L1Calo rack layout in USA15

Rationale

The main considerations for the calorimeter trigger are:

• The receivers and PPMs need plenty of space in front of the racks to accommodate bulky cabling. Therefore they cannot be in the central area where the wall is only 1m away. For safety, we leave an extra 1 rack width from the wall area.
• The CPM and JEM racks are located as centrally as possible while still leaving the required(?) three rack gap. This minimises the length of the longest LVDS cable.
• The longest latency is (was?) expected to be in Jet/Energy processor, so the JEM rack is placed closest to the proposed location of the CTP on the floor below.

Other comments:

• It would be helpful in organising the receiver-PPM cabling if those racks could have a slightly increased spacing between them. Adding an extra 10cm into three rack boundaries on each side would effectively mean losing one rack width from the whole row. This would presumably be lost from the centre. (But Im not sure this is possible).
• Unallocated racks in the centre (and probably many of those at the outside) of the row are required for computers and other test equipment.

Estimate of cable lengths

The basic assumptions are:

• 0.6 metre rack width
• maximum horizontal routing between racks (ie from furthest edges)
• about 2.5m down from upper crate to uppermost underfloor cable tray
• about 0.5m each end for cable management

The lengths of the long signal cables from the detectors will be
determined by Technical Coordination. For the purposes of looking
at various rack layouts, it is useful to estimate the length from
the holes to the outermost racks in a particular layout.

Signal cables from holes to outermost receivers:
(A side) horiz:   5.5*0.6 = 3.3m    (C side) horiz:   4.5*0.6 = 2.7m
         forward:           1.5m             forward:           1.5m
         upwards:           2.5m             upwards:           2.5m
                            ----                                ----
                            7.3m                                6.7m

Signal cables from holes to TileCal patch panels (1 rack further):
(A side) horiz:   6.5*0.6 = 3.9m    (C side) horiz:   5.5*0.6 = 3.3m
         forward:           1.5m             forward:           1.5m
         upwards:           2.5m             upwards:           2.5m
                            ----                                ----
                            7.9m                                7.3m

TileCal patch panels to receivers (depends on receiver layout):
(both sides) horiz: 3.0*0.6 = 1.8m
  fwd/back/up/down:          ~4.2m
                              ----
                              6.0m 

Receivers to PPMs:
(both sides) horiz: 4.0*0.6 = 2.4m
  fwd/back/up/down:          ~3.6m
                              ----
                              6.0m 

PPMs to CPMs (LVDS cables):
(A side) horiz:     9*0.6 = 5.4m    (C side) horiz:     6*0.6 = 3.6m
         fwd/back:          1.0m             fwd/back:          1.0m
         up/down:   2*2.5 = 5.0m             up/down:   2*2.5 = 5.0m
                            ----                                ----
                           11.4m                                9.6m

PPMs to JEMs (LVDS cables):
(A side) horiz:     7*0.6 = 4.2m    (C side) horiz:     7*0.6 = 4.2m
         fwd/back:          1.0m             fwd/back:          1.0m
         up/down:   2*2.5 = 5.0m             up/down:   2*2.5 = 5.0m
                            ----                                ----
                           10.2m                               10.2m

Between CMMs:
(a) via hole between racks:    CP: 1.5m,  JEP: 1.0m
(b) via underfloor cable tray: CP: 4.0m,  JEP: n/a

NB option (a) was mentioned to Mark Hatch a long time ago without
raising strong objections, so the shorter path is assumed below.

CMM to CTP (with present CTP position on niveau 1):
CP system:  horiz:  3*0.6 = 1.8m    JEP system: horiz:  2*0.6 = 1.2m
            fwd/back:       1.0m                fwd/back:       1.0m
            downwards:     ~4.0m                downwards:     ~4.0m
                            ----                                ----
                            6.8m                                6.2m


Longest cable lengths from the Receivers to the CTP:
CP system:   6.0 + 11.4 + 1.5 + 6.8 = 25.7m (ie about 5.1 BCs)
JEP system:  6.0 + 10.2 + 1.0 + 6.2 = 23.4m (ie about 4.7 BCs)

Including the TileCal patch cables:
CP system:   6.0 + 6.0 + 11.4 + 1.5 + 6.8 = 31.7m (ie about 6.3 BCs)
JEP system:  6.0 + 6.0 + 10.2 + 1.0 + 6.2 = 29.4m (ie about 5.9 BCs)

The present layout is optimised to reduce the total cable length for
the JEP system. If the CP system has the worst latency, the positions
of the JEP and CP racks could be swapped. This would reduce the total
CP cable from ~26m to ~25m and increase the JEP from ~23m to ~25m
(plus TileCal patch cables).


The Glink cables dont affect the latency, but estimates of the lengths
we expect are required for bit error rate testing of these lengths.
The cablse from the PPMs come from the back of each module, so may
have to go down under the floor at some point (certainly from the A
side) whereas those output from the front of the CPMs/JEMs can go
directly across the front of the racks to the ROD front panels.

Glink cables to RODs:
PP (A side):    10*0.6 + 1.0 + 5.0 = 12.0m  (under floor)
PP (C side):     4*0.6 + 1.0 + 5.0 =  8.4m  (under floor)
CP/JEP (max):    4*0.6 + 0.5 + 2.0 =  4.9m  (across front)