one APV In case of a possible defect the capacitance of the strip changes and so the measured noise changes So the information about the noise helps to understand the quality of the detector An exception is the behavior of edge strips of the APV Here the noise is always higher The reasons therefore are not completely understood but the effect can be reduced by a improved grounding scheme 14 23 The measured signal Sen n of channel ch in event n is defined as Senn DRenn Pen 2 P aL c 14 gt 5 L 8 8 F g oF g 200 a 12 z A 8 150 E 3 5 08 100 0 6 04 50 E 0 2 C L L 1 1 1 1 1 1 1 1 1 1 L 1 i l l l L L l 1 1 1 1 1 1 1 1 1 1 1 1 l 100 200 300 400 500 100 200 300 400 500 channel number channel number Figure 3 1 Pedestal and common mode subtracted CMS noise of a ring 3 module Obviously the noise of the APV edge channels 1 128 256 384 is higher The noise of channel 94 indicates a str
110. t all petals have been assembled into both TECs A pedestal run in peak and deconvolution mode has been performed 30 In this section the data taken with TEC in deconvolution mode is compared to the LT test data SAPV channels 1 2 127 and 128 63 6 4 Comparison between LI and sector test Chapter 6 Analysis of LT measurements Total Number of Channels 3975680 Total Number of Channels 3851440 gt 72 2 3000 3000 3 3 A A N F D gt g D 7 5 2000 gt 5 2000 ey Ga g gt 7 E 1000 E 1000 3 2 468 42 amp 0 0 only LT LT ARC only ARC only LT LT ARC only ARC a with APV edge channels b without APV edge channels Figure 6 19 Number of channels flagged during LT and ARC test split into channels flagged only by LT and ARC or by both test systems 51 In Fig 6 20 the noise distribution of the TEC petals in deconvolution mode can be found It is split into the rings 1 to 4 and 5 to 7 The red histogram indicates strips which are Ring1234 cmsnoise_peddec_lengthscaled Ring567 cmsnoise_peddec_lengthscaled 10 Number of Strips Number of Strips 0 500 1000 1500 2000 2500 3000 3500 4000 0 500 1000 1500 2000 2500 3000 3500 4000 Noise After CM Subtraction Noise After CM Subtraction a rings 1 4 b rings 5 7 Figure 6 20 Scaled noise distribution measured in TEC Channels flagged by the LI test are marked in red APV edge channels and bad APVs are excluded 30
111. t results show an unexpected enhancement of the edge channels The fact that the ARC setup does not combine several tests to classify a defect but uses just the worst test of all could be an explanation for the increased rate A detailed look into the ARC data shows Fig 6 16 that the surplus of defective strips mainly has the flag short and affects mostly the APV channels 1 and 2 or rather module channels 1 and 2 not shown in this plot To explain this behaviour in Fig 6 17 the peak height distribution of one module is shown as an example The first two channels of this module are flagged as short This distribution obviously shows that the defect is not a problem of the strip but a problem of the classification scheme 23 The reference of the classification scheme is a flat distribution of the peak height for each APV A channel is flagged if the deviation from the median exceeds a certain value But the peak height distribution versus the channel number is not always flat for an APV and even not linear Therefore the channels 1 and 2 are flagged more often than the other channels as the deviation to the expected reference value is largest for those channels even larger than for channel 129 and 130 The classification scheme of LT is different as it needs more than one test to declare a channel to be defective and uses only noise tests 60 Chapter 6 Analysis of LT measurements 6 3 Comparison between ARC an
112. t will be possible to determine the bunch crossing and to get a good trigger decision 9 1 3 2 The Hadron Calorimeter The hadron calorimeter HCAL 11 is the outermost component inside the solenoid It is divided into two parts The central calorimeter including barrel and end cap region reaches up to n 3 0 and the forward calorimeter with n lt 5 0 The HCAL measures the energy and direction of hadrons Through its large hermetic coverage it is also possible to get a hint for neutrinos This can be done by measuring the missing transverse energy The HCAL also helps to identify electrons photons and muons together with the tracker electromagnetic calorimeter and muon system Chapter 1 Introduction 1 3 The CMS Experiment eta 0 8 RPC 1 04 oe no oe oe no s cos no os 5 CODE ro 5 er o E Bie 0 200 400 600 800 1000 1200 Z cm Figure 1 3 A quarter of the muon system The different technologies are labeled and coloured 10 Barrel and end cap calorimeter are sampling calorimeters with brass alloy absorber plates which are interleaved with scintillator sheets The first and last absorber plate of the barrel region are made of stainless steel As the barrel HCAL inside the coil is not sufficiently thick to absorb all t
