文档库 最新最全的文档下载
当前位置:文档库 › Available on CMS information server CMS NOTE 1997-076 The Compact Muon Solenoid Experiment

Available on CMS information server CMS NOTE 1997-076 The Compact Muon Solenoid Experiment

Available on CMS information server CMS NOTE

1997-076

The Compact Muon Solenoid Experiment

Mailing address: CMS CERN, CH-1211 GENEVA 23, Switzerland

CMS Note

10september1997

T est measurements on thermal,electrical and optical properties of the CMS/ECAL-Prototype

setup

J.Badier,H.Hillemanns,A.Karar,J.C.Vanel

LPNHE,′Ecole Polytechnique,F-91128Palaiseau Cedex

Abstract

Beside the intrinsic characteristics of the CMS electromagnetic calorimeter components like crystal quality and APD performance,also electromagnetical shielding,thermal and optical behaviour of the alveole and their supporting mechanics have a large impact on the overall calorimeter performance.

The results of test measurements,which have been performed on these items in order to evaluate their

in?uence and to conclude on possible improvements,will be reported and discussed.

1Introduction

The CMS e.m.calorimeter performance is strongly related to intrinsic parameters of the crystals for example light

yield,radiation hardness and the quality of the readout system(i.e.,quantum ef?ciency,noise,gain stability,radia-

tion hardness,etc.).Beside these parameters,the characteristics of the mechanical support structure,in which these components are mounted together,can affect the overall calorimeter performance.Thus the intrinsic calorimeter

parameters described above require a mechanical support providing a good electromagnetic shielding behaviour

for noise protection and high thermal stability in order to stabilize the light production of the crystal and the gain of the APD.Furthermore,the light collection uniformity of each crystal should not be changed by the supporting

alveole structure.Different test setups have been made in order to evaluate the in?uence of these effects.The results obtained with these tests are described in this note.

2Thermal behaviour of the crystal mechanical support structure

The need for thermal stability of the mechanical support structure is given by the temperature dependence of

both the light production in the crystals and the temperature dependence of the APD gain.The APD temperature stability of T0.1C over one temperature calibration period and a temperature gradient of less than 1C over all the crystal length is necessary in order to avoid visible effects in the energy resolution of the e.m.

calorimeter[1].As a basic condition for that,the temperature of the cooling water certainly has to be stable to a level of0.05C.Furthermore,the absolute temperature difference between the APD and the cooling water has to be less than1.0C.Otherwise,e.g.,a10%?uctuation in the heat?ow near the APD would be a temperature ?uctuation of the APD greater than0.05-0.1C.Fluctuations in the heat?ow through the crystals,the APD’s and the mechanics can be caused by inhomogeneities in the temperature distribution in the calorimeter.This in turn can be caused by heat production?uctuation of the VFE-Board or by the APD’s itself in case of high dark currents as a consequence of radiation damages.Another reason could be the network of thermal resistances between the crystal and the APD on one side and the cooling system on the other https://www.wendangku.net/doc/058215380.html,rge thermal resistances results in high APD or crystal temperature?uctuations even in case of small heat?ow?uctuations.Consequenctly,to have a small sensitivity of crystal and APD temperature against?uctuations in the heat?ow througout the calorimeter,the thermal resistance of the supporting structure between the APD and the cooling pipes has to be suf?ciently small. This can be achieved by geometrical optimisation of the support mechnics(i.e.short distances and large sections) and by the use of metals with good thermal conductivity,like copper.The latter is usually correlated with high density materials,which leads to unwanted absorption effects for the hadronic calorimeter.Therefore a solution with geometrically optimized and light materials like aluminium is preferable.

2.1Test Setups for Cooling Optimisation

To investigate the thermal resistance between the APD and the cooling water,two mechanically different test setups have been studied.They have been equipped with temperature sensors at different positions.Fig.1and?g.15.1 show a technical drawing of the?rst mechanical support structure(two pipe solution)as foreseen for the1997 prototype.It consists of an alveolar structure holding the crystals,an aluminium block(“chocolate bar”)housing the APD’s and the VFE-boards.The heat transport out of this“APD”-block is provided by an aluminium bar on the top of this block ending up with a cooling pipe.The thermal resistance between the APD and cooling pipe consists of two parallel resistors.The?rst cooresponds to the system APD-thermal screen-dowel-chocolate bar and is planned to be about10C/W,the second corresponds to the system APD-crystal-chocolate-bar,estimated to be about150C/W.Both values are not measurable in an isolated manner.

