|
|
Crates
- 6 LAr receiver crates (2 EM barrel, 2 EM endcap, 2 Hadron endcap plus FCAL)
- 2 Tile receiver crates (as above)
- 8 Receiver summing patch panels (4 quadrants * 2 ends)
- 8 PreProcessor crates (organisation roughly as the receiver crates)
- 16 PPMs (15 in some crates)
- 1 TCM (with VME64x adapter card)
- 1 CPU
- 4 Cluster Processor crates (phi quadrants)
- 2 Jet/Et Processor crates (quadrants 0+2, 1+3)
- 2 ROD crates (suggested organisation)
- 4 PP slice data RODs
- 2 CP slice data RODs
- 2 CP RoI RODs
- 1 JEP slice data ROD
- 1 JEP RoI ROD
- 1 TCM (with VME64x adapter card)
- 1 BUSY module (?)
- 1 CPU
- 1 TTC/BUSY crate
- 1 TTCvi
- 1 TTCvx (or similar)
- >=1 BUSY module
- 1 DCTPI module (double width)
- 1 CPU
- 64 channels
- 4 LAr signal cables (16 pairs)
- 8 Tile signal cables (4 with 9 or 10 pairs, 4 with 6 or 7 pairs)
- 4 output cables to PPMs (but see below)
- 16 channels (4*4 towers) per cable
- 1 output cable to summing interconnect (4 towers per cable)
- 16 channels
- 4 input cables of 4 channels each (from receiver sum outputs)
- 1 output cable of 16 channels each (to normal receiver input)
NB these are used to bring EM barrel signals to EM endcap
receivers for summing across the transition region.
- 64 channels
- 4 input cables from receivers
- 16 channels (4*4 towers) per cable
- 4 analog input daughter cards (1 per cable)
- 16 PPrMCMs, each containing
- 4 channels
- 4 FADCs
- 1 PHOS4 timing chip
- 1 PPrAsic
- 3 LVDS transmitters (2 CP, 1 JEP)
- 1 Readout/merger FPGA
- 10 (8 + 2 fanout) 4-channel LVDS cables to CPMs (8 BCmuxed towers per cable)
- 5 (4 + 1 fanout) 4-channel LVDS cables to JEMs (4 0.2*0.2 jet sums per cable)
- 1 (fanout) 2-channel LVDS cables to JEMs (2 0.2*0.2 jet sums per cable)
- 1 Glink output (slice data)
- 80=10*4*2 signals input (40 each BCMuxed EM and Hadronic)
- 80=20*4 towers direct input (76=19*4 used)
- 64=16*4 towers processed
- 60=20*3 towers sent to neighbours
- 60=20*3 towers received from neighbours
- 80 LVDS receivers
- 20 serialiser FPGAs
- 8 CP algorithm FPGAs
- 2 Glink outputs (1 slice data, 1 RoI)
- 88=11*4*2 signals input (44 each EM and Hadronic 0.2*0.2 sums)
- 44=11*4 jet cells direct input (summed EM+Hadronic)
- 32=8*4 jet cells (summed EM+Hadronic) processed
- 33=11*3 jet cells sent to neighbours
- 33=11*3 jet cells received from neighbours
- 88 LVDS receivers
- 11 input/Et FPGAs
- 1 jet algorithm FPGA
- 2 Glink outputs (1 slice data, 1 RoI)
- 16 CPM/JEMs processed
- 24 bits per CPM/JEM
- 3 input cables to system CMM from other crate CMMs (CP system)
- 2 input cables to system CMM from other crate CMM (JEP system)
- 1 output cable from crate CMMs to system CMM (CP system)
- 2 output cables from crate CMM to system CMM (JEP system)
- 2 output cables to CTP (only from system CMMs)
- 64 LVDS pairs per cable
- 2 merger FPGAs (1 for crate, 1 for system)
- 2 Glink outputs (1 slice data, 1 RoI)
- 1 TTC optical fibre input
- 1 CANbus output (via ELMB?)
- 18 electrical backplane outputs
- 4 electrical front panel outputs
See backplane spec.
- 18 Glink inputs
- Up to 4 Slink outputs (via rear transition module)
- 1 BUSY output
- 1 AUX output?
ROD allocation
One 18 channel ROD will handle either the slice data or the RoIs
from all the modules in one crate, ie there are two RODs per crate
(for CP and JEP subsystems).
The slice data requires up to four Slinks, one per four or five inputs,
to provide the required bandwidth.
For the RoIs the bandwidth is much smaller. One Slink to each of
the two destinations (ROS and RoIB) is adequate.
Source | N.RODs | N.inputs/ROD |
N.outputs/ROD |
PP data | 8 | 16 PPMs | 4 |
CP data | 4 | 16 = 2 CMMs + 14 CPMs | 2 |
CP RoIs | 4 | 16 | 2 |
JEP data | 2 | 18 = 2 CMMs + 16 JEMs | 4 |
JEP RoIs | 2 | 18 | 2 |
Total | 20 | | |
Slinks
- 32 Slinks to ROS from PP RODs
- 12 Slinks to ROS from CP RODs (8 data, 4 RoIs)
- 10 Slinks to ROS from JEP RODs (8 data, 2 RoIs)
- Total of 54 Slinks to ROS
- 4 Slinks to RoIB from CP RODs
- 2 Slinks to RoIB from JEP RODs
- Total of 6 Slinks to RoIB
Full details of all input and internal cabling and patch panels
are given in our
cabling document
(in future this will be in EDMS). Some summary numbers are given below.
