Bunch timing differences

1
Bunch timing differences

• Warm 192 x 1.4 ns spacing _at_ 120 Hz
• 23,000 bunches /second

• Cold 2820 x 337ns spacing _at_ 5 Hz
• 14,000 bunches /second

Implication for daq

• Warm integrate over bunches, tag time in
  tracker, ecal and hcal (duty cycle 1/900)
• Separately record bunches in lumcal

• Cold analog or digital pipeline needed,
  separately record bunches, no time tagging (duty
  cycle 1/200)

2
Luminosity assumptions

• Pairs/crossing (Jaros Paris Talk)

Occupancies/Train8600 ee- pairs 35k ?s
(MeV) 154 ??- pairs 56 had events
3
Physics Impact Higgs in WW-fusion _at_ 1 TeV
Effect of pt cut on signal and on dominant (ZZ)
background
(K. Desch talk at LCWS 2004)
t
4
Physics Impact 3 Higgs in WW-fusion _at_ 1 TeV
normalized to 1BX _at_ TESLA (effect smaller at
warm machine due to lower lumi/bunch)
5
0 BX
1 BX
4 BX
0.0031
0.0033
0.0035
T. Barklow studies at 1 TeV, relative error on
yield h to bb, Includes all gamma gamma
background (1ab-1)
6
Toshinori Abe Study of Higgs Mass Resolution
exclude low pt tracks (Pt gt 1.0 GeV), use
expected resolution in kinematic fit
7
Abes conclusion on Higgs mass error
8

• Stau search -- must veto 250 GeV electrons in
  forward calorimeter,
• No need for precision measurement of the electron
  energy
• Unlike previously discussed studies, this
  require
• a measurement of every bunch.
• Results from takashi, graf and Bombade

9

• Conclusion from physics studies
• Modest impact from integrating over 1-4 bunch
  crossings for most physics
• We should do our best to build a detector which
  can associate tracks and clusters with correct
  beam crossing
• Only the far forward detectors will need single
  bunch integration

10
Occupancies are not a problem. Simple time
tagging works everywhere except in lumcal (T.
Abe)
11
Timing in Silicon

• Charge collection time depends on bias voltage
• For 300 micron silicon and bias voltages 40V or
  more volts above depletion all charge can be
  collected in less than 25ns.
• Can reduce collection time to less than 10ns by
  raising the bias voltage
• In most situation timing will be limited by
  electronics
• The charge collection time can be further reduced
  by decreasing carrier lifetime (see diamond)

12
Expected resolution depends on signal to noise
and on the shaping time of the amplifier
where SN is the signal to noise and t is the
shaping time of the amplifier. Assume 3 sigma
threshold.
SN of 15 and t 40 ns or better have been
demonstrated in many silicon systems Layer
resolution of better than 5ns possible
13
Existence Proofs
Laboratory (D. Strom U Oregon) measurements of
resolution from cosmic ray MIPs from pad
detectors similar to LC prototypes
Electronics parameters t 25 ns SN 15
After correction for time walk (right) layer
resolution is 3.4ns
14
Time Walk Correction
Time walk correction determined from infrared
laser data. Correction good to 1ns
15
Time measurement for showers
Use infrared laser light to simulate pixels in
GeV showers
Resolution for 6 MIP signal better than 1 ns
16
Measurements in Babar
6.3 ns average timing resolution obtained for
tracks. Good result due to imperfect phasing of
clocks. Shaping time is 100 ns -- not originally
designed for precision time measurements. Time
walk correction essential
Gerry Lynch and H. Sadrozinski
17
Many scintillation system have achieved sub
nanosecond time resolution

• KLOE fiber calorimeter (200ps) (NIMA 482, 364)
• Mark II time of flight (180ps) (NIM 221, 503)
• RPC time-of-flight (90ps)(IEEE TNS
  481658-1663,2001
• http//dustbunny.physics.indiana.edu/FSUwork/TOF.h
  tml

18
Simulation of ECAL timing of MIPs

• ECAL timing measurements depend on distance of
  pixels from the readout chip and Landau
  fluctuations
• A detailed simulation has been developed to
  estimate time resolution for the 30 samples in
  the calorimeter.
• Many effects considered, including
• Variation in capacitance of traces
• Finite resistively of traces
• Channel-to-channel threshold differences (5)
• Time walk calibration errors (5)
• Landau fluctuations

19
Reasonable (but not final) parameters have been
used in the simulation of the electronics, e.g.
the transconductance of the input FET was assumed
to be only 1.5mS and shaping time 50ns.
Truncated mean for 30 calorimeter samples has
resolution 0.7ns
Sub-nanosecond systematics possible
if disciplined -- Radeka
20
Bunch-by-bunch recording in the lumcal
In the far forward region it will be necessary to
separately record data in each of the bunches in
the bunch train because of the very large
background from low energy pairs (see slide 1).

• This detector will be used to veto very large
  signals 250GeV (gt1000 MIPs) from electrons in
  gamma-gamma events.
• Special low electronics will not be needed,
  signals will be very large pC rather than fC.
• Electronics can be located outside of the detector

21
Diamond existence proof 1
22
Diamond existence proof (2)
23
Other materials are possible

• Quartz fibers
• Thin low resistivity silicon -- mobility of
  carriers in diamond is only slightly faster than
  in silicon
• Other semiconductors (eg GaAs)