makoto taguchi

Development of Multi-Pixel

Photon Counters and

readout electronicsMakoto Taguchi High Energy Group

Contents

T2K experimentMulti-Pixel Photon Counters (MPPC)Basic performanceLaser testReadout electronics of MPPC Conclusion

T2K experiment

• search for appearance

•precise measurement of disappearance

T2K experiment

J-PARCSK

μνeν

ν

main goals

Photosensor for T2K

• # of channels ~ 60,000 and space is limited

compact & low cost • under 0.2T environment tolerance to magnetic

field• efficiency for the

detection of particles large light yield

target0.2T magnetMagnet use scintillator+wave lengt

h shifting fiber for the near detectors

MPPC is chosen as the photosensor for T2K that satisfies these requirements

requirements to photosensor

T2K near detectors

Multi-Pixel Photon Counters(MPPC)

Multi-Pixel Photon Counter(MPPC)New(~2years ago) photosensor produced by

Hamamatsu Photonics

~5mm

Geiger-APD pixels 100u

m

100 or 400 avalanche photodiode(APD) pixels in 1mm2

Each pixel works in Geiger mode above breakdown voltage

output from each pixel is independent of # of created p.e. within the pixel

The output from MPPC is a sum of output charge from all APD pixels

output from MPPC is proportional to the # of fired pixels

photon

high (~106) gain Geiger mode

insensitive to magnetic field semiconductor

compact, low cost (~2000Yen?)

excellent photon counting

attractive feature

Measurement of basic performance of MPPC

Test samples ・・・ latest (Oct. 2006) 100 and 400 pixel samples Test items ・・・ raw signal gain noise rate Photon detection efficiency crosstalk-rate Linearity recovery time

presented here

Basic performance of MPPC satisfies the T2K requirement?

MPPC is a new photodetector

Motivation

Raw signalblue LED

1p.e.2p.e.3p.e.

output charge oscilloscope or

ADC

1p.e.

2p.e. 3p.

e.

gate

photon

MPPC

Excellent photon counting capability!

pedestal

pedestal

Gain

pedestal

1p.e.

Gain = Q/e

Q

bias voltage

15deg.20deg.25deg.

•Gain = 1.0x106 ~ 3.0x106 and increases with lower temperature

•linear dependence on bias voltage, G = C (V-Vbd)

ADC distribution

C : capacitance

V: bias voltage

Vbd : breakdown voltage

Noise rate• MPPC emits thermal noise without external light• count the rate above thresholds of 0.5p.e. and

1.5p.e.

bias voltage

(kHz)

15deg.20deg.25deg.

1.5p.e. th.

0.5p.e. th.

•noise rate at 0.5p.e. th. <500kHz and becomes higher with higher temperature

•noise rate at 1.5p.e. th. <100kHz and becomes higher with lower temperature

cross-talk effect

Cross-talk• “photons generated during an avalanche trigger

another avalanche in neighboring pixel”crosstalk rate =

festimated : estimated fraction of 1p.e. events from that of pedestal events assuming Poisson

fobserved : observed fraction of 1p.e. events

1-festimated

fobserved

bias voltage

15deg.20deg.25deg.

ADC dist.

Crosstalk rate 0.2 ~ 0.4 and increases with lower temperature

Photon Detection Efficiency(PDE)• PDE = Geigerεgeomε x QE x

geometrical efficiency ~70%

quantum efficiency of APD ~70%

Geiger probability (V,T) ~90%

MPPC

PMT

1mmφslit

WLS fiber

blue LED

PDE(MPPC)/QE(PMT)

setup

15deg.20deg.25deg.p.e.(MPPC)

p.e.(PMT)V

PDE of MPPC is 2~3 higher than that of PMT and increases with lower temperature

Summary of basic performance

100pixel 400pixel Requirement for T2K

Gain 1.0~3.0x106 ~1.0x106 ~106

Noise rate at 0.5p.e. th.

100~500kHz

100~500kHz

<1000kHz

Noise rate at 1.5p.e th.

10~100kHz 10~100kHz <50kHz

Cross-talk rate

0.2~0.4 0.2~0.4

PDE 20~45% 20~30% >15%

Linearity ~40p.e. ~120p.e. ~100p.e.

@20deg.

Performance of MPPC satisfies the requirements for T2K!

