1 tof electronics overview j. schambach university of texas doe review, bnl, 25 sep 2006
TRANSCRIPT
1
TOF Electronics Overview
J. Schambach
University of Texas
DoE Review, BNL, 25 Sep 2006
2
Outline
• Electronics “Essential Model”
• Trigger & DAQ Interfaces
• Board Status
• Production & Testing
3
TOF Electronics Top Level
DAQ Room
STAR Detector WestSTAR Detector East
Start Detector
Electronics
60 Trayswith RPCs
Start Detector
Electronics
Fiber
60 Trayswith RPCs
Fiber
HV HVLV LVControl PC
withCANBUS I/F
CA
Nbu
s
Rack Electronics
DAQ
22
2
CA
Nbu
s
2
TCD
Trigger
2 2
L0 Trigger
60
Slow Controls Slow Controls
HV HV
Slow Controls
L0 Trigger
60
Trigger
TriggerLevel-2
Fib
er
2
Fiber
2
4
Electronics for One Side
Start Detector with
19 Scintillators &19 PMTs
Start DetectorElectronics
Tray
32 RPCWith 6 channels
eachOn-Tray Electronics
Data, Clk, Ctrl
CANbus
30Trays
L0 Trigger
Data, Clk, Ctrl
Dual Fiber DAQTHUB
Master ClockC
AN
bus
Tray
32 RPCWith 6 channels
eachOn-Tray Electronics
Data, Clk, Ctrl
CANbus
30Trays
L0 Trigger
Dual Fiber DAQTHUB
Slave Clock
CA
Nbu
s
OneDetector Side
CA
Nbu
s
Clk
, Ctr
l
To other THUB’s
Clk, Ctrl
Clk, Ctrl
Control PCwith
CANBus I/F
TCD & BusyLogic
Dual Fiber
Dual Fiber
L2
L2
5
Tray Level Electronics
Start Detector
TDIGTDIGTPMDTPMDPMTsPMTs
TOF Tray
TDIG TDIG
TCPU
DATA, CLOCK& CANbus
TINO TINO
MRPC MRPC
96 CHAN
TRAY CAN BUS
MULTIPLICITY
THUB
COPPER: DATA, SAMPLE CLOCK, RESET TRIGGER STROBE & DATA
TOP LEVEL CAN BUS
COPPER LINKS TO 29 TRAYS
SIU FIBER
CAN BUS TO 29 TRAYS
TCD
DAQ
96 CHAN
4
L0 Trigger
4
PMTs TPMD TDIG TCPUData, Clock& CANbus
Cop
pe
r:D
ata
, S
am
ple
Clk
, R
eset
Trg
Str
obe
& D
ata
Reset & ClockFrom other THUBs
Reset & ClockTo other THUBs
To Control PC
L2TDIG TDIG
TINO TINO
MRPC MRPC
44
SIU FIBER
Clock & Commands
Busy
Fast Z-Vertex Electronics
3 3
6
Front-End Electronics “TINO”TINO
NINO C
NINO B
NINO A(8 ch Amp / Disc)
MRPC 4(6 ch)
MRPC 1(6 ch)
MRPC 3(6 ch)
MRPC 2(6 ch)
Multiplicity to TDIG PLD(1 bit LVDS)
Hits to HPDTC(8 LVDS)
Threshold(All Common from TDIG)
MR
PC
Inte
rfac
eM
RP
C In
terf
ace
MR
PC
Inte
rfac
eM
RP
C In
terf
ace
6
6
6
6
6
6
2
4
4
2
8
8
TD
IG Interface
TD
IG Interface
TD
IG Interface
8 per tray
960 total
7
CERN/LAA NINO Chipdeveloped for ALICE
Parameter Value
Peaking time 1ns
Signal Range 100fC – 2pC
Noise (with detector) < 5000 e- rms
Front edge time jitter <25ps rms
Power consumption 30 mW/ch
Discriminator threshold 10fC to 100fC
Differential Input impedance 40Ω< Zin < 75Ω
Output interface LVDS
8
Digitizer Board “TDIG”8 per tray
3 per start side
966 total
Altera FPGA
HPTDC(LE & TE)
HPTDC(LE & TE)
HPTDC(LE & TE)
TINO Interface
8
88
3
Inte
rfac
e to
pre
viou
s T
DIG
MicroController
CANbus
Temperature sensors
External memory
Out
put i
nter
face
to n
ext
TD
IG o
r T
CP
U
Upstream data outbound
For Multiplicity
PLD control and status monitoring
Trigger & clocks inbound
Config & Status (3 TDCs)
8 p
air
fa
st m
ulti
plic
ity,
seri
al d
ata
rd
o s
ign
