Large scale test of wedge shaped micro strip gas counters

*Corresponding author. Tel.:#32 2 629 3208; fax:#32 2 6293816.

E-mail address: [email protected] (C. Vander Velde)1On leave of absence from Universita di Genova and Sezione

INFN, Genova, Italy.

2Research Director FWO.3Aspirant FNRS, presently at IRES, Strasbourg.4On leave of absence from Moscow State University.

Nuclear Instruments and Methods in Physics Research A 436 (1999) 313}325

Large scale test of wedge shaped micro strip gas counters

M. Ackermann!, S. Atz!, V. Aulchenko", S. Bachmann#, B. Baiboussinov",S. Barthe$, W. Beaumont%, T. Beckers%, F. Bei{el#, Y. Benhammou&,

A.M. Bergdolt$, K. Bernier', P. BluK m!, A. Bondar", O. Bouhali), I. Boulogne*,M. Bozzo+,1, J.M. Brom$, C. Camps#, V. Chorowicz,, J. Co$n$, V. Commichau#,

D. Contardo,, J. Croix$, J. De Troy%, F. Drouhin&, H. EberleH $, G. FluK gge#,J.-C. Fontaine&, W. Geist$, U. Goerlach$, K. Gundl"nger#, K. Hangarter#,

R. Haroutunian,, J.M. Helleboid$, Th. Henkes$, M. Ho!er$, C. Ho!man$, D. Huss$,R. Ischebeck#, F. Jeanneau&, P. Juillot$, S. Junghans!, M.R. Kapp$, K. KaK rcher!,D. Knoblauch!, M. KraK ber!, M. Krauth$, J. Kremp#, A. Lounis$, K. LuK belsmeyer#,C. Maazouzi$, D. Macke#, R. Metri!, L. Mirabito,, Th. MuK ller!, V. Nagaslaev",

D. Neuberger!, A. Nowack#, A. Pallares!, D. Pandoulas#, M. Petertill#, O. Pooth#,C. Racca$, I. Ripp$, E. Ruo! !, A. Sauer!, P. Schmitz#, R. Schulte#,

A. Schultz von Dratzig#, J.P. Schunk$, G. Schuster$, B. Schwaller&, L. Shektman",R. Siedling#, M.H. Sigward$, H.J. Simonis!, G. Smadja,, J. Stefanescu),

H. Szczesny#, A. Tatarinov", W.H. ThuK mmel!, S. Tissot,, V. Titov", T. Todorov$,M. Tonutti#, F. Udo), C. Vander Velde),*, W. Van Doninck),2, Ch. Van Dyck%,

P. Vanlaer),3, L. Van Lancker), P.G. Verdini-, S. Weseler!,B. Wittmer#, R. Wortmann$, A. Zghiche$, V. Zhukov%,4

!IEKP Universita( t Karlsruhe, Postfach 3640, D-76021 Karlsruhe, Germany"BINP, RU - 630090 Novosibirsk, Russian Federation

#RWTH Aachen I & III, Sommerfeldstrasse 26-28, D-52056 Aachen, Germany$IReS, 23 rue du Loess, BP28-67037 Strasbourg Cedex 2, France%UIA Antwerpen, Universiteitsplein 1, B-2610 Wilrijk, Belgium&GRPHE-UHA, 61 r. Albert Camus, F-68093 Mulhouse, France

'UCL, Chemin du cyclotron 2, B-1348 Louvain-la-Neuve, Belgium)IIHE-CPB 230, ULB-VUB, Boulevard de la plaine 2, B-1050 Brussels, Belgium

*UMH, av. Maistriau 19, B-7000 Mons, Belgium+CERN, CH1211 Geneva 23, Switzerland

,IPN Lyon, 43 Bd du 11 novembre 1918, F-69622 Villeurbanne Cedex, France-INFN, via Livornese, 582/A, San Piero a Grado, I-56010 Pisa, Italie

Received 8 April 1999

0168-9002/99/$ - see front matter ( 1999 Elsevier Science B.V. All rights reserved.PII: S 0 1 6 8 - 9 0 0 2 ( 9 9 ) 0 0 5 1 4 - 8

Fig. 1. A detector's cell showing the anode and cathode ge-ometry for the wedge shaped counters. P is the anode pitch andG the anode}cathode gap.

