ATLAS Pixel Optolink FDR

1
ATLAS Pixel Optolink FDR

• System Tests
• 12. February 2003
• Peter Gerlach

2
VCSEL irradiations

• VCSELs degrade due to normal operation and
  radiation induced damages
• good annealing when current gt10mA is applied
• accelerated lifetime studies predict small losses
  if temperature is 0 /- 5 Co

• MITEL
• 41015 neq/cm2 mit 1MeV n --gt ok
• 81014 neq/cm2 mit 1MeV n --gt ok
• 11015 neq/cm2 mit 30MeV p --gt ok
• 31015 neq/cm2 mit 1 MeV n --gt ok
• Truelight
• 60MRad gammas --gt ok
• 21014 neq/cm2 mit 30MeV p --gt ok
• 11015 neq/cm2 mit 30MeV p --gt ok
• 31015 neq/cm2 mit 24GeV p --gt ok
• 5.71015 neq/cm2 mit 24GeV p --gt

3
Truelight VCSELs with 24GeV protons

• increased competition between stimulated
  radiative recombinations and non-radiative
  recombinations introduce by protons
• results in lower light yield and higher threshold
  currents

gt optical link rad. level
protons converted into neq using NIEL of GaAs
4
Truelight VCSELs with 30MeV protons

• comparison 24GeV and 30MeV (1.5E15 neq/cm2)
• threshold current rises to about 7mA
• light output drops to 50
• Radiation damage of VCSELs scales with NIEL !!

30MeV protons converted into neq using NIEL of
GaAs
5
Single Event Upsets

• Flux at PP0 2106 1/cm2sec
• 85 pions
• PiN is a detector, not only for light
• large area to pick up all light coming from
  fibre
• Bit Error Rate will be dominated by this!
• SEU in Opto-ASICS negligible
• Only a concern for TTC link
• Setup data may be read back and repeated
• Command (LVL1-trigger) may be corrupted

6
Single Event Upsetin optical link

• Impact of single particle
• current pulse sensed in the receiver circuit
• decay with RC time constant of DORIC input stage

• a) 1 is not corrupt
• b) 0 may be corrupted
• c) 0 is corrupt

7
DORICI4 in T7 irrad.
worse case working point
µA
8
SEU in PiN
9
Redundant Coding

• Even a low BER might disturb fast commands
• Coding of commands redundantly

Special care has been taken for the command that
switches to the configuration mode.
10
Wuppertal Optoboard Driver And iNterface
11
WODAN main board
BERT
12
BER measurement
BER lt 10-11 _at_ Ch2 13.5 µA Ch3 16.1 µA Ch4 17.5
µA Ch6 16.9 µA Ch7 15.6 µA Ch8 13.8 µA
13
Module driven by OLink
PLL
PLL opto interface (Siegen)
Opto Board (OSU)
clk
BPM
DORIC (D3)
daisy chain
data in
data out
DRX
VDC (D2)
Clock return (XCKR) generated on this card
Microcable (LVDS)
PCC
pigtail
LVDS driver card
module
Sync, MCC TM, power...
Flex support card
power
14
Setup
FE-B module with pigtail
microcables
optical fibres
PLL opto interface (Síegen)
opto board (OSU)
15
Results Module without optical readout

• chips 11, 12 never worked on this module
• untuned module

threshold (3800? 350)e-
noise (320? 60)e-
16
Results Module with full optical readout

• at 100µA PIN current
• chip 9 stops working due to ripped off wire-bond

threshold (4460? 330)e-
noise (270? 50)e-
17
Results Module with full optical readout

• at 40µA PIN current

threshold (4470? 380)e-
noise (340? 50)e-
18
Results chip 4 optical readout

• no significant change visible
• all changes are usual fluctuation seen in FE-B
  modules

measured with standard test setup
19
Opto-Link Production Plan
Version Nov. 2002 K.K. Gan / P. Gerlach / M.
Ziolkowski QA quality assurance
OSU
OSU Siegen
Ac.Sinica
OPTO BOARD blank
DORIC/VDC on wafer
OPTO PACKAGE
passive components mounting
QA
QA
Ac.Sinica
QA
OSU Siegen
dicing
TX/RX
Wuppertal (Ericsson)
OPTO BOARD
QA
PP1-USA15 fibers
PP0-PP1 fibers
Cambridge (Wuppertal)
QA
BOC
Wuppertal
QA
Wuppertal (Ericsson)
QA
QA
ATLAS detector
mounted on PP0
20
Opto-Link Production Plan
version 02.12.2002
Fibre cables at CERN Oct.04 Installation
beginning 05
21

• from now on
• backup transparencies

22
SEU in optical link

• Prestudies with 14 MeV n (NPL) indicated SEU in
  optical link
• SEU studies at Paul Scherrer Institut

• Dipole magnets to select momenta in the range
  300 to 465 GeV/c
• quadrupoles focussing particle beam
• adjust slits to control flux

23
Flux monitoring

• multi-wire proportional chambers for beam profile
• scintillator counters for flux measurements
  (S1,S2)
• S2 on computer controlled xy stage to measure
  beam profile
• Al foil for cross check

24
Testing Procedure

• VDC, DORIC and opto-package (VCSEL and PiN)
• PiN photodiode is most sensitive element to SEU
• beam focused perpendicular onto surface of PiN

SEU rate as a function of the mean photocurrent
ltIPiNgt
25
BER scans with and without beam
???

• No measurable BER if ltIPINgt larger than 30 mA
  with no beam
• BER above this current is introduced by SEU
• BER decrease with increasing ltIPINgt
• SEU appear in analog part of optical link

• sensitive volume of PiN gtgtsensitive volume of
  transistors ? dominant SEU are due to energy
  deposited in PiN which is larger than the
  threshold

26
SEU cross sections

• define SEU cross section
• convert IPiN to eq. energy deposition in
  photodiode

• 300 MeV p --gt D
• use worst case results (300MeV p) for
  estimation of SEU

27
SEU Analysis

• Expected value ltIPiNgt75mA
• sSEU 1.210-8 cm2 (Fit)

• --gt BER of 3.610-10 (SCT)
• better than specs. (110-9)
• and what about Pixels?
• SEU is depending on signal size in the
  photodiode --gt independent of DORIC Version

28
SEU in Pixel optical link

• recalculate Emin for different input stage of
  DORIC-D2

• use PSI data to calculate BER

• possible to establish optical link with
  acceptable SEU