- Email: [email protected]
Design and performance of the large HERMES drift chambers
Nuclear Instruments
and Methods in Physics Research
A 367 (1995) 96-99
ELSEVIER
Design and performance of the large HERMES drift chambers S. Bernreuther”,
H. B6ttcherb, Harderb, D. Haschb, M. A. Miller”, F. Neunreithe?, F.
A.B. Borissovb, W. Briicknerd, M. Ferstl”, A. Gute”, U. Kirscha, B. Krauseb, W.A. Lachnit”, F. Mei13nerb, W.-D. Nowakb, M. Pohlb, H.O. Roloffb, H. Russoa, Schmidt”, A.E. Schwindb’*
“Universitiit Erlangen-Niirnberg, 91058 Erlangen, Germany “DESY-IfH Z&then, 15735 Zeuthen, Germany ‘TRIUMF, Vancouver, BC. Canada, V6T-2A3 dMPl fiir Kerphysik, 69029 Heidelberg. Gemany
Abstract Big planar drift chambers built for the downstream tracking system of the HERMES spectrometer are described. Using the fast non-flammable gas mixture Ar/CO,/CF, (901515) average spatial resolutions of about 180 pm per plane at efficiencies above 96% have been obtained from test run data analysis. 1. Introduction The HERMES apparatus is designed as a fixed target spectrometer to investigate the spin structure of the nucleon by measuring doubly polarised lepton-nucleon scattering with an internal target in the HERA lepton ring. The spectrometer is split into two halves installed above and below the two HERA beam pipes. Each spectrometer half is equipped with the same sequence of detectors. In front of the spectrometer magnet, the tracking system is composed of two 3-plane ,microsttip gas counters [l] and a pair of 6-plane drift chambers. Three proportional chambers are mounted within the magnet to reduce tracking ambiguities. Behind the magnet, tracking is based upon two pairs of large 6-plane drift chambers (Back Chambers), which will be described in this paper. Particle identification is accomplished by a threshold Cherenkov counter, a transition radiation detector and an electromagnetic calorimeter, the latter also being used for triggering.’ A complete description of the spectrometer can be found in Ref. [2]. In Section 2 the Back Chamber design and construction are described, followed by some practical experience presented in Section 3. Test beam results are discussed in Section 4. The on-line chamber calibration procedure is outlined in Section 5 and in Section 6 a summary is given.
2. Design and construction Each chamber or module consists of six sense wire planes interleaved with cathode foils. As schematically * Corresponding author. E-mail [email protected] Elsevier Science B.V. SSDI
0168-9002(95)00755-5
shown in Fig. I, sense wires are tilted by ?30° in the first and last pair of these six planes with each second plane having the sense wires staggered by half a drift cell. The cathode foils and potential wires are at negative high voltage. The spectrometer layout requires two different Back Chamber sizes. For both types some parameters are given in Table 1. Both sense wire and cathode foil frames were made of GFK (glass fiber-reinforced epoxy) by Stesalit (Zullwil, Switzerland). To avoid frame distortion the GFK frames are sandwiched between two steel frames of 38 mm in thickness each. In the mechanical design special attention was paid to high stability and high wire positioning accuracy in the sense wire frames. The deformation of the latter due to wire tension was taken into account using a finite element stress analysis. Wire positions are defined by locating the wires against small pins made of polyoxymethylen (diameter 1.5 mm) imbedded in the anode frame. They are precisely located initially along a curved trajectory which is deflected into a straight line once the wire tension is applied. All wire positions were measured by an optical surveying system built for this purpose and found to deviate less than 30 Frn from nominal positions [3], i.e. within the specifications. The fast non-flammable gas mixture Ar (90%), CO, (5%), CF., (5%) yields drift velocities ranging up to 70 pmlns at room temperature and atmospheric pressure. After one year of running no ageing effects were seen in a prototype chamber. The drift chambers are read out via cards mounted on the chambers containing preamplifiers (transconductance 60 mV/p,A), comparators, and ECL-drivers. The LeCroy 1877 Fastbus TDCs are located some 30 m away from the
S. Bemreuther et al. I Nucl. Instr. and Meth. in Phys. Res. A 367 (1995) 96-99
GAS
97
IN’
EC-MODULE
Fig. I. Schematic view of the large BC module. spectrometer. The total number of channels amounts to 7680.
