March 16th
QUALITY CONTROL PROTOCOL OF THE PHOTOMULTIPLIER
TUBES FOR THE ATLAS TILE CALORIMETER
Warning: This document is not a technical documentation on the PMT testbench but a draft on the planned protocol for PMT quality control. It will
be updated when first test-bench will be finally assembly. The time needed
for each measurement is just an indicative estimation. The exact time
needed for the protocol will be adjusted after qualification batch
experience. At that time a complete and detailed technical documentation
will be available
1
Introduction
Delivery will be in batches as agreed in the contract schedule. Once all quality
control checks have been performed and provisional acceptance granted after
conformity of the material agreed upon and signed, CERN will authorize
payment of the batch to Hamamatsu.
Table 1 summarize Tilecal specifications as agreed between CERN and
Hamamatsu
The acceptance-rejection of each batch will be notified by CERN Hamamatsu
within 8 weeks from the arrival of the tubes.
The size of a sub-batch that will be used to qualify the average characteristics
will be 250 photomultipliers. Provisional acceptance will be done in subbatches.
The basic production unit, in a sub-batch, that will be tested and qualified for
provisional acceptance is a photomultiplier.
The provisional acceptance of the tubes will be based on following criteria:






Visual inspection.
480 N.M. quantum efficiency.
Nominal voltage to get the Tilecal nominal amplification.
Dark current at 800 and 900 Volts.
Photocathode linearity.
Short term stability (ON/OFF drift).
The results of the measurements performed by ATLAS and the values that the
contractor has measured will be filed in a quality acceptance database and
quality control sheets. It will result in an acceptance-rejection certificate that will
be issued by CERN to Hamamatsu as shown in Table (2) and (3).
If a tube do not conform to some of the minimum specifications, it will be
rejected.
If a sub-batch of tubes does not conform to the average quantum efficiency at
480 nm, to the average dark current and to the average short drift requirements,
the sub-batch will be rejected. Hamamatsu should replace some of the worse
tubes so that the whole sub-batch fulfill average specifications
The voltage repartition of the photomultiplier used for all measurements of the
PMT’s quality control is the nominal voltage repartition that had been designed
to optimize tubes’s linearity.
The nominal amplification used for all measurements of the PMT’s quality
control is 10**5
2
The room temperature used for all measurements of the PMT’s quality control
is defined as 25°C ± 3°C
Out of the first batch of tubes (qualification batch), a set of 50 pieces should be
delivered by Hamamatsu with the following data:

480 nm quantum efficiency

short term stability with nominal voltage repartition

pulse linearity with nominal voltage repartition

current amplification at 800 and 900 Volts with nominal voltage repartition

dark current at 800 and 900 Volts with nominal voltage repartition
From these data Hamamatsu and Tilecal measured data the final procedure
established to control the quality of the tubes will be agreed
A set of 7x5 tubes will be chosen from these 50 pieces. This batch of 35 tubes
will be defined as the batch of reference photomultipliers. This reference tubes
will be used in order to :

provide a long term monitoring of each bench by measuring all along the
PMTs quality control period the characteristics of 4 reference PMTs

achieve a cross calibration of the 7 test benches used for the PMTs
quality control. Such a calibration will be repeated at regular periods
during the production phase.
This quality control comprises two parts:

those tests that should be carried to achieve selection-rejection of the
tubes. These tests will be called hereafter STEP1 tests

those that should be carried to characterize the pair PMT-voltage divider.
These tests will be called hereafter STEP2 tests
STEP1 measurements
Here photomultipliers are coupled with the special dividers that allow to switch
between the photocathode mode and the anode one:
 in photocathode mode, one can apply a desired voltage value (up to 100V)
between the photocathode and the first dynode. and measure the
corresponding current
 in anode mode, one can apply a desired high voltage value and measure the
corresponding current between the photocathode and the last dynode.
3
The light used is a DC one at 480nm.
There are 25 locations in the test bench grid: one is taken by the large area
central photodiode and four by reference photomultipliers. It means that 20
photomultipliers can be tested at the same time. As there are 250 of them in
each sub-batch , the whole sub-batch test should divided in 12 grids of 20 tubes
and one of 10 for a total of 13 successive periods of test.
Prelude to PMTs measurements
Between each successive period, some manual operation should be done. This
is the only part of the all STEP 1 which needs human actions. It seems that it
could take one hour with two people involved:
 First, one has to remove the grid from the box, remove the photomultipliers
from the grid, dismount each of them from its special divider, and replace
them in the stocking area.
 One can then take the new set of 20 tubes, associate each photomultiplier
with a special divider, mount them in the grid, and replace the grid in the
box.
 After having introduced the serial number of the new batch of
photomultipliers in the controlling computer, one can start the bench autotest.
Bench auto-test is done by comparison with expected values and lasts about
two hours. Tested systems are:
 light (and therefore preamplifiers),
 programmable voltages,
 LECROY HV4032,
 movements of the X-Y table (for channel intercalibration),
 filter-wheels (movements, and attenuation measurements for current
amplification determinations).
The temperature regulation system is autonomous. It is able to regulate the
temperature of the photomultipliers box to (25.0  0.2)C which is very important
in particular for PMTs stability measurements. For that, the testing room
temperature had also to be somehow regulated. The PMT box temperature
stabilisation is obtained few hours after closing the box.
By moving the photodiode in front of each channel using the X-Y table, we
determine the correction coefficient to apply to each of them for quantum
efficiencies measurement.
The tested PMTs are then heated for 10 hours without any illumination.
After these autotest and calibration period, the different PMTs chararcteristics
involved in the selection-rejection criteria are measured, that are:
4





