1
Multigap glass RPCs in HARP
Design
Running experience
Results
I. Boyko, G.Chelkov, D. Dedovich, A. Elagin, M.Gostkin, Y.Nefedov, K. Nikolaev, A. Zhemchugov (JINR Dubna),
V. Ammossov, V. Koreshev (IHEP Protvino),
F. Dydak, J. Wotschack (CERN
Presented by: Joerg Wotschack (CERN)
RPC2005, 10–12 Oct. 2005 Joerg Wotschack (CERN) HARP RPCs / 2
RPCs in the HARP detector
RPCs:-30 barrel (around TPC)-16 f/w (before 1st DCH)
Total number of readout channels: 368Area covered: 8 m2
Time-of-flight over 0.4–2m for e/π separation below 300 MeV/c . Design goal:- Time resolution: 200 ps- High efficiency
Hadron production experiment at CERN PSHARP data taking: 2001 & 2002
RPC2005, 10–12 Oct. 2005 Joerg Wotschack (CERN) HARP RPCs / 3
RPC design - glass stack 4-gap glass RPC
Glass stack:1920 mm x 106 mm x 7.6 mm 6 glass plates: 0.7 mm; gap size: 0.3 mm (spacer: fishing line) HV: -6 kV (over two gas gaps) Central readout electrode for all four gas gaps
RPC2005, 10–12 Oct. 2005 Joerg Wotschack (CERN) HARP RPCs / 4
RPC design - layout barrel
Glass stack Pre-amplifier
30 RPCs in two layers100% coverage
Looking upstream
RPC2005, 10–12 Oct. 2005 Joerg Wotschack (CERN) HARP RPCs / 5
Implementation in HARP
Barrel: 30 RPCs in 2 layers• Length: 2 m• Width: 150 mm• Thickness: 10 mm
preamplifiers
RPC2005, 10–12 Oct. 2005 Joerg Wotschack (CERN) HARP RPCs / 7
RPC design - pad structure
RPC2005, 10–12 Oct. 2005 Joerg Wotschack (CERN) HARP RPCs / 8
RPC readout scheme
QDCTDC
Trigger
RPC modulesPA
(12 QDCs)(12 TDCs)
Splitter
8 channels
16 channels80 m twisted pair
80 m twisted pair
5 m coax8 channels
8 summing preamplifiers per RPC (on chamber)
PA connected to 8 strips (strip = 30 x 104 mm2)
Splitter at 5 m distance Timing: discr. thr.: 5 mV Charge
80 m twisted pair cables TDC: CAEN V775 (35 ps) QDC: CAEN V792 (0.1pC)
RPC2005, 10–12 Oct. 2005 Joerg Wotschack (CERN) HARP RPCs / 9
RPC operating conditions Gas: C2F4H2:iCH10:SF6 (90:5:5); ~1 volume change/hr HV: -6 keV over 2 gas gaps Random hits (noise + cosmics)
Monitored over two years - stable Typical rates: 200–300 Hz/RPC (2000 cm2) ≤ 0.1 Hz/cm2
Low particle rates: ≤ 1 Hz/cm2
Beam intensity: < 20000 per spill (400 ms) Typically 1000 interactions per spill (0.05 target) Average multiplicity: 4 (in barrel RPC acceptance) Rate: ~10 kHz/5 m2 ≤ 1 Hz/cm2 (barrel)
Temperature: 20–35 ºC in experimental area Barrel RPCs temperature stabilized: 27–30 ºC (±0.5 ºC ) Forward RPCs exposed to hall temperature
RPC2005, 10–12 Oct. 2005 Joerg Wotschack (CERN) HARP RPCs / 10
Data sets Scan of four RPCs exposed to 12 GeV/c π- beam
Global time-slewing correction (time measured vs charge) Time & charge response vs impact position (x and y) Efficiencies and time resolution
(Results presented earlier and not covered here, see RPC2003)
Physics tracks with RPCs in HARP detector (2002) Corrections for electronics effects Corrections to global time-slewing correction t0 for each pad Charge response as function of impact angle Time resolution & efficiencies in HARP
RPC2005, 10–12 Oct. 2005 Joerg Wotschack (CERN) HARP RPCs / 11
Steps from raw RPC time to TOF Convert TDC counts to picoseconds Correct for temperature effects Subtract arrival time of beam particle in target
(measured by beam line instrumentation) Apply global time-slewing correction Correct for impact point dependence of timing
Strip number Hit position along strip -> modification of global time-
slewing correction Determine and apply pad specific t0 constants
RPC2005, 10–12 Oct. 2005 Joerg Wotschack (CERN) HARP RPCs / 12
Temperature effects I Time response is strong
function of temperature in experimental area
Channel dependent day-night variations of t0 of up to 900 ps (!)
