Transcript Rad-Hard ASICs for Optical Data Transmission in the

Radiation-Hard ASICs for
Optical Data Transmission in the
ATLAS Pixel Detector
Richard Kass
K.E. Arms, K.K. Gan, M. Johnson, H. Kagan, R. Kass, C. Rush,
A. Rahimi, S. Smith, R. Ter-Antonian, M.M. Zoeller
The Ohio State University
A. Ciliox, M. Holder, S. Nderitu, M. Ziolkowski
Universitaet Siegen, Germany
Outline
l
l
l
l
Introduction
Results from IBM 0.25 mm Prototype Chips
Results from Proton Irradiations
Summary
1
LECC03, 9/30/2003
Richard Kass/OSU
ATLAS Pixel Detector
2 disks/end
2 layers in barrel
l
l
l
l
Inner most charged particle tracking detector
Pixel size: 50 mm x 400 mm
~ 100 million channels
Dosage after 10 years:
middle barrel layer:
optical link:
50 Mrad or 1015 1-MeV neq/cm2
30 Mrad
2
LECC03, 9/30/2003
Richard Kass/OSU
ATLAS Pixel Opto-link
VCSEL: Vertical Cavity Surface Emitting Laser diode
VDC:
VCSEL Driver Circuit
PIN:
PiN diode
DORIC: Digital Optical Receiver Integrated Circuit
Opto-board: this board holds the VDCs, DORICs, PINs, VCSELS
use BeO as substrate for heat management
3
LECC03, 9/30/2003
Richard Kass/OSU
VCSEL Driver Circuit Specs
lConvert LVDS input signal into single-ended signal appropriate to
drive the VCSEL diode
lOutput (bright) current: 0 to 20 mA,
controlled by external voltage
lStanding (dim) current: ~ 1 mA
to improve switching speed
lRise & fall times: 1 ns nominal
(80 MHz signals)
lDuty cycle: (50 +/- 4)%
l“On” voltage of VCSEL: up to 2.3 V at 20 mA for 2.5 V supply
l Constant current consumption!
lCurrent design uses TRUELIGHT “high power oxide” VCSELs
4
LECC03, 9/30/2003
Richard Kass/OSU
Digital Optical Receiver IC Specs
Decode Bi-Phase Mark encoded (BPM) clock and command
signals from PIN diode
l
l
Input signal: 40-1100 mA
40 MHz
clock
l
Extract: 40 MHz clock
command
l
Duty cycle: (50 +/- 4)%
l
Total timing error: < 1 ns
l
Common Cathode PIN array
l
Bit Error Rate (BER):
< 10-11 at end of life
l
BPM
Training period: ~1 ms of 20 MHz clock (BPM with no data)
Input transitions ] leading edges
Internal delays
LECC03, 9/30/2003
] trailing edges
Richard Kass/OSU
5
Status of VDC and DORIC
VDC and DORIC produced using IBM 0.25mm CMOS process
Migrated ATLAS SCT’s VDC & DORIC design in AMS 0.8mm Bipolar to DMILL
DMILL (0.8mm CMOS) process met e-specs but was not rad hard enough
VDC-I5e and DORIC-I5e: our fifth IBM submission
Use CADENCE program
5 layers, enclosed layout transistors and guard rings
4 channels per chip (VDC: 1.4mm4mm DORIC: 2.3mm4mm)
engineering run, produced enough chips for production
Received chips in June 2003
Electrical measurements performed on these chips:
detailed measurements of many parameters:
rise/fall time, duty cycle, bright/dim current…
VDC-I5e and DORIC-I5e are acceptable for use in pixel detector
Need to certify that these chips are radiation hard
previous VDC and DORIC version (I4) rad hard to > 50Mrad
detailed plan to certify radiation hardness
LECC03, 9/30/2003
Richard Kass/OSU
6
DORIC and VDC Layout
DORIC-I5 (2.3 x 4mm)
VDC-I5 (1.4 x 4mm)
Ch 4
Ch 3
Ch 2
Ch 1
Ch 1
Ch 2
Signal_in
GndA
Noise_in
VddAmp
VddA
Ch 3
Input Pads:
Ch 4
Preamp / Digital Test Pads
7
LECC03, 9/30/2003
Analog / Digital Bias, Resets
Richard Kass/OSU
VDC-I5e Measurements
VDCI5e I-dim and I-bright vs Iset
15
bright
Current (mA)
10
Eight VDCI5e chips with four
channels per chip were measured
5
dim
0
0
0.5
1 Duty Cycle
VDCI5
1.5
2
2.5
Current (mA)
54
VDCI5e duty cycle
+ Duty Cycle (%)
53
VDC-1
VDC-2
VDC-3
VDC-4
52
VDC-5
VDC-6
VDC-7
Spec is 50±4%
VDC-8
51
50
0
1
LECC03, 9/30/2003
2
3
Channel #
4
5
8
Richard Kass/OSU
VDC-I5e Rise and Fall Times
VDCI5e Rise Times at Iset = 0.6 mA
2.0
rise time
VDC-1, I_set = .6 mA
VDC-2, I_set = .6 mA
VDC-3, I_set = .6 mA
VDC-4, I_set = .6 mA
VDC-5, I_set = .6 mA
VDC-6, I_set = .6 mA
VDC-7, I_set = .6 mA
VDC-8, I_set = .6 mA
Time (ns)
1.5
1.0
0.5
0.0
0
1
2
3
4
5
spec is <1 ns
VDCI5e
Fall Times
Channel
# at Iset = 0.6 mA
0.8
fall time
VDC-1, I_set = 0.6 mA
VDC-2, I_set = 0.6 mA
VDC-3, I_set = 0.6 mA
VDC-4, I_set = 0.6 mA
VDC-5, I_set = 0.6 mA
VDC-6, I_set = 0.6 mA
VDC-7, I_set = 0.6 mA
VDC-8, I_set = 0.6 mA
Time (ns)
0.7
0.6
0.5
0
1
2
3
4
5
9
Channel #
LECC03, 9/30/2003
Richard Kass/OSU
Status of BeO Opto-Board
converts
electricaloptical
6-7 links/board
First BeO opto-boards
arrive in April.
