Transcript MPPC mass check for the GLD calorimeter

A proposal of new simple system for
testing a large number of MPPC
for the R&D phase of GLD
calorimeter
2007/Feb/6
ACFA ILC Workshop
ICEPP , University of Tokyo
Hidetoshi OTONO
On behalf of the GLD calorimeter group
1
Summary as Contents
A large number of MPPCs are made in coming years
in R&D. Checking them would be required for sharing
the task among many universities; thus simple, robust,
cheap and precise system is preferable.
Improvement points;
-Hardware
Developed a serializer (with optical relay) in order to reduce the
number of expensive devices such as amplifiers and ADCs.
-Software
Simplification of measurement method for basic properties;
A way to simultaneously evaluate the
• dark noise rate
• noise rate due to cross-talk (double pulse height)
• gain of each MPPC .
2
of each MPPCs.
In near future, we would hopefully
test a prototype with order
of 100K MPPCs.
100k
Now
1k
10k
R&D Period
10M
Experiment
Num. of MPPCs
Motivation
Time
• Several calorimeter beam tests with
increasing number of MPPCs are expected
before the real ILC experiment.
• We have to check basic properties (gain, noise rate,
etc.) of brand-new MPPCs in each phase of R&D.
•The simple, robust and cheap measurement method
3
is preferable.
Hardware :
A development of
signal serializer in order to reduce
the number of expensive amplifiers and so on.
•Present method
•A proposal of new method
4
Issue of Present Method
HV
RS232C
Control PC
ＭＰＰＣ
AMP
ＭＰＰＣ
AMP
ＭＰＰＣ
ＭＰＰＣ
MultiAMP
plexer AMP
Present method uses parallel
AMPs and ADCmodules.
Sequential process
Modules
Module
AMP
5
Measurement
target
HV
RS232C
Our Method
For this purpose
Control PC We have chosen optical relay.
DC Supply
ＭＰＰＣ
relayOFF
ＭＰＰＣ
relayOFF
relayOn
ＭＰＰＣ
relayOFF
About 5$
AMP
ＭＰＰＣ
relayOFF
modules
6
Picture of the Circuit with Relay
MPPC output after a relay
MPPC output
before a relay
AMP
Supply voltage for a relay
OFF
OFF
OFF
ON
Supply voltage for a relay
7
Requirement and Choice of Relay
Isolation (how well the pulse can be stopped when a relay OFF)
?
relayOFF
High isolation is required.
Insertion loss (how well the pulse can be passed when a relay ON)
10ns
10dB
GOOD
[dB]
Insertion loss
Isolation
[dB] Isolation
Low insertion loss
is required.
?
relayOn
Insertion loss 0.6dB
Orange’s properties
between
0.3dB
100MHz ~ 1GHz
GOOD
30dB
100MHz
1GHz
100MHz
1GHz
Isolation:
10dB~30dB
Insertion loss:
8
0.3dB~0.6dB
 10 %
No relay [kHz]
Noise rate due to
Cross-talk
With relay[10^5]
Noise rate
With relay [kHz]
With relay [kHz]
The effects of serializer on 10 samples
Gain
 10 %
 10 %
No relay [kHz]
A few %
 3%
No relay [10^5]
There is no effect on noise rate and Cross-talk rate.
Gain measurement with a relay is shifted a few percents,
because of insertion loss
→Correction is needed for gain measurement.
Noise rate and cross-talk rate spreads are about 10%,
gain spread is 3%, due to large temperature variation.
9
Plan
• The relay elements is quite effective in
reducing the number of AMP and costs.
• Our 4ch circuit is still noisy, we need more
improvement.
• We are going to make a circuit with many
relays.
• In parallel, we are searching another
candidate for serializer.
10
Software : We applied MPPC’s photon
counting capabilities to simplify and
the measurement of noise rate, cross-talk rate
and gain.
Moreover we raise the precision of measurement.
•Present method
•A proposal of new method
11
Present Measurement
Scaler counting for
Noise rate measurement
(double pulse height) Setting MPPC
LED on
Adapting threshold
for noise rate
(double pulse height)
Gain
measurement
LED off
Threshold searching
Adjusting threshold for noise rate
Scaler counting for
Noise rate
measurement
12
A proposal of new method
Change MPPC
•Using dark noise for gain
measurement
•Threshold is fixed for noise rate
Taking ADC histogram OFFLINE analysis
•Noise rate
•Noise rate
(double pulse height)
13
•Gain
Our Circuit
For the measurement
of DAQ dead-time
The number of MPPCs
will be increasing .