All chamber positions in HEWS will be on-line monitored relative to other tracking detector components by means of a laser alignment system with 50 pm design accuracy.
sense wires by electrical discharge, it- is important to limit the capacitance of the chamber. The total capacitance including additional capacitors for high voltage buffering must be kept low, e.g. in case of 25 cl;m sense wire diameter it should be smaller than 8 nF [5]. In addition, the high voltage power supply needs a current limiting kill mode.
3. Practical experience
4. Test beam data analysis - Dark current. The chambers exhibit low dark currents of 0.1 PA per module at a high voltage of 1800V. This
was achieved by a high voltage training procedure at the beginning of the chamber’s operation using alternative polarities at limited currents not higher than 5 p,A per module. It is reasonable to assume that in case of reverse polarity the sense wires are cleaned by ion bombardement r41. -Grounding.The quality of grounding strongly influences the chamber performance. It is essential to ground in a plane-like manner the readout cards and the high voltage connectors to the steel frame, and to realize a c&d Faraday cage around the chamber. - Wire protection. To avoid damage of the delicate Table 1 Some parameters for two different Back Chamber sizes Module active area (mm*) Sense wires per plane Channels per module Cell width X gap Anode wires Potential wires Cathode foils Rad. length per module
1880 x 520
2890 x 710 192 1152 15mmX 16mm 25.4 pm Au coated W 127pm Au coated Cu-Be 25 pm C coated Kapton 0.26%
124 768
During 1992-1994 test runs were made at the ,DESY test beam facility to study a small 6-plane and a production-size 3-plane prototype, and finally the first two serial production chambers. Silicon microstrip detectors with 20 p,rn internal resolution [6] interleaving the drift chambers were used to precisely define incoming electrons with energies between 2 and 4 GeV. On average one full drift cell was illuminated at the given beam profile. The results discussed in the following are not final and rely on test run data obtained with ‘the first serial production module using tracks crossing perpendicularly. Data taken for planes with inclined wires and with non-perpendicular tracks are still to be analysed. The working regime of the drift chambers under test run conditions was adjusted to optimise resolution and efficiency together with an acceptably small amount of crosstalk. The latter was measured in a given channel as percentage of signals induced in the neighbouring channels. Note that both inter-channel and inter-plane crosstalk were studied, the latter being small compared to the former. For optimum conditions the electronics threshold had to be chosen as low as possible to detect the earliest drift electrons and to minimize the amplifier walk, and high enough to prevent the electronics from oscillating.
III. DRIFT CHAMBERS
98
S. Bemreuther
et al.
I Nucl. Instr.
and Meth.
The high voltage had to be chosen high enough to achieve stable chamber operation and as low as possible to prevent the chamber from premature ageing. The gross features of the chamber behaviour were investigated utilising a simple linear space drift time relation (SDTR). In Fig. 2 the chamber resolution in the central drift region is shown as a function of high voltage and electronics threshold. The best value was obtained for the setting lSOOV/5OmV. Here the crosstalk is rather large, as can be seen from Fig. 3. While the resolution deteriorates only slightly when changing the high voltage within 250 V or the electronics threshold within ? 10 mV, the crosstalk changes considerably. Hence by e.g. choosing 1750 V/50 mV a crosstalk of about 6% can be reached at almost no loss in resolution. A non-linear space drift time relation was extracted to allow the precise determination of the chamber parameters over the drift range and to serve as initial calibration for HERMES running in 1995. In Fig. 4 the drift times are plotted against the eternal reference coordinates for a full drift cell. The data correspond to 50 000 tracks recorded at 1800 V/60 mV. In the left half of Fig. 4 a least squares fit is superimposed using a 4th order polynomial which yields a very satisfactory parametrisation of the SDTR with a reduced x2 of about 1.3. The right half of the same figure shows the corresponding result from on-line calibration which will be discussed in the next section. The chamber resolution, averaged over the whole drift cell and calculated from the same data with the same non-linear SDTR, is shown in Fig. 5. While this average is about 180 pm the differential behaviour of the resolution over the drift cell has its best value below 150 pm and deteriorates significantly at the potential wire. The variation of the chamber efficiency is presented in Fig. 6. Variable bin sizes were chosen to clearly demon-
y: -1 in Phys. Res. A 367 (1995)
Es? 2.5
’
3 w
20
: :
96-99
:
HV =; 18OO’V + ,, ,..,:
"
175o;v /
A
E
V
15
10
5
1 t
o30
40
50
60
70
80
90
Threshold
100
110
( mV )
Fig. 3. Crosstalk as a function of high voltage (HV) and electronics threshold. The curves are drawn IO guide the eye.