From the LAr calorimeters, 16 pair cables are used throughout
although not all parts of the calorimeter need all 16 pairs.
From the TileCal, 16 pair cables are also used throughout.
The muon signals are carried on the same cables and split off
at special patch panels.
Source | N.pairs required | N.Cables (C side) |
N.Cables (A side) | N.Cables (total) |
LAr EM Barrel | 15 | 64 | 64 | 128 |
LAr EM Endcap | 16 | 56 | 56 | 112 |
LAr HEC | 16 | 48 | 48 | 96 |
LAr FCAL | 16 | 12 | 12 | 24 |
Tile Barrel | 9 (10) | 64 | 64 | 128 |
Tile Ext.Barrel | 6 (7) | 64 | 64 | 128 |
Total | | 308 | 308 | 616 |
The 256 cables from TileCal come to two patch panels (C side and A side)
where calo and muon signals from barrel and extended barrel cables
(128 of each) are merged. There are 128 combined output cables to
the TileCal Receivers and another 128 outputs to muon discriminators.
In general, one Receiver output cable carries 16 signals to one PPM input.
However there are four regions where special arrangements are required
either combining or splitting cables. These are at the inputs of PPM_4,
PPM_7 and hadronic PPM_9s, where the PPM numbering is taken from the
PPM specification.
- PPM_4 (EM): the EM barrel/endcap transition requires that two receiver
outputs are combined to a single PPM input.
There are 8 EM instances of PPM_4 (2 ends * 4 quadrants)
making 32 special PPM input cables from 64 receiver
outputs. Each receiver output provides 8 signals.
- PPM_4 (had): the TileCal/HEC transition is more complicated. Here one
HEC output cable must be split into four sets of 4 signals
and combined with four TileCal receiver outputs (each
providing 12 signals) making four PPM inputs. This assumes
we will be able to arrange the TileCal receivers like the
EM barrel which should be OK in principle I think, as long
as we can get all the cables into them.
There are 8 hadronic PPM_4s, making 32 PPM inputs coming
from 40 receiver outputs.
- PPM_7 (both): here one receiver output is split into two PPM inputs
in both EM and hadronic layers, ie 64 special PPM input
cables from 32 receiver outputs.
- PPM_9 (had): each hadronic FCAL PPM input comes from combining two
receiver outputs providing 8 signals each. There are two
hadronic PPM_9s in the system, so this represents 8 PPM
input cables from 16 receiver outputs.
The total numbers of normal and special cables are as follows:
| PPM Inputs | Connection | Receiver Outputs |
4 * (5 * 16 + 8 + 2) | 360 | normal 1:1 cables | 360 |
4 * (8 + 2) | 40 | Rec:PPM 2:1 connections | 80 |
4 * 8 | 32 | Rec:PPM 5:4 connections | 40 |
4 * 16 | 64 | Rec:PPM 1:2 connections | 32 |
| 496 | Total input/outputs | 512 |
There are a total of 124 PPMs (16 each of PPM_1 to PPM_7, 8 PPM_8s and 4 PPM_9s).
There are (probably) 32 TileCal Receivers and 96 LAr Receivers:
32 for EM barrel, 32 for EM endcap, 26 for HEC and 6 for FCAL (2 EM, 4 hadronic).
PPM-CPM Cabling
- From one PPM:
- 8 4-channel LVDS cables to CPM in same quadrant
- 1 4-channel LVDS cable to CPM in quadrant+1
- 1 4-channel LVDS cable to CPM in quadrant-1
- To one CPM:
- 16 4-channel LVDS cables from 2 PPMs (EM+had) in same quadrant
- 2 4-channel LVDS cables from 2 PPMs (EM+had) in quadrant-1
- 2 4-channel LVDS cables from 2 PPMs (EM+had) in quadrant+1
PPM-JEM Cabling
- From one PPM:
- 4 4-channel LVDS cables to JEM in same quadrant
- 1 2-channel LVDS cable to JEM in quadrant+1
- 1 4-channel LVDS cable to JEM in quadrant-1
- To one JEM:
- 16 4-channel LVDS cables from 4 PPMs (EM+had, 2*0.4 in eta) in same quadrant
- 4 2-channel LVDS cables from 4 PPMs (EM+had, 2*0.4 in eta) in quadrant-1
- 4 4-channel LVDS cables from 4 PPMs (EM+had, 2*0.4 in eta) in quadrant+1
- NB FCAL cabling is more complex: see PPM and JEM module specs and/or
the cabling document.
CMM-CMM Cabling
See CMM module spec.
CMM-CTP Cabling
See CMM module spec.
Last updated on 22-Jul-2004
by Murrough Landon
|