Abstract

In order to check the system aspects of the forward}backward MSGC tracker designed for the future CMS experimentat LHC, 38 trapezoidal MSGC counters assembled in six multi-substrates detector modules were built and exposed toa muon beam at the CERN SPS. Results on the gain uniformity along the wedge-shaped strip pattern and across thedetector modules are shown together with measurements of the detection e$ciency and the spatial resolution. ( 1999Elsevier Science B.V. All rights reserved.

Keywords: Gaseous particle detectors; Tracking detectors

1. Introduction

CMS is one of the two multipurpose detectorsthat will be installed at the future LHC proton}proton collider at CERN. The outermost part ofthe CMS tracker will be equipped with MSGCs. Inthe forward}backward region they are arranged ondisks perpendicular to the beam pipe. Eleven suchdisks are foreseen on either side of the MSGCbarrel, covering the radial region from 700 to1160 mm and extending in z from $1215 to$2760 mm. On these disks the MSGC counters arearranged in four rings made of modules containingseveral trapezoidal MSGC counters put side byside in a common gas volume. A full description ofthe CMS forward}backward MSGC tracker can befound in Ref. [1].

The main design criteria are a full radial cover-age of the entire forward}backward area for eachdisk, minimization of dead space between the de-tector modules, rigidity of the disk assembly consis-tent with material budget constraints, and accessfor services. The detector modules are brie#y de-scribed in Section 3 after a description of the indi-vidual counters in Section 2.

The aim of the present test was to learn how tobuild such detector modules, keeping in mind thefuture mass production and to compare variousalternative designs. Another aspect of the test wasto learn how to run all these modules together andstudy the uniformity of the response. For that pur-pose the six detector modules were operated at theCERN SPS in the X5 muon beam. The experi-mental set-up is described in Section 4 and the dataanalysis procedure in Section 5. The results are

presented in Section 6 before drawing the con-clusions.

2. MSGC counters

As it is convenient for pattern recognition, insingle-sided counters, the anode and cathode stripspoint towards the beam pipe leading to a trap-ezoidal shaped electrode geometry. The anodewidth is kept constant. In order to keep a constantgain over the full length of the substrate, both theanode}cathode gap and the cathode width varywith the radial position on the detector module,following an homothetic rule proposed by NIK-HEF:

G"P/8#20 lm (1)

314 M. Ackermann et al. / Nuclear Instruments and Methods in Physics Research A 436 (1999) 313}325

Table 1Characteristics of the masks used to print the electrode patternonto the MSGC substrates

Outer ring Inner ring

Anode pitch 250}212 lm 200}185 lmAnode width 10 lm 7 lmCathode width 139}110 lm 100}90 lmAnode}cathode gap 51}46 lm 47}44 lmCentral anode length 170 mm 40 mmTotal length 186 mm 52 mmLarge base width 128 mm 102 mm

5SRON, Space Research Organization, Utrecht, The Nether-lands.

6NPP VOSTOK, Novosibirsk, Russian Federation.7OPTIMASK, 12 av. Ferdinand-de-Lesseps, 91420 Moran-

gis, France.8Baumer IMT Industrielle Messtechnik AG, Im Langacher,

CH-8606 Greifensee, Switzerland.

where G represents the anode}cathode gap andP the anode pitch (see Fig. 1).

For the present beam test two wedge-shapedmasks have been designed, with 512 anodes each,corresponding to the outermost and innermostrings, respectively. Their characteristics can befound in Table 1.

The substrates are made of bare DESAG 263glass, 300 lm thick, with either aluminium orgold strips printed on them by various manufac-turers: SRON,5 VOSTOK,6 OPTIMASK7

and IMT.8 For the "nal production of the CMSMSGC tracker, coated substrates with advancedpassivation are foreseen in order to allow them tosustain the high radiation level that will be presentin the LHC trackers [1].

The drift planes are made of either a 3 mm thickhoneycomb structure coated with thin layers ofcopper and nickel or of 300 lm thick DESAG 263glass, coated with a thin layer of either chromiumor nickel. The gas gap between the drift plane andthe substrate is 3 mm.

As the "nal chip foreseen for the readout ofMSGCs in CMS is not yet available, the counters

were equipped with four PreMux128 chips [2]per substrate. These chips hold 128 preampli"ersand shapers with 45 ns shaping time combined witha 1 MHz multiplexer. The sensitivity of the pre-ampli"ers is 2.75 V/pC. These chips were connectedto the anodes at the large side of the counters.