Laser test for the old samples,• gain in the edge of pixel is higher than that in the center of pixel • breakdown voltage is different in each pixelcheck the response within one pixel /of each pixel for the new samples

• test items ・・・ gain, efficiency, cross-talk

green laser

movable stage

MPPC

•uniformity within one pixel

•pixel-to-pixel uniformity

setupefficiency =

# of total events

# of events > 0.5p.e.

Motivation

presented here

10um

gaingain

efficiency

uniformity within one pixel

pixel-to-pixel uniformity

efficiency

RMS/mean=2.0%

RMS/mean=2.0%

RMS/mean=3.3%

RMS/mean=2.5%

Response within one pixel/of each pixel is well uniform!

Readout electronics of MPPC

Readout electronics of MPPC with “Trip-t” chip

use ~60,000 MPPCs in T2K and compact&multi-channel electronics is necessary

establishment of test system for mass production of MPPC is also needed

we have developed the readout electronics with Trip-t• ASIC produced at Fermilab• 32 channel inputs • for negative charge1) serialized analog output corresponding

to the amplitude of input charge2) serialized analog output corresponding

to the timing of input charge3) discriminated output for each channel

Motivation

Trip-t

14mm# of readout channels 321

14mm

Trip-t

front end

Pipeline

Pipeline

Digital multiplexer

analog multiplexer

analog multiplexer

A_OUT (charge)

T_OUT (timing)

D_OUT(digital)

•amplifier (gain is adjustable)

•generate digital signals

store signals before readout (depth 1~48)

serialize 32ch signals

input charge

charge

timing

digital

ch1

ch2

ch3

ch1ch2

ch3

Readout of MPPC with Trip-t

1p.e.2p.e.

output from Trip-t

4MPPC

3p.e.

LED

test board(4ch)

AD conversion by flash ADC

100pixel 400pixel 400pixel 400pixel

readout 4 MPPCs simultaneously

succeed in developing the multi-channel readout electronics of MPPC!

Dynamic range of Trip-t

charge(-pC)

ADC count

saturation

> >Trip-t gain

•Dynamic range ~40p.e. with the lowest gain of Trip-t, assuming MPPC gain of 7.5x105

OK for test of large number of MPPCs

not OK for T2K (requires ~100p.e.)

MPPC

100pF

10pF

high gain channel

low gain channel

high/low gain method for real type elec.

•high gain channel ・・determine gain w/ photopeaks

•low gain channel ・・accommodate large signal

Trip-t can be used for the readout electronics of MPPC in T2K

Conclusion• MPPC is a new photodetector produced by

Hamamatsu Photonics and chosen as the photosensor for T2K

• Basic performance of MPPC satisfies the requirements for T2K

• Response within one pixel/of each pixel is well uniform

• Trip-t which was produced at Fermilab can be used for the readout electronics of MPPC in T2K

• future development ・・ test of large number of MPPCs with 32ch Trip-t board

• Our study is an important step not only for T2K but also for wide use of MPPC

backup

Principle of APD•high reverse bias voltage applied to a pn junction

multiplication region, where created e- -e+ pairs cause an avalanche multiplication

•Normal mode - operate below the breakdown voltage(Vbd) - gain < ~ 100 - have linear output to # of injected photons

•Geiger mode - operate above the breakdown voltage(Vbd)

- gain ~106 - does not have linear output to # of injected photons

E

reverse bias

Gain

100pixel

400pixel

bias bias

15deg.20deg.25deg.

gain(2)

100pixel

400pixel

15deg.20deg.25deg.

ΔV ΔV

ΔV =V-Vbd

Gain is a function of only ΔV

Device-by-device gain variation

bias V

Device-by-device gain variation comes from the device-by-device variation of Vbd

ΔV =V-Vbd

•400pixel

#2#1

#3#2#1

#3

Noise rate

100pixel

bias bias

(kHz) (kHz)

15deg.20deg.25deg.

1.5p.e. th.

0.5p.e. th.

0.5p.e. th.

1.5p.e. th.

400pixel

Device-by-device variation of noise rate at 0.5p.e. th.

bias V

Device-by-device variation of noise rate comes from the device-by-device variation of Vbd

(kHz) (kHz)

•400pixel

#2#1

#3

#2#1

#3

ΔV =V-Vbd

Cross-talk rate

bias bias

100pixel

400pixel

15deg.20deg.25deg.

Cross-talk rate(2)

100pixel

400pixel

ΔV ΔV

15deg.20deg.25deg.

cross-talk rate is a function of only ΔV

ΔV =V-Vbd

Device-by-device variation of cross-talk rate

bias V ΔV = V-Vbd

Device-by-device variation of cross-talk rate comes from the device-by-device variation of Vbd

•400pixel

#2#1

#3#2#1

#3

PDE(MPPC)/QE(PMT)

bias bias

100pixel

400pixel

15deg.20deg.25deg.

PDE(MPPC)/QE(PMT)(2)

ΔV ΔV

100pixel

400pixel

15deg.20deg.25deg.

