CERN/CMS optical links radiation CERN/CMS optical links radiation damagedamage
K. Gill, J. Troska, F. VaseyR. Grabit, C. SigaudM. Axer, S. Dris
CERN EP-MIC/OE
[email protected] www.cern.ch/cms-tk-opto RADWG 29-11-2005
OutlineOutline
• Optical links
• QA programme
• Selection of test results
[email protected] www.cern.ch/cms-tk-opto RADWG 29-11-2005
Optical link for CMS Tracker readout/controlOptical link for CMS Tracker readout/control
40k fibres (40Msamples/s analogue 8 bit)7k fibres (80Mbit/s digital)9k fibres (1Gbit/s digital)
Links ~70m long, multiple breakpoints
[email protected] www.cern.ch/cms-tk-opto RADWG 29-11-2005
Analogue link architecture (1996)Analogue link architecture (1996)
9612
CMS Tracker
Laser Transmitters on optohybridRuggedizedRibbon
Dense Multi-ribbonCable
Distributed PP In-line PP
Rx-Module
1
12
Back-end PP>40k fibrechannels
System specINL 1% typS/N 48dB typBW>70 MHz
[email protected] www.cern.ch/cms-tk-opto RADWG 29-11-2005
Implementation: Technology choice (1996)Implementation: Technology choice (1996)
• Developed analogue link system first (most links + most difficult)
Requirement Technology choice
Linearity Edge emitting Laser
Dynamic Range Single mode system, 1310nm wavelength
Settling Time Fast electronics (CMOS-Sub)
Gain 10bit ADC with equalization
Magnetic Field Non-magnetic connectors and packages
Radiation Extensive qualification of COTS-based components
Density
Low massSemi-customized laser package
Fibre ribbon & array connectors
Customized multi-ribbon cable
Semi-customized Rx-module
• Control link and ECAL readout link developed later using many of same parts
[email protected] www.cern.ch/cms-tk-opto RADWG 29-11-2005
Implementation: Components (2000-02)Implementation: Components (2000-02)
ST/CERN-MIC/Kapsch/G&ASumitomo
Diamond
Ericsson
NGK Optobahn
Front-endoptohybrid
Distributedpatch panel In-line
patch panel
Cable
Back-end A-RX
Ericsson
Ericsson
• Many COTS/COTS-based parts • Each component also has CERN and supplier specification• Long procurement and qualification process, complicated logistics • Current status - production complete or very nearly complete for all parts• Collaboration including Perugia, Vienna, Minnesota university groups
[email protected] www.cern.ch/cms-tk-opto RADWG 29-11-2005
Charged hadron fluence (/cm2 over ~10yrs)(M. Huhtinen)
Requirements: radiation environmentRequirements: radiation environment• High Energy 7+7TeV• High rate
– Large radiation field• mainly pions (few hundred MeV) in Tracker
[email protected] www.cern.ch/cms-tk-opto RADWG 29-11-2005
COTS issues COTS issues
• Extensive use of commercial off-the-shelf components (COTS) in CMS optical links
– Benefit from latest industrial developments
• cheaper• reliable qualified devices
– However COTS not made for CMS environment
• no guarantees of long-life inside CMS
• validation testing of COTS mandatory before integration into CMS
LLD ASIC
RX40 ASIC Pin photodiodes
Laser diodes
Fibres
e.g. control links digital opto-hybrid
[email protected] www.cern.ch/cms-tk-opto RADWG 29-11-2005
Quality Assurance programmeQuality Assurance programme
Prototype sample validation
Pre-production Qualification
Pre-productionAdvance validation test (AVT)
Lot acceptance
FunctionalityRadiation hardness
FunctionalityRadiation hardnessReliability
Radiation hardnessReliability
Functionality
1995-2000
2000-2002
2003-2004
2003-2005
Integration2005-2006 Functionality
[email protected] www.cern.ch/cms-tk-opto RADWG 29-11-2005
