Transcript Power Point file

TGC LVL1 Trigger System Link
C.Fukunaga/Tokyo Metropolitan University
TGC electronics group
System link
• Data transfer for EC muon LVL1 trigger processing is
important because system is sliced to four parts in two
areas
• On detector
• In USA15 cavern
PS pack
~15m
~90m
Patch Panel
Slave Board
Hi-pT Board
(Bunch X
ID)
(Low pT ID)
(High pT ID)
On detector
Sector
Logic
USA15
• Data Link for DAQ between SB and Star Switch
14/6/2000 End Cap Muon Trigger
Review at CERN
1
Trigger Data Flow
Optical fibre
20 bit
G-Link(90m)
9 bit
LVDS
(15m)
14/6/2000 End Cap Muon Trigger
Review at CERN
2
Long Distance LVDS link
• For Trigger data of ~10 bit data transfer between SB and
Hi-pT, we plan to use differential LVDS signal protocol.
– Length ~ 15m
– Bandwidth ~10bit * 40MHz =50MByte/s
– 10 bit data are serialized at the output of an SB and
deserialized at Hi-pT input
• Problems foreseen:
– Serializer/deserializer chips
– Long distance transfer
– Magnetic field
• Experience of LVL1 Cal. Group is very helpful for us
14/6/2000 End Cap Muon Trigger
Review at CERN
3
Issues for LVDS
•
10 bit Serializer/Deserializer from National Semiconductor Co.
DS92LV1021/1210 (16-40 MHz) will be ideal one for our purpose as
LAr Group does.
– They pointed out, however, the chipset has not been easy to use; careful
board layout is needed, it must be well separated from the power-supply
lines to protect noise pick up.
•
There is an evaluation board recently released for this chipset from
the company.
– With this board, we check the chipset with long distance signal xfer.
•
New 10 bit Serializer/Deserializer DS92LV1025/1224 chipset but
works for 40-66 MHz is also released
– Since the pin allocation and size are the same as present ones, we can check
this with the same test setup.
14/6/2000 End Cap Muon Trigger
Review at CERN
4
Issues for LVDS (cont.)
• Long distance driving of LVDS (~15m)
LAr group has envisaged high frequency losses of cable over
long distance. Losses cause jitter at Receiver side (> 100ps) .
For reduction this value, we may need some compensation
circuits.
• Magnetic Field
500-1KGauss field is foreseen, it will make problems owing to
– Small driving currents of LVDS
– Nature of differential signal transfer (two reciprocal currents in
a pair)
14/6/2000 End Cap Muon Trigger
Review at CERN
5
G-Link
for Hi-pT and Sector Logic
• G-Link (Gigabit Rate Transmission Protocol) is developed by
Hewlett-Packard.
• Serializer/deserializer with transceiver/receiver chipset
HDMP-1022/1024 by HP is useful for data transfer from a
Hi-pT board to Sector Logic in USA15
– Length ~90 m
– Bandwidth: 20 bit*40 MHz ~ 0.8 Gbyte/s (Gbit Ethernet)
14/6/2000 End Cap Muon Trigger
Review at CERN
6
G-Link Structure
HP
HDMP-1022
VCSEL
(LASER
EMITTER)
HP
HDMP-1024
Optical
Fibre
14/6/2000 End Cap Muon Trigger
Review at CERN
7
Issues
• VCSEL & Optical fibre
– VCSEL (Vertical Cavity Surface Emitting Laser)
• High Band Width (1-10Gb/s)
• Surface emitter – good connectivity with optical fibre
– Low divergence, circular beam profile from VCSEL is
appropriate for multimode fibre
• High output power/Low power consumption
• Yet a few products for 850nm light emission (short wave)
– Optical fibre
• Recent development of GB/10GB Ethernet makes easier to
select appropriate multimode fibres
– for IEEE 802.3z 1000BASE-SX
– Core/Outer 50/125um, Graded Index
14/6/2000 End Cap Muon Trigger
Review at CERN
8
Schedule
• LVDS link
– Till summer test with an evaluation board will be completed
including compensation circuits issues
(Evaluation board is new product of National – it takes 2 months
for delivery)
– Then we will check the newer chipset with the same evaluation
board
• G-Link
– In autumn after LVDS test we will start extensively the G-link
study
14/6/2000 End Cap Muon Trigger
Review at CERN
9