HAL Id: in2p3-00010953
http://hal.in2p3.fr/in2p3-00010953
Submitted on 17 May 2001
HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or
L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires
Test bench of the barrel calorimeter modules
J. Boniface, H. Bonnefon, J. Colas, G. Dromby, M. Jevaud, B. Mansoulie, N.
Massol, F. Molinie, M. Moynot, P. Perrodo, et al.
To cite this version:
J. Boniface, H. Bonnefon, J. Colas, G. Dromby, M. Jevaud, et al.. Test bench of the barrel calorimeter modules. 2001, pp.22. �in2p3-00010953�
ATL-LARG-2001-007
26/02/2001
Testbenchofthebarrelcalorimeter
modules
February20,2001 ATLASInternalNote
LARG-NO-xxx
February2001
writtenby 1
MassolN.
1
BonifaceJ., 1
BonnefonH., 1
ColasJ., 1
DrombyG., 2
JevaudM., 3
MansoulieB.,
3
MolinieF., 1
MoynotM., 1
PerrodoP., 2
SauvageD., 3
SchwindlingJ., 3
TeigerJ.,
1
ZolnierowskiY.
|||||||||
1
Laboratoired'Annecy-le-VieuxdePhysiquedesParticules,IN2P3-CNRS,Annecy-
le-Vieux
2
CentredePhysiquedesParticulesdeMarseille,IN2P3-CNRS,Marseille
3
CEA,DSM/DAPNIA,Centred'EtudesdeSaclay,Gif-sur-Yvette
Abstract
Asystematicproceduretoqualifythebarrelcalorimetermodules
isanessentialsteptoguaranteea0.7%constantterm,whichisthe
collaborationobjective.Theproceduredetailedinthisnoteconsists
ofqualitymonitoringduringmechanicalassemblingandofasetof
electricaltestssuchaselectricalcontinuity,cellandcross-talkcapaci-
tancemeasurement,andhigh-voltagebehaviour.Forthewholetest,it
hasbeennecessarytodevelopdedicatedelectronicboards,todevelop
measurementmethods,andthewholeoperationsoftware.Makingthe
procedureautomaticwillguaranteethequalityofeachmoduleduring
assembling,cabling,andtestinliquidargon.
1
The ATLAS barrel calorimeter [1] is composed of two half-barrels, each of
them containing 16modules (gure1). Placed insidea cryostat, it works at
1 module
Figure1: Half-barrel calorimeter
the liquidargon temperature. Each module is made of 64 absorbers and 64
electrodes(gure 2),andweights 3.3tons. These moduleswillbeassembled
one by one, and only six of them will be fully scanned at CERN with an
electron beam. So to reach the goals driven by the physics requirements, a
complete test procedurehas to be developed to qualify allcalorimeter mod-
ules before starting the experiment. To guarantee a small overall constant
term in the energy resolution, the mechanical anelectricalcharacteristics of
the detector are carefully monitored. Similar tests are used to qualify the
end-cap calorimetermodules.
These tests run in three steps:
during assembling, we verify that the readout electrodes are not dam-
agedduetotheirhandling(TBF-1test),thenwemonitorthecleanness
of each component (HV test), and nally we check the mechanical as-
sembling quality (capagap test).
After cabling of the whole module, at room temperature, we check
connectics between HV bus and electrodes (TBF-2 test),we check ifit
holds high voltage (HV test),and weexercise each cell(TPA test).
sampling 2 collision point
...
η r
sampling 2
1 2 3
sectors
sampling 3
Figure2: Readoutelectrode
Thesetestsarereproducedattheliquidargontemperaturetoeliminate
problems induced by the thermal contraction.
The qualication procedure of the electromagnetic calorimeter modules is
performed by a test bench to study the calorimeter behavior at room tem-
peratureand at liquidargon temperature, and toobtain amap of all cells.
All informationsconcerningthe modulesare saved inadatabase: itincludes
the measurementson the absorbers and onthe electrodes beforeassembling
and thetestsresultsduringassemblingandaftercabling. Moreoverwemake
the procedure automatic tominimize errors.
