Trigger system for the T600 detector: applications to
Supernova detection
G. Fiorillofor the ICARUS Collaboration
CRYODET WorkshopLNGS, March 13-14, 2006
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 2
View of the inner detector
T600 semimoduleCryostat (half-module)
Readout electronics
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 3
ICARUS DAQ
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 4
Physics considerations
• Events in T600 at LNGS:– Cosmic rays muons– Atmospheric neutrinos– Solar neutrinos + neutrons E ~ 5 MeV– Neutrinos from Supernova (burst) E ~ 20 MeV– Proton decay– CNGS neutrinos external timing– Beam muons
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 5
Event topologies
Cosmic-ray shower Muon
Low energy electrons
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 6
Data volume (A. Rubbia ICARUS IM Sept.02)
event/year wires/event drift factor of average total size number DLTin T600 samplescompression size per year equivalent(order of 4 in in GB (40GB/DLT)magnitude) Mbytes
solar neutrino 3000 27648 2500 4 32,96 97 2,4neutron capture 6000000 27648 2500 4 32,96 193119 4828,0muon crossing 2,00E+06 27648 2500 4 32,96 64373 1609,3side rock event 1,00E+05 27648 2500 4 32,96 3219 80,5atmospheric-like event 1,00E+03 27648 2500 4 32,96 32 0,8supernova event 0 27648 2500 4 32,96 0 0,0proton decay 0 27648 2500 0,0
TOTAL w/o low energy 67720 1693,0TOTAL with low energy 260743 6518,6
– 1.5 m drift or 1 ms is 2500 samples of 400 ns
– 27130 wires with 2500 drift samples
– 32 wires per board x 18 boards per crate x 48 crates= 27648 readout channels of 2500 samples
Total volume: 260 TB per year (25% CMS or ATLAS experiment!)
Volumes are dominated by physics backgrounds: neutron capture and muon crossing
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 7
Neutrinos from Supernovae
€
ν e + 40Ar→40Cl* + e+
• Elastic scattering from neutrinos (ES)
• Electron-neutrino absorption (CC)
• Elastic scattering from antineutrinos (ES)
• Electron-antineutrino absorption (CC)
(νe)+0.15 (ν + ν)
νx +e− → νx +e−
νe+40Ar→ 40K * +e−
(νe) Q=5.885 MeV
€
ν x + e− → ν x + e−
(νe)+0.34 (ν + ν)
(νe)Q≈8 MeV
low energy reactions in Argon
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 8
Shock breakout
collapse
OscillationNo
oscillation
757
730
27
Non adiabatic
1036
995
41
Adiabatic
Expected events from SN at 10 kpc in 3 kton
244TOTAL
203absorption
41Elastic
reaction
A.Rubbia, I.Gil-Botella
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 9
Supernova burst• Specific time structure
– Ex.: ~ 100 SN triggers in T300 in 1 sec(expected for a SN at 10 kpc, with an energy release of 1053 erg)
• Global trigger (full detector readout): bandwidth + storage problem– 1 event = 27648 ch 2500 samples 2 bytes
~ 130 MB 13 GB total
– DAQ throughput limit: < 40 MBytes/s (or >5 minutes to empty buffers)
• Local trigger: SN events are localized and limited to 1 crate per view– 5 events per crate (on average) in COLL + IND2 views ~ 40
MB/crate– 13 events per crate (on average) in IND1 view ~ 60 MB/crate
Each crate can be read-out as a separate event
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 10
Segmented trigger• Intrinsic granularity of the trigger is
a crate
• The requirement of the Supernova trigger introduces two types of triggers
– Global triggers: an entire T300 is readout as a single event
– Local triggers: the various crates are readout independently as separate events Overlapping in time local triggers
can be reconstructed offline (or online) using the absolute time information
• The implementation of local triggers presupposes the development of a small “local trigger control unit” card in each crate
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 11
Segmented trigger pixel definition
Rack 1Rack 20
Rack 11Rack 13
T600 Half Module – 1 chamber viewed from cathode
1 pixel area: ~ 0.6 m2
Total Number of Pixels ~ 80
32 x 9 Induction II
wires
32 x 9 Collection
wires
864 mm
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 12
MC simulation: low energy eventsICAFLUKA
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 13
MC simulation: high energy eventsICAFLUKA
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 14
Preliminary considerations• Trigger rate is dominated by physics background
– Neutron capture rates expected in T600 (ICARUS/TM-
2002/13): • 210-4 s-1 from natural radioactivity of the rock• 0.030.1 s-1 from Al container
• Segmented trigger potentially solves bandwidth and storage problems
• Event pre-classification data streams– extraction of solar neutrino data from low energy
stream
• Test bed for larger LAr detector low energy trigger
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 15
Trigger Input
• PMTs (global and local triggers)
• DAEDALUS (Based on single wire information)
• AWS (Analog 32 wires sum, exists on the front-panel of each analogue card, local triggers)
• External (beam profile chambers, cern-spill, …global)
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 16[x 50 ns]
PMT signal
The high light yield measured in T600 provides the natural solution for T0 definition and (in part) for trigger
PMT system
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 17
A PMT based trigger
• The PMT system allows for the design of a fast triggering system able to select the events on the basis of the scintillation light produced in liquid argon.