als
, cl
ock
s,a
nd
re
sets
Sa
me
as
ou
tpu
t in
terf
ace
TEMP & Power
IDTray position
switch
9
INL Correction
• Sigma = 0.9 timebins = 22 ps
• Implies single channel resolution of 16ps
10
Tray Controller “TCPU”
1 per tray
1 per Start Detector
122 total
MCU
TRAY LEVEL CAN BUS I/F
TOP LEVEL CAN BUS I/F
THUB-TCPU Interface
Trigger Data, reset inboundDATA outbound
TDIG CABLE I/F
TDIG CABLE I/F Trigger & clock, reset outboundTDC & Mult. Data inbound
Clock select and distribute
Local osc
To TDIG cables
RAM
TRAY ID
Mu
ltip
licity
Su
m
AlteraPLD
L0 TriggerInterface
Sa
mp
le C
lock
11
DAQ/Trigger Interface “THUB”
2 per Detector side
4 total
AlteraFPGA
MCU
TCDInterface
External ClockInput
ClockGeneration
Clock Outputs To 3 other THUB
BufferRAM
TCPU – THUBInterface
TCPU – THUBInterface
TCPU – THUBInterface
31
DDL SIU
Local Oscillator
System LevelCANbusInterface
TDC & Mult DATA
3
PR
EC
ISIO
N C
LOC
KS
TO
TC
PU
s
31
Trigger Data
& Strobe, Resets
31
Command, Token & Clock
BUSY
3 Reset InputFrom other THUBs
3
Reset OutputTo other THUBs
DDL SIU
DAQ
L-2
12
THUB Design
SERDES0-3
SERDES0-3
SERDES0-3
SERDES0-3
SERDES0-3
SERDES0-3
SERDES0-3
SERDES4-7
SERDES0-3
SERDES0-3
SERDES0-3
SERDES8-11
SERDES0-3
SERDES0-3
SERDES0-3
SERDES12-15
SERDES0-3
SERDES0-3
SERDES0-3
SERDES16-19
SERDES0-3
SERDES0-3
SERDES0-3
SERDES20-23
SERDES0-3
SERDES0-3
SERDES0-3
SERDES24-27
SERDES0-3
SERDES0-3
SERDES0-3
SERDES28-31
CYCLONE FPGA
672 pin
0.5M RAM0.5M RAM0.5M RAM0.5M RAM
CYCLONE FPGA
672 pin
0.5M RAM0.5M RAM0.5M RAM0.5M RAM
CYCLONE FPGA672 pin
0.5M RAM0.5M RAM0.5M RAM0.5M RAM
CYCLONE FPGA
672 pin
0.5M RAM0.5M RAM0.5M RAM0.5M RAM
CYCLONE FPGA
672 pin
0.5M RAM0.5M RAM0.5M RAM0.5M RAM
CYCLONE FPGA
672 pin
0.5M RAM0.5M RAM0.5M RAM0.5M RAM
CYCLONE FPGA
672 pin
0.5M RAM0.5M RAM0.5M RAM0.5M RAM
CYCLONE FPGA
672 pin
0.5M RAM0.5M RAM0.5M RAM0.5M RAM
CYCLONE FPGA
672 pin
36
36
36
36
36
36
uC
SIU
TCD
CAN
CLOCKS
Power supplies
CPLD
RESET OUT(3)
RESET IN
36
36
Clocks OUT(3)
Clock IN
SIU
13
ALICE DDL LinkFront-end electronicsFront-end electronics
DetectorData Link
DDL SIUDDL SIU
DDL DIUDDL DIU
RORCRORC
SourceInterfaceUnit
DestinationInterfaceUnit
ReadOutReceiver Card
PCPCData
Acquisition PC
Optical Fibre~200 meters
14
PMT Input Board “TPMD”3 per Start Detector
6 total
DISCRIMINATOR(1/2 MAX9601)
PMT in (x8)
TDIG Connectors
8
From TDIG dac
THRESHOLD Buffer (& Internal set)
External In
ISOLATION BUFFER
(AD 8009)
Z Vertex out
Leading EdgeTrailing Edge
Z Vertex out
Terminator
**
x8
INPUT PROTECTION
CLAMP
15
Interface to L0 trigger
• Provides multiplicity at 9.4 MHz with <~700 ns latency.• The multiplicity range is 0-24 for each tray, where one is added to the sum if
any of the 8 TOF channels in a NINO chip is above threshold.• TDIG receives the multiplicity information from the NINO and passes it to
TCPU. Each TCPU will send the multiplicity over 5 pairs to the Level 0 DSMI on the south platform. These cables all need to be the same length (+- 1 ns).