Radiation testingRadiation testing• e.g. validation tests on lasers
irradiation irradiation
annealingageing
(in-system) lab tests Market Survey
(in-system) lab tests
• Measurements:– Damage: different sources, different T, bias– Annealing rates, acceleration factors– Wearout
• >500 laser samples tested in total
n irradiation
qualification
AVT
[email protected] www.cern.ch/cms-tk-opto RADWG 29-11-2005
• In-situ measurements allows confident extrapolation/comparison of accelerated tests– Avoid before/after tests unless damage kinetics understood– Few changes to test-procedure since 1997 for consistency
• Very similar system used for fibre and photodiodes
Irradiation test systemIrradiation test system
• Measurement setup (lasers)
Opticalreceiver
array
Currentgenerator
array
PC
Datalogger
GPIB
laser undertest
fibre
irradiation cellor oven
[email protected] www.cern.ch/cms-tk-opto RADWG 29-11-2005
SCK-CEN Co-60 2kGy/hr
underwater
Irradiation at SCK-CEN and UCLIrradiation at SCK-CEN and UCL
neutrons
deut
eron
sUCL (T2 source)~20MeV neutrons
flux ~ 5x1010n/cm2/s
Samples stackedinside cold box (-10°C)
Interested to use these sources?please contact me…
[email protected] www.cern.ch/cms-tk-opto RADWG 29-11-2005
Radiation Damage in lasers – 20MeV neutronsRadiation Damage in lasers – 20MeV neutrons
• A lot of damage from neutrons
– Increase in laser threshold current
– Decrease of laser efficiency
Typical L-I characteristics before and after 5x1014n/cm2
1000
800
600
400
200
0
Ou
tpu
t lig
ht
po
we
r, L
(W
)
50403020100
Current, I (mA)
before irrad
after 5x1014
n/cm2
25
20
15
10
5
0
thre
sho
ld in
crea
se (
mA
)
1.0 2.0 3.0 4.0
~20MeV neutron fluence (1014
n/cm2)
1.00
0.95
0.90
0.85
0.80
0.75
0.70
rela
tive e
fficiency, E
/E(0
)
efficiency
threshold
• Damage proportional to fluence– Degradation of carrier lifetime
• After 4.5x1014n/cm2
– Threshold increase ~20mA– Efficiency loss ~20%
• All wafers tested with this source in advance of final production (2002-2004)
– 300 lasers in total
[email protected] www.cern.ch/cms-tk-opto RADWG 29-11-2005
Laser damage mitigationLaser damage mitigation
• LLD specified to compensate for laser damage
• for threshold up to 45mA
– Recall worst-case CMS-Tracker
Ithr~5.3mA after 10 years
• Large safety margin
(almost 10x)
LLD ASIC Laser diodes
[email protected] www.cern.ch/cms-tk-opto RADWG 29-11-2005
Annealing in lasersAnnealing in lasers
• Significant amount of annealing
– Proportional to log(tanneal)
• distribution of activation energies for annealing
• Similar rate for efficiency
– damage mechanism same as Ithr
• Expect at least 70% annealing after 10 years
102 105103 104
1.0
0.8
0.6
0.4
frac
tion
of d
amag
e re
mai
ning
1.0
0.8
0.6
0.40.1 1 10 100
annealing time (hrs)
wafer E
threshold damage
efficiency damage
0.4
0.2
[email protected] www.cern.ch/cms-tk-opto RADWG 29-11-2005
Relative damage factors
Valduc 0.75MeV n (=1)
UCL 20MeV n (=4.5)
PSI 200MeV (=8.4)60Co (~0)
• Coverage of various parts of CMS particle energy spectrum– Pions most important
• Similar factors for other 1310nm InGaAsP/InP lasers
Damage different sourcesDamage different sources
• Laser threshold Ithr with different sources (averaged and normalized) (1998-2000)
[email protected] www.cern.ch/cms-tk-opto RADWG 29-11-2005
• Based on damage factors and annealing rate at close to -10°C
• Take worst-case– radius=22cm in Tracker– pion damage dominates
Ithr = 6mA in 10 years
E = 6% in 10 years
• Damage decreases rapidly with distance from beam interaction point
– ~50% at r=32cm
