Overview of the new CMS ECAL electronics - PowerPoint PPT Presentation

About This Presentation
Title:

Overview of the new CMS ECAL electronics

Description:

... the Trigger Primitive Generation(TPG) for one trigger tower in the barrel ... sends the encoded and time aligned TPG to the Regional Trigger via a dedicated ... – PowerPoint PPT presentation

Number of Views:55
Avg rating:3.0/5.0
Slides: 28
Provided by: philipp125
Category:

less

Transcript and Presenter's Notes

Title: Overview of the new CMS ECAL electronics


1
Overview of the new CMS ECAL electronics
  • Outline
  • The TDR era
  • - The new architecture
  • - Front-End Electronics
  • Off-Detector Electronics
  • Summary

2
The TDR era
  • Each crystal is equipped with
  • a photo-transducer (APD or VPT)
  • a floating point preamplifier (FPPA) ie 1
    preampli,4 amplis, logic to select online the
    gain (1.,5.,9.,33.)
  • a 12-bit sampling ADC (40 Msamples/s)
  • a bit-serialiser and optical link _at_ 800 Mbits/s
  • Digital data are taken out to the counting room
    housing the Upper Level Readout (ULR)
  • Benefits high level of flexibility, minimal
    amount of radiation hard electronics
  • Drawbacks high volume of data transfer and
    storage, high volume of ULR electronics (more
    than 1000 boards into 60 crates)

3
Glossary
  • in the barrel the trigger towers are 5 X 5
    crystals (ie 0.087 X 0.087 in eta X phi). They
    are formed by 5 strips. Each of these strips is
    1(eta) X 5(phi) crystals.
  • in the endcap the crystals are grouped by 5 X 5.
    Each of these group is a supercrystal. The
    trigger towers geometry (eta/phi) do not follow
    the supercrystal geometry (x/y). A trigger tower
    (ie 0.087 X 0.087 in eta X phi) is made of
    pseudo-strips (groups of 5 crystals with any
    shape). A trigger tower can be made of 1,2,3,4,5
    pseudo-strips

4
The TDR era architecture
transmit all data at 40 MHz
single digitization for both readout and trigger
5
The new architecture on and off detector
principle of a single digitization for readout
and trigger is kept
-transmit trigger data _at_ 40 MHz -transmit readout
data _at_ LV1 rate
6
The new architecture block diagram
  • Glossary
  • VFE identical ie (APD/VPT, FPPA, sampling ADC)
    (On-Detector)
  • FE Front-End (On-Detector)
  • CCS ClockControl System (Off-Detector)
  • DCC Data Concentrator Card (Off-Detector)
  • SRP Selective Readout Processor (Off-Detector)
  • TCC Trigger Concentrator Card (Off-Detector)

7
Front-End function
  • receives the digital signals from the Very
    Front-End ie 12 bits from sampling ADC and 2 bits
    for coding the gain_at_ 40 MHz
  • stores the data during the LV1 latency
  • performs the Trigger Primitives Generation(TPG)
  • formats and sends the data to the DCC upon a LV1
    accept
  • sends the TPG to the TCC _at_ 40 MHz

8
Front-End board barrel
GOL for trigger path 0.8 Gbits/s
GOL for data path 0.8 Gbits/s
FENIX in Trigger Cell Processor mode (25
crystals)
FENIX in Strip mode (5 crystals)
FENIX in DAQ mode (25 crystals)
9
FENIX chip characteristics
  • inputs 5 X 16 bits
  • outputs 16-bit TRIGGER bus, 16-bit DAQ bus
  • 4 operation modes
  • STRIP creating filtered strips
    (barrel)/pseudo-strips (endcap) sums for the
    Trigger Cell Processor(TCP)
  • TCP finalizing the Trigger Primitive
    Generation(TPG) for one trigger tower in the
    barrel
  • DAQ trigger tower (barrel)/supercrystal(endcap
    ) readout state machine, event encapsulation
  • MEMs reading out the laser monitoring system
  • ASIC in 0.25 um CMOS technology
  • FPGA demonstrator

10
FENIX chip the trigger path
linearization from 122 to 18
used in barrel endcap
used in barrel only
total ET
strip or pseudo strip signal
Fine Grain Veto Bit
Fine Grain Veto Bit
time
  • VHDL description verification is in progress
  • implementation in a XC2V1000-4FG256 takes 9 clock
    cycles (40 MHz) to generate the TPG (barrel). The
    target is to run the TPG (barrel) algorithm in 8
    clock cycles in the ASIC version

BCID FIR filter Peak Finder
11
Optical links
  • 3100 Data links (FE to DCC) at 0.8 Gbits/s
  • 4500 Trigger links (FE to TCC) at 0.8 Gbits/s
  • 3000 Clock and Control links (FE from/to CCS)
    at 0.8 Gbits/s
  • instead of 80000 in the TDR approach
  • same technology as the tracker data links
  • 12-fiber ribbons, packed in cables of 8 ribbons
  • packing greatly impacts on the DCC and TCC
    architectures