A second cooling system,thermically isolated to the chocolate bar cooling,will transport the heat coming from the VFE-boards thermically connected with a second block on top of the chocolate bar.The ef?ciency of this cooling system will strongly depend on the thermal isolation between the APD and the VFE-board.The thermal resistance of the bonding wires is about1300C/W and the heat exchange between the two aluminium bars transporting heat to the cooling pipes is about1mW in case of1C temperature difference.Simulations using the analogy to electrical circuits and thus allowing for SPICE-simulations show a temperature difference of about1.2C between the VFE and his corresponding cooling pipe,whereas the APD temperature remains stable at the level of about 0.1C.

However,due to his mechanical complexity a detailed understanding of heat?ow in the1997prototype is dif?cult to obtain.Therefore the thermal resistances have been measured using a much simpler setup(?g.2).In this second test setup only one cooling circuit has been used and the mechanics on the top of the chocolate bar has been replaced by a simple aluminium block.

2

Figure1:Technical drawing of the PROTO’97mechanical support structure.A larger and more detailed drawing can be obtained from?g.15.1

Figure2:Schematic drawing of the simpli?ed mechanical support structure for temperature tests

3

In all setups the alveole containing 12crystals shown in ?g.1was replaced by two crystals covering the bottom side of the chocolate bar in order to simulate a realistic thermal resistance between the APD and the chocolate bar.The APD’s were simulated by small metallic slices of typically APD size glued on the crystals.

To correct for heat energy picked up by the setup due to his limited thermal isolation versus room temperature it was necessary to produce different heat levels in the chocolate bar.This was done for all test setups by implemented resistor

elements.

b)

= Temperature Sensors

a)

Figure 3:Schematic drawing of thermal resistance measurement setup for a contact between (a)a cylindrical and a plane surface,(b)a small plane surface instead of a cylindrical surface

As a fourth test in order to measure the thermal resistance of a ?exible mechanical connection of the chocolate bar to the grid,a test setup has been made of an aluminium block in thermal contact with a cylindrical surface (?g.3a).This allows a movement in one direction along the dashed line required due to different spatial orientations of the subunits in the ?nal detector.The equivalent thickness between both surfaces in this case is given by:

(1)

The cylindrical surface can also be replaced by a surface as illustrated in ?g.3b,allowing the same one directional movement like the cylindrical solution.

2.2Test Results

In ?g.4the temperature difference between the APD and the cooling water is plotted for two con?gurations as a function of created heat energy with a straight line ?t overlayed.

In order to correct for pickup heat energy by the surrounding air and by thermal contact of the hole setup with the test bench,the temperature difference between APD and cooling water has also been measured with no power creation in the chocolate bar.This leads in all cases to an offset of the curve of about 0.6-0.7C.

The curve for the two pipe con?guration shows the largest thermal resistance between the APD and the cooling water of about R=0.55C/W and therefore the highest sensitivity to heat production ?uctuations.An improvement is achieved by using the single pipe aluminium cooling system,leading to a 40%smaller thermal resistance of about R=0.37C/W.This is due to the much shorter distance and much larger section for heat ?ow,resulting in

4

00.5

1

1.5

2

2.5

3

3.5

44.55

preamplifier power [W ]

t e m p . d i f f .[0C ]Figure 4:Temperature difference between APD and cooling water for different heat levels

a remarkably improved thermal conductivity.The thermal resistance in case of using a setup as shown in ?g.3a yields a thermal resistance of about R=0.7C/W between the two

surfaces.