strate the efficiency behaviour when approaching sense or potential wire. The chamber was considered efficient if for a given reference track position a signal was found within a corridor of ?~cT, where u is the average chamber resolution given above. As can be seen, the efficiency has a stable plateau over almost the full drift range. This plateau value does not significantly depend on the threshold and improves slightly with high voltage (not shown). When approaching the sense wire the efficiency decreases to 84%, whereas at the potential wire it drops to 22%. The average efficiency over the whole drift range is 96%.
60
0
140
’
30
I
I
I
I
I
50
60
70
80
Fig. 2. Resolution electronics
threshold.
-6
-4
-2
Reference
40
Threshold
.a
( mV )
as a function of high voltage (HV) and The curves are drawn to guide the eye.
0
2
4
coordinate
6
8
( mm )
Fig. 4. Measured drift time versus reference coordinate of the silicon microstrip detector. Both curves represent Ieast squares fits to 4th order polynominals (left: off-line analysis; right: on-line monitoring).
S. Bemreuther et al. I Nucl. Instr. and Meth. in Phys. Res. A 367 (1995) %-99
f 8
with six measurements. A least squares fit to this track allows to determine up to two free parameters, as e.g. trigger time offset and average drift velocity or average resolution. The coefficients of a higher order polynomial describing a non-linear SDTR can be obtained when the number of free parameters is increased by a common fit to several tracks. This procedure was applied to the same data set as used in the previous section. When comparing the right curve (on-line) in Fig. 4 with the left one (off-line) it becomes evident that the on-line result is compatible with the exact off-line analysis. Note that the on-line selfcalibration procedure averages over all six planes of a module. The algorithm is able to cope with a counting rate of 50Hz as expected for 1995 HERMES running.
700$ x plane
600 SW
~
100
I I
zlm
loo
0
-3
-2
-I
I
0
99
2
residual of measured coordinate ( mm ;
6. Summary
Fig. 5. X plane residual distribution (resolution) at the working point: high voltage/threshold
= 1800 V/60 mV.
Excluding 500 p,m around the potential wire increases the average efficiency to 96.5%.
5. On-liue chamber calibration A fast and alignment independent on-line monitoring of the chamber status is desired during operation of the detector. It can be real&d by self-calibrating a given module without using external reference tracks. A fully efficient chamber module responds to an incoming particle
Including spare modules, 10 large 6-plane drift chamber modules were built for the HERMES downstream tracking system. Several prototypes and the first two serial production chambers were extensively tested in an electron beam. Using the non-flammable gas mixture Ar/CO,/CF, @O/5/5) average spatial resolutions of about 180 pm have been measured at efficiencies above 96%. The optimum working regime of the chamber as derived from test run conditions requires the high voltage between 1750 and 1800 V and the electronics threshold between 50 and 60 mV. A non-linear space drift time relation was determined to serve as initial calibration for 1995 HERMES running.
Acknowledgements The authors express their gratitude to U. Gensch, J. May, I? Sting, and A. Wagner for their continuous support of this project and for many helpful discussions. The skilful and engaged participation of the mechanical and electronics workshops in Erlangen and Zeuthen was absolutely essential to build the whole system according to schedule and with high quality.
References
0
I
2
3
4
Drift
5
distance
6
7
( mm )
Fig. 6. Efficiency as a function of the drift distance at the working
111 J. van den Brand et al., NIKHEF-H/94-36. 121 The HERMES Collaboration, Technical Design Report, DESY-PRC 93/06 (1993). I31 A. Gute, Diploma thesis, Universitit Erlangen (1994). PI R. Henderson, TRIUMF, Vancouver, private communication. [5] P Steffen, DESY, Hamburg, private communication. [61 J. B%r et al., Nucl. Instr. and Meth. A 324 (1993) 145.