At the short side of the counters, the high volt-ages are brought to the cathode strips by groups of16 and to the drift plane. The anodes weregrounded and most detectors were run with about!530 V on the cathode strips and !3 kV on thedrift plane. The drift "eld across the gas gap is veryhigh, of the order of 8 kV/cm.

Some counters had fuse resistors integrated inthe bonding pads of the pitch adaptors at theinput of the PreMux128 chips. These were meantto allow remote disconnection of anode stripsdeveloping a short during operation, by applyinga positive voltage through a high-voltage diodebypassing the cathode group protection resistor.The fuses have a resistance of &400 ) and consti-tute an additional impedance at the input of thePreMux128.

3. Detector modules

Several trapezoidal MSGC counters are placedside by side in a common gas volume to forma detector module. In two set-ups, it was shownthat it is possible to operate safely counters withtheir neighbouring anodes as close as 400 lm andthat no loss of e$ciency is observed for minimumionizing particles traversing the region between thetwo counters [3,4]. Di!erent designs have beeninvestigated for these detector modules and six ofthem were built, labelled O1 to O4, C1 and C2 inwhat follows. Only the detector module O4 hasinner ring substrates.

In the design now adopted by the CMS colla-boration (see Ref. [1]), the modules consist of threeframes supporting four substrates, four drift cath-odes and the electronic hybrids (see Fig. 2). For thepresent test, the frames were machined out of fullmaterial; for the bottom and top frame, Stesalitwas used, while the middle frame was made of

M. Ackermann et al. / Nuclear Instruments and Methods in Physics Research A 436 (1999) 313}325 315

Fig. 2. The various elements building up an MSGC detectormodule.

Fig. 3. Cross section of an MSGC detector module.

PEEK, avoiding splinters from the Stesalit on thesubstrates.

The substrates are glued to the 2.5 mm thickbottom frame after relative alignment. Sti!eningbars inside the frame are foreseen to avoid deforma-tion of the substrates through sagging. Holes inthese bars allow the counting gas to pass through.

The distance frame, without any internal bars, sep-arates the substrates and the drift cathodes. Witha 3 mm height it de"nes the sensitive detection gap.It is glued onto the bonding pads of the substrates,which are bonded to the readout and high-voltagehybrids outside the gas volume, after the moduleshave been closed. The top frame supports the driftcathodes. The entire module is closed by a Kaptonfoil that is coated with copper on the outside.A cross section of a module is shown in Fig. 3. Forthe present test, the C1 module is made of two suchmodules mounted side by side on the same alumi-nium support plate, leading to eight substrates intotal.

Module C2 follows a variant of the above designand is made of a single four-fold detector module.In the four remaining modules: O1}O4, there areup to eight substrates in the same gas volume. Theindividual counters are assembled and tested,including their electronics, before mounting andaligning them inside a module.

4. Experimental set-up

The six MSGC detector modules were tested ina tertiary 100 GeV/c muon beam of low intensity,obtained from the primary SPS proton beam at theCERN X5 beam facility. Beam particles were se-lected by means of a scintillator trigger and theirtrajectory reconstructed in two orthogonal direc-tions (X and >) owing to a silicon microstrip


Fig. 4. Top view of the experimental set-up with the de"nition of the reference frame. The boxes left in white are other silicon detectorsunder test for the CMS tracker, not used in this paper.

Fig. 5. Perspective view of the bench with the six MSGC de-tector modules.

telescope. Fig. 4 shows the experimental set-up con-sisting of two mechanically independent benches,one carrying the six MSGC detector modules andone carrying the telescope, the trigger scintillatorsand other silicon detectors under test for the CMStracker (not used in this paper and shown in whiteon Fig. 4). The reference frame used in this paper isalso shown in Fig. 4, with the Z-axis along the beamdirection, the X-axis vertical and the >-axis hori-zontal; the various detector planes are, in "rstapproximation, parallel to the X> plane.

The trigger is performed by a coincidence of twoout of three scintillators. The "rst one is alwaysa 12]12 cm2 scintillator and the second one iseither a 6]6 cm2 or a 2]2 cm2 scintillator.