Figure 5:Technical drawing of the Proto’97test beam setup

Finally the cooling system setup for the Proto’97beam tests has been tested.The hole setup consists of 3cooling circuits.The ?rst circuit cooling the box keeps the Proto’97in a well thermalized environment in order to avoid external heat pickup.The second one surrounding the 36crystals,mounted in 3submodules,simulates the ther-malized crystals,which would surround this con?guration in the ?nal detector.The last circuit builds the cooling system of the Proto’97mechanics,cooling ?rst the chocolate bar with the APD’s inside and transporting after the heat produced by the VFE boards.The temperature regulation of the cooling water was done with a LAUDA re-frigerator,providing a water ?ux of several l/min.For the temperature tests the middle submodule in the protoype setup (?g.5)has been equipped with temperature sensors at various places on the chocolate bar and the cooling mechanics on top of it.

As a ?rst test we run the system without VFE power for 3days in order to follow the thermalization characteristics.Independent of room temperature,the temperature was kept stable at all sensor positions within 0.1C during the 3days.

Switching on the VFE power for the 12channels of the middle subvolume,we observed an (expected)1.3C temperature rise in the VFE supporting structure of the alveole mechanics.On the other side no change in the APD

temperature (measured by the dowel)has been observed (T

0.1C).The results are indicating on one side the Proto’97cooling design is ful?lling its requirements.On the other side the comparison with other more simple cooling scenarios using only one cooling pipe shows that there is a

5

certain possibility for simplifying the mechanical support structure of the alveole and thus reducing remarkably the production costs without loosing performance in terms of APD temperature stability.

3Electromagnetic Shielding

The noise characteristics of the e.m.calorimeter readout chain (i.e.APD’s,VFE’s,interface cards,FPU’s)usually depends on intrinsic parameters like the APD capacity,dark current,input capacity of the VFE and so on.Beside that also pickup noise can have a major impact on noise characteristics.Pickup noise is caused either by internal currents in all mechanical components of the alveole support structure or by capturing emitted e.m.radiation from other detector components near the e.m.calorimeter.The latter is due to the antenna characteristics of all metallic components of the e.m.calorimeter mechanics (i.e.stray capacities,inductivities and resistors).

The electromagnetic shielding behaviour simulation of the ECAL electronics is quite dif?cult due to the complicate mechanical design of the calorimeter in terms of antenna characteristics [2].In addition the radiation character-istics of the other detector components,which could have an in?uence on the ECAL noise performance,will be more or less unknown until the experiment starts.However to give an idea,whether different mechanical designs and shielding scenarios can in?uence the electromagnetic shielding behaviour,the response of the VFE boards,mounted in different prototype mechanics,has been measured during exposure to external e.m.radiation from 100kHz up to 500MHz.

3.1Experimental setup

H.Hillemanns, LPNHE, Paris

,

Figure 6:Schematic drawing of the setup for measuring the electromagnetic shielding behaviour

All EMI measurements have been carried out in a large shielded anechoic room,which acts as a Faraday cage in order to avoid signal distortion by external radiation and which has a low re?ective walls for internal incident e.m.radiation.The test setup consisted of an alveole structure equipped with the chocolate bar and a cooling block https://www.wendangku.net/doc/058215380.html,ing a spectral analyzer the signal response was measured from VFE-boards equipped with PIN-diodes (gain 1and 8)in order to simulate the correct input capacity and the connecting wire geometry.The signal height is measured in db/mV ,i.e.the signal attenuation is proportional to the signal height.The connection between VFE-boards and analyzer was done with standard Lemocables thus providing a certain shielding from behind the VFE-boards.The whole setup was exposed to parallel e.m.radiation created by the antenna system scetched in

6

?g.6.The spectrum analyzer reading the boards is associated with a tracking generator,which output is ampli?ed using a high bandwith ampli?er before injection into the antennas.A PC had been used for further data analysis.

