The CALICE DAQ2 system for Si-W ECAL is based on the following parts:
- CCC form UK modified with a 10-100 ethernet module instead of the RS232 interface and put into a triple width 6U VME mechanical module
- standard NIC board into a PC
- commercial G ethernet switch
- CAT6 cables
- GDCCs put into a triple width 6U VME mechanical module
- optional DCCs (single with 6U VME module)
- optional reset module based on Adruino (recommanded for LDA)
- DIF from UK together detector SLABs
- separate power supplies (-5, +5, +3.3 for CCC ; +5 for LDA ; +3.3 for DIF ; +5 for SLABs)
- pulse generators (SPILL, charge injection, etc...)
VME modules are put into a VME crate with J1 backplane. The VME bus is
not used for data but for power supplies only.
It is recommended to assemble a custom crate with a 12 or 15 slots J1 in order to include the CCC module which has a larger depth. A list of schroff references is given below:
Description |
Schroff |
Radiospares |
BACKPLANE 12SLOT VME J1 |
23000042 |
500-2152 |
Subrack EuropacPRO, 6U 355D shielded |
24563444 |
721-2988 |
6U/3U 63HP |
24562-463 |
487-602 |
GUIDE P160 10X |
24560-351 |
484-8418 |
MODULE 6U |
20817459 |
437-1889 |
Mechanical modifications for power supplies, their connectors and internal cabling have to be foreseen by users.
Users have also to build their own modified CCC and reset modules.
GDCC can be replaced by an LDA (an equivalent firmware can be given) in the case of simple setups (one or two detector modules).
It is reminded that the system is sensitive to cabling and power supplies (in particular at start-up).
Minimal desktop setup
- no VME mechanics, boards laid down on the table
- 1 PC (scientific linux 6) + 10/100/1000 NIC
- 1 CCC
- 1 LDA (GDCC when available)
- separate power supplies (-5, +5, +3.3 for CCC ; +5 for LDA ; +3.3 for DIF ; +5 for SLABs)
- 1 pulse generator for 10 Hz SPILL signal (TTL 5V, active low pulse during ~2ms)
KNOWN ISSUES
- power on sequence (seen on AHCAL & SI-W ECAL with 2 different systems)
- undeclared SLABS must be removed or not powered (unless: DIF will send right data but unexpected by the software)
- packet loss (<1/10000 somewhere in the switch, NIC or SW driver)
--
RemiCornat - 19 Nov 2013