A special support structure has been built to holdthe six forward MSGC detector modules, as shownin Fig. 5. Each MSGC module is mounted on a6 mm thick aluminium plate supporting also thegas and cooling connectors, the high-voltage linesand the electronics service boards. The aluminiumbare plate had a hole of the size and shape of thedetector module milled in it behind each module, tominimize multiple scattering.

The whole platform is mounted on a movablesupport enabling translations of the six modulestogether, in X and > directions. The maximumexcursion allows to scan a whole MSGC module,

with a total possible translation of 32 cm in X and110 cm in >.

The six modules were #ushed in parallel witha gas mixture of Ne/DME-30/70% delivered bya single gas rack.

A silicon microstrip telescope, built in Bari [5],was used to reconstruct in space the trajectory of


triggered beam particles. It consists of four double-sided silicon microstrip detectors with 50 lmreadout pitch, gathered in two arms of twodetectors.

Unfortunately, as during the present tests thesilicon telescope and the MSGC modules were ondi!erent benches, they underwent uncorrelated vi-brations, especially during working hours. The situ-ation is very unfavourable to an alignment of thetwo systems. This situation is made still worse bythe geometry of the set-up (see Fig. 4), with a smalldistance between the two telescope arms anda large distance between the telescope and theMSGC detector modules.

The data acquisition is an updated version of thedistributed system described in Ref. [6]. For thesake of performance and modularity, the system'shardware is divided into a front-end and an eventbuilder part. The front-end part itself is segmentedinto three VME crates which communicate withthe event builder via a daisy chained fast datalink.

During beam spills, the multiplexed analog out-put of the detector front-end electronics is digitized,demultiplexed and stored by means of several VMEFADC modules which reside in the front-endcrates. The readout cycle is controlled by one VMEsequencer module per crate which provides thetiming signals for the front-end electronics and theFADCs. For the triggering of the whole system,custom designed modules are used [7]. All front-end crates are equipped with a real-time processorrunning OS9 which starts the readout cycles, col-lects the data from the FADC modules and rear-ranges it for transmission to the event builder.

5. Data analysis procedure

The retrieval of the particle related MSGC sig-nals from raw data requires a subtraction of thepedestals and a correction for the common modeshifts. The pedestals were computed for each ADCchannel as the average charge value for the 500 "rstevents of the run. An iterative procedure was usedin order to reject events in which the consideredstrip may have been hit by a particle. After correc-tion of the raw data for the pedestals, the common

mode was computed for each event, separately forgroups of 128 strips physically connected to thesame readout chip. It is given by the average chargevalue for the group of strips considered. After thesetwo corrections, the strip noise p

iwas estimated as

the standard deviation of the distribution of theremaining signal S

i, in the absence of an incident

particle.To "nd impact points of ionizing particles with

the MSGC counters, clusters of strips with a signalwere formed. A strip is considered to have a signal ifSi/p

i'3. Adjacent strips satisfying this criterion

were added together till no more adjacent stripsabove threshold were found. To allow for deadstrips, one strip below threshold was accepted with-in a cluster. The cluster charge Q#-645%3 was com-puted as the sum of the signals of the stripsaccepted in the cluster. The cluster noise p#-645%3 wasde"ned as

p#-645%3"SN+i/1

p2i

(2)

where N is the number of strips belonging to thecluster. The cluster signal-to-noise ratio was thende"ned as

S/N"Q#-645%3/p#-645%3. (3)

Fig. 6 shows a typical signal-to-noise distributionobtained for one of the MSGC counters operatedon the e$ciency plateau with a drift voltage of!2700 V and a cathode strip voltage of !530 V.One can clearly see the separation between noise andsignal. In order to reject the clusters due to the noise,observed in Fig. 6, a cluster was eventually accepted ifits signal-to-noise ratio exceeds 5. This distributioncorresponds to an average signal over noise ratio ofabout 24, after the cut rejecting noise clusters.

