IFAEATLAS,

1
Using TileCal as a Luminosity Monitor
I.Korolkov Sep05

• Outline
• Luminosity measurements at ATLAS and LHC
• Min.Bias Monitoring System of TileCal
• MBMing for Luminosity Monitoring

2
Luminosity measurements at LHC

• Luminosity needed for
• Precision comparison with theory
• e.g. ?bb, ?tt, ?W/Z, ?jet, ?H, ?SUSY,
• Cross section gives additional physics
• information
• Precision comparison with other expts
• at LHC, TeVatron, LEP, HERA,
• Luminosity from
• LHC machine parameters beam
• instrumentation (5-10)
• Rate measurements of precisely-calculable
  processes (1-2)

3
Goals for the Luminosity measurements at LHC

• ? L/L goals
• LHC machine 5-10 ? measurements of beam
  profiles and positions
• ATLAS 2 ? normalization to Coulomb scattering
  (Roman Pots)
• TOTEM/CMS 1-2 ? using Ntotalextrapld,
  dNelastic/dt t0, and the OT (forward
  trackers 3lt?lt7 and Roman Pots)
• LHCb 5-6 ? Nvtx0/Nvtx?1 rates, MC tuned with
  dN/d?, ?tot, ?inel, ?SD ? 2 ? using ?b?
• ALICE pp 5 ? using Ninel (A86), MC
  extrapolation, and ?tot, ?inel HI lt10
  (standard) ? 2 ? QED
• L Monitoring
• using (dedicated) detectors (TileCal) clean
  physics signals e.g. W/Z ?leptons,

4
Luminosity Topics in ATLAS

• Absolute Luminosity
• Elastic scattering and Roman Pots
• Options
• W/Z counting
• Machine Parameters
• Machine parameters and size via vertex
• Luminosity monitors (calibration issue)
• LUCID
• Options
• Existing Hardware of Forward Calorimeter
• Existing Hardware of TileCal
• Minimum bias counter
• Diamond

5
Monitoring by Counting

• Dedicated Luminosity Monitors
• LUCID counts primary forward tracks
• Counting Physics Signatures
• g g ? m m-
• With kinematic and vertex fit requirements plus
  Trigger ? a statistical error of
• DL/L lt 8 per day (J.Pinfold, Rome05)
  (nominal Lum?)
• W ? ln
• At nom Lum the Rate is 60Hz (HLT bandwidth?) ?
  a statistical error of
• DL/L lt 1 per 3min (J.Pinfold, Rome05)
• Z ? ll-
• Similar to W-gtlv ?
• TileCal min.bias Monitoring
• a statistical error of DL/L lt 1 per
  0.1sec, nom Lum

6
Roman Pots in ATLAS
Fiber detector in RP
242m
Roman Pots One station per side, Two RP units
per station
Roman Pots
7
Fiber Detector
Design 1200 fibers glued and tilted at 45
degrees on both sides of the 170um base plate.
200 PMTs for the signal collection. Final
prototype by March 2006
8
CDF CLC vs LUCID (J. Pinfold)
150 cm
Solution CDF
Cerenkov light
Collector
Particle
PMT
Isobutane or C4F10 gas
Al Mylar Cone
Quartz fibre optic cable to transmit light to a
remote PMT
Solution LUCID
Particle
C4F10 gas
Thin aluminium or carbon fibre tube
9
CDF CLC L Monitoring

• Pointing Cherenkov cones
• sensitive to charged IP tracks,
• blind to non-pointing secondaries
• Real-time feedback to LHC control
• Based on existing CDF design/operation
• Good linearity shown by CDF

CDF superimposed dataL21032, 6
interactions/crossing
10
LUCID (J.Pinfold Alberta)

• Dedicated detector
• bundle of projective Cherenkov cones 5 layers of
  40 tubes each
• low mass (6 kg), rad hard (C4F10 ), quartz fiber
  readout
• 40 MHz capability no large extrapolation from
  L1027 ? 1034
• linearity required over 1-30 interactions/crossing
  
• Counts primary particles
• Mostly insensitive to non-primary particles
• 11 primary tracks per interaction
• total signal ? charged primaries ?
  interactions/crossing ? L
• C-photon statistics rad induced photons ?
• proof of principle CLC at CDF (e.g. S. Klimenko
  et al., NIM A441 (2000) 266)
• late on prototyping (2005?)

