k.k. ganus atlas pixel meeting1 opto-ic/board k.k. gan the ohio state university wbs 1.1.1.3 & 4...
TRANSCRIPT
K.K. Gan US ATLAS Pixel Meeting 1
Opto-IC/Board
K.K. Gan
The Ohio State University
WBS 1.1.1.3 & 4
July 9, 2003
K.K. Gan US ATLAS Pixel Meeting 2
Outline
Introduction
VDC/DORIC
Opto-board
Cost and schedule summary
Conclusions
K.K. Gan US ATLAS Pixel Meeting 3
ATLAS Pixel Opto-link
K.K. Gan US ATLAS Pixel Meeting 4
OSU Responsibilities
VDC and DORIC:
design testing of 50% of final production
Opto-board:
design and fabrication testing of 50% of final production?
K.K. Gan US ATLAS Pixel Meeting 5
OSU ATLAS Personnel
FacultyK.K. Gan, Harris Kagan, Richard Kass
Research AssociatesAmir Rahimi, Mike Zoeller
Graduate StudentsKregg Arms, Rouben Ter-Antonian
EngineersMark Johnson, Chuck Rush
TechniciansJim Burns, Jason Moore, Shane Smith, Bob Wells
new
new
K.K. Gan US ATLAS Pixel Meeting 6
VDC: convert LVDS signal into single-end signal to drive VCSEL
DORIC: extract clock and command signal from bi-phase mark encoded signal received at PIN
five IBM prototype runs with various improvements andchanges in response to opto-pack evolutionfour MPW runs plus one with FE/MCC-I1
VDC/DORIC-I5e were submitted with MCC-I2 as engineering runconvert from 3-metal MPW layout to 5-metal layoutminor improvements in DORIC-I5e
this will be production run if performance is satisfactory
Status of VDC/DORIC
K.K. Gan US ATLAS Pixel Meeting 7
bright current is similar to I5 dim current is similar to I5 and close to target 1 mA rise/fall times, duty cycle, current consumption within specs
VDC-I5e: Bright and Dim Currents vs. Iset
VDC-I5e
VDC-I5
high power oxide VCSEL
0
4
8
12
16
0.0 0.4 0.8 1.2 1.6 2.0
Iset (mA)
IVCSEL
(mA)
K.K. Gan US ATLAS Pixel Meeting 8
started with 1-channel VCSEL
adapted 8-channel p+ VCSEL to use with MT connector
converted to oxide VCSELbecause p+ has been discontinued by Truelight
converted to high-power oxide VCSELbecause normal oxide has been discontinued by Truelight
VCSEL Opto-pack Evolution
K.K. Gan US ATLAS Pixel Meeting 9
p+/oxide VCSELs have similar optical power in operating range (10-20 mA)
oxide VCSELs have larger effective serial resistance
smaller VCSEL current when driven by VDC
I-V Curves/Optical Power ofP+, Normal & High Power Oxide VCSELs
high power oxide (3 arrays)normal oxidehigh power oxide (2 arrays)p+
0.0
1.0
2.0
3.0
0 5 10 15 20
Current (mA)
Voltage (V)
0
1000
2000
3000
Optical Power (
μ )W
K.K. Gan US ATLAS Pixel Meeting 10
some VCSEL arrays contain a channel with much larger effective serial resistance
maximum VCSEL current of 10 mA when driven by VDCsignificantly below target of 20 mA annealing current
I-V Curve of High Power Oxide VCSELs
normal oxide
high power oxide
0.0
0.5
1.0
1.5
2.0
2.5
3.0
0 2 4 6 8 10 12 14 16 18 20
Current (mA)
Voltage (V)
0
1000
2000
3000
Optical Power (
μ )W
p+ with 10
high power oxide
0
5
10
15
20
0.0 0.4 0.8 1.2 1.6 2.0
Iset (mA)
IVCSEL
(mA)VDC6 Channel 1, NormalVDC6 Channel 2, NormalVDC6 Channel 3, NormalVDC6 Channel 4, NormalVDC6 Channel 1, High PowerVDC6 Channel 2, High PowerVDC6 Channel 3, High PowerVDC6 Channel 4, High PowerVDC1 Channel 1, P+, with 10 ohmVDC1 Channel 2, P+, with 10 ohmVDC1 Channel 3, P+, with 10 ohmVDC1 Channel 4, P+, with 10 ohm
K.K. Gan US ATLAS Pixel Meeting 11
some links have twice as worse PIN current thresholds for no bit errors
measured with all clocks running asynchronically
High Power Oxide VCSEL with Negative Common Cathode
0
20
40
60
1 2 3 4 5 6 7
Link
IPIN
(μ)A
20 mAVc @ -0.2 V
10 mAVc @ gnd
other links @ 80 μA
other links off
K.K. Gan US ATLAS Pixel Meeting 12
0
10
20
30
40
50
60
1 2 3 4
DORIC-I5e #1DORIC-I5e #2DORIC-I5 #6DORIC-I5 #3
Threshold (uA)
Link
PIN current thresholds for no bit errors measured with 44-pin packages
thresholds are low and similar to DORIC-I5
expect lower thresholds on opto-board as DORIC-I5
upper thresholds > 1200 μA successfully converted reset from active high to low
DORIC-I5e: PIN Current Thresholds for No Bit Errors
K.K. Gan US ATLAS Pixel Meeting 13
control circuit boards: designed/built/tested
LabView programs: written/tested
VDC/DORIC QA plan: currently use packaged chips to exercise QA system have received probe cards perform QA with probe station