A coordinate of the incident particle impact pointcan be given only in units of strip number, n

453*1, by the

charge centre of gravity method. Indeed, due to thetrapezoidal shape of the counters leading to varyinganode pitches, knowledge of the coordinate along thestrips is needed to transform n

453*1into a spatial coor-

dinate. For this purpose use will be made of theinformation provided by the silicon telescope asthere are no MSGC stereo strips in the presentprototypes.


https://www.researchgate.net/publication/2460261_The_trigger_system_of_the_CMS_barrel_and_forward_milestones?el=1_x_8&enrichId=rgreq-58fa1232ec801882ed219c3bc7ebff93-XXX&enrichSource=Y292ZXJQYWdlOzIyMzgzNjQ3NDtBUzoxMDQyNDU4MzE0MDU1NzFAMTQwMTg2NTU1MTM5OA==

Fig. 6. Typical signal over noise distribution for an MSGC counter.

For the telescope data, the pedestal and commonmode corrections together with the cluster "nding areperformed in a way similar to the one used forMSGCs. However no strip with a signal below thestrip threshold is accepted within a cluster and thecluster signal-to-noise ratio must exceed 15. In orderto avoid ambiguities, events with more than onecluster in a telescope plane were rejected. The tele-scope planes were aligned with respect to each other,using beam particles and a "tting procedure basedon the minimum s2 method. Assuming thesame spatial resolution for the four planes provid-ing a given coordinate, they are found to be 7.0 and3.5 lm for the X and > coordinates, respectively.These di!erent resolutions are due to di!erent inter-mediate strips between the readout strips in thesilicon layers measuring the X and> coordinates (seeRef. [5]).

6. Results

6.1. Ezciency plateaus

The particle detection e$ciency of a given moduleis measured using only tracks reconstructed by theeight planes of the silicon telescope. Tracks with animpact point prediction outside the MSGC sensitivearea as well as those hitting a region with dead stripswere rejected. A cluster reconstructed in an MSGCwas accepted only if the n

453*1value measured by the

MSGC module is in agreement with the one pre-dicted by the telescope, after relative alignment of theMSGC module with respect to the telescope.

The diagrams in Fig. 7 show the detection e$cien-cy as a function of the applied cathode strip voltage(a) and as a function of the average cluster signal-to-noise ratio (b), for two modules operated at di!erent


Fig. 7. Detection e$ciency as a function of the cathode stripvoltage (a) and as a function of the cluster signal-to-noise ratio(b), for two modules operated at di!erent drift voltages.

Fig. 8. Average signal over noise ratio versus drift voltage at two cathode strip voltages: !530 and !510 V.

drift voltages. In Fig. 7a, it is observed that the cathodestrip voltage at which a full e$ciency of 97.5% isreached depends on the drift voltage. It is of the orderof!520 V. On the contrary, for both counters, the fulle$ciency starts at a same signal-to-noise ratio of about18 (see Fig. 7b). Indeed, this minimum signal-to-noiseratio is expected to depend only on the gas mixture, fora given gas thickness. As a result, to compare theuniformity of response of all counters, the signal-to-noise ratio, easier to calculate than the detectione$ciency, will be used instead (see Section 6.2).

A cluster multiplicity distribution, taken witha random trigger, between beam spills, shows anaverage value of 0.05 only. Rising the cluster thre-sholds does not decrease that number signi"cantlybut decreases the detection e$ciency. It was esti-mated that this number of random clusters corres-ponds to a strip occupancy of 0.03%, negligiblecompared to the beam related background occu-pancy of a few percent expected at LHC.


Fig. 9. Average signal-to-noise ratio as a function of the anode pitch, for modules O3, O1 and C1 (a), for modules O2 and C2 (b) and formodule O4 (c).

Full e$ciency can be reached at a lower cathodestrip voltage with a higher drift voltage. This isillustrated by Fig. 8 showing the variation of theaverage signal-to noise ratio with the drift voltagefor a "xed cathode strip voltage of !530 V. Onepoint at !510 V is also shown. It demonstratesthat an increase of about 700 V on the drift allowsto decrease the strip voltage by 20 V. This is inter-esting regarding possible sparks as their energy isproportional to the square of the strip voltage.A lower strip voltage thus decreases the risk ofdamaging the strips with sparks.

6.2. Gain uniformity study

All MSGC counters were scanned with the beamin various regions of the substrates in order tocheck the uniformity of the response.

6.2.1. Scan along the stripsAs the distance between the anodes and cathodes

varies along the strips, it is important to verify thatthe gain does not vary too much along the strips.Figs. 9a}c show the variation of the average signal-to-noise ratio as a function of the anode pitch,separately for the six modules.