113. tes A deeper investigation with the electronics engineer showed that the FEC adapter card used by all test centers was misdesigned Ground and hence the reset line were not under control This leads to the fact that the communication with the petal was sometimes possible but very instable After correction of the adpater card it was possible to establish a stable communication During the LT test phase several software problems were found like tick marks which were not saved incorrect leak measurement or slow test routines Most of those problems lead to an reinstallation of the software As we updated our software regulary we had the role of a beta tester and run often into such software problems Obvious problems were reported directly to the responsible persons and corrected 5 2 Defective components Out of 297 petals 51 were tested in Aachen Due to a design problem not enough clearance between ICB and sensor some of those petals were rebuilt later and retested mainly in Strass bourg 22 In addition petals which were disassembled from the TEC at CERN were retested mechanican D Jahn and electronics engineer F Beissel with the help of Th Hermanns and W Beaumont 3the card wich translates LVTTL into LVDS 45 5 2 Defective components Chapter 5 Problems found with long term test later at CERN petals exchanged from the TEC in Aachen were retested in Aachen Therefore Aachen did the final test for 32 petals
114. the AOH is plugged to the ICB The AOH is connected to the ICB and to the petal body Petal body and ICB have different thermal expansion coefficients Cooling the petal causes a shear force inside the connector and hence increases the chance to loose the electrical connection between AOH and ICB This problem was found approximately ten times and could always be solved by removing the screw Usually a redesign of the petal mechanics would have been necessary but as this problem was found at the end of the production phase it was not possible to reprocess all petals and even not to remove all screws In addition removing the screws introduces the risc that an AOH gets completly loose and scratches over the modules The amount of AOHs which will loss their communication due to this problem during normal operation of the tracker is hard to predict as it is unknown at which temperature exactly the tracker will be operated and how the temperature distribution inside the tracker will be Hence it was decided to remove the screws only if the problem was found In later tests of the TECs this problem was found for about ten AOHs again Until today it is unclear how large the impact of lost AOHs due to the shearing forces will be 48 Chapter 6 Analys s of LT measurements This chapter gives a detailed analysis of the measurements of the LT test Therefore the LT test data of all petals are used A strategy to find defects is developed using the ARC
115. the environment of the system It measures the temperature of the cooling plant the fridge and of the sensors mounted in the fridge and on the cooling pipes Further it measures currents and voltages of LV and HV circuits every 30 seconds It controls the voltages and temperature of the cooling plant and the fridge During the long term test the slow control was controlled remotely by the DAQ software Nevertheless the slow control is able to take precautions if the dew point or the currents are to high In this case the slow control initiates a shutdown of the system The same happens if the readout fails for more than 2 minutes or if the fridge door is opened In Figs 4 5 and 4 6 two screen shots of the slow control software are shown As every PIC uses slightly different power and monitoring systems an adaption to the locally circumstances has to be done by each PIC itself In case of the PIC Aachen the cooling of the fridge and the read out of the temperature probes have to be adapted to the software Both steps were performed with help of the Cooli 44 In addition the controlling and monitoring of the HV provided by depp boards 45 have to be implemented 4 6 DAQ The DAQ software 47 is the control center for the test It controls the readout of the petal and the tests which should be performed It sends commands to the slow control to change 37 4 6 DAQ Chapter 4 Long term test i main_panel vi File Edit Operate Tools Browse Win
116. thickness of 320 um for thin and 500 um for thick sensors The sensor consists of n doped bulk material with p type strip implants on the front side Fig 2 14 Single sided modules have 512 strips and double sided 768 strips each The p strips are AC coupled to aluminum strips A separation of both is given by multiple layers of SiO and SiN providing the dielectric for the capacitors made of each pair of p and aluminum strip The width of the aluminum strips is about 15 larger compared to the width of the p implants to avoid high fields at the edges of the p implants The width of the implanted strips depends on the strip pitch a constant width pitch 5 of 0 25 is used for all sensor geometries The total capacitance of a strip per unit length Crot depends on the interstrip capacitance and the backplane capacitance In a range of 300 um to 500 um silicon thickness Crot can be parameterized as Crop 0 8 1 7 pF cm 2 1 P Iy 2 4 TEC Chapter 2 The Silicon Strip Tracker wire bond bias ring guard ring bias resistor DC pad AC pad aluminium strip 400 VDC oxide thin layers of SiO and Si N n layer p implants below 7 a p strips bias and e a p implants below w p guard ring bias and guard ring aluminium backplane Figure 2 14 Schematic design of one corner of a silicon strip sensor It is worth mentioning that Cto in this range is independent
