November 2001 g.hall@ic.ac.uk www.hep.ph.ic.ac.uk/~hallg/ 1

Electronic requirements for detectors

Use LHC systems to illustrate

physics technical

Tracking high spatial precisionlarge channel countlimited energy precisionlimited dynamic range

low power ~ mW/channelhigh radiation levels ~10Mrad

Calorimetry

high energy resolutionlarge energy rangeexcellent linearityvery stable over time

intermediate radiation levels ~0.5Mradpower constraints

Muons very large areamoderate spatial resolutionaccurate alignment & stability

low radiation levels

November 2001 g.hall@ic.ac.uk www.hep.ph.ic.ac.uk/~hallg/ 2

Generic LHC readout system

Amplifier

Pipeline

memory

Electrical

or

optical link

DAQ Driver

DAQ

Comparator

MUX

A/D

Receiver

A/D

DSP

...

Clock

Trigger

Control

(optional) (optional)

Amplifier

Pipeline

memory

Electrical

or

optical link

DAQ Driver

DAQ

Comparator

MUX

A/D

Receiver

A/D

DSP

...

Clock

Trigger

Control

(optional) (optional)

Amplifier

Electrical

or

optical link

MUX

A/D

Pipeline

memory

DAQ Driver

DAQ

Receiver

A/D

DSP

...

Clock

Trigger

Control

Amplifier

Electrical

or

optical link

MUX

A/D

Pipeline

memory

DAQ Driver

DAQ

Receiver

A/D

DSP

...

Clock

Trigger

Control

•functions required by all systemsamplification and filteringanalogue to digital conversionassociation to beam crossingstorage prior to triggerdeadtime free readout @ ~100kHzstorage pre-DAQcalibrationcontrol monitoring

•CAL & Muons special functionsfirst level trigger primitive generation

•optionallocation of digitisation & memory

November 2001 g.hall@ic.ac.uk www.hep.ph.ic.ac.uk/~hallg/ 3

“Deadtime free” operation

•Pipeline memory buffer depth and trigger rate determine deadtimedata often buffered in pipeline

queueing problem

APV25 NB ≈ 10, NP =192 @100kHz

compare with deadtime from maximum trigger sequence = 1001… = 50ns/10µs = 0.5%

November 2001 g.hall@ic.ac.uk www.hep.ph.ic.ac.uk/~hallg/ 4

Basic radiation effects on electronics

• Bipolaratomic displacement

carrier recombination in basegain degradation, transistor matching,dose rate dependence

• CMOS oxide charge & trap build-upthreshold (gate) voltage shift, increased noise,…change of logic state = SEU

• All technologiesparasitic devices created => Latch-up can be destructive

p+nn++emittercollectorbaseIBICIEp+nn++emittercollectorbaseIBICIE

p typedrainsourcegaten typedepleted regionoxideinverted channelsubstratep typedrainsourcegaten typedepleted regionoxideinverted channelsubstrate

November 2001 g.hall@ic.ac.uk www.hep.ph.ic.ac.uk/~hallg/ 5

Why 0.25µm CMOS?

•by 1997 some (confusing) evidence of radiation toleranceextra thin gate oxide beneficial

tunnelling of electrons neutralises oxide charge

• negative effects attributed to leakage paths around NMOS transistors

cure with enclosed gate geometry

1Mrad VT vs toxide

November 2001 g.hall@ic.ac.uk www.hep.ph.ic.ac.uk/~hallg/ 6

First results from 0.25µm CMOS (1997)

•technology thought to be viable for intermediate radiation levels (~300krad)but results much better than expected

November 2001 g.hall@ic.ac.uk www.hep.ph.ic.ac.uk/~hallg/ 7

Tracking systems

•ATLAS•Innermost: Pixels•Inner: Silicon microstrips 6M channels

Occupancy 1-2%

•Outer: Transition Radiation tracker gas filled 4mm diameter straw tubes 420k channelsx-ray signals from e- above TR thresholdoccupancy ~ 40%

•CMS•Innermost: Pixels•Remainder: Silicon microstrips 10M channels

Occupancy 1-2%

•Radiation hardness is a crucial point for trackers

November 2001 g.hall@ic.ac.uk www.hep.ph.ic.ac.uk/~hallg/ 8

ATLAS TRT readout

•ASDBLR amplifier/shaper/discriminator•key points

speed and stability, since high occupancy peaking time 7-8ns => reduce pileupbaseline restorer => maintain threshold levelstwo level discriminator => electron identification

November 2001 g.hall@ic.ac.uk www.hep.ph.ic.ac.uk/~hallg/ 9

ATLAS TRT ASDBLR front end

•Amplifier =>tail cancellation and baseline restorationselectable for CF4 and Xe gas mixtures

4mm straw + Xenon based gas

November 2001 g.hall@ic.ac.uk www.hep.ph.ic.ac.uk/~hallg/ 10

ATLAS SCT front end

•Amplifier/discriminator + pipeline/sparse readout ABCD (BiCMOS)•Binary readout

simplesmall data volumebutmaintain 6M thresholdsvulnerable to common mode noise

•SpecificationsENC < 1500eEfficiency 99%Bunch crossing tag 1 bunch crossingNoise occupancy 5x10-4

Double pulse resolution 50ns after 3.5fC signalDerandomising buffer 8 deepPower <3.8mW/channel

November 2001 g.hall@ic.ac.uk www.hep.ph.ic.ac.uk/~hallg/ 11

CMS microstrip tracker readout

•10 million detector channels•Analogue readout

synchronous systemno zero suppressionmaximal informationimproved operation, performance and monitoring