PDE is a function of only ΔV

ΔV =V-Vbd

Device-by-device variation of PDE

bias V ΔV = V-Vbd

Device-by-device variation of PDE comes from the device-by-device variation of Vbd

•400pixel

#2#1

#3#2#1

#3

Vbd vs TVbd

Vbd

degree degree

Vbd is proportional to the temperature

100pixel 400pixel

Comparison of latest and old samples (gain)

latest

old

latest

old

100pixel 400pixel

ΔV ΔV

Comparison of latest and old samples (noise rate)

latest

old

100pixel

latest

old

400pixel(kHz) (kHz)

ΔV ΔVnoise rate of latest sample is lower

Comparison of latest and old samples (cross-talk rate)

latestold

100pixel

ΔV ΔV

latest

old

400pixel

cross-talk rate of latest sample is higher increase of geometrical efficiency

Comparison of latest and old samples (PDE)

latest

old

100pixel

ΔV ΔV

latest

old

400pixel

PDE of latest sample is higher increase of geometrical efficiency

Linearity

paper

setup •# of injected p.e. to MPPC is estimated by the p.e. detected by a monitor PMT

•expected response:MPPC

PMT

))N

c)x(1exp((1NN

00fired

Nfired : # of fired pixels

N0 : # of pixels

c : Cross-talk rate

x : # of injected p.e.

LED

100pixel

400pixel

+10%

+20%

-10%

-10%

injected p.e.

injected p.e.

injected p.e.

injected p.e.

# of fired pixel

# of fired pixel

(Data-exp.)/Data(%)

(Data-exp.)/Data(%)

Data

Data

expectation

expectation

Linearity(3)

100pixel 400pixel

# of injected p.e.

# of injected p.e.

(Data-Fit)/Data(%) (Data-Fit)/Data(%)

-20% -20%

Recovery time

• “time to quench an avalanche and then reset the applied voltage to its initial value”

• fire all pixels by the light from LED1 and check the response to the light from another LED(LED2) with changing the time difference between the LED1 and LED2

Recovery time(2)

•100 pixel

•400 pixel

All pixels are recovered 100ns after all pixels are fired

Recovery time < 100ns

uniformity of cross-talk rate within one pixel

cross-talk rate = # of events > 0.5p.e.

# of events > 1.5p.e.

100pixel 400pixel

Pixel-to-pixel uniformity of cross-talk rate

100pixel 400pixel

Measurement of active area inside one pixel

100pixel

400pixel

100um

85um

50um

38um

um um

laser spot

scan

efficiencyefficiency

geomε =72%

geomε =58%

Correction of MPPC signal

MotivationGain, PDE, crosstalk of MPPC are all sensitive to the temperature and bias voltage

It is necessary to correct the variation of gain, PDE,crosstalk when temperature or bias voltage changes

MPPC Signal ∝ Gain(T,V) x PDE(T,V) x 1-crosstalk(T,V)

1

T : temperature V : bias voltage

I have studied two correction methods 　　　　　　　　　

Set up

1/2inch PMT

cosmic-ray

1mm φfiber

MPPC2(100)

MPPC1(100)

MPPC3(400)

MPPC4(400)

scintillator

blue LED

• put scintillators in four layers• inserted fibers are connected

by four MPPCs(two are 400 pixel and two are 100pixel)

• change temperature intentionally between 20 and 25 degree

• The same bias voltage is applied to four MPPCs

• two triggers(cosmic,LED)

temperature chamber

With this setup we have traced the variation of light yield for cosmic-ray

Correction method Again

cross-talk rate PDE

ΔVΔVΔV

1) monitor the variation of gain

1)

2) 2) 2)

2) estimate the variation of ΔV

3) estimate the variation of crosstalk,PDE

3)

3)

15deg.20deg.25deg.

Correction method A(2)

calibration constant=

gain x PDE x

MIP ADC counts

1- crosstalk1

•this value must be constant if we can correct the variation of gain, crosstalk ,PDE

Correction method B

MIP ADC count 　∝ gain(T,V)×PDE(T,V)×

　 LED ADC count ∝ gain(T,V)×PDE(T,V)×1

1- crosstalk(T,V)

1- crosstalk(T,V)

1

MIP ADC count

LED ADC count

calibration constant =

MIP ADC count

LED ADC count

dist.taken by cosmic trig.

dist.taken by LED trig.