117. this constraint for front and back petals at the same time Interchanging two complete ribbons solves this problem In this case just one 9 Analog Digital Converter 10K arlsruhe Multiplexer 39 4 4 Cooling Chapter 4 Long term test BE Control signals electrical Data P optical Data LV Low Voltage HV High Voltage Figure 4 3 Schematic view of the communication and readout of a petal connection has to be redone as it is possible to interchange two ribbons directly Details on the final mapping can be found in App B 4 4 Cooling To cool down the petal a combination of an active and a passive cooling system is used For the latter a fridge which allows to cool down the ambient temperature to 30 C is used The fridge is equipped with an input for dry air This allows to dry the air to a dew point of 40 C A system of five temperature and two humidity sensors allows to monitor the atmosphere inside the fridge The active cooling is performed by a cooling plant It circulates the cooling fluid CgF 4 through the petal and cools down the fluid to 25 C Two temperature probes 36 Chapter 4 Long term test 4 5 Slow control Figure 4 4 Photograph of a full cabled K MUX glued to the cooling tube monitor the input and output temperature of the fluid The cooling plant was designed to run the system between 17 C and 25 C 4 5 Slow control The slow control 43 steers
118. trip is disconnected from the readout system 2R 3 3 Defect types Chapter 3 Single module test T peak height E gy 35 3 35 S gt 8 8 g a 20 2 25 25 2 2 20 20 peak time l 1 1 L l 1 1 1 L L 1 1 I L L 1 1 1 j Ji 1 250 300 50 100 150 200 250 300 time ns time ns a PA S open b faultless I L 1 1 J 1 1 I 1 L 1 50 100 150 200 Figure 3 5 Calibration profile of a PA S open a and a faultless channel b The PA S open has compared to the faultless channel an earlier peak time and a raised peak height po ri Mil Fa MERE 4 j T nr a pui os IR qu EE Pi Fi f uk E i m si beanie LILEL ER AIR I cited URIN Mae Figure 3 6 Photograph of touched and destroyed bonds between two sensors Figure 3 7 Microscopic view of a scratch on the sensor surface 29 28 Chapter 3 Single module test 3 3 Defect types 3 3 2 Saturated channel A strip affected by this defect shows always a saturated signal height The origin is a defective chip on the hybrid No traversing particle can be measured by this channel Due to the fact that the channel is always saturated no variation can be measured Furthermore the peak during the CalProf test is very small Consequently noise and peak height are very low Fig 3 8 height ADC counts l 1 1 l 1 1 1 1 1
119. ts Stockholm Sweden 2001 http accms04 physik rwth aachen de cms Tracker Electronics L Borello et al Sensor Design for the CMS Silicon Tracker CMS Note 2003 020 2003 V Zhukov and W Beaumont Qualification tests of Silicon strip detectors during the mass production CMS IN 2003 41 CERN 2003 A Linn PhD Thesis to be published T Franke Development and Evaluation of a Test System for the Quality Assurance during the Mass Production of Silicon Microstrip Detector Modules for the CMS Experiment PhD thesis RWTH Aachen 2005 E Gatti and P F Manfredi Processing the signals from solid state detectors in elementary particle physics Riv Nuovo Cim 9N1 1 146 1986 M Friedel The CMS Silicon Strip Tracker and it electronic Readout PhD thesis Institute of High Energy Physics Austrian Academy of Sciences 2001 W R Leo Techniques for Nuclear and Particle Physics Experiments Springer Verlag 1994 G Lutz Semiconductor Radiation Detectors Springer Verlag 1999 S Gadomski et al The deconvolution method of fast pulse shaping at hadron colliders Nuclear Instruments and Methods A 320 pages 217 227 1992 M Poettgens Development and Evaluation of a Test Station for the Quality Assurance of the Silicon Microstrip Detector Modules for the CMS Experiment PhD thesis RWTH Aachen 2007 R Brauer Integration of the End Cap TEC of the CMS Silicon Strip Tracker PhD thesis RWTH Aachen 2008 R Brauer Fram
120. urther comparison of ARC and LT test leads to a defect classification algorithm based on a simple cut strategy For those strips that have been noticed defect by both ARC and LT setup the classification indicates in 83 the same kind of defect The combination of the defect identification and classification algorithm shows an efficiency to classify a defect found by ARC with the same defect with LT to 76 The combination of defects seen by LT and ARC test yields in a defect rate of 0 09 Shells and discs of TIB and TID have a single strip defect rate of 0 07 and the rods of TOB a defect rate of 0 13 53 52 So the result of the LT test is in the same range Results have been cross checked with the results of TEC sector test data to verify and to validate the presented method 95 of the strips that are declared as defect based on the LT test results have also been identified as defect based on the sector test data About 0 002 strips have been noticed as defect based on the sector test data and thus not flagged before This indicates the high accuracy of the presented algorithms A further 0 23 of all APVs have been excluded from the analysis of the sector test data due to defects which are supposed to be caused during the integration of the petals into the TEC 30 The combination of ARC LT and sector test results of TEC yields a final defect rate of 0 33 The determined defect rate of 0 33 for strips inside TEC is far bel

Download Pdf Manuals

Related Search

Related Contents

MODEL #500 OWNERS MANUAL  Manual-alarma auto- Tomahawk-TW-9010    LE SYNDICAT CANADIEN DE LA FONCTION PUBLIQUE (SCFP  

Copyright © All rights reserved. Failed to retrieve file