Transcript Active Splitter/ShowerMax Monitoring code readiness(Chris), ppt

ShowerMax Input Path Status
Christopher Neu, Daniel Whiteson
L2 Upgrade
Installation Readiness Review
27 September 2004
Outline: •Brief review
•Splitting of input signals
•Status of monitoring
•Plan
XCES Input Path: Description of Run 2A System
•Shower Maximum (CES) data used in L2 to reduce trigger rates for electron and photon triggers
•reduces detector background from single-phototube discharge
•track-matching to shower max reduces combinatoric background for electrons
•CES detector: within CEM calorimetry
•32 wires at |z|<200cm, 32 wires at |z|>200cm for each wedge
•CES charged pulse -> digitized ->
•4 neighboring wires are summed in some time slice(4 CDF clocks): 16 wire groups per wedge
•Compare sum to low and high thresholds: Tlow = 1.5 GeV, Thigh = 3 GeV (changed in ~May)
•Low and high trigger bits are set if wire group sum exceeds threshold
•16 wire groups x 2 thresholds = 32 data bits per wedge/evt (called XCES bits)
•Data flow:
•SMXR board performs sum and threshold comparison; one SMXR per wedge
•Single fiber transmits 32 data + 8 for buffer # = 40 total bits from each wedge/evt
Why 8? Is buffer number transmitted on a 32 bit word?
Fixed length, which is good
Mostly zeroes --- lends itself to zero-suppression to reduce overhead in readout, transmission
•48 total cal wedges => 48 XCES fibers into L2
•4 RECES boards accept these 48 fibers; use Taxi technology; latch according to L2 buffer
•RECES boards are aux cards in L2 decision crate (b0l2de00)
•Alpha reads out XCES bits when needed (for L2 decision making and/or when making TL2D)
September 27, 2004 L2 Upgrade Meeting
Christopher Neu Penn/CDF
Page 2
Review From 3/12 Meeting
• We had an internal mini-review on 3/12 and another in June
• The status of each sub-task/input path was reviewed
• ShowerMax input path significant issues:
– Splitting of incoming optical signals for parasitic commissioning
– Need for development of monitoring tools
• Much progress has been made on both of these fronts - see below
• A few new issues need attention as well
• Much remains to be done
September 27, 2004 L2 Upgrade Meeting
Christopher Neu Penn/CDF
Page 3
Splitting XCES Input Channels
•
Optical splitting problem:
– Existing XCES channels driven by
TAXI protocol through Agilent
optical drivers
• Significant spread of optical power
across system
– Optical splitting easy solution is
passive splitter
• Simple prism
• But incurs 1-to-2 -3dBm optical
strength hit (50% reduction in signal
power)
– Many (about half) of the 48 input
channels are too weak to split
passively
•
Need for an optical splitter that drives
its outputs without significant loss
•
Looked for commercial product – but
found nothing that matched our needs
(somewhat meager compared to
bandwidth of typical current
applications)
September 27, 2004 L2 Upgrade Meeting
ch 0
1
2
3
4
5
6
7
8
9
10
11
Christopher Neu Penn/CDF
West 1
West 2
-15.20 dBm -15.50
-16.94
-16.86
-17.74
-19.41
-14.67
-16.00
-19.15
-15.12
-13.22
-17.07
-17.45
-17.21
-15.33
-15.69
-16.42
-18.29
-15.08
-14.82
-17.20
-16.53
-17.89
-17.20
East 1
-15.91
-17.01
-17.01
-16.44
-17.62
-18.82
-16.55
-19.07
-17.90
-16.15
-18.20
-18.40
East 2
-15.56
-15.47
-15.73
-15.47
-18.35
-15.52
-19.30
-15.68
-15.66
-16.27
-18.22
-16.80
Page 4
Active Splitter: Circuit Design
This is not a schematic! Just a visualization tool…schematic can be found at
http://www.hep.upenn.edu/HEP_INST/OpticalSplitter.jpg
For all the gory details see CDF 7152
Taxi Rx HFBR 2416
AD96687
light
MC10H115
Taxi Tx HFBR 1414
photodiode
light
VEE
GND
LED
Taxi Tx HFBR 1414
light
VEE
LED
VCC
September 27, 2004 L2 Upgrade Meeting
VBB
Christopher Neu Penn/CDF
NB: both +5V and -5V are needed for the circuit.