Laser damage prediction in CMS Tracker Laser damage prediction in CMS Tracker
• Even without thorough understanding, can predict damage evolution over a 10-year lifetime inside Tracker
80
60
40
20
0
dam
age
(% d
efec
ts le
ft a
fter
10
yrs)
1086420
LHC operating time (years)
LHC luminosity profile:
year 1: 10%year 2, 33%year 3, 66%
years 4-10, 100%
total damage annual components
[email protected] www.cern.ch/cms-tk-opto RADWG 29-11-2005
• leakage current and responsivity Fermionics InGaAs 80um diameter (1996-2000)
Photodiode leakage and responsePhotodiode leakage and response
• RX40 ASIC designed to compensate for damage– dc current offset up to 500uA – automatic gain control (signals of 10-500uA)
[email protected] www.cern.ch/cms-tk-opto RADWG 29-11-2005
Photodiodes - responsePhotodiodes - response
• Photocurrent (InGaAs, 6MeV neutrons, 1998)
1.0
0.5
0.01050
neutron fluence, (1014
n/cm2)
1.0
0.5
0.0
1.0
0.5
0.0Fermionics
Epitaxx (FI)
Epitaxx (BI) /Italtel
Tests A and B
Test B
Test C
• Significant differences in damage• depends mainly if front or back-
illuminated– front-illuminated better
[email protected] www.cern.ch/cms-tk-opto RADWG 29-11-2005
PD SEU PD SEU
• photodiodes in control links sensitive to SEU– Bit-errors in control commands, clock and trigger signals
• strong dependence upon particle type and angle
[email protected] www.cern.ch/cms-tk-opto RADWG 29-11-2005
Photodiode Single-event-upsetPhotodiode Single-event-upset
• Bit-error-rate for 80Mbit/s transmission with 59MeV protons in InGaAs p-i-n (D=80m)• (UC Louvain la Neuve, 2000)• 10-90 angle, 1-100W optical power • flux ~106/cm2/s (similar to that inside CMS Tracker)
10-12
10-10
10-8
10-6
10-4
10-2
100
Bit
-err
or-
rate
-30 -25 -20 -15 -10
Optical power amplitude at PD (dBm)
p+ (beam off)
p+ (59MeV, 90°)
p+ (59MeV, 45°+)
p+ (59MeV, 30°+)
p+ (59MeV, 10°+)
• Ionization dominates for angles close to 90
• nuclear recoil dominates for smaller angles
– Confirmed with neutron tests
• BER inside CMS Tracker similar to rate due to nuclear recoils
• should operate at ~100W opt. power
45° 10°90°beam
[email protected] www.cern.ch/cms-tk-opto RADWG 29-11-2005
Tests of fibre attenuationTests of fibre attenuation
• Gamma damage in various COTS single-mode fibres at 1310nm (1997-8)
[email protected] www.cern.ch/cms-tk-opto RADWG 29-11-2005
Fibre attenuation vs fluenceFibre attenuation vs fluence
• damage actually most likely due to background
• ‘Neutron’ damage (1997)
[email protected] www.cern.ch/cms-tk-opto RADWG 29-11-2005
Fibre annealingFibre annealing
• damage recovers after irradiation (e.g. gamma)
• Damage therefore has dose-rate dependence
[email protected] www.cern.ch/cms-tk-opto RADWG 29-11-2005
ConclusionsConclusions
• We are building large systems of COTS (tele/datacoms) based optical links for CMS
• Challenges include• Radiation damage and reliability• Large number of parts and density of links• Low mass, non magnetic packages• Also long project timescale, procurement, logistics, manpower….
• Our approach – Extensive quality and reliability assurance program
• CMS radiation effects • Reliability (Bellcore standard GR 468)
– Build good relationships with suppliers and partners– Validation of sensitive parts (lasers, photodiodes, fibres) before production
• Prototypes, pre-production qualification, advance validation in production
– Compensation for damage effects built into system
• Production phase essentially complete – good confidence of reliability, large safety margins