opto-board QA plan: perform QA in temperature controlled chamber
QA system almost ready for production
VDC/DORIC/Opto-board Quality Assurance
K.K. Gan US ATLAS Pixel Meeting 14
converts: optical signal electrical signal
contains 6-7 optical links:
layers 1 and 2 optical link:
2 DORICs, 2 VDCs, PIN/VCSEL (opto-pack)
B layer optical link:
2 DORICs, 4 VDCs, PIN/2 VCSEL (opto-pack)
use BeO for heat management but prototype initiallyin FR-4 for fast turnaround and cost saving have four FR-4 prototype runs to accommodate
various design changes in chips/opto-packs
Opto-Board
K.K. Gan US ATLAS Pixel Meeting 15
First batch: 30 boards delivered in April several open vias on each board due to insufficient gold filling
repair with wire-wrap wires
Status of BeO Opto-board
VDC-I5 opto-pack
K.K. Gan US ATLAS Pixel Meeting 16
PIN Current Thresholds for No Bit Errors
low PIN current thresholds for no bit errorsindependent of activity in adjacent channels no design errors
K.K. Gan US ATLAS Pixel Meeting 17
Second batch: 31 boards delivered in June vias overfilled and excess metal ground away
no open vias 17 boards have shorts between power and ground lines attempt to populate a second BeO board:
opto-pack housing fits wellone DORIC mounted backwardno visible damage to adjacent chip after replacementboard failed to workreplaced both DORICsopto-board now works!
Status of BeO Opto-board
K.K. Gan US ATLAS Pixel Meeting 18
BeO Opto-board from 2nd Batch
VDC housingDORIC opto-pack latch broken
34.7 oC 36.4 oC
Ambient:22.5 oC
K.K. Gan US ATLAS Pixel Meeting 19
both opto-boards have low PIN current thresholds for no bit errorsindependent of activity in adjacent channels will produce opto-boards with second vendor
PIN Current Thresholds for No Bit Errors
0
10
20
30
40
0 1 2 3 4 5 6 7 8
Link
IPIN (
μ)A
, #1Other Links with no input BeO batch
40 , #1All Other PINs Have uA BeO batch
, #2Other Links with no input BeO batch
80 , #2All Other PINs Have uA BeO batch
2nd batch
1st batch
OSU
K.K. Gan US ATLAS Pixel Meeting 20
Opto-Link Proton Irradiations four 24-GeV proton irradiations at CERN in last three years
OSU designed and fabricated all test systems: cold box: electrical test of DORIC/VDC shuttle: test of opto-link on opto-boards
BER testerfibersopto-board
K.K. Gan US ATLAS Pixel Meeting 21
duty cycle increases by ~ 2% after 58 Mrad
VDC-I4: Clock Duty Cycle vs. Dosage
44
46
48
50
52
54
56
0 10 20 30 40 50 60
Dosage (Mrad)
Duty Cycle (%)
link1
link2
link3
link4
link5
link6
link7
link8
K.K. Gan US ATLAS Pixel Meeting 22
VCSELs annealed with 20 mA during indicated periodsPIN current thresholds for no bit errors remain constant up to 21 Mrad
Opto-Board Bit Error Threshold vs. Dosage
0
5
10
15
20
25
30
IPIN
(μ)A
Start
3.1Mrad9hours12hours6.2Mrad15hours8.3Mrad12hours12.2Mrad
14hours13.8Mrad
9hours13hours16.9Mrad
12hours21.0Mrad
3Link
4Link
5Link
6Link
Dosage/VCSEL Annealing Time
K.K. Gan US ATLAS Pixel Meeting 23
opto-link with DORIC/VDC-I3 and p+ VCSEL opto-pack meets radiation specs
August 2003 irradiation: double irradiation statistics:
cold box: four 4-channel DORIC/VDC shuttle: four opto-boards
Proton Irradiation Summary and Plan
K.K. Gan US ATLAS Pixel Meeting 24
cost adjustments:
BeO opto-board prototype run with new vendor: $15K
increase cost for production BeO boards: $30K
increase cost for SMD mounting of production BeO boards: $3K
technician for opto-board production: $57K
no funds allocated in FY04 irradiation:need irradiation of random sample of production opto-boards?
WBS 1.1.1.4 FY04ETC03: $78K?ETC04: $78K+$105K
Changes in ETC04
K.K. Gan US ATLAS Pixel Meeting 25
Opto-Board Milestones
WBS Task_Name Baseline Date ETC02 Date ETC03 Date1.1.1.4.1 Optical FDR 31-Jan-02 10-Oct-02 28-Feb-03 Optical PRR 5-Mar-03 12-Jun-03 12-Jun-03 Release initial MC for optical components 19-Mar-03 26-Jun-031.1.1.4.3 First optical boards 20-Aug-03 20-Nov-03 Optical production complete 24-Dec-03 24-May-04
???
??? ???
K.K. Gan US ATLAS Pixel Meeting 26
preliminary results indicate that DORIC/VDC-I5ecan be used for production opto-boards
BeO prototype opto-board can operate with low PIN current thresholdsno technical problem is foreseen but first vendor is unsatisfactory
Summary