3.2Results

Cable Effect -100

-90-80

-70

-60

-50

-40

-30-20-100103104

105

open cable cable 50? term.gain 1gain 840 MHz

frequency [kHz ]a t t e n u a t i o n [d b /m V ]Figure 7:Signal response of the VFE-board in gain 1and 8in comparison with the readout cable only

As a consistency check the relative sensitivity of the VFE-board has been studied for gain 1and 8versus a simple cable,which is more or less the “minimum sensitivity”one has in this test setup (?g.7).Between gain 1and 8one can see a difference of about 20db/mV in the band region of the ampli?er (between about 10MHz and 40MHz),well re?ecting the factor 8in signal height between gain 1and 8.The VFE signal response has also been measured while changing the alveoles position relative to the e.m.wave propagation (?g.8).No position dependence could be observed.In a next step the VFE response has been measured outside the alveole a)with a PIN-diode directly mounted in front of the VFE and b)a 15cm cable between the diode and the VFE-board.As plotted in ?g.9an additional nois of about 10db/mV is picked up in case of connecting the diode with a cable.On the other side no additional pickup noise has been observed in case of measuring both con?gurations mounted on the alveole mechanics due to the shielding inside the alveole (?g.10).

In a next series of measurements a 6x2-submodule fully equipped with crystals,APD’s and VFE-boards has been

7

Gain 8 at different positions -100

-90-80

-70

-60

-50

-40

-30-20-100103104

105

pos 1pos 2pos 340 MHz

frequency [kHz ]a t t e n u a t i o n [d b /m V ]Figure 8:Signal response of the VFE-board with gain 8for several positions

8

Gain 1 preamp outside alveole -100

-90-80

-70

-60

-50

-40

-30-20-100103104

105

preamp-diode preamp-cable-diode 40 MHz

frequency [kHz ]a t t e n u a t i o n [d b /m V ]Figure 9:Signal response of the VFE-board with gain 1with and without cable between PIN-diode and VFE-board

9

Gain 1 preamp inside alveole -100

-90-80

-70

-60

-50

-40

-30-20-100103104

105

preamp-diode preamp-cable-diode 40 MHz

frequency [kHz ]a t t e n u a t i o n [d b /m V ]Figure 10:Signal response of the VFE-board with gain 1with and without cable between PIN-diode and VFE-board mounted on the alveole mechanics

10

exposed to external e.m.radiation.With respect to the previous measurements the VFE-boards now were con-nected by capton cables with the T-card.The T-card was connected to the readout by Lemocables.Both capton cables and T-boards weren’t shielded.In ?g.11the signal response of one Proto’97readout channel is compared to Proto97, FE with PIN and cable, gain 8/1-20

-10

010

20

30

40

50607080103

104

105

Proto97 g1Proto97 g8FE, cable, diode, g1FE, cable, diode, g8

frequency [kHz ]a t t e n u a t i o n [d b /m V ]Figure 11:Signal response of the VFE-board in the Proto’97mechanics in comparison with a VFE-cable-PIN system mounted in the alveole

the response of VFE plus a 15cm cable plus a PIN-diode mounted in the alveole for both gain values.The larger sensitivity of the Proto’97to external e.m.radiation in comparison with the VFE-cable-PIN-system is clearly visible.This is due to the fact that the capton cable and the T-card behind the VFE in the Proto’97setup weren’t shielded and thus picking up noise.This effect is demonstrated in ?g.12,where the signal response is plotted for successively connection of each component in the Proto’97readout chain.Beginning with the readout cable,where one has the smallest sensitivity,one recognizes a large step after the connection of the T-card to the readout chain.Another contribution appears after connection of the capton cable plus the VFE electronics.This is due to the unshielded capton cable.

11

Proto97, https://www.wendangku.net/doc/058215380.html,ponents, gain 8-20

-10

010

20

30

40

50607080103

104

105

cable only plus 30cm plus T-piece everything

frequency [kHz ]a t t e n u a t i o n [d b /m V ]Figure 12:Signal response of the Proto’97components by successively connecting of each component for gain 8

12

3.3Conclusion on EMI Measurements

These results are certainly emphasizing the need of an ef?cient shielding of the capton cables and the T-cards. Both shielding could reduce the pickup noise down to a acceptable level.On the other side the results obtained with VFE-cable-diode system clearly demonstrate a smaller sensitivity to external e.m.radiation than the Proto’97 mechanics in its actual design.This could open further possibilities towards a simpli?cation of the mechanical design without dramatic noise performance reduction.