point: high vokage/threshold = 1800 Vi60 mV
III. DRJFI’ CHAMBERS
and Methods in Physics Research
A 367 (1995) 96-99
ELSEVIER
Design and performance of the large HERMES drift chambers S. Bernreuther”,
H. B6ttcherb, Harderb, D. Haschb, M. A. Miller”, F. Neunreithe?, F.
A.B. Borissovb, W. Briicknerd, M. Ferstl”, A. Gute”, U. Kirscha, B. Krauseb, W.A. Lachnit”, F. Mei13nerb, W.-D. Nowakb, M. Pohlb, H.O. Roloffb, H. Russoa, Schmidt”, A.E. Schwindb’*
“Universitiit Erlangen-Niirnberg, 91058 Erlangen, Germany “DESY-IfH Z&then, 15735 Zeuthen, Germany ‘TRIUMF, Vancouver, BC. Canada, V6T-2A3 dMPl fiir Kerphysik, 69029 Heidelberg. Gemany
Abstract Big planar drift chambers built for the downstream tracking system of the HERMES spectrometer are described. Using the fast non-flammable gas mixture Ar/CO,/CF, (901515) average spatial resolutions of about 180 pm per plane at efficiencies above 96% have been obtained from test run data analysis. 1. Introduction The HERMES apparatus is designed as a fixed target spectrometer to investigate the spin structure of the nucleon by measuring doubly polarised lepton-nucleon scattering with an internal target in the HERA lepton ring. The spectrometer is split into two halves installed above and below the two HERA beam pipes. Each spectrometer half is equipped with the same sequence of detectors. In front of the spectrometer magnet, the tracking system is composed of two 3-plane ,microsttip gas counters [l] and a pair of 6-plane drift chambers. Three proportional chambers are mounted within the magnet to reduce tracking ambiguities. Behind the magnet, tracking is based upon two pairs of large 6-plane drift chambers (Back Chambers), which will be described in this paper. Particle identification is accomplished by a threshold Cherenkov counter, a transition radiation detector and an electromagnetic calorimeter, the latter also being used for triggering.’ A complete description of the spectrometer can be found in Ref. [2]. In Section 2 the Back Chamber design and construction are described, followed by some practical experience presented in Section 3. Test beam results are discussed in Section 4. The on-line chamber calibration procedure is outlined in Section 5 and in Section 6 a summary is given.
2. Design and construction Each chamber or module consists of six sense wire planes interleaved with cathode foils. As schematically * Corresponding author. E-mail [email protected] Elsevier Science B.V. SSDI
0168-9002(95)00755-5
shown in Fig. I, sense wires are tilted by ?30° in the first and last pair of these six planes with each second plane having the sense wires staggered by half a drift cell. The cathode foils and potential wires are at negative high voltage. The spectrometer layout requires two different Back Chamber sizes. For both types some parameters are given in Table 1. Both sense wire and cathode foil frames were made of GFK (glass fiber-reinforced epoxy) by Stesalit (Zullwil, Switzerland). To avoid frame distortion the GFK frames are sandwiched between two steel frames of 38 mm in thickness each. In the mechanical design special attention was paid to high stability and high wire positioning accuracy in the sense wire frames. The deformation of the latter due to wire tension was taken into account using a finite element stress analysis. Wire positions are defined by locating the wires against small pins made of polyoxymethylen (diameter 1.5 mm) imbedded in the anode frame. They are precisely located initially along a curved trajectory which is deflected into a straight line once the wire tension is applied. All wire positions were measured by an optical surveying system built for this purpose and found to deviate less than 30 Frn from nominal positions [3], i.e. within the specifications. The fast non-flammable gas mixture Ar (90%), CO, (5%), CF., (5%) yields drift velocities ranging up to 70 pmlns at room temperature and atmospheric pressure. After one year of running no ageing effects were seen in a prototype chamber. The drift chambers are read out via cards mounted on the chambers containing preamplifiers (transconductance 60 mV/p,A), comparators, and ECL-drivers. The LeCroy 1877 Fastbus TDCs are located some 30 m away from the
S. Bemreuther et al. I Nucl. Instr. and Meth. in Phys. Res. A 367 (1995) 96-99
GAS
97
IN’
EC-MODULE
Fig. I. Schematic view of the large BC module. spectrometer. The total number of channels amounts to 7680.