• The study is based on the MC simulation of real events inside the T600 detector.
• The contribution due to the dark counts and noise is taken into account.
• The segmentation of the sensitive volume according to the PMT layout permits the definition of local triggers.
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 18
Trigger Efficiency PlotsThe plot summarizes the trigger efficiency (triples case) as a function of the
electron kinetic energy, for different threshold values (G.Raselli IM Jan06).
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 19
Trigger Efficiency Tables
10
7
0.5 1.5
5
2
15
3
31Threshold
(phe)Energy(MeV)
98.3%
6.2 Hz
Trigger efficiency / Stochastic rate
(triples case)
99.7%
99.9%
99.9%
99.9%
95.2%
99.1%
99.8%
99.9%
99.9%
86.1%
97.9%99.1%
99.8%
99.9%
73.6%
93.3%97.1%
99.4%
99.9%
46.2%
80.2%
92.8%
97.9%
99.6%
6.2 Hz
6.2 Hz
6.2 Hz
6.2 Hz
1.1 Hz
1.1 Hz
1.1 Hz
1.1 Hz
1.1 Hz
0.2 Hz
0.2 Hz
0.2 Hz
0.2 Hz
0.2 Hz
<0.1 Hz
<0.1 Hz
<0.1 Hz
<0.1 Hz
<0.1 Hz
«0.1 Hz
«0.1 Hz
«0.1 Hz
«0.1 Hz
«0.1 Hz
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 20
Basic design requirements
• Redundancy important to measure efficiency• Global trigger:
– Generated by PMTs or external– drift deadtime GLOBAL_DRIFT (1ms)– Read-out deadtime GLOBAL_BUSY (1s) vetoes new
global triggers– Local triggers vetoed during GLOBAL_DRIFT
• Local trigger:– Generated by AWS + PMT– LOCAL_DRIFT (1ms) vetoes new local triggers
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 21
Trigger System Architecture
LTCU: discriminates the inputs, has one independent
threshold for each input, gives 1st level trigger
proposals as output; TCU:
performs coincidences between LTCUs proposals,
processes the fired pixels to study and label the event topology, produces global or local trigger proposals;
Trigger Supervisor: monitoring of the trigger and
the DAQ system, statistical functions.
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 22
Trigger System Architecture
LTCU TCUTRIGGER
SUPERVISORDAQ
CERN or any other external request
20 boards per chamber
(80 boards for T600)
from v791
n-bit request
v816
96
9
4 boards per T600
9
Trigger
Distribution
T120
T220
T11
T21
PMT LTCU
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 23
• LTCU discriminates the 18 inputs comparing them with remotely controlled thresholds. Noisy inputs can be masked
• Gives as output two separate trigger proposals (T1 and T2), corresponding to the OR of INDII or COLL boards respectively
• Test mode to check the working status
• Counting capabilities to check rates of trigger proposals for each input
• Board functionalities are driven by a remote controller
TCULTCU
9 T1
T29
from v791
• 1 board per crate - 20 boards per chamber
• each board receives as input 9 + 9 coming from the v791 analog boards (generally 9 of 2nd induction and 9 of collection)
1st trigger level: LTCU
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 24
• 1 board per chamber
• VME standard
• each board receives as input 40 signals from the AWS LTCUs (2 T signals x 20 boards) and N signals from PMT LTCU
• TCU performs coincidences between:
• the wire planes (in a ~3 s time window),
• the PMT signals,
• the external requests
• checks Majority, Minority, 2D Pattern logic conditions
• event is labelled according to topology
• An n-bit-word trigger request is sent to the Trigger Supervisor, defining the fired crates
2nd trigger level: TCU
CERN or any other external request
TS
T11
LTCU
TCUT1
20
T220
T21
n-bit request
PMT LTCU
AWS
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 25
TRIGGER SUPERVISOR
DAQ
n-bit request
v816
96
Trigger Distribution
• 1 or more boards
• VME standard
• TS receives 4 input n-bit-words from the TCU and the ABS CLOCK
• TS evaluates if the trigger request from TCU corresponds to a Local Trigger or a Global Trigger
TCU
• Monitoring of the system and validation of Trigger requests.