Page 5
Active Splitter: Circuit Design
This is not a schematic! Just a visualization tool…schematic can be found at
http://www.hep.upenn.edu/HEP_INST/OpticalSplitter.jpg
820 nm light
comes in via
optical fiber
For all the gory details see CDF 7152
Taxi Rx HFBR 2416
AD96687
light
MC10H115
Taxi Tx HFBR 1414
photodiode
light
VEE
GND
LED
Taxi Tx HFBR 1414
light
VEE
LED
VCC
September 27, 2004 L2 Upgrade Meeting
VBB
Christopher Neu Penn/CDF
NB: both +5V and -5V are needed for the circuit.
Page 6
Active Splitter: Circuit Design
This is not a schematic! Just a visualization tool…schematic can be found at
http://www.hep.upenn.edu/HEP_INST/OpticalSplitter.jpg
Analog voltage
is For
input
to gory details see CDF 7152
all the
comparator
820 nm light
comes in via
optical fiber
Taxi Rx HFBR 2416
AD96687
light
MC10H115
Taxi Tx HFBR 1414
photodiode
light
VEE
GND
LED
Taxi Tx HFBR 1414
light
VEE
LED
VCC
September 27, 2004 L2 Upgrade Meeting
VBB
Christopher Neu Penn/CDF
NB: both +5V and -5V are needed for the circuit.
Page 7
Active Splitter: Circuit Design
This is not a schematic! Just a visualization tool…schematic can be found at
http://www.hep.upenn.edu/HEP_INST/OpticalSplitter.jpg
Analog voltage
is For
input
to gory details see CDF 7152
all the
comparator
820 nm light
comes in via
optical fiber
Taxi Rx HFBR 2416
AD96687
light
MC10H115
Taxi Tx HFBR 1414
photodiode
light
VEE
GND
LED
Taxi Tx HFBR 1414
light
Refernce is set
with variable
resistor
VCC
September 27, 2004 L2 Upgrade Meeting
VEE
LED
VBB
Christopher Neu Penn/CDF
NB: both +5V and -5V are needed for the circuit.
Page 8
Active Splitter: Circuit Design
This is not a schematic! Just a visualization tool…schematic can be found at
http://www.hep.upenn.edu/HEP_INST/OpticalSplitter.jpg
Analog voltage
is For
input
to gory details see CDF 7152
all the
comparator
820 nm light
comes in via
optical fiber
ECL output is
fed to repeater
Taxi Rx HFBR 2416
AD96687
light
MC10H115
Taxi Tx HFBR 1414
photodiode
light
VEE
GND
LED
Taxi Tx HFBR 1414
light
Refernce is set
with variable
resistor
VCC
September 27, 2004 L2 Upgrade Meeting
VEE
LED
VBB
Christopher Neu Penn/CDF
NB: both +5V and -5V are needed for the circuit.
Page 9
Active Splitter: Circuit Design
This is not a schematic! Just a visualization tool…schematic can be found at
http://www.hep.upenn.edu/HEP_INST/OpticalSplitter.jpg
Analog voltage
is For
input
to gory details see CDF 7152
all the
comparator
820 nm light
comes in via
optical fiber
Dual parallel
gate outputs
drive 2 optical
transmitters
ECL output is
fed to repeater
Taxi Rx HFBR 2416
AD96687
light
MC10H115
Taxi Tx HFBR 1414
photodiode
light
VEE
GND
LED
Taxi Tx HFBR 1414
light
Refernce is set
with variable
resistor
VCC
September 27, 2004 L2 Upgrade Meeting
VEE
LED
VBB
Christopher Neu Penn/CDF
NB: both +5V and -5V are needed for the circuit.