=> Corrtemp= 60 ±10 ps/C Not a detector effect:
barrel RPCs are temp.stabilized (±0.5C)
Threshold shift in splitter-discriminator electronics
Barrel RPCs
600 ps
+8.9 GeV Be
RPC2005, 10–12 Oct. 2005 Joerg Wotschack (CERN) HARP RPCs / 13
Temperature effects II Time response f/w RPCs
Similar as in barrel
t/T ≈ 54 ± 6.5 ps/ C.
Forward RPCs are fully exposed to T
Suggests: small contribution from detector itself
Charge response
No temperature variation of charge for barrel RPCs
Clear effect in forward RPCs
Q/T≈ 3%/ C
Ch
arge
(0.
1 p
C)
Temperature
Charge vs temperaturef/w RPCs
t/T slopesf/w RPCs
t/T slopes
RPC2005, 10–12 Oct. 2005 Joerg Wotschack (CERN) HARP RPCs / 15
Time response along strip I Time response is function of
impact point distance to pre-amplifier (PA)
Time difference b/w near and far ends of strip (x ~ 90mm):
200 ps for large Q 0 ps for small QExpect: 450–500 ps
Effect explained by pulse reflection on not-terminated strip end and superposition of signals (simulation agrees with data)
Requires charge-dependent modification of global time-slewing correction
Charge (0.1 pC)
Impact point position along strip (mm)
Slo
pe
(ps/
mm
)T
ime
resp
onse
(p
s)
PAPA
Width of RPC
Small charges Large charges
Scan data
RPC2005, 10–12 Oct. 2005 Joerg Wotschack (CERN) HARP RPCs / 17
pad ring 6
--- far–– near
Time-slewing correction - revisited Pragmatic approach:
Measure difference in time response b/w far and near end of strips as function of charge
Use results as effective correction for impact positions at strip ends
For impact points along strip use an interpolation based on an analytical model calculation
Results in modification of global time-slewing correction which is different for the eight pad rings
RPC2005, 10–12 Oct. 2005 Joerg Wotschack (CERN) HARP RPCs / 18
Pad specific t0 constants t0 normalization depends on cable
delays and has to be determined for each pad individually from physics data
Method: photon conversions Determine 50% point of rising
edge of time spectrum (t50% ) Calculate its relative position wrt
time expected for hits in pad centre (tcorr, analyt. simul.)
Relate to nominal time of flight b/w target and centre of pad
TOF() = t50% + tcorr – tz0 – t0
( tz0 = beam arrival time in target)
Estimated uncertainty: ~30 ps
Typical time spectra for + neutrals ( conv.) – tracks
Pad 151(pad ring 7)
t50%
RPC2005, 10–12 Oct. 2005 Joerg Wotschack (CERN) HARP RPCs / 19
Stability of t0 constants Coherent shifts of t0
constants for different run periods3 Ta - 8.9 Be: t0 = -250 ps
H2O - 8.9 Be: t0 = +70 ps
but: temperature slopes agree (!)