• GLOBAL/LOCAL DRIFT/BUSY Logic
• Trigger signal distribution to the v816
• Event trigger tag (trigger number, time, type, acquisition dead time…)
• Statistics for the trigger system
ABS CLOCK
Trigger Supervisor
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 26
TS GLOBAL/LOCAL Logic
• Trigger classification depends on:
• Energy deposition/number of PMT fired.
• Detector occupancy
• 2D/3D Pattern of fired pixels
• TS gives GLOBAL trigger when a MAJORITY condition is met in PMT logic
• Otherwise the trigger is LOCAL
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 28
The trigger generation algorithm
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 29
Time DiagramPMT
Induction II signal
Collection signal
LTCU T1 signal
LTCU T2 signal
TCU Trigger request
TS Trigger signal
T 3 s
T 1.5 ms Trigger dead time
n-bit
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 30
FPGAInput stage
RS232 interface
10 MHz oscillatorPower supply
18 inputsTrigger outputs
DAC
The LTCU prototype v1.0
IN
VDAC
Discriminator
Voltage follower
RC filterOUT
IN
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 31
VME INTERFACE Xilinx Spartan XC2S100 INPUT STAGE Xilinx Spartan XC2S100 SYNCHRONOUS FIFO IDT72V263
80 programmable input / output
2nd level Prototype TCU
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 32
Summary
• The readout of SN events, due to their burst nature, requires the implementation of local triggers to overcome data volume problems
• The segmentation of the sensitive volume permits the definition of local triggers potentially solving bandwidth and storage problems
• Event pre-classification by the trigger allows for the definition of data streams (useful for the analysis)
• Redundancy in the trigger system is important in order to measure the efficiencies
Segmented Trigger architecture based on PMTs and Wires Global trigger: the full detector is read-out as a single event Local trigger: only active “pixels” are read-out as separate events Events are labelled according to pixel topology
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 33
LTCU prototype for AWSv1.0
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 34
LTCU v1.0
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 35
The LTCU functionalities v1.8i
• Mask the input channels; • Read the mask status;• Set the thresholds;• Monitor the trigger rate
for each input;• Discriminator test mode;• Select one discriminator
output put on front panel.
All the board functionalities are remotely controlled via RS232 interface.
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 36
V1.0 test on 50 liters LAr TPC
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 37
Trigger rate test chain
• LTCU inputs = 4 signals from collection plane;
• Trigger generated from only one input (no FastOR);
• LTCU trigger output distributed to V816 module.
LTCU Trigger OUT
V789
LTCU
V791Ind Coll
IN IN IN IN IN IN IN IN
OUT OUT OUTOUT OUTOUT OUT OUT
IN
OUT
IN
Analog OUT
V816
trigger
PC (RS232)
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 38
Trigger Rate Plateau (I)Trigger rate for v791 n°1 in LTCU ch0
0
5
10
15
20
25
0 20 40 60 80 100 120 140
Threshold (mV)
Trigger rate (Hz)
Trigger rate for v791 n°2 in LTCU ch1
0
5
10
15
20
25
0 20 40 60 80 100 120 140
Threshold (mV)
Trigger rate (Hz)
Plateau zones
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 39
Trigger rate for v791 n°4 in ch5
0
5
10
15
20
25
30
0 20 40 60 80 100 120 140
Threshold (mV)
Trigger rate (Hz)
Trigger rate v791 n°3 in LTCU ch4
0
5
10
15
20
0 20 40 60 80 100 120 140
Threshold (mV)
Trigger rate (Hz)
Trigger Rate Plateau (II)
Plateau zones
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 40
Trigger efficiency test chain
• LTCU inputs = 4 signals from collection plane;
• One LTCU IN and Trigger OUT digitalized by a modified V791;
• PMT trigger distributed to V816 module.