Page 10
Active Splitter: Implementation
• Assembled by hand at Penn
– Need 12, have 14 in hand
• Use vector boards
– One side has ground plane
– Top side has placement field with
isolated pads
• Four channels per board
• 96 pin connector for power
• Current draw:
– +5V: 40mA/board
– -5V: 700mA/board
• This means total power consumed per
board is less than 4W
• Each input voltage is fused on each
board
September 27, 2004 L2 Upgrade Meeting
Christopher Neu Penn/CDF
Page 11
Active Splitter: Implementation
AS Card 1
Tx Mezz
•
•
-12.85
ch0-1
-12.42
-13.69
ch1-0
-12.41
-12.97
ch 0
-18.40
-15.06
ch1-1
-13.07
-11.98
1
-17.74
-16.02
ch2-0
-11.51
-13.46
2
-17.93
-16.98
ch2-1
-12.20
-12.82
3
-18.03
-14.13
ch3-0
-15.15
-12.69
ch3-1
-13.58
-11.40
Original design of active splitter had significantly higher output power than the Tx’s on TAXI
mezzanine cards used in the system
Determined by current across LED inside the Agilent optical Tx part
AS: ~-12dBm
Pulsar TAXI Tx mezz cards: -15 to -19dBm
AS cards designed to be within the specification of the Agilent and TAXI recommended operating
points
Some risk in exceeding those limits
Overload downstream Rx parts
Adjusted pull-up resistor on Tx to achieve -16dBm output strength across all AS channels
–
•
-12.38
Card 06
–
•
ch0-0
Card 06
–
–
–
•
Tx Mezz
AS Card 2
Amounted to changing a 30ohm to 60ohm
Modification entails changing 96 surface mount resistors
–
Not hard just takes time - Will be completed in the coming weeks
September 27, 2004 L2 Upgrade Meeting
Christopher Neu Penn/CDF
Page 12
Active Splitter: Implementation
•
•
•
•
The active splitter boards will reside in a
custom-built crate
Crate resides in trigger room rack 2RR31C
15 slots
Power provided by two 5V, 20A Lambda
supplies
– Per recommendation from B. DeMaat and S.
Chappa, crate is completely enclosed, like a
black box
– Allows us to use unique backplane to provide
power
•
Fully driven (15 boards) current draw and
power:
– -5V: 10.5A
– +5V: 600mA
– 55.5W total dissipated
•
•
•
Each supply is fused independently
Power supplies will be monitored with
standard CDF tools
Design and implementation approved by
DeMaat, Chappa, and Dervin Allen
September 27, 2004 L2 Upgrade Meeting
Christopher Neu Penn/CDF
Page 13
Active Splitter: Implementation
48 fibers to
Alpha
•
•
Fiber routing recommendation from Dervin
– Need to protect fibers from CH
– Need a flexible configuration for
commissioning
– Recommendation:
• Move CH fibers once
• Couple them to jumper fibers that feed
AS crate
• 48x2 output fibers routed to Run 2A
and 2B systems
– Virtues:
• CH fibers untouched
• Only local fibers move during AS
commissioning
• Easy to work on/remove the AS crate
from data path during commissioning
and down the road
48 fibers to 3
Pulsar boards
2RR31C
(front)
Active
Splitter
Crate
Need to purchase 48 new “jumper” fibers
Patch panels
September 27, 2004 L2 Upgrade Meeting
Christopher Neu Penn/CDF
48 fibers
from CH +
16 spares
Page 14
Active Splitter: Testing
Active splitter cards would endure minimally 1-2
hours of VLTN testing upon arrival at FNAL
from Penn as part of checkout.
Only 3 out of 56 channels were identified as
faulty.
September 27, 2004 L2 Upgrade Meeting
• TAXI protocol is just a method of
encoding data for optical link
– Useful data is packaged in
TAXI transmitter side into
frames
– 5b/4b
– The framing is then interpreted
on the receiver side and data
extracted
• TAXI looks for this framing
constantly –
– an error line is pulled high
when missing
– VLTN signal
• Frames come in at 80MHz
– High rate testing
– 1hr = 3E11 frames
Christopher Neu Penn/CDF
Page 15
Active Splitter: Beam Testing
• Crate with four active splitter cards installed for beam testing in July
• Lack of power supply monitoring restricted tests to data transmission to
upgrade electronics only (did not want to endanger the real system)
• This obstacle contributed to fully split ShowerMax input path not participating
in August commissioning of system (returning real L2 decisions with CPU…)
• But some test runs were collected
• TP2D v. TL2D
– Two pulsar boards receiving copies of split data
– Should be identical
– Looked at all available events with TP2D and TL2D – in the 10s of
thousands
– Two failure mechanisms:
• Buffer number problem (10 events)
• Missing data problem(2 events)
September 27, 2004 L2 Upgrade Meeting
Christopher Neu Penn/CDF
Page 16
Active Splitter: Beam Testing
•
Buffer number problem:
– Of the 16 channels coming into the Pulsar Rx, 7 have L2 buffer number that does not match
its neighbors (which are correct wrt TL2D).