All chamber positions in HEWS will be on-line monitored relative to other tracking detector components by means of a laser alignment system with 50 pm design accuracy.
sense wires by electrical discharge, it- is important to limit the capacitance of the chamber. The total capacitance including additional capacitors for high voltage buffering must be kept low, e.g. in case of 25 cl;m sense wire diameter it should be smaller than 8 nF [5]. In addition, the high voltage power supply needs a current limiting kill mode.
3. Practical experience
4. Test beam data analysis - Dark current. The chambers exhibit low dark currents of 0.1 PA per module at a high voltage of 1800V. This
was achieved by a high voltage training procedure at the beginning of the chamber’s operation using alternative polarities at limited currents not higher than 5 p,A per module. It is reasonable to assume that in case of reverse polarity the sense wires are cleaned by ion bombardement r41. -Grounding.The quality of grounding strongly influences the chamber performance. It is essential to ground in a plane-like manner the readout cards and the high voltage connectors to the steel frame, and to realize a c&d Faraday cage around the chamber. - Wire protection. To avoid damage of the delicate Table 1 Some parameters for two different Back Chamber sizes Module active area (mm*) Sense wires per plane Channels per module Cell width X gap Anode wires Potential wires Cathode foils Rad. length per module
1880 x 520
2890 x 710 192 1152 15mmX 16mm 25.4 pm Au coated W 127pm Au coated Cu-Be 25 pm C coated Kapton 0.26%
124 768
During 1992-1994 test runs were made at the ,DESY test beam facility to study a small 6-plane and a production-size 3-plane prototype, and finally the first two serial production chambers. Silicon microstrip detectors with 20 p,rn internal resolution [6] interleaving the drift chambers were used to precisely define incoming electrons with energies between 2 and 4 GeV. On average one full drift cell was illuminated at the given beam profile. The results discussed in the following are not final and rely on test run data obtained with ‘the first serial production module using tracks crossing perpendicularly. Data taken for planes with inclined wires and with non-perpendicular tracks are still to be analysed. The working regime of the drift chambers under test run conditions was adjusted to optimise resolution and efficiency together with an acceptably small amount of crosstalk. The latter was measured in a given channel as percentage of signals induced in the neighbouring channels. Note that both inter-channel and inter-plane crosstalk were studied, the latter being small compared to the former. For optimum conditions the electronics threshold had to be chosen as low as possible to detect the earliest drift electrons and to minimize the amplifier walk, and high enough to prevent the electronics from oscillating.
III. DRIFT CHAMBERS
98
S. Bemreuther
et al.
I Nucl. Instr.
and Meth.
The high voltage had to be chosen high enough to achieve stable chamber operation and as low as possible to prevent the chamber from premature ageing. The gross features of the chamber behaviour were investigated utilising a simple linear space drift time relation (SDTR). In Fig. 2 the chamber resolution in the central drift region is shown as a function of high voltage and electronics threshold. The best value was obtained for the setting lSOOV/5OmV. Here the crosstalk is rather large, as can be seen from Fig. 3. While the resolution deteriorates only slightly when changing the high voltage within 250 V or the electronics threshold within ? 10 mV, the crosstalk changes considerably. Hence by e.g. choosing 1750 V/50 mV a crosstalk of about 6% can be reached at almost no loss in resolution. A non-linear space drift time relation was extracted to allow the precise determination of the chamber parameters over the drift range and to serve as initial calibration for HERMES running in 1995. In Fig. 4 the drift times are plotted against the eternal reference coordinates for a full drift cell. The data correspond to 50 000 tracks recorded at 1800 V/60 mV. In the left half of Fig. 4 a least squares fit is superimposed using a 4th order polynomial which yields a very satisfactory parametrisation of the SDTR with a reduced x2 of about 1.3. The right half of the same figure shows the corresponding result from on-line calibration which will be discussed in the next section. The chamber resolution, averaged over the whole drift cell and calculated from the same data with the same non-linear SDTR, is shown in Fig. 5. While this average is about 180 pm the differential behaviour of the resolution over the drift cell has its best value below 150 pm and deteriorates significantly at the potential wire. The variation of the chamber efficiency is presented in Fig. 6. Variable bin sizes were chosen to clearly demon-
y: -1 in Phys. Res. A 367 (1995)
Es? 2.5
’
3 w
20
: :
96-99
:
HV =; 18OO’V + ,, ,..,:
"
175o;v /
A
E
V
15
10
5
1 t
o30
40
50
60
70
80
90
Threshold
100
110
( mV )
Fig. 3. Crosstalk as a function of high voltage (HV) and electronics threshold. The curves are drawn IO guide the eye.