Clock
Coin.
Scaler
latch
MPPC
relayOn
AMP
x 63
Dark Box
HV
RS232C
AMP
x 10
Disc.
Output
G.G veto
register
latch
DC Supply
Control PC
Disc.
40ns gate
NIM
Charge
ADC
CCNET
14
CAMAC
Num. of event
Our Method
Gate = 40ns
GAIN
Noise rate
(to be corrected
for DAQ dead time)
Noise rate (double pulse height)
(to be corrected
for DAQ dead time)
THRESHOLD
Accidental coincidence rate of Noise is negligible.
So the second peak is due to Cross-talk.
Channel
15
Noise rate
• OURS
• HPK
Gain [10^5]
Bias Voltage
Gain
• OURS
• HPK
Bias Voltage
Noise rate [kHz]
Noise rate [kHz]
Results
Noise rate
(double pulse hight)
• OURS
• HPK
Bias Voltage
We compared our measurement
with HPK data
•Noise rate and Gain is good correlation
with HPKdata.
•Difference can be seen
for noise rate (double pulse height)
→We are studying now.
16
The Picture of Pulses
We show several pulses which we are
studying now.
• Very closing pulses
• Curious pulses like discharge whose
reason is completely unknown.
We took all pictures at 78.8V which is typical
voltage for MPPC uses and they can be
frequently shown .
17
Very closing pulses
40ns
40ns
40ns
40ns
18
Curious Pulses
40ns
40ns
40ns
40ns
19
Plans
• Our method can be use for very closing pulses,
and ours is more precise than present method.
• Accidental coincidence of noises is negligible at
100kHz noise rate, thus the closing pulses can’t
explain the difference between measurement method.
• We guess that the closing pulses are due to crosstalk, though cross-talk induces ONLY the double pulse
height output by present view. we are studying now.
• The origin of curious pulses is unknown,
thus we are studying too.
20
Summary
Improvement points;
-Hardware
Developed a serializer (with optical relay) in order to reduce the
number of expensive device such as amplifiers and ADCs
-Software
Simplification of measurement method for basic properties;
a way to simultaneously evaluate the dark noise rate, noise rate
(double pulse height) and gain .
A large number of MPPCs are made in coming years
in R&D. Checking them would be required for sharing
the task among many universities; thus simple, robust,
cheap and precise system is preferable.
Our system is suitable for this purpose.
21
backup
22
Result (cont’d)
• OURS
• PRESENT
• HPK
Noise rate due to
cross-talk
Comparison with the result
of the same MPPC
by present system
at another university
A significant difference
can be seen .
The difference is about
2 times.
23
It’s our guess that the reason is understanding of cross-talk.
Our Idea of Cross-talk
Present idea
MPPC signal
Our idea
MPPC signal
MPPC signal
OR
(c.f)
normal noise
MPPC outputs two types of pulse due to cross-talk;
two times pulse height of normal noise and
double closed normal noise
24
The present method judges the latter as single normal noise
A Candidate of Difference between
Measurement Methods
Present measurement method
MPPC signal
THR
Disc. out
scaler
2 noises
Our measurement method
MPPC signal
ADC date
ADC
3 noises
Accidental coincidence rate of Noise is negligible.
25
These can’t perfectly explain the difference.
Typical Pulse
DarkBlue : raw pulse
LightBlue : pulse via ONrelay
Pink
: pulse via OFFrelay
26
Gain Measurement Method Test
Measurement with light source = 2.75 (10^5)
Gain
We checked 10 MPPCs
Measurement without light source
_ 0.07 (10^5)
= 2.7 +
Thus we decide to use
Dark noise for Gain.
27
Gain(10^5)
1p.e rate
Dependence on temperature
•
•
•
•
•
•
•
30 oC
25 oC
20 oC
15 oC
10 oC
0 oC
-20 oC
•30oC
•25oC
•20oC
•15oC
•10oC
•0oC
•-20oC
28