K.K. Gan ATLAS Pixel Week 1

New Results onVDC-D2 and DORIC-D2

K.K. GanThe Ohio State University

February 13, 2001

K.K. Gan ATLAS Pixel Week 2

Outline

Introduction

Results on VDC-D2

Results on DORIC-D2

Plans

K.K. Gan ATLAS Pixel Week 3

VCSEL Driver Chip (VDC): convert LVDS signal into single-ended signal

appropriate to drive VCSEL Digital Opto-Receiver Integrated Circuit (DORIC):

decode clock and command signals from PIN diode

Introduction

K.K. Gan ATLAS Pixel Week 4

Ohio State University: Kregg Arms, K.K. Gan, Mark Johnson, Harris Kagan,

Richard Kass, Chuck Rush, Michael Zoeller

Siegen University: Michael Kraemer, Joachim Hausmann, Martin Holder,

Michal Ziolkowski

Opto-electronics Team

K.K. Gan ATLAS Pixel Week 5

Bright/Dim Currents of VDC-D2

two dice have dim currents < 1 mA

K.K. Gan ATLAS Pixel Week 6

Duty Cycle vs VCSEL Current

deviate significantly from 50% duty cycle for low VCSEL current

42

44

46

48

50

52

0 4 8 12 16

IVCSEL (mA)

Tunepad open = 1.25 V

Tunepad = 1.72 V

K.K. Gan ATLAS Pixel Week 7

42

44

46

48

50

52

0 0.5 1 1.5 2

Iset (mA)

tunepad open = 1.25 Vtunepad = 1.72 V

Duty Cycle vs I_set Current

deviate significantly from 50% duty cycle for low I_set current can reduce deviation significantly with tune-pad connected

K.K. Gan ATLAS Pixel Week 8

Improvement in VDC-I

ideal design: dim current depends on tune-pad but not I_set VDC-D2: dim current depends on I_set due to

imperfect matching of currents in two transistors! submit two versions of VDC-I:

DMILL version and a version with bright current controlledby I_set added to dim current controlled by tune-pad

K.K. Gan ATLAS Pixel Week 9

VCSEL Current vs Tune-pad Current

20

10

0

I

(

mA

)V

CS

EL

amplitude of VCSEL current remains constantas tune-pad current is increased from 0 to 1 mA

K.K. Gan ATLAS Pixel Week 10

VCSEL Current vs I_set

dim current remains constantas I_set is increased from 0 to 2 mA

I

(

mA

)V

CS

EL

K.K. Gan ATLAS Pixel Week 11

DORIC-D2 with LVDS Reset at VDD (Default)

Input data

Decoded clock

Decoded data

K.K. Gan ATLAS Pixel Week 12

DORIC-D2 with LVDS Reset Grounded

Input data

Decoded clock

Decoded data

K.K. Gan ATLAS Pixel Week 13

BER/Crosstalk Measurement with DORIC and VDC

SCALER

40 MHzAsynchronous

CLOCK

XilinxXC4005SE CMOS

LVDS

CMOS LVDS

VDC

VDC

LVDS CMOS DORIC

VCSEL

VCSEL

PIN

Opto-package

Data Data

fiber o

ptic lin

k

K.K. Gan ATLAS Pixel Week 14

Binary Error Rate

DORIC-D2 has significantly more cross talk than DORIC-4A

placing DORIC-D2 in close proximity to PIN yields similar BER

10-10

10-8

10-6

10-4

10-2

100

10 20 30 40 50 60 70 80

BER

I PIN

(µA)

Packaged DORIC-D2Inner VCSEL on

Packaged DORIC-D2Inner VCSEL off

Packaged DORIC-4AInner VCSEL on

Packaged DORIC-4AInner VCSEL off

(90% CL)

DORIC-D2Opto-board

Truelight VCSEL

K.K. Gan ATLAS Pixel Week 15

observe spikes at edge-detector and pre-amp spy points pulsing test spy circuit induces a signal at LVDS ouputs floating and grounding guard-ring has no effect on pre-amp

back side silicon is highly resistive: ~ 1 MΩ

reduce to ~ 100 KΩ when scratched currently investigating possibility of grinding and metalizing

back side of bare dice

Investigation of Cross-talk

K.K. Gan ATLAS Pixel Week 16

600 mV different at pre-amp +/- spy point outputs pre-amp has low gain for small signal and output distorted pre-amp offset estimates to be a few mV has designed a dc feedback circuit for DORIC-I

simulation shows that it will fix offset problem need to include stray capacitance of traces in simulation

to verify that circuit will not oscillate

Pre-amp Offset

K.K. Gan ATLAS Pixel Week 17

Pre-amp with Feedback

K.K. Gan ATLAS Pixel Week 18

Pre-amp with Feedback

new

K.K. Gan ATLAS Pixel Week 19

Simulation of DORIC-D2 Pre-amp

no output signal for small input signal if there is an offset

K.K. Gan ATLAS Pixel Week 20

Simulation of DORIC-I Pre-ampDORIC-D2, zero offset

DORIC-I, zero offset

DORIC-I, 5 mV offset

DORIC-I, 10 mV offset

produce output signal similar to an pre-amp with no offset

K.K. Gan ATLAS Pixel Week 21

current consumption consistent with expectations 7-channel opto-board:

1.0 W @ 10 mA VCSEL current 1.4 W @ 20 mA VCSEL current

Power Consumption

0

5

10

15

20

25

I (m

A)

DORIC-D2 VDC-D2

VDD = 3.3 V

K.K. Gan ATLAS Pixel Week 22

VDC-D2 and DORIC-D2 work but with some deficiencies VDC-D2: some have low dim currents DORIC-D2:

some pre-amps have small offset cross-talks inside die

will emphasize irradiation of DORIC-D2 in April a new rad-hard bias circuit without

requiring a reset is needed for any DMILL submission

Plans