11
The LHC operation modes (in no way official)

• (short?) Machine tuning
• (0.01-gt10)x1032
• Phase-0, 1st year to the ultimate performance
• Tuning the machine performance from physics
  official start-up to the ultimate values
  (0.1-gt2.3)x1034
• Runs dedicated to the Luminosity calibrations
• (0. 1-gt1.0)x1028
• Phase-1, machine upgrade
• IR upgrade is estimated to take place once a
  certain rad. damage of the quadrupoles at the IR
  is reached. Possible increase in luminosity to
  1035 (SLHC). Several scenarios were proposed,
  some more to come.
• Phase-2, machine upgrade
• Possible energy upgrade 1035 (VLHC).
• 
  1027 - 1035

12
ATLAS Detector
13
TileCal Geometry
The TileCal, the barrel part of the hadronic
calorimeter of ATLAS, is a sampling device made
of steel and scintillating tiles (41). Due to
the LAR EM-calorimeter in front mainly the
response to hadrons is optimized. Sensitivity to
muons is used to enhance the muon
tagging. An example distributions
will be given for the cells A13 the most
exposed to the min.bias events cell of TileCal,
BC5 typical TileCal cell, D0 the least
exposed to the min.bias events cell of TileCal.
Hermeticity ? coverage ?1.7 cracks at
transition region Intermediate Tile
Calorimeter Gap Scintillators
Segmentation Each 0.1 azimuthal slice
consists of 73 cells arranged into projective
towers with ???0.1. Each of 64X3 modules has
three depth segmentations called samples A,
BC, and D.
y
D0
D
D0
D
BC5
BC
BC
A
A
A13
z
A13
BC5
5.6 degree azimuthal slice
?1.7
14
TileCal Single Channel FE Readout
Cesium
In-situ Physics
Tile
Minimum Bias
Fibre
TileCal cell
Laser
Mixer
HV
PMT
PMT Block
HV Opto
HV Micro
Canbus
Divider
Charge Injection
3-in-1
L H
Mother Board
Digitizer
Analog
Integrator ADC-I
Canbus
Adder
Digital
Drawer
Optical Interface
TTC
ROD
µ Trigger
Had Trigger
Energy
15
Min.Bias Signals in TileCal (MC)
Min.Bias events inelastic pp collisions at low
momentum transfer. -gt Expected 23MB events per
BXing at nom Lum (1034) -gt Energy distribution is
symmetric in ? -gt Mean Energy deposited in a
given cell per collision is small -gt
Fluctiations of the deposited energy are rather
large Simulations of the Min.Bias events in
the TileCal are available at the cell
level. This is a critical step to proceed with
the system evaluation.
16
Min.Bias in the TileCal Cells per collision (MC)
Cell occupancies (chances for the shower to reach
the cell) are lt1. Mean Energy is small and
driven by the occupancy. Means in Sample A
(0.2-1.1)MeV Means in Samples B,D are much
smaller RMSs are much higher than the means
A13
Occupancy in the TileCal Cells per Collision
Occupancy ()
BC5
D0
pseudo rapidity
mean(E) in the TileCal Cells per Collision
A13
RMS(E) in the TileCal Cells per Collision
A13
RMS MeV
mean(E) MeV
BC5
BC5
D0
pseudo rapidity
pseudo rapidity
D0
17
Min.Bias in the TileCal Cells per bx (nom.Lum, MC)
A13
Cell occupancies and mean Energy roughly scale up
by 23. Means in Sample A (5-25)MeV Means in
Samples BC,D are smaller RMSs scale up roughly
by sqrt(23).
Occupancy in the TileCal Cells per bx
Occupancy ()
BC5
D0
pseudo rapidity
mean(E) in the TileCal Cells per bx
RMS(E) in the TileCal Cells per bx
A13
A13
RMS MeV
mean(E) MeV
BC5
BC5
D0
pseudo rapidity
pseudo rapidity
D0
18
Min.Bias Anode Currents in the Tilecal Cells
(nom.Lum, MC).
I f x k x Re / (e/pi) x E ? I(Lum) 28nA/MeV
x E(Lum) where E(Lum) is the energy deposited
in the given cell per bx, f is bx frequency
40MHz, k 2808/3564 0.79 is correction for
empty bunches other coefficients are measured at
the TBs Re 1.157pC/GeV is (pC/GeV) ratio for
electrons at 20o e/pi 1.3 is the ratio to scale
Re to the pion level
DC currents from TileCal Cells
A13
I, nA
BC5
D0
I
pseudo rapidity
By setting the RF to (1-100)MO one will obtain
measurable Vout around 1V.
19
FE Slow Current Integrator (G.Blanchot)
X.Portell