2 General description of the electrical tests
2.1 Test bench equipment
The test benchis composed of (gures3 and 4):
electronic boards to generate several test signals: a low frequency si-
nusod (TBF board [2] + MUXCAPA board) and step pulses (TPA
board [3]),
a set of multiplexers (MUX board [4] + MUXCAPA board) to drive
the output channels towards the readoutsystem,
CCCCCC CCCCCC CCCCCC CCCCCC
PC
HH
HH VME Crate
5v,10 v BUS GPIB
HH VME Crate
BUS GPIB Power:+5v
GPIBmux
1 2
1 2 3 14
GGGGGGG
11 MUX Board
HT_3KV LHH
interface L
RS232/IEEE
HH
DIGITAL SCOPE 2 Channels
HH
VME Crate BUS GPIB Power:+5v 1 2 3 11
14 MUX Board .... ns
-40v,+40v
60v/1Hz
CALIB
.... mv
IGPIB
TPA TBF
MUX
IGPIB IGPIB
LABVIEW
Front 50 Ohms
Middle+Back 25 Ohms
MUX
mux
Figure3: Electrical test bench
an interface board between GPIB bus and VME-like crate (IGPIB
board [5]),
a digitaloscilloscope toproceed the output signals acquisition,
a high voltage module (1469 board [6]),
a precisioncapacimeter [7],
a PC to controlthe dierent boards and devices through a GPIB bus
(drivers written with LabVIEW [8] software), and to analyse all the
measurements.
a set of cables for each test: lemo cables (10 m), HV cables (10 m),
25 and 50 Axon cables with ATI connectors (7 m for assembling
tests and 4m +2 mfor cabling tests).
2.2 Electrical continuity test (TBF test)
2.2.1 Principle
This test veries the continuity of the electrical circuit to be sure that the
high voltage is distributed everywhere. It checks connectors and resistors
chains on the electrodes outer layers. The principle is to send (gure 5)
a low frequency sinusodal signal on the high voltage lines and to read the
response from the signal layer by capacitivecoupling. The signal amplitude
collected oneachcelldependsonthe decouplingcapacitance. Thecapacitive
part ofthesignal isextractedbyacarefulchoice offrequenciestodetectany
damage inthe resistor chain.
000000 000000 000000 000000 000000 000000 000000 000000 000000 000000 000000 000000 000000 000000 000000 000000 000000 000000 000000 000000 000000 000000 000000 000000 000000 000000 000000 000000 000000 000000 000000 000000 000000 000000 000000 000000 000000 000000 000000 000000 000000
111111 111111 111111 111111 111111 111111 111111 111111 111111 111111 111111 111111 111111 111111 111111 111111 111111 111111 111111 111111 111111 111111 111111 111111 111111 111111 111111 111111 111111 111111 111111 111111 111111 111111 111111 111111 111111 111111 111111 111111 111111
000000 000 111111 111
1 ΜΩ
00000000 0000 11111111 1111
TBF
Oscilloscope HV bus
S1 S2
C
S3
A P A K A P T O N
½ electrode Absorber
Figure5: Electrical continuity test
2.2.2 Hardware for this test
The sinusodal signal is provided by the MUXCAPA board. The frequency
is programmable between 0.1 Hz and 100 Hz, by step of 0.1 Hz, and the
amplitude is about 70 V. Another board, TBF, is used to distribute the
signal on the HV lines. It has 16 channels remote controlled by a PC via a
GPIB interface. Both boards use the VME mechanical standard. A digital
oscilloscope reads the signal coming from the electrodes via a multiplexer,
and a PC computes all parameters. The data are compared with reference
values.
2.3.1 Principle
Thistestdetectsanyhighvoltageproblemlinkedtoacleannessproblem. For
this,the leakagecurrentofallchannelsismonitored. Itisatestofabsorbers,
electrodes,andhoneycombscleanness,oranysurfacedefault. Wecanchoose
the voltage ramp to adapt the charge current. It should not exceed a given
threshold (< 1 A per channel and per electrode) to avoid damage on the
electrodes.