Inject the light from LED with the similar light intensity as MIP light yield

Variation of calibration constant

+3%

-3%

method A

+3%

-3%

-3%

+3%

method Bcalibration constant

calibration constant

• only the errors of MIP ADC count and gain are included

hour hour

C20○

C20○C25○ C25○

•detector response can be corrected within 3% level by both methods

MPPC1(100)

MPPC2(100)

MPPC3(400)

MPPC4(400)

Method A 2.5% 2.3% 3.8% 3.1%

Method B 2.5% 1.3% 2.4% 1.4%

Summary of correction methods

RMS/mean of calibration constant

•Required precision is a few % (This depends on the type of detector)

Both correction methods satisfy the requirement!

Requirement to MPPC

Item Requirement From where

Area 1.2x1.2mm2 To match 1.0mm fiber

No. of pixel 100/400 To keep dynamic range up to ~100p.e.

Gain ~106 To match readout electronics

Noise rate <1MHz To reduce accidental hits

Crosstalk <5% To reduce the noise rate with 1.5p.e. threshold

PDE >15% Light yield

Timing resol. 2-3ns Not so meaningful requirement

Specification of Trip-t

• Size : 14mm x 14mm• Power supplies : +2.5V• Power consumption < 10mW per

channel

Trip-t gain

pipeline

1pF

3pF

preamp gain x1 or x4

opamp gain x2,x4,x8…

preamp

input

opamp

•Trip-t gain can be changed by programming the registers

Setup for readout of MPPC

FADCA_OUT

-+

opampTrip-t

+5V

-5V

Digital wave generator

Control signals

Trigger

MPPC

LED

4m flat cable

AD conversionTrigger

Readout sequence of Trip-t A_OUT(charge)

MPPC signal (ch1)

preamp integrate

signal after preamp

A_OUT

Multiplexer clock0 1 2 33

ch1

A_OUT

charge = -1.5pC

charge = -0.5pC

A_OUT linearityADC count

charge(-pC) charge(-pC)

Trip-t gain

> >

Trip-t gain

> >

nonlinearity in the low pulse amplitude will be fixed for the new version of chip

(Data-linear)/Data(%)

Crosstalk of A_OUT• signal observed in the channel where

the charge is not injected

ch

0.4%

15

•crosstalk(i) =

ADCcharge(i)-ADCnocharge(i)

ADCcharge(15)-ADCnocharge(15)

•ADC(no)charge(i) = ADC count in the channel i when the charge is (not) injected

crosstalk<0.4%

Variation of Trip-t gain

• variation is ~4%ch gain (AD

C count/-pC)15 730

16 710

31 735

32 732

Readout sequence of Trip-t T_OUT(timing)

MPPC signal (ch1)

preamp integrate

signal after preamp

T_OUT

Multiplexer clock0 1 2 33

pipeline clock

ch1

T_OUT

time difference = 200ns

time difference = 100ns

T_OUT linearity

ns ns

ADC count

(Data-linear)/Data(%)

saturation

T_OUT conversion factor

• variation is ~25%ch Conversion factor (ADC count/ns)

15 8.52

16 8.60

31 8.00

32 9.97

D_OUT(Digital)

Trip-t1pF

charge injection

10ns •Delay inside Trip-t is

~10ns

100pixel 400pixel

MPPC gain measured by CAMAC and Trip-t

bias bias

Trip-t gain is well calibrated

Trip-t

CAMAC

Trip-t

CAMAC

Current design of real type electronics for T2K

16 MPPCs

16MPPCs

Trip-t

ADC ADCFPGA

Trip-t

16MPPCs

16 MPPCs

Trip-t

Trip-t

•64MPPCs per board with 16 high/low gain channels per chip

•control of Trip-t and ADC by FPGA

•temperature monitoring

•HV trimming DAC

spill(8bunch)

integrate(~300ns)

reset (~50ns)

Chip time structure (preliminary)

5.6us

readout catch late signal

~3us ~50us

beam

10

2

3

4

5

678

100

2

3

4

5

678

1000

2

3

4

AD

C u

nits

5 6 7 80.1

2 3 4 5 6 7 81

2 3 4 5 6 7 810

2 3 4

total external charge injected [pC]

high/low gain method

high

low

~300p.e.

~30p.e.

ADC distribution with lowest gain of Trip-t

•MPPC gain = 7.7x105

•S/N=3

VA chip• ASIC used for the K2K SciBar detector - 64ch inputs - VA does not have pipeline - VA itself cannot issue discriminated output - VA gain cannot be changed

Readout sequence of VAMPPC signal (ch1)

signal after shaper amp

outm

0

ch1

1 2

hold_b

Multiplexer clock

Readout of MPPC with VA•MPPC gain = 2.7x106, noise rate at 0.5p.e. th. = 240kHz

pileup of MPPC noise

S/N=1.7