• Each DSMI can handle 128 bits of input. We plan to use 100 bits on each DSMI, representing 20 trays. Each DSMI has 8, 32-pin connectors. 6, 32-pin connectors will each serve 3 trays, 1 connector will serve 2 trays, and 1 will be unused.
• We plan to time in the signal to trigger by trial and error. We plan to scan the trigger info in ~10 ns steps using various cable lengths to find the sweet spot. The logic pulse should be >30 ns.
• Initially (this year), we will “self-latch” the data. Each TDIG will send a value between 0 and 3 to TCPU, TCPU sums the data and sends it to the DSMI
• If “self-latching” doesn’t work, we are considering modifying the DSMI to send the RHIC strobe to each tray’s TDIGs, learn the phase of this clock at each TDIG, and use it to latch the data on each TDIG
16
Interface to Level-2 Trigger• Interface is the identical DDL link that will be used to
connect TOF to DAQ.• Initially, it was planned to provide a 23k bit map of the
TOF hits to L2 for each L0 trigger, 192 bits per tray, where each bit represents a hit channel.
• In principle, same (timing) data presented to DAQ can also be provided to L2, needs just a change in firmware
• Data volume is big:– Fully occupied TOF: 385,888 bits per event per THUB – Assuming <20% occupancy: 77,178 bits per event per THUB – 10kHz L0 rate: 736 Mbits/sec per fiber ~ 92MBytes/sec per fiber – Total for 4THUBs: 2944 Mbits/sec ~ 368 MBytes/sec, four fibers
• Data is uncorrected, needs INL correction (lookup table?) and slewing correction
17
Interface to DAQ
• The TOF system needs to be faster than the upgraded TPC so as not to introduce any additional dead time. The TOF information is only useful in a STAR event if the TPC is also readout in that event (for tracking).
• The system is able to handle L0 accept commands at >10 kHz.
• The system will not process L2-accept or Abort commands, but rather pass those on to DAQ (as separate events) over the fiber for DAQ to process.
• The current system design foresees sending all events to DAQ independent of any higher level trigger decisions for each Level-0 trigger. This design does not require much memory on THUB other than a small amount (arranged as a FIFO) to decouple the clock domain on the TCPU (SERDES) side from the clock domain of the SIU interface.
18
DAQ Data RatesOverall Specifications
Trigger Level-0 Input Rate 10,000 Hz Input Rate into Level-2 10,000 Hz Number of Fiber pairs to DAQ 4 Number of channels per MRPC 6 Number of MRPCs per tray 32 Number of Channels per tray 192 Number of trays 120 Total Number of MRPC channels 23,040 Number of Start Detector Channels 19 Number of Start Detectors 2
Tray Specifications
Data Volume per (black) Event per Tray in bits 12,448 (192*2*32 + 3*32 header + 2*32 geographical)
THUB Specifications
Number of TCPU-THUB Interfaces 31 L0-rate per tray @10kHz in bits/sec (“black event”) 124,480,000 (“Black”) Event Size per THUB (31 trays) in bits 385,888 L0-rate assuming < 20% occupancy (bits/sec) 771,780,000 DDL Fiber Bandwidth @ 66MHz 250MB/s = 2Gb/s DDL Fiber Bandwidth @ 40 Mhz (* 40/66) 150MB/s = 1.2Gb/s
19
Current System Status
20
TINO Status• Motivation: Replace Maxim Amplifier & Comparator of TAMP with
custom ASIC “NINO” incorporating both functions:– Lower Cost– Power (no negative supply & lower power, fewer power supplies)– Fully differential: better match to HPTDC– Pulse stretching: one TDC can measure both leading & trailing edge
• Decision to use TINO instead of TAMP was reached in Feb 2006 based on cosmic ray testing (see following slides) with both TAMP and TINO giving similar timing resolutions
• Automated TINO prototype production has been achieved, fabricated 35 circuit boards, 15 out of 15 boards successfully assembled
• Minor revisions made to design and layout for part availability and assembly reasons
• Mechanical compatibility with tray design verified• Will install & test 8 TINO on tray in Run 7
21
UT Cosmic Ray Test Setup
• Readout through TAMP/TINO, TDIG, and TCPU
• dT = t3 – (t1+t2+t4)/3• Same INL correction
for all HPTDCs
S1
S2
S3
Gasbox
MRPC 1
MRPC 2
MRPC 3
MRPC 4
3.00
2.00
22.00
8.50
22
UT Cosmic Ray Test Results
Time-over-Threshold (ToT)
Slewing Corrections
Final Timing Resolutions
average σ(delta t) ≈ 91ps
TA
MP
TIN
O
23
TINO Test Plan
• NINO chips were fully tested @ CERN after packaging• Confirm timing & crosstalk performance of TINO design