2.3.2 Hardware for this test
The collaboration has chosen a high voltage board from Lecroy with the
characteristics given below:
programmableDC voltage between 0 and 3500 V,
independent current measurementfor each channelwith a precisionof
1% forcurrent higher than 10nA,
programmabletrip for each channel between 0 and 100 A.
This equipment is characterized by a current measurement threshold par-
ticularly low. It is controlled by a PC via a GPIB bus and a GPIB-RS232
convertor [9]. The high voltage cablesare connectedto aCQHT (SECUrity
HT) board to discharge rapidly the whole module in case of emergency.
2.4 Gap thickness measurement (capagap test)
2.4.1 Principle
The ATLAS goal is to obtain a constant term contribution of 0.15% from
the dispersionof the gap thickness, equivalent toa dispersion of 50 microns
for a4 mmgap. Totest the gap thickness uniformity duringassembling, we
want to measure the gap capacitance with a precisionof 0.1%.
2.4.2 Hardware for this test
We use a capacimeter controlled by a PC. It sends a sinusodal signal on a
cell, and it measures the current and the voltage at its terminals to deduce
the impedance module and phase. We deduce the distance between two
absorbers from the capacitance measurement. Tomeasure automatically all
the capacitances foranelectrode, we use the secondpart of the MUXCAPA
boardwhich multiplexes the dierent sectors.
2.5.1 Principle
The purpose ofthistest istodetectcablingerrors,andtotestthe wholecal-
ibrationchain: boards,cables,connectors,... Todothat,wesend steppulses
on the calibration lines and we record the cells response. The ouptut signal
islteredbysoftware,and theresultingamplitude, linkedtothecapacitance
value, allows to detect the bad channels (grounded cells, short-circuits be-
tween channels, excessive cross-talk,...). The principle is described on the
gure 6.
injection resistor
TPA
detector capacitance
oscilloscope
50 Ω (1 ΚΩ)
resistor (50 Ω) adaptation
Figure6: Cell capacitance measurement
2.5.2 Hardware for this test
10Vstep pulseswitha2nsrisetimeare providedby theTPA board. It has
64 relays remote controlled by a PC via a GPIB interface in order to pulse
on calibration harness. The output channels are connected toa multiplexer
system, and only one channel atatime is analysedon adigital oscilloscope.
The ltering part is done by software. Multiplexers and pulse generator
modules use aVME mechanical standard.
3 Tests development and measurement results
A more detailed explanation about tests development and the results on
prototypes can be found in [10].
3.1 Electrical continuity test during assembling
The maindiÆculty forthis test comesfromthe outputsignalsummation on
the electrodes. Duringassembling,thechannelsare summedintodecrease
channels from sampling1, for the back, we sum 4 channels fromsampling2
with 2 channels fromsampling 3. This summationimplies to distinguish an
output signal variationproportionalto the number of defective channels.
3.1.1 Test condition
Therststepistodeterminetheoptimalfrequencyforthistest. Weconsider
only the sampling1: it is the most complicated case, because this sampling
is excited via another sampling.
First we choose a high frequency to obtain a negligible impedance of
detector capacitanceincomparisontothe 1Mresistor onthe HV bus. We
obtain theconditions described on thegure 7. Wesuppose thatthe second
E S1
SIMPLIFICATION
E
R1
1 ΜΩ S1 1 ΜΩ Z <<1 ΜΩ
Ck1
Ck1 Ck2
Ck2 kapton layer 1 kapton layer 2
1 ΜΩ
negligible
Cki: kapton capacitance of sampling i negligible
1 ΜΩ
R
R
Oscilloscope R1
Figure7: TBFtest at \high"frequency (>1 kHz).
HV face is oating. This system issensitive onlyto the resistive part of the
electrode. However, if a resistor is broken (R1 on gure 7), the capacitance
between HV layer and signal layer must decrease. That is why we must be
sensitive tothe capacitivepart.