12
Off-Detector electronics layout
  • 4 TTC/TTS partitions 2X 1/2 Barrel 2 Endcaps
  • Barrel 2 X 18 supermodules Endcap 2 X 4
    quadrants
  • Supermodule 68 trigger towers Quadrant 78
    supercrystals
  • Barrel
  • 1 DCC 1 TCC (double width) 1 CCS per
    supermodule
  • 1 crate 5 supermodules
  • Barrel (ie 1584 trigger towers) electronics fit
    in 8 crates
  • Endcap
  • 2 DCC 10 TCC (simple width) 1 CCS per
    quadrant
  • Endcap (ie 624 supercrystals) electronics fit in
    8 crates
  • SRP 1 crate

13
CCS function
  • interface to Trigger Control System (TCS) and
    Trigger Throttling System (TTS)
  • distribution of clock and control signals DCC and
    TCC and front-end system
  • fan-in the TTS signals from the DCC, TCC and SRP
    and transmission to TTS
  • configuration of the Front-End system

14
CCS block diagram
OD
FE
15
DCC function
  • receives formatted digital data from the
    Front-End boards and TCC
  • verifies integrity of the input event fragments
  • reduces the data volume performing the zero
    suppression using the Selective Readout
    information provided by the SRP
  • formats events
  • transmits events to the global DAQ
  • transmits monitoring and spying events to the
    local DAQ

16
DCC data volumes and data rates
  • 38.1 kBytes for input crystal event (10 time
    samples)
  • 156 Bytes (barrel) or 300 Bytes (endcap) TCC
    event
  • 2 kBytes average output event size
  • data reduction factor 21 in average
  • 200 Mbytes/s average output data rate
  • 528 Mbytes/s maximum output data rate
  • 32 events stored (maximum)

17
DCC block diagram
moves iFIFOs to oFIFOs
68 inputs
2 inputs (monitoring)
Receiver Block 2 channels each
to the DAQ
18
DCC input block
From Selective Readout Processor
demultiplexer 1 to 16
Zero suppression 10-tap FIR filter
19
TCC function
  • receives and deserializes the digital data from
    the Front-End boards
  • performs the geometrical mapping between the
    supercrystals and the trigger towers in the
    endcap
  • finalizes the TPG computations for the endcap
  • encodes the TPG using a non-linear scale for the
    total transverse energy
  • sends the encoded and time aligned TPG to the
    Regional Trigger via a dedicated daughter board
    (Synchro and Link Board)
  • classifies the trigger towers into high, medium
    or low interest for the SRP
  • stores the encoded TPG during the L1 latency
  • transmits the stored TPG to the DCC for each L1
    event

20
TCC block diagram (barrel)
optical to electrical converters (12 channels)
SynchroLink Board to the Regional Trigger (8
channels)
demultiplexer 1 to 16 (12 channels)
to the Selective Readout Processor
processing storage unit (12 channels)
21
TCC block diagram (endcap)
single routing for any position in the endcap
different programs for building the trigger
towers
22
SLB block diagram
4 channels 2 Gbits/s ie 8 trigger towers to the
Regional Trigger
- build online histogram for survey of the LHC
bunch crossings structure - align the channels
BC0
40.08 to 120.24 MHz
23
TCC preliminary implementation (barrel)
double width module to fit the connectors
6 FPGAs for processing data storage
I/Os
24
Selective Readout Processor function
  • selects dynamically (L1 rate) the meaningful ECAL
    information which is spread in 1.5 Mbytes (ie
    number of crystals X 10 samples X 2 bytes)
  • performs suppression in order to reduce the ECAL
    data volume (ie crystals and trigger towers data)
    to 100 kBytes/event (average)

25
Selective Readout Processor principles
  • organizes the readout using readout towers of
    25 crystals (barrel trigger towers,endcap
    supercrystals)
  • classifies the readout towers in term of total
    transverse energy (high, medium or low interest)
  • establishes correspondence between classification
    and readout state (no reading, full reading,
    reading with Zero Suppression)

26
Selective Readout Processor 2 possible
implementations
  • TCC produces the classification _at_ 40 MHz, SRP
    produces the readout states during the L1 latency
    requiring a very high (80 GBytes/s) transfer rate
    between TCC and SRP.
  • TCC produces the classification _at_ L1 rate
    (100kHz), SRP produces the readout states after
    the L1 accept signal requiring extra buffering in
    TCC and DCC for data storage

27
Summary
  • New ECAL electronics architecture has been
    presented
  • Major reduction (40) of the total cost of the
    system with exactly the same functionality as
    described in the TDR
  • Design of a new radiation hard ASIC (FENIX chip)
    in 0.25um is already started
  • Reduced number of data links between the On and
    Off Detector electronics using the same
    technology as the tracker
  • (re)Design of the less costly Off-Detector
    electronics has recently started
Write a Comment
User Comments (0)
About PowerShow.com