strate the efficiency behaviour when approaching sense or potential wire. The chamber was considered efficient if for a given reference track position a signal was found within a corridor of ?~cT, where u is the average chamber resolution given above. As can be seen, the efficiency has a stable plateau over almost the full drift range. This plateau value does not significantly depend on the threshold and improves slightly with high voltage (not shown). When approaching the sense wire the efficiency decreases to 84%, whereas at the potential wire it drops to 22%. The average efficiency over the whole drift range is 96%.
60
0
140
’
30
I
I
I
I
I
50
60
70
80
Fig. 2. Resolution electronics
threshold.
-6
-4
-2
Reference
40
Threshold
.a
( mV )
as a function of high voltage (HV) and The curves are drawn to guide the eye.
0
2
4
coordinate
6
8
( mm )
Fig. 4. Measured drift time versus reference coordinate of the silicon microstrip detector. Both curves represent Ieast squares fits to 4th order polynominals (left: off-line analysis; right: on-line monitoring).
S. Bemreuther et al. I Nucl. Instr. and Meth. in Phys. Res. A 367 (1995) %-99
f 8
with six measurements. A least squares fit to this track allows to determine up to two free parameters, as e.g. trigger time offset and average drift velocity or average resolution. The coefficients of a higher order polynomial describing a non-linear SDTR can be obtained when the number of free parameters is increased by a common fit to several tracks. This procedure was applied to the same data set as used in the previous section. When comparing the right curve (on-line) in Fig. 4 with the left one (off-line) it becomes evident that the on-line result is compatible with the exact off-line analysis. Note that the on-line selfcalibration procedure averages over all six planes of a module. The algorithm is able to cope with a counting rate of 50Hz as expected for 1995 HERMES running.
700$ x plane
600 SW
~
100
I I
zlm
loo
0
-3
-2
-I
I
0
99
2
residual of measured coordinate ( mm ;
6. Summary
Fig. 5. X plane residual distribution (resolution) at the working point: high voltage/threshold
= 1800 V/60 mV.
Excluding 500 p,m around the potential wire increases the average efficiency to 96.5%.
5. On-liue chamber calibration A fast and alignment independent on-line monitoring of the chamber status is desired during operation of the detector. It can be real&d by self-calibrating a given module without using external reference tracks. A fully efficient chamber module responds to an incoming particle
Including spare modules, 10 large 6-plane drift chamber modules were built for the HERMES downstream tracking system. Several prototypes and the first two serial production chambers were extensively tested in an electron beam. Using the non-flammable gas mixture Ar/CO,/CF, @O/5/5) average spatial resolutions of about 180 pm have been measured at efficiencies above 96%. The optimum working regime of the chamber as derived from test run conditions requires the high voltage between 1750 and 1800 V and the electronics threshold between 50 and 60 mV. A non-linear space drift time relation was determined to serve as initial calibration for 1995 HERMES running.
Acknowledgements The authors express their gratitude to U. Gensch, J. May, I? Sting, and A. Wagner for their continuous support of this project and for many helpful discussions. The skilful and engaged participation of the mechanical and electronics workshops in Erlangen and Zeuthen was absolutely essential to build the whole system according to schedule and with high quality.
References
0
I
2
3
4
Drift
5
distance
6
7
( mm )
Fig. 6. Efficiency as a function of the drift distance at the working
111 J. van den Brand et al., NIKHEF-H/94-36. 121 The HERMES Collaboration, Technical Design Report, DESY-PRC 93/06 (1993). I31 A. Gute, Diploma thesis, Universitit Erlangen (1994). PI R. Henderson, TRIUMF, Vancouver, private communication. [5] P Steffen, DESY, Hamburg, private communication. [61 J. B%r et al., Nucl. Instr. and Meth. A 324 (1993) 145.
point: high vokage/threshold = 1800 Vi60 mV
III. DRJFI’ CHAMBERS