PMT
V789
LTCU
V791Ind Coll
IN IN IN IN IN IN IN IN
OUT OUT OUTOUT OUTOUT OUTOUT
IN
OUT
IN
Analog OUT
V816
trigger
PC (RS232)
OUT
IN
IN
Modified V791
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 41
Test Results
Problems Solutions
High frequency noise on input signal
Band-pass filter
(f-3dBH = 2 MHz , f-3dB
L = 1 kHz);
Offset for negative input isn’t a stable solution
Inverter amplifier (G=1) in the input stage;
More than 20mV white noise
New pcb (with 8 layers) and EM/RF screening
Not stable DAC threshold
Stable VREF circuit
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 42
LTCU prototype v2.0
Power supply an filter
Input stage (for one channel)
or
IN
Band-pass filter Selectable inverter (G=1)
VTest
VTH
OUT
IN
VTH
DAC (for one channel)
VREF
VREF for DAC
Filter for low noise performance
Stable 259mV tension circuit
VREF
EM/RF Screening for input stage
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 43
PWR and GND distribution for LTCU v2.0
±5A, AGND (V791 preamp. stage)
+5A1, AGND1 (V791 mux stage)
VCC, DGND (V791 digital stage)
Four GND and two PWR planes
The second trigger level
Trigger Control Unit
Preliminary design
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 45
TCU Features
• ≥1 boards per chamber (depending on segmentation)• VME standard• checks Majority, 2D/3D Pattern logic conditions• event is labelled according to topology• Each TCU module receives as input
– Naws signals from the 20 AWS-LTCU boards of a chamber– Npmt signals from PMT-LTCU – Next from external (spectrometers, beam, ...)
• TCU performs coincidences between:– The wire planes (in a 3s time windows)– PMT signals– External requests
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 46
Pattern RecognitionNeed to store the evolution in time of the number of fired pixel
A long track fires consecutive pixels (at different times) according to direction. In this case the event is global can be acquired as many local events.
In case of a Supernova burst several pixels could be fired with a big spacing in position and in time.
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 47
General Idea on Partial majority
Rack 11Rack 13
• Window runs over the detector
• Majority condition is checked for each window
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 48
General scheme of TCU
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 49
WiRe coincidences FPGA Features
• The detector is subdivided in four sections• Each section is monitored by a different WR-
FPGA• Each WR-FPGA
– Makes coincidences between wires of different directions
– Checks the occupancy of the section– Detects spots, tracks.
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 50
PMT FPGA Features• receives an N bit Word which defines
PMT status• makes concidences between PMT and
wires• makes global trigger proposals
VME interface Features
Directly interfaced to VME CPU. Establishes communication between VME and the rest of the board.
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 51
Wire CoincidencesFPGA
preliminary
SYNCBLK
SYNCBLK
ANDBLK
9
13
INA
INB
MASK
9
13
MASK 9
AA
BB
13
A0_B[4:0]
A1_B[5:1]
A2_B[6:2]A3_B[7:3]
A4_B[8:4]
A5_B[9:5]
A6_B[10:6]
A7_B[11:7]
A8_B[12:8]
FLASH
ADR[4:0]
ENC
REG_A
REG_B
REG_C
SUMBLK
VTH
InternalXilinx
Memory
WholeCarpet
Majority
ENCA
ENCA2
ENCA3
MJLOCAL
MJGLOBAL
ADDRESSGENERATOR
FLASH
CLK PIPELINE
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 52
A possible implementation: the Uniboard
ICCMASTER
ICCSLAVE1
ICCSLAVE2
ICCSLAVE3
Spartan II
FAUXSpartan II
VME(J2)
Spartan II
VME(J1)
Spartan II
3 3 3
FIFOA
FIFOB
FIFOA
FIFOB
FIFOA
FIFOB
FIFOA
FIFOB
18 18 18 18 18 18 18 18
FF FF FF FF
EF
G. Fiorillo Cryodet Workshop, LNGS 13 March 2006 53
UNIBOARD