E
S1
SIMPLIFICATION
R
Oscilloscope Ck1
Ck2
1 ΜΩ 1 ΜΩ
Ck1
Ck1
S1 negligible
(compared to 1 Mohms)
E Zeq
1 ΜΩ
1 ΜΩ R
Figure8: TBFtest at low frequency with the other face grounded.
at atime, weneed tobe sensitive toonlyone HV layer. Todothat, the HV
bus of the otherouter layerisgrounded. Thetypicalvalue foracapacitance
of a front sampling channel is 8 pF 200. To neglect the 1 M resistor
in comparison with the impedance of the kapton capacitance, we choose a
frequency suchas:
1=C
k1
! >10 7
=)f <10 Hz
For the front, the output signalis given by the relation:
S
1
=
1
2+1=jR C
k1
! E
and after simplication:
S
1
=jR C
k1
!E
1 capacitance which is proportional to the number of cells. So in these
conditions, we are able to distinguish adefective cell.
3.1.2 Signal variation versus the number of cells
For this measurement, we take the conditions of gure 8. We measure for
the sampling1the signalcorrespondingtoonestrip,then twostrips,...until
eight strips to reproduce the eect of summation boards in . The result is
given on gure 9. We obtain a linear variation between the output signal
Number of strips
0 1 2 3 4 5 6 7 8
Voltage (volts)
0 0.5 1 1.5 2 2.5 3 3.5
Amplitude versus the number of cells
Figure 9: Output signalamplitude versus the numberof cells at 5 Hz.
and the numberof connectedstrips, asexpected. In caseof problem,we are
able to determine the number of missing cells.
3.1.3 Final conguration
The prototype electrodes have highlighted a real problem concerning the
inked resistors. The dispersion of measured values is so high that it is im-
possibletowork withthe same frequency onthe whole electrode. Moreover,
if we use a very lowfrequency for all channels, the test length can strongly
increase. Thatis why it isnecessary to modifythe test tobeless dependent
to the electrode quality. Of course, we must keep the sensibility to the ca-
pacitive part but we use a couple of frequencies to extract capacitance and
resistance. Withsuchamethod,wedon'tmakeanyapproximationanymore,
and wekeep an automatic test.
For this test, the inner layer of each electrode is grounded via a 10 M
resistor on the summation boards. the table 1 summarizes the main used
parameters.
Parameters Typical value
Nominal voltage 1800 V(in air,humidity <50%)
Voltage ramp 5V/s
Trip 17A for1 face (7sectors)
Charge current 0;5<I
avg
<1A per sector(/ surface)
Leakage current <200 nA after 30mn for a wholeelectrode
Table 1: Main parametersfor the HV test.
An exampleof charge currents on one electrode is given ongure 10.
2.0
0.02.0
0.02.0
0.02.0
0.02.0
0.02.0
0.02.0
0.0
240 0
secteur 1 secteur 2 secteur 3 secteur 4 secteur 5 secteur 6 secteur 7 courants HT2
2.0
0.02.0
0.02.0
0.02.0
0.02.0
0.02.0
0.02.0
0.0
240 0
secteur 1 secteur 2 secteur 3 secteur 4 secteur 5 secteur 6 secteur 7 courants HT1
µA µA
Figure 10: Chargecurrents for eachsector and foreach faceversus time.
These currents are dominated by the decoupling capacitance charge. We
have:
i
charge
=
"
o
"
r S
e
V
t
The high voltage software uses a linear voltage ramp. Moreover, a feed-
back loop on the current of each channel avoidinopportune trips in case of
discharge. Aboveagiventhreshold,thevoltageisstabilizeduntilasignicant
decrease of the whole currents.
At the nominal voltage, we measure the leakage currents (gure 11).
0.1
0.00.1
0.00.1
0.00.1
0.00.1
0.00.1
0.00.1
0.0
1500 480
secteur 1 secteur 2 secteur 3 secteur 4 secteur 5 secteur 6 secteur 7 courants HT2
0.1
0.00.1
0.00.1
0.00.1
0.00.1
0.00.1
0.00.1
0.0
1500 480
secteur 1 secteur 2 secteur 3 secteur 4 secteur 5 secteur 6 secteur 7 courants HT1
µA µA
Figure11: Leakage currentsfor each sector and for each face versus time.
These currents decrease inabout one hour to the detection threshold of the
HV system (10 nA).
So this test veries not only the cleanness of the module (discharge or
shortcircuitduringtheramp),butalsotheelectrodeinnerstructure(leakage
current).