• DC coupled to PMT with permanent coupling to the
  shaper
• Build around input stage operational amplifier
• Gains (RF) are selected
• remotely through 3in1 logic
• Gains, offsets and linearities can be
• calibrated by a dedicated charge
• injection system (Vdac)
• Local output switch
• controlled by the 3in1 logic
• Radiation Tolerant
• Part of the 3in1 card controlled
• through the TTC and CAN

20
FE ADC (G.Blanchot)

• 12 bits digitizer, (0-5)V dynamic range
• Global Pedestal Control Feature
• Local Integrator Enable Input for
• distributed readout
• CANbus port.
• Radiation tolerant.

INTG_SEL
INTG_OUT
3in1 Logic
INTG_GND
3in1 Logic
3in1 Logic
Pedestal Control
Micro
DAC
Analog Bus
ADC
CANBus Port
3in1 Cards Control Logic
Differential Input Stage
3in1 Cards Controls
21
Accuracy of a single measurement of the min.bias
anode currents from a given Cell (nom.Lum, MC).
Accuracy of a single measurement of the min.bias
anode current for a given TileCal cell can be
estimated from the energy spread per bx and
number of bxs per the integration time. For the
nominal Luminosity Accuracy in Sample A
1 Accuracy in Sample B (1-3) Accuracy in
Sample D (2-9) If systematic effects, such
as beam-gas interactions, are not taken into
account, this would give an estimate on the stat.
error of the relative luminosity measurement.
Accuracy of a single measurement of the min.bias
anode current for a given TileCal cell
D0
Acc ()
BC5
A13
pseudo rapidity
22
Luminosity reach for a single channel accuracy
of a single measurement
Luminosity X 1034
Anode current (nA) as a function of luminosity
A13
Luminosity reach for a single channel is limited
by the ADCGains dynamic range ADC saturation at
1770 nA 10ADC counts 0.12 nA Overall
performance is optimized for the nominal
luminosity. At the low luminosities the
statistical sum of many channels of the TileCal
should improve the accuracy. The lt1 accuracy
should not be taken seriously for the systematic
effects are not considered.
BC5
I (nA)
D0
Luminosity X 1034
D0
A13
Acc ()
Stat accuracy () of a single measurement as a
function of luminosity
BC5
23
Partitions, Readout chain
DCS
Four Partitions of the TileCal four TTC
Crates (in USA15) CCT VME CPU TTCvx,
TTCvi modules Read-Out Buffer (RB) per crate
/ partition
Local network

• There are 4 MBMing Partitions in the TileCal.
• Each partition consist of
• CCT, TTCvi, RB.
• Each RB controls 4 CAN lines configured to
  250kbps.
• There are 16 ADC-I CAN nodes per line.
• ADC per TileCal module.
• There are 45 (36) channels controlled by each
  ADC-I in the Barrel (EB) module.