Likely explanation: long-term threshold shifts in the discriminator/splitter electronics
Requires t0 calibration for each run period May 2002 (-8 GeV/c)
August 2002(+8.9 GeV/c)
pad 24
t0(Ta) - t0(Be)t0(H2O) - t0(Be)
Be(-) run: May 2002
Ta run: June 2002
Be(+) run: Aug 2002
H2O run: Sept 2002
RPC2005, 10–12 Oct. 2005 Joerg Wotschack (CERN) HARP RPCs / 20
Results
Charge vs track length System efficiency Time resolution Particle identification
RPC2005, 10–12 Oct. 2005 Joerg Wotschack (CERN) HARP RPCs / 21
Charge response vs track angle
Charge deposited in RPCs for charged particles as function of pad ring (= track impact angle)
Charge deposited in RPCs for charged particles as function of path length in detector
RPC2005, 10–12 Oct. 2005 Joerg Wotschack (CERN) HARP RPCs / 22
System efficiency Intrinsic RPC efficiency was
measured in scan (at high particle rate) to be 97–98%
System efficiency is expected to be lower absorption in material in
front of RPCs large energy-loss for low
momentum protons Measured values in HARP:
Eff = 97–98%
Is a lower limit on intrinsic RPC efficiency
positive tracks + negative tracks
RPC2005, 10–12 Oct. 2005 Joerg Wotschack (CERN) HARP RPCs / 23
Time resolution - physics data t from physics tracks
through pad overlaps Same track is measured
twice Independent of beam timing Peak position checks t0s and
time-slewing correction Width measures convoluted
resolution of the two pads Result (for all pads in barrel)
/√2 = 145 ps Narrow Gaussian (85%) on
top of a wider distribution
Barrel RPCs
RPC2005, 10–12 Oct. 2005 Joerg Wotschack (CERN) HARP RPCs / 24
Time resolution vs pad ring
Time resolution vs pad ring
pad rings correspond to track inclination wrt RPCpad ring 2/3: ≈ 90°pad ring 6/7: ≈ 30°
RPC2005, 10–12 Oct. 2005 Joerg Wotschack (CERN) HARP RPCs / 25
Time resolution - the tails Noise: genuine low charge hits
for which threshold is passed too early because of overlayed noise. Results in enhancement at low-charges.
Knock-on: low-energy particles kicked out from RPC material and trapped in magnetic field; they move slowly in RPC adding charge some ns after genuine hit. Results in correct time signal but too large charge and therefore wrong time-slewing correction(effect only present in magnetic field)
t > 400 ps
t > 400 ps
Data points normalized to same shape as t < 400 ps spectrum
Tracks through pad overlapsCharge spectra for peak and tails
RPC2005, 10–12 Oct. 2005 Joerg Wotschack (CERN) HARP RPCs / 26
Particle identification (I)+8.9 GeV/c 0.05 Be target – pad ring 5 (average t)
positive tracks
pπ
e
Bet
a
Momentum (GeV/c)
negative tracks
Momentum (GeV/c)
p
π
e
Pad ring 5 Pad ring 5
RPC2005, 10–12 Oct. 2005 Joerg Wotschack (CERN) HARP RPCs / 27
Electron enriched sample: photon conversion candidatestwo tracks with same origin and production angle
Particle identification (II)
+8.9 GeV/c on Be target (0.05 )
<> = 0.995 ± 0.003 = 8.3%
RPC2005, 10–12 Oct. 2005 Joerg Wotschack (CERN) HARP RPCs / 28
What have we learned? Multigap glass RPCs are great detectors: fast, precise,
efficient, and robust Detector design OK, but …
Strip termination would have made our life much easier. Threshold drifts of discriminators with temperature and ‘time’. Differences in signal transmission b/w strips and preamplifier.
Small fraction of wrong time measurements Low threshold => noise correlated with hits (wrong time, early) Knock-on particles => right time, wrong charge = wrong TS corr.
Overall: the system worked extremely well, final result
time = 145 ps; system efficiency ≈ 98%