Title: Presentazione di PowerPoint
1SDD DAQ CARLOSrx
INFN Bologna Samuele Antinori Davide
Falchieri Alessandro Gabrielli Enzo Gandolfi
Massimo Masetti
2Bologna
Torino
Final SDD DAQ chain 24 CARLOSrx
1
busy
LTU
TTCrq
TTCvi
TTCex
2
CARLOSrx
DRORC
12
3CARLOSrx logic architecture
16
16
16
DESER
REC
LOGIC
FIFO
16
16
16
DESER
REC
LOGIC
FIFO
16
16
16
DESER
REC
LOGIC
FIFO
16
16
16
DESER
REC
LOGIC
FIFO
16
16
16
DESER
REC
LOGIC
FIFO
queue manager event builder
16
16
16
data from 12 CARLOS
DESER
REC
LOGIC
FIFO
32
SIU
16
16
16
DESER
REC
LOGIC
FIFO
16
16
16
DESER
REC
LOGIC
FIFO
16
16
16
DESER
REC
LOGIC
FIFO
16
16
16
DESER
REC
LOGIC
FIFO
16
16
16
DESER
REC
LOGIC
FIFO
16
16
16
DESER
REC
LOGIC
FIFO
clock serial link for 12 CARLOS
24
serial link
VME64x
24 TRANSM
LOGIC
LOGIC
ck
L0, L1, L2
TTCrx QPLL
TTC
busy
4CARLOSrx tasks
- What shall CARLOSrx do ?
- high bandwidth data concentration and buffering
- from 12 CARLOS end ladder boards to 1 DDL
- 12 x 640 Mbit/s in input, 1.28 Gbit/s in output.
- error checking and status monitoring
- clock distribution and build the busy signal
towards the LTU - program the front-end chips when receiving the
related command from the SIU and when some errors
occur. - Data buffer capacity
- A SDD event is 128 Kbytes without compression
- The same event takes 7 Kbytes with compression
- CARLOSrx has a buffer gt 256 Kbytes per channel
5Final CARLOSrx board design
- Each CARLOSrx board has to manage
- 12 optical fibres (800 Mbit/s) from 12 SDD end
ladders ? data - 12 optical fibres (80 Mbit/s) towards the SDD end
ladders ? clock - 12 optical fibres (40 Mbit/s) towards the SDD end
ladders ? serial link - TTCrq interface ? timing trigger information
(including L0) - SIU interface ? front-end config and data
transmission - VME interface ? loading firmware into on-board
PROMs - LTU interface ? busy signal
6(No Transcript)
7May 05 first CARLOSrx design (one 6U board)
Layout scheme presented at ITS week in May 05
where it was foreseen to use the
parallel optical receiver / transmitter for 12
optical fibres from NGK electronics
BUT
8June 05
- PROBLEM No parallel optical transceiver was
available any longer on the market. - SOLUTION Single optical transceivers have to be
used. - PROBLEM 24 optical transceivers cannot be hosted
on only one VME board. - SOLUTION CARLOSrx breaks in 2 9U PCBs
- CARLOSrx for data processing
- CARLOSrx for clock distribution
BUT the deadlines shift because the realisation
is much more sophisticated
9CARLOSrx for data processing
40 MHz
80 MHz
80 MHz
80 MHz
16
16
DESER
TRANSC
XC2VP20
R
FIFO
32
32
16
16
DESER
TRANSC
FIFO
R
FIFO
16
16
DESER
TRANSC
FIFO
R
72T36125
16
16
FIFO
R
DESER
TRANSC
32
16
16
32
FIFO
R
FIFO
DESER
TRANSC
16
16
FIFO
R
DESER
TRANSC
32
1 MB
SIU
40 MHz
FIFO
80 MHz
DESER
TRANSC
16
16
40 MHz
FIFO
R
100 KB
DESER
TRANSC
32
32
16
16
FIFO
R
FIFO
DESER
TRANSC
16
16
FIFO
R
DESER
TRANSC
16
16
72T36125
R
FIFO
32
16
16
32
DESER
TRANSC
FIFO
R
FIFO
16
16
DESER
TRANSC
FIFO
R
serial link
L0, L1 L2
XC2VP7
clock
TTCrq
TTC
10CARLOSrx for data processing
- 9U x 400 mm PCB
- 12 Optoway optical transceivers for I/O (dealing
with 24 I/O fibers) - 3 Virtex II PRO FPGAs
- 1 Spartan II FPGA
- Link to the SIU
- 12 internal bus, each 16 bit at 40 MHz
- Power consumption
- 6 A from 5V
- 8A from 3.3V
11data
dataflow
TTCrq
SIU
12trigger distr.
serial link
SIU
13clock distr.
clock
SIU
14JTAG distr.
serial link
TTCrq
SIU
15VME control
TTCrq
SIU
16(No Transcript)
17CARLOSrx for clock distribution
18Foreseen CARLOSrx test setup
19CARLOSrx test
20CARLOSrx test
21VME crate
Delivery date 1 Feb 2006
22VME crate
Card-Cage Wiener 0B06.021JK Vario Local
4x6U, 17x9U, FT400mm Total Height 11U
Backplane VME64x 6U slots 4 Power
supply Wiener 0P06.0720 6U (5V/200A,
3.3V/100A, 12V/10A) Cooling air Control
CANbus Fan-tray Wiener 0F00.070A 9U FM (400mm)
23Boards disposition in the VME crate
CARLOSrx CLOCK
CARLOSrx
FANIN
CPU
CARLOSrx CLOCK
CARLOSrx CLOCK
CARLOSrx CLOCK
CARLOSrx CLOCK
CARLOSrx CLOCK
CARLOSrx
CARLOSrx
CARLOSrx
CARLOSrx
CARLOSrx
CARLOSrx CLOCK
CARLOSrx CLOCK
CARLOSrx
CARLOSrx
empty
empty
empty
6 U
9 U
24Work plan in Bologna
End November 2005 PCB final design submission
for CARLOSrx data processing, while CARLOSrx
clock will be delayed if the clocks have to be
shifted with programmable steps. January 2006
Start of the test of the prototype with the two
boards linked together inside the crate. March -
April 2006 Start of the production if the test
is OK.
25(No Transcript)