For this scan looking at gain variations insidea given counter, it was not important to operate thedi!erent modules at the same voltages. The variousoperating conditions can be found in the "gure.They explain most of the signal-to-noise variationsfrom one module to another. However, in the caseof O3, the 60% lower signal-to-noise ratio, is alsodue to the high current (a few lA) drawn by thesubstrates of this module, leading to a signi"cantvoltage drop in the protection resistor. This highcurrent was due to a bad cleaning procedure of the


Fig. 10 . Variation of the average signal-to-noise ratio versus the counter number in module O2 (a), C1 (b) O1 (c) and C2 (d). The square(round) marks correspond to beam positions in a left (right) corner of the substrates. The open (closed) marks correspond to beampositions in a bottom (top) corner of the substrates. The full lines correspond to the average value of all measurements. The dashed anddotted lines to one and two standard deviations from that mean.

substrates. In the case of module C2, operating ata higher drift voltage than the other modules, thesignal-to-noise ratio is still 20% lower. This is dueto a bad connection to the readout electronics,leading to important common mode #uctuationsand an average noise twice as large as for the othermodules. It should also be noted that the inner ringmodule O4 is operated at lower voltages due to thesmaller anode width.

For modules shown in Figs. 9a and c, it isobserved that the pulse height is indeed constantalong the strips. The gain variations are less than10% which is below the usual gain variation ob-tained from one substrate to the next for MSGCswith parallel strips. These substrates were producedby IMT, SRON and VOSTOK. On the contrary,for modules in Fig. 9b, for which the substrateswere produced by OPTIMASK, a decrease of up to


20% is observed with increasing pitch, although thelarger pitches are towards the readout side.

6.2.2. Scan perpendicular to the stripsThe uniformity of the response in a direction

perpendicular to the strips was also checked.Fig. 10 shows the variation of the average signal-to-noise ratio for the eight MSGC counters ofmodules O2, C1 and O1, and for the four sub-strates of module C2. Modules O3 and O4 werenot in the beam during this scan. For this scanall counters were operated at the same voltages,!530 V on the cathode strips and !2700 Von the drift plane, except for module C2, operatedwith !3500 V on the drift plane. This higherdrift voltage for C2 was needed to obtain the samesignal-to-noise ratio as in the other modules,due to the higher noise in C2 (see Section 6.2.1).The di!erent marks correspond to di!erent posi-tions of the beam, in each of the four cornersof the substrates. The statistical errors on thesemeasurements are negligible. The full linescorrespond to the average value of 25 obtained forall measurements. The dashed and dotted linescorrespond to one and two standard deviationsfrom that mean respectively (p"2.8).

This gain uniformity study indicates that itmight be possible to operate all MSGC countersin a given module at the same high voltages.Gain variations smaller than 25% are thenexpected. It should however remain possibleto modify the voltages on some counters show-ing some defect. The relation 1, between the anodepitch and the anode}cathode gap allows toobtain a uniform gain along the strips of the trap-ezoidal MSGCs but will need to be checked andpossibly be adapted, depending on the substrateprovider.

6.3. Spatial resolution study

Due to the uncorrelated vibrations of the MSGCand telesope benches and to the unfavourable leverarm, the prediction of the beam impact positioninto the MSGC detector modules is only of theorder of 100 lm. To estimate the spatial resolutionof the MSGC modules, tracks were reconstructedusing the measurements in two MSGC modules

just before and after a third module. The predictionfor the impact point in the central module is thencompared to the measurement in that module. Inthis analysis, the information from the silicon tele-scope is only used to determine the coordinatealong the strips in order to "nd the appropriateanode pitch. This procedure is applied after align-ment of each concerned MSGC detector with re-spect to the telescope.

Fig. 11a shows the cluster size, expressed in num-ber of strips, as a function of the anode pitch formodules O1, O2, O3, C1 and C2. Except for O3working at a lower signal to noise because of theirimportant leakage current, the cluster size is 2.3strips in average, as usual with the Ne/DME-30/70% gas mixture used here. For all modules,the cluster size does not vary signi"cantly withthe anode pitch. This result is surprising. Indeed,a constant cluster size in space should lead todecreasing cluster size in number of strips whenthe anode pitch is increasing, going from 2.3 at212 lm to 2.0 at 240 lm. The e!ect should still bemore important for O2 and C2 having a decreasingsignal-to-noise ratio with increasing anode pitch(see Fig. 9b). A simple Monte Carlo simulationshows that the sampling step and, to a lesser extent,the threshold e!ect reduce the discrepancy by afactor two.