3.3 Gap thickness measurement
To dothis test, we must takeintoaccount several constraints:
thereisabigdispersionofresistorsvaluesontheelectrodesHV layers.
Sothe measuredcapacitance varies with the frequency. It isnecessary
to become independent of this resistors eect.
solution impliesan increase of test duration and money.
we must use a long cable (' 10 meters) which introduce a parasitic
capacitance (100pF/m). This is not acceptablefor precisionmeasure-
ment oncells of several hundreds of pF.
To avoidthese problems, weadopt the principles given below:
we measure the three samplings in the same time (gure 12). In this
condition, we are not sensitive to the capacitance variation with fre-
quency.
CAPACIMETER
Absorber
C2
C3 C1
Borne Low
1 ΜΩ HV channel
High Borne 64 C1
8 C2 + 4 C3
Figure 12: Measurement of capacitance between electrode and absorber by
sector.
we measure the capacitance by sector of = 0:2 to decrease the
numberof cables.
we use a four terminal pair conguration to be insensitive to some
parasiticseects.
we choose a frequency of 100 kHz. A higher frequency is not a good
choice because our system becomes sensitive to the cables inductive
eect.
3.4.1 Principle
In this test,we don'twant totest the electricalcontinuity onthe electrodes,
but the HV cables and boards. Moreover, in the liquid argon, it is impera-
tivetoverify the connecticsbetween electrodes and HV boards. Likeduring
assembling, we send a sinusodal signal on the HV channels and we anal-
yse signals induced by capacitive coupling. Nevertheless, several dierences
modify the test conditions:
one HV channelisconnected to32electrodesinonone faceand one
sector in.
anoutputchannelconcerns one cell, i.e. asum of4 electrodes infor
sampling 2. Soour system includes 4dierent resistors incomparison
to 1at assembling stage.
due tothe motherboard, each cell is connected tothe ground through
an injectionresistor and an adaptationresistor.
Weworkoncells ofsampling2becauseofthe summationinby groupof4
insteadof16onsampling1,andbecauseofthe littlecapacitancevariationin
incomparisonwiththecellsofsampling3. Weobtainthesystemdescribed
on gure 13.
Inthisconguration,theoutputsignalSdependsonkaptoncapacitances
and inked resistors. The maindiÆculty isdue tothe bigdispersiononthese
resistors value. We can't work with 2 frequencies to extract resistive and
capacitive parts like during assembling. So we must choose a frequency as
lowaspossibletoneglectthe resistances. But adecrease offrequency causes
a decrease of signal amplitude. The measured signals are very noisy and it
becomesdiÆculttodetect them. A solutionisto use anFFTtoextract the
fundamental amplitude.
3.4.2 Results
Withthis method,welook atthe linearity between the outputsignalampli-
tude versus the number of electrodes. We work at1Hz toobtain aphase of
=2. Weconsider a complete celland we disconnect successively the elec-
trodes. Weobtainthe resultgiven ongure14. We areable todistinguisha
missing electrode inside a cellwithout ambiguity. Moreover, on sampling2,
we have 8cellsin fora given sector,allof themconnected tothesame HV
channel. So for a given position in , we calculatethe mean value on these
S
ρ =1 ΜΩ
(in phi) E
25 Ω R1
Ck R2
Ck R3
Ck R4
Ck
R5
Ck R6
Ck R7
Ck R8
Ck
Calibration Ri
50 Ω
terminated in MUX
x7
Figure 13: Electrical continuity test after cabling, for a cell of sampling 2.
The detector capacitancesare not represented.
8 cells. We obtaina dispersionof about 1-2%, excepted in case ofa bad cell
with a dispersion of 9%.
Withthe same test, thanks to its good sensibility, we are able todetect
a wrong cabling between the electrode outer layers. On gure 15, we have
the amplitude of the output signal versus the cell number in , for a given
sector. The cells
0 to
5
have 4 electrodes, while the cell
6
is empty and
the cell
7
has only 1 electrode. The two curves correspond to the two HV
layers. The observed dierenceis due to the dierence of the glue thickness
between the copperlayers.