Four CAN lines per RB lt170m (including daisy
chain), 250 kbps, 16 nodes. LV_CAN_PS to be
produced by CF.
Sixteen ADCs (CAN nodes) per CAN line
Mezzanine board
TTC control
CAN control
3in1 bus
Control over 3in1
45 integrator cards per ADC (barrel)
PMT block
24
Readout Cycle, Stat Accuracy per Sweep per
Partition
The system reads out 1 channel per module
following single trigger, which takes place every
50msec, as defined by SHAFT calibration board of
TileCal. gt 256 channels are read out per single
trigger. Max Luminosity update rate 50msec.
Luminosity X 1034
The cycle to read out all channels in the
partition (45chs/module (36chs/m) in the Barrel
(EB) partition) is called sweep. TileCal is
divided into 2 Barrel and 2 EB partitions. Every
cell is read out twice (from two sides) per
sweep. Hence, once the readout bandwidth is
equally divided between all the channels, one
sweep will take 2.5sec (2.0) for the Barrel (EB)
partition.
Barrel Partition
accuracy ()
stat accuracy () from single sweep of barrel
partition
If the monitoring task of the TileCal will
permit, an extra readout bandwidth can be given
to the sample A channels (more sensitive to the
min.bias interactions) at the expense of the
cells in the samples BCD. This will enhance the
stat accuracy, Lum lt 1032.
25
Related Part of the TileCal DCS
The CCT CPUs, located in the TTC crates, one per
TileCal partition, will read the data from the
RBs. The functions of further data
transmission, analyses and data storing will be
spread between the CCTs and 1-2 dedicated TileCal
DCS stations depending on their
performance. The DCS related tests scheduled to
start in September 2005. Only short summary
will come to the TileCal DCS station in the ATLAS
control room.
26
Components Summary
10k integrator cards produced (Chicago),
burned-in, QC-ed, installed in the drawers,
tested, gt85 installed in the modules, tested
certified. 270 ADC-I cards v.5 produced
(Barcelona), burned-in, QC-ed , installed in the
drawers, tested, gt85 installed in the modules,
tested certified. 20 VME-CAN compressor (RB)
boards of v.2 are produced (Barcelona) and QC-ed.
6 RBs are currently used at CERN. The CAN cable
length seems currently under control (lt170m), The
cable length defines communication speed (250kbs)
which in turn limits the read-out rate. CAN
power supplies are in production
(Clermont-Ferrand). The TCal calibration board,
known as SHAFT, that has to provide suitable
triggers to the three online Calibration systems
of the Tile Calorimeter (CIS, LAS, MBMing), is
under design.
27
Tests
The performance of the integrator and ADC-I
cards, and the RB boards was extensively tested
since 1999 in many TileCal groups, during the Cs
certification of the modules and during ten
TileCal TB periods. System functionality test
was performed at the CTB
M.Volpi
Normal and 25nsec beam profiles as seen by
the system
28
TileCal WP12 (barrel, 2005)
The commissioning of the mim.bias monitoring
system is a part of the TileCal phase-1
commissioning plan as Work Pakage 12. We aim
for three goals (for the TileCal barrel) by the
end of 2005 1) Integrate all final readout
components into the system and certify
communication with ADC-Is. 2) Integrate the
min.bias code with the TDAQ and DCS of
TileCal. 3) Integrate MBMig system with CIS and
LAS systems.
29
Calibration at 1027-1028 ?

• Below Lum5x1029, less than a single min.bias
  hit/cell/10msec is expected in every cell of the
  TileCal, including the sample A cells. The
  histogram of the values measured from the
  integrator becomes of a peculiar type containing
  two distinguished parts
• 1) Pedestal events when no hits happened in the
  cell during the integration time, with the
  Gaussian shape directly measurable on pedestal
  events.
• 2) Signal tail spanning several ADC counts
  (80MeV/ADC count), with the shape that is result
  of convolution of
• - min.bias deposition spectrum - not known (but
  from MC)
• - RC response function - directly measurable
• - noise Gaussian as in the pedestal events -
  measurable
• Neither of two parts change shape as the Lum go
  lower. What does change is the ratio of the
  entries. If one could decouple the part of the
  signal tail (we have to do studies on it. D0 will
  be in this mode upto 1032), one can use its
  mean value for the absolute calibration on given
  channel.
• - 1 stat error on the mean with the data from 30
  (300) hours at Lum1028 (1027), 240 channels at
  20Hz.
• - Gains can be set to x3 (x8 may be) by HV
  system.
• - Have to study if OP amplifier gains can be
  increased.

X.Portell
Example of the energy deposition by
min.bias events per collision in a given TileCal
cell (MC)
A1
Calibration of the system at L(1027-1028)
would be very challenging but may be possible. We
need to do dedicated studies on the options.
30
Low Frequency noise
Sigrid Iacopo
31
Summary
The min.bias monitoring system, dedicated to
monitoring the performance of the TileCal optics
and read-out, might be used as one of the lum
monitors of ATLAS. The system is based upon
anode current integrators that are attached to
every readout channel of the TileCal. The
read-out of the integrators is not a part of the
standard ATLAS readout. The system performance
is optimized for the nominal luminosity
(1034). The stat. error on the relative
luminosity measurement per sweep (2.5sec) is
expected to be 12 for L1030, 4 for L1031,
lt1 for L(1032-1035). Calibration of the
system at L(1027-1028) would be very challenging
but may be possible. We need to do dedicated
studies on the subject. The frequency of the
update on the relative luminosity from the system
will depend on the demands and computing
resources available.
32
Summary
Most of the hardware components used in the
system are produced and were tested over the
years. The readout software is under
development, a prototype had been tested at the
Combined Test Beam 2004. The commissioning of
the monitoring system is part of the TileCal
phase-1 commissioning Work Packages.