Fig. 11b shows the spatial resolution measuredfor module O2 and C1 as a function of the anodepitch. The result of 38 lm obtained for O2 ata pitch of 218 lm is compatible with the spatialresolution measured for counters with parallelstrips and an anode pitch of 200 lm, "lled witha Ne/DME-50/50% gas mixture: 36$2 lm [8].The spatial resolution degrades only to 44 lm forthe largest anode pitch measured, of 240 lm, whichis still quite acceptable for the tracking in CMS.However, this increase of the spatial resolution ofO2 with increasing anode pitch, could be a conse-quence of the decreasing signal-to-noise ratiofor that module. Indeed, for C1, having a constantsignal-to-noise ratio along the strips, the spatialresolution also seems to be constant except forthe smallest pitch. The slightly worse spatial resolu-tion of C1 compared to this of O2 may be explainedby their lower signal-to-noise ratio (see Figs. 9aand b).


Fig. 11. Variation of the cluster size (a) and of the spatial resolution (b), as a function of the anode pitch.

7. Conclusions

Six sector modules of the CMS forward back-ward MSGC tracker were built by various institu-tions. The di!erent designs investigated allproved to be feasible and to produce similar per-formances. The design of C1 eventually adoptedby the CMS collaboration was chosen because itpromises easier mass production and substratealignment. The six modules were all operatedtogether in the X5 muon beam in CERN. A numberof 38 substrates corresponding to a total of19 456 channels were present. A response witha gain dispersion of 11% standard deviation wasobserved both when going from one substrate tothe next in a given module and when going from

one module to another, excluding modules with anelectronics problem (C2) or drawing too much cur-rent (O3). However, this result should be testedwith the "nal CMS substrates. Indeed the presenceof coating could deteriorate the uniformity of theresponse.

A scan along the strips of the trapezoidalcounters also shows a constant gain, within 10%,for all modules except for two modules for whichthe substrates were produced by the same com-pany. This shows that the rule adopted for therelation between the anode}cathode distance andthe varying anode pitch, allows to maintain aconstant gain along the strips but might need modi-"cations depending on the substrate productionprocedure.


A detection e$ciency of 97.5% is measured foran average signal-to-noise ratio starting at about18. The spatial resolution in the trapezoidalcounters is measured. For substrates with a con-stant gain along their strip, it seems that there is nosigni"cant variation of the spatial resolution. Onthe contrary, it depends on the signal-to-noise ratioat which the counter is operated. All results ob-tained here vary from 38 to 44 lm, which full"llsthe requirements for the CMS tracker.

Acknowledgements

We wish to express our thanks to A. Peisert,R. HammerstroK m, O. Runolfsson and their tech-nical sta! for their many contributions. Their e$-cient manufacturing and testing procedure for thePreMux hybrids used to read out the detectors isacknowledged here. We are also indebted to the

technical sta!s of all contributing institutions fortheir help at various levels of this project.

References

[1] CMS Technical Design Report Tracker, CERN/LHCC98-6, CMS TDR 5.

[2] L.L. Jones, PreMux128 speci"cation, version 2.3, Ruther-ford Appleton Laboratory Internal Document, January1995.

[3] O. Bouhali et al., Proceedings of the International Work-shop on MSGCs, Lyon 1995, p. 101.

[4] J.J. van Hunen, Nucl. Instr. and Meth. A 409 (1998) 95.[5] L. Celano et al., Nucl. Instr. and Meth. A 381 (1996) 49.[6] F. Drouhin, B. Schwaller et al., A Unix SVR4-OS9 Distrib-

uted Data Acquisition for High Energy Physics, UHAGRPHE, IEEE Trans. Nucl. Sci. NS-5 (1998) 4.

[7] B. Schwaller, F. Drouhin, A. Pallares, J.C. Fontaine,Y. Benhammou, F. Charles, D. Huss, The trigger systemof the CMS barrel and forward milestones, CMS Note1998/029.

[8] O. Bouhali et al., Nucl. Instr. and Meth. A 413 (1998)105.






Large scale test of wedge shaped micro strip gas counters

Documents