•0.25µm CMOS technology intrinsic radiation hardness

•Off-detector digitisation analogue optical data transmission reduce custom radiation-hard electronics

November 2001 g.hall@ic.ac.uk www.hep.ph.ic.ac.uk/~hallg/ 12

Impulse deconvolution at LHC•High speed signal processing is required to match the 40MHz beam crossings

Low power consumption is essential - 2-3mW/channelPerformance must be maintained after irradiation

•Start from CR-RC filter waveformform weighted sum of pulse samples

zero response outside narrow time window small number of weights (>3)implementable in CMOS switched capacitor filter

w(t)= a1.h(t)+a

2.h(t-Δ )+t a3. ( -2h t Δ )t

Ideal CR-RC

Sampled CR-RCwaveform

Deconvolutedwaveform

November 2001 g.hall@ic.ac.uk www.hep.ph.ic.ac.uk/~hallg/ 13

2000

1600

1200

800

400

02520151050-5-10

chan 2 chan 43 chan 107

closed symbols: peak mode: 270 + 38/pFopen symbols:deconvolution: 430 + 61/pF

2000

1600

1200

800

400

02520151050-5-10

chan 2 chan 43 chan 107

closed symbols: peak mode: 270 + 38/pFopen symbols:deconvolution: 430 + 61/pF

Pulse shapes & noise APV25

• 125

100

75

50

25

0

250200150100500

2pF 4p1 8p1 10p7 14p5 17p5 20p5

125

100

75

50

25

0

250200150100500

2pF 4p1 8p1 10p7 14p5 17p5 20p5

1 MIP signal

EN

C [

ele

ctro

ns]

Input capacitance [pF]

t [ns]

•System specification

Noise <2000 electrons

for CMS lifetime

•System specification

Noise <2000 electrons

for CMS lifetime

November 2001 g.hall@ic.ac.uk www.hep.ph.ic.ac.uk/~hallg/ 14

Calorimeter systems

•ATLAS ECAL/Endcap HCALLiquid Argon 190k channels

signal: triangular current ~500ns fall (drift time)

CD ~ 200-2000pF

•ATLAS Barrel HCALScintillating tiles 10k channels

•CMS ECALPbWO4 crystals + APDs (forward: VPT)

80k channels

fast signal ~ 10ns CD = 35-100pF

•CMS Barrel/Endcap HCAL Cu /scintillating tiles with WLS 11k channels HPD readout

Requirements

large dynamic range 50MeV-2TeV = 92dB = 15-16bits

precision ≈ 12bits and high stability

precise calibration~ 0.25%

Radiation environment few 100krad - Mrad +high neutron fluxes (forward)

November 2001 g.hall@ic.ac.uk www.hep.ph.ic.ac.uk/~hallg/ 15

CMS crystal ECAL

•Amplifier close to photo-detector (APD or VPT)4 gain amplifier + FPU gain selection12bit 40MHz digitisationcommercial bipolar ADC - rad hard

•1Gb/s optical transmission 12bit (data) + 2bit (range) custom development using VCSELs 80,000 low power links

•Recent substantial changes in philosophy

November 2001 g.hall@ic.ac.uk www.hep.ph.ic.ac.uk/~hallg/ 16

Optical links in LHC experiments

•Advantages c.f. copper:low mass, no electrical interference, low power, high bandwidth

•LHC requirementsdigital control ~40Ms/s digital data transmission ~1Gb/s analogue: 40Ms/s CMS Tracker

•Fast moving technological area driven by applications

digital telecomms, computer linksanalogue cable TVrequirements c.f. commercial systems

bulk, power, cost, radiation tolerance ?? possible for some applications?

November 2001 g.hall@ic.ac.uk www.hep.ph.ic.ac.uk/~hallg/ 17

Semiconductor lasers

•Now dominate market, over LEDsnarrow beam, high optical power, low electrical power, better matched to fibres

•Direct band gap materialGaAs ~ 850nmGaAlAs ~ 600-900nm In, Ga, As, P ~ 0.55-4µm

•Forward biased p-n diode -> population inversionoptical cavity => laser at I > Ithreshold

often very linear response

•Fibres and connectorssufficient rad hardnesstrackers require miniature connectorscare with handling compared to electrical

8

6

4

2

02520151050

Current (mA)

un-irradiated after 2x10

14n/cm

2

8

6

4

2

02520151050

Current (mA)

un-irradiated after 2x10

14n/cm

2

November 2001 g.hall@ic.ac.uk www.hep.ph.ic.ac.uk/~hallg/ 18

CMS Tracker analogue optical links

•Edge emitting 1.3µm InGaAsP MQW laser diodesminiature devices requiredsingle mode fibre ~50mW/256 detector channels

Si-submount

PIN photodiode

Fibre

2 mm

1.5 mm

Si-submount

PIN photodiode

Fibre

2 mm

1.5 mm

Laser die

2 mm

1.5 mm

Si-submount

Fibre

Laser die

2 mm

1.5 mm

Si-submount

Fibre

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0.0

-0.150454035302520151050

Time [ns]

2.5

2.0

1.5

1.0

0.5

0.0

-0.5

input output

5 ns 5 ns20 ns

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0.0

-0.150454035302520151050

Time [ns]

2.5

2.0

1.5

1.0

0.5

0.0

-0.5

input output

5 ns 5 ns20 ns

TxTx

Rx

same components for digital control BER << 10-12 easily achievable