– Dump from PulsarMon:
CES_Monitor::validateCESData() - CCN - simbits (portion of 162 CES words):
03030303030303030303030303030303
03030303030303030303030303030303
00000000000000000000000000000000
CES_Monitor::validateCESData() - CCN - realbits (portion of 162 CES words):
03000303030303000000000003000000
03000303030303000000000003000000
00000000000000000000000000000000
– Problem seen in runs with and without active splitter (same exact channels)
– Similar problem reproduced in teststand (Tx-Rx)
• Needs attention
• This is priority task for ShowerMax path
September 27, 2004 L2 Upgrade Meeting
Christopher Neu Penn/CDF
Page 17
Active Splitter: Beam Testing
•
Missing data problem
•
CES_Monitor::validateCESData() - CCN - simbits (portion of 162 CES words):
00000000000000000000000000000000
00000000000000000000000000000000
00000000000000000000000000000000
CES_Monitor::validateCESData() - CCN - realbits (portion of 162 CES words):
0000000000000000000000
0000000000
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 30000 0 0 0 0 0 0 0 0 0 0
0000000000000000000000
0000000000
– Similar problem seen previously
– Used CESD info to link this error to Alpha side
– Need to verify that this is the culprit in August runs
September 27, 2004 L2 Upgrade Meeting
Christopher Neu Penn/CDF
Page 18
Active Splitter: What’s Next
•
•
Study and repair L2 buffer number problem
Look deeper into the missing data issue
•
•
•
•
Get needed jumper fibers
Make modification to AS boards to throttle back current to Tx
As mods are completed, install in system and test with L2 torture
– Expect help from Penn
– Electronics lab technician coming to participate in the work
– 1-2 week commitment
Fix and diagnose three broken channels
•
Verify the power supply monitoring for crate works
•
Make sure the monitoring is complete and working well
September 27, 2004 L2 Upgrade Meeting
Christopher Neu Penn/CDF
Page 19
ShowerMax Path Monitoring
CESD
CDF
TL2D
TP2D
CESD
PulsarMon
TL2D
TP2D
•Like other Pulsar boards, DAQ buffers in the input and output stage which must be monitored
Can toggle on/off the input and output monitoring depending on task at hand
•Must monitor data from multiple boards
System designed to have 3 boards receiving the 48 XCES channels
Can toggle on/off monitoring on each of these boards
•Comparison is done between simulated TP2D and real TP2D:
•Sources for simulated TP2D:
Upstream bank: CESD
Parallel to TL2D
• Nice complimentarity between these two options
Can isolate errors up/downstream of upgrade electronics
Can identify whether discrepancies between Alpha and Pulsar are 2A or 2B problem
September 27, 2004 L2 Upgrade Meeting
Christopher Neu Penn/CDF
Page 20
ShowerMax Path Monitoring
Using TL2D as reference
Number of missinghigh and falsehigh bits
•Output of monitoring contains many plots
1D stats plots for each input and output channel
2D comparison plots of simulated TP2D versus real TP2D
Global summary histograms
These will be the useful ones for COs
Still developing what exactly will be reported for shift crew
•Error messages also sent to ErrorLog
Also still working on the verbosity level
•Still some open issues but PulsarMon for ShowerMax is maturing
September 27, 2004 L2 Upgrade Meeting
Christopher Neu Penn/CDF
Page 21
Summary
•Things are progressing on the ShowerMax input path
•Much work remains to be done
•Major open issues, manpower and timeline:
Present – Oct 15: - understand data integrity issues from beam runs
- order jumper fibers
Oct 15 – Nov 15: - complete AS modifications
- test each card for VLTN and in system with L2 torture
- system build-up
- finishing touches on XCES portion of PulsarMon and start
running XCES monitoring online
Nov. 15: - personal deadline for fully split 48 ShowerMax channels
- monitoring complete and capable of running online
- consistent L2 testing
Nov. 22: - beam in the machine
Dec. 1: - collisions
September 27, 2004 L2 Upgrade Meeting
Christopher Neu Penn/CDF
Page 22