Find problems:
many vias not filled.
Send boards back to
company for repair.
Company only partially
successful with repairs
(~50% success rate)
We find a new company
to produce the boards
two populated
opto-boards used for
rad-hardness tests
(more $$$ too)
LECC03, 9/30/2003
10
Richard Kass/OSU
Radiation Hardness Measurements
CERN August 2003
Important to measure/certify radiation hardness of optical components
VDC
DORIC
VCSEL
Fibers
Glues, etc…
Use CERN’s T7 beam (24 GeV protons) for radiation hardness studies.
We perform two types of radiation hardness tests:
Cold Box: read out VDC and DORIC electrically (copper wires)
Shuttle: readout VDC and DORIC via optical fibers
VDC and DORICs mounted on BeO opto-board
shuttle can be moved in and out of beam remotely
For both tests we monitor the VDCs and DORICs in real time.
11
LECC03, 9/30/2003
Richard Kass/OSU
CERN Irradiation Tests
System for Irradiation in Cold Box
PC
BER Tester
20 MHz
0.2 m Ribbon
Coaxial
DORIC Distribution
25m Ribbon
+
25 m Coaxial
VDC Distribution
Beam Area Distribution
0.5m Ribbon
Test Boards
VDC or
DORIC
Testing system at OSU before shipping to CERN
12
LECC03, 9/30/2003
Richard Kass/OSU
Irradiation Tests
Opto-board
System for Irradiation in Shuttle
Bit error test board in
T7 control room
Bi-phase marked
optical signal
25 m fibers/wires
Decoded data
Opto-pack
clock
PIN
DORIC
data
VCSEL
VDC
VCSEL
VDC
An irradiation test
Opto-link
measure power of clock and data Vs dose
LECC03, 9/30/2003
Richard Kass/OSU
13
Setup for CERN’s T7 Shuttle
shuttle test electronics
at OSU prior to shipping to CERN
25m of optical fibers
opto-boards
Optical fibers
Remotely moves
in/out of beam
14
LECC03, 9/30/2003
Richard Kass/OSU
Results from Irradiation
Threshold for zero bit errors vs Dose
Opto-board #1
anneal
anneal
anneal
anneal
anneal
all other links: pseudo-random bit string and Ipin=100 uA
Thresholds independent of dose
LECC03, 9/30/2003
Richard Kass/OSU
15
Results from Irradiation
Clock Optical Power Vs Dose
Irradiate for ~4-6 MRad (~4-5 hours)
Move boards out of the beam and anneal for 12-15 hours
LVCSELs need time to anneal
Monitor the optical power of each link in real time.
Opto-board #2(BeO#4)
Opto-board #1(BeO#3)
anneal
anneal
anneal
anneal
anneal
anneal
anneal
anneal
anneal
anneal
Beam Iset = 0.6 mA (=10mA) for beam, 1.5mA (=1314mA) for annealing
16
LECC03, 9/30/2003
Richard Kass/OSU
Summary and Conclusions
VDC and DORIC chips satisfy ATLAS Specs
BeO Opto-board:
Original vendor not able to produce quality boards
Have found new vendor updated layout submitted
Radiation studies:
Cold Box: VDC and DORIC are more than hard enough!
Shuttle: VCSEL radiation hardness an issue
will VCSELs anneal enough to meet power specs ?
Move to Production in 2004
~500 VDCs, ~400 DORICs, ~200 opto-boards (includes spares)
17
LECC03, 9/30/2003
Richard Kass/OSU