in TDIG RFI environment• Verify input to output integrity of each PCB assembly• Verify PCB assembly, current drain• Measure, with TDIG readout, for each channel:
– Threshold input voltage required to produce “0” discrimination threshold
– Overall gain, (input discrimination level @ max. threshold)– Output pulse width “stretch”
24
TINO Test Items Needed
• Power supply with accurate current metering• Pulser with
– MRPC input attenuator / shaping / connector jig– PC interface (for automated testing)
• Production TDIG• PC with readout / control software to implement
desired degree of test automation
25
TPMD (start side FEE)
• New replacement for “TPMT” to match start detector electronics to tray electronics
• Design considerations:– Map 4 channels max. per HPTDC to minimize
crosstalk effect
– Outputs provided for (“fast”) Z-vertex FEE (not part of the TOF project)
• Status:– Design and layout completed– Printed Circuit Boards and parts ordered
26
THUB Status• Needed to concentrate data before sending to DAQ in order to
reduce the number of DAQ-fiber interfaces• Also provides interface to Level-2• Prototype designed @ UT, first board delivered in April 2006 with 4
SERDES links assembled• TCPU-THUB interfaces implemented as SERDES on daughter
cards; distributes trigger, clock, and resets to TCPU, receives data from TCPU
• SERDES link works reliably up to ~25 feet @ 40MHz, > 50 feet @ 20 MHz. Need about 30 feet for start detector cable run
• A cable delay test performed with a TCPU clock driven by a SERDES derived clock and a Run 5 TDIG shows no difference in timing resolution. This indicates derived clock has small enough jitter to operate HPTDC reliably.
• Probably use 20MHz on SERDES link, multiply to 40MHz on TCPU. Will try to add buffer to see if cable length at 40MHz can be increased.
27
THUB Plans• Need another revision for final THUB
– Add Level-2 interface (SIU)– Correct minor issues with prototype– Possibly place SERDES on main board (no daughter cards)?
• Need about 3 weeks of circuit design work and about 4 weeks for layout of final revision
• Production and assembly takes about 4 weeks (only 4 boards are needed)
• Board locations on the STAR magnet have been documented within STAR
• Cable runs were discussed with R.Brown• Two prototype THUB (one for East and one for West
side) will be ready Nov 2006 for installation in Run 7.• Final revision of THUB is not needed until Run 8. Final
design review will depend on TCPU design, probably combined with TCPU Final Design Review
28
TDIG & TCPU Status
• New design for TDIG almost completed• Plan to build 4 to confirm functionality, then
order about 20 for Run 7 and bench testing• TCPU design will start as soon as TDIG design
is finished• Details presented in Lloyds talk• See also Geary Eppley’s talk regarding the Blue-
Sky contract for these boards
29
Production & Testing
30
Schedule
• TDIG Design close to being finished, prototype production to start soon.
• TINO design finished and first 15 boards produced and assembled. Final production can start as soon as compatibility with new TDIG has been bench tested
• TCPU R&D should be finished by Jan 07• TPMD design finished and production of 15 cards has
started. Delivery expected in 5 weeks.• THUB design should be ready for final design review in
Jan 07.• Plan for a system test with all electronics components• Final Design review shortly after first TCPU prototypes
have been bench tested (Jan 2007)
31
Production
• TDIG production rate: ~100 boards per month. Total production should start in Dec 06 and be finished in less than 18 months (~Apr 2008)
• TINO production will stay ahead of TDIG production at about the same rate
• TCPU production will be in parallel to TDIG production, but need fewer boards
• TPMD production can be finished with one order• THUB production can be finished in one order
32
Electronics Testing
• Bare board testing and simple stuffed board testing at Vendor
• Simple single board functionality testing
• Test board set of 8 TINO, 8 TDIG, and TCPU (all tray electronics) as a set on a real tray top
• Test installed electronics on a tray in a cosmic ray setup
33
Electronics Test Plans
Bare board test
Automated board stuffing
Board Level acceptance
test
Tray level board test (17 boards + cables)
Tray assembly
Tray test
Automated board test (“flying probe”)
Burn-in
Purchase bare boards and components
Vendor
Rice
UT
UT
Vendor
Vendor
Vendor