4Light Collection with Alveoles

Usually the energy resolution of the CMS e.m.calorimeter is characterised by a stochastic and a constant term. The stochastic term is more or less related to intrinsic calorimeter parameters like photostatistics and shower containment.The constant term depends strongly on the light collection uniformity of a crystal.Since the crystals are usually subjected to certain?uctuations concerning their light collection uniformity,several crystal treatments are used presently in order to achieve a light collection uniformity in the order of less than0.5%/X.This can be done by an appropriate crystal wrapping,depolishing,painting or by a combination of all these methods. Nevertheless it has to be guaranteed,that the light collection uniformity once achieved for a crystal does not change by putting the crystal into an alveole with its aluminium layer inside.In addition using the alveoles should not drop down dramatically the absolute amount of light leaving the crystal.This can be achieved using specially treated aluminium for the alveoles,which have an additional re?ective layer.Possible solutions are actually under study in view of optical performance and large series production compability.

To study the light collection uniformity behaviour using alveoles,several crystals have been uniformized using depolishing methods and Tyvek wrapping[3].The light collection uniformity has been studied using the Bldg.27 optical test bench.Fig.13shows the light collection uniformity for crystal1338using Tyvek wrapping.A unifor-

Figure13:Light collection uniformity of crystal1338 using Tyvek wrapping Figure14:Light collection uniformity of crystal1338 using an alveole

mity of about0.3%/X can be observed in region of the shower maximum,suf?ciently good to not to deteriorate the constant term in the energy resolution.Fig.14shows the same crystal using the alveole.One observes the same light collection uniformity as for the Tyvek wrapping,but on the other side also a degradation in the absolute light yield of about20%of the Tyvek wrapping light yield.This example is representative for all crystals investigated and was con?rmed by repeating the same measurements at PSI,giving the same relative light losses and unifor-mity characteristics.Tab.1summarizes the measurements carried out at both places.The LY loss values indicate, that there is less LY loss in case of measuring polished crystals i.e.,before a uniformisation procedure.Therefore “natural uniform”crystals would loose less light in case of putting them in the alveole than uniformized crystals. The light collection uniformity observed also roughly was reproduced by the PSI measurements.

13

Crystal Homogeneity

[%/X]

PSI

Alveole

0.760.42

0.840.26

0.66 1.5

0.77

0.77-0.24

stable by the crystal,which itself will be cooled by the front side of the ECAL.The heat energy created by the APD of about3mW will be transported by the crystal.Nevertheless this option requires a cable between the APD and the VFE board between2cm and5cm length.Measurements carried out on electromagnetic shielding discussed above are indicating,that this option could be realized using a coaxial cable nearly without additional pickup noise. Furthermore this option turns out to be the cheapest one of our proposals,providing in addition a radial dimension reduction of about4cm with respect to the present Proto’97design.

Despite their remarkable cost reduction potential all3proposals are showing more or less possible drawbacks either in terms of thermal or mechanical behaviour or in noise performance.Further and more detailed studies have to be carried out on these(or other)options in order to evaluate,wether or not possible performance reduction could be accepted.

Acknowledgments

The authors want to thank all technicians and engineers of′Ecole Polytechnique,who were involved in the prepa-ration of the test measurements.Thanks also to J.P.Walder,F.Martin et F.Zach for their help in doing the measure-ments in electromagnetic shielding.Furthermore we want to thank E.Auffray,P.Lecoq,M.Schneegans,S.Paoletti from Bldg.27for their help in preparing and doing light yield measurements.Also thanks to K.Deiters and D.Renker from PSI for their help in measuring the light yield of several samples.

6References

References

[1]A detailed discussion of the impact of the cooling system on the calorimeter performnace will follow in a

further Technical Note.

[2]H.W.Ott,Noise reduction techniques in electronic systems,J.Wiley

[3]M.Schneegans,E.Auffray,S.Paoletti,P.Lecoq,Light response uniformisation stu?es on lead tungsten crystals

for the CMS electromagnetic calorimeter,to be published

15

Figure15:Schematic drawing of four proposals for a the simpli?ed mechanical support structure

16

相关文档