Dark current measurement
Photocathode linearity
480 nm quantum efficiency
Current amplification
Short term stability
Each measurements is performed automatically under the control of a functional
block of the Labview program. Each functional block is independant of the
others but could chained together to do only one global measurement block.
For each measured characteristics, corresponding data are recorded in one
excel file, but the whole global measurements data will be also summarized in a
global file.
Rejection-acceptation criteria will be introduced directly in the software to finaly
reject PMTs out of specification
Dark current measurement
The photomultipliers dark current will be measured with the ATLAS voltage
repartition and at room temperature defined as 25°C ± 3°C. Special divider is
set in anode mode by the logic of the DAQ and control electronic.
Dark current will be measured over the whole voltage range from 500V to 900V.
Selection-rejection specifications agred with Hamamatsu in the contract are:
 The maximal dark current at 800 Volts should be less than 2nA.
 At 800 Volts the average value should not exceed 250 pA.
 At 900 Volts the maximal value of dark current should be less than 8 nA,
with a ratio of the dark current values between 900 Volts and 800 Volts
smaller than 10.
Photocathode linearity
The photocathode linearity will be measured as a function of the photocathode
to dynodes voltage. Special divider is set in photocathode mode by the logic of
the DAQ and control electronic. The photocathode current is measured between
the photocathode and the ground. The voltage is applied between the dynodes
and the ground.
A voltage of 100 Volts is first applied between the photocathode and the
dynodes to be sure to be at the plateau of the PMTs collection efficiency.
The light is then adjusted to get a photocathode current of 10 nA for one of the
reference PMTs.
Then, we measure the currents for voltages from 0V to 100V and deduce the
photocathode linearity curves.
Selection-rejection specification agreed with Hamamatsu in the contract is that
tubes have not reached 90 % of the plateau value at a voltage of 50 Volts will
be rejected
480 nm quantum efficiency
The tube quantum efficiency will be measured by ATLAS at a wavelength of
480  5 nm. The tubes will be illuminated over all the effective photocathode
5
area using the standard light mixer (18x18x43) mm and a continuous light
source (LED) filtered by an interferential filter (ANDOVER 480FS10-50).
The light flux will be measured using large area photodiodes (Hamamatsu
S6337-01) set in place of the tubes, in front of the light mixer with a gap less
than 1mm.
The geometry of the implementation in the test bench of one of the standard
PMT channel and of the large area photodiode will result in a different light flux
illuminating a tube and the large area photodiode. Such a difference is taken
into account by a calibration factor for each PMT channel measured in the
preliminary bench auto-test period
The large area photodiodes used in the 7 control sites will be calibrated by a
reference large area photodiode. ATLAS and Hamamatsu will agree on the
calibration of this reference photodiode and accept that measurements done by
the set of large area photodiodes after calibration corrections correspond to the
tube’s 480 nm quantum efficiency.
Special divider is set in photocathode mode by the logic of the DAQ and control
electronic.
A voltage of 100V is first applied between photocathode and dynodes. Then we
adjust the light using filters to get a current of 20nA at one of the reference
photomultipliers. Finally, we acquire the currents of the photodiode and of the
photomultipliers and deduce the quantum efficiencies from the ratios taking into
account the light yield non uniformity for each channel (channel calibration
coefficients)
Selection-rejection specification agreed with Hamamatsu in the contract is that
tubes with a 480 nm quantum efficiency below 15 % will be rejected and the
contractor agree to replace them by tubes with the correct quantum efficiency. If
the averaged 480 nm quantum efficiency of a sub-batch of tubes is below 18%,
the sub-batch will be rejected.
Current amplification
The tube current amplification will be measured with a 480 nm DC light source
and the ATLAS voltage repartition in a two steps procedure:

First the photocathode sensitivity is measured by measuring the current
between the photocathode and all the dynodes electrically connected
together. In this photocathode mode, the voltage between the photocathode
and the dynodes is set first to 100V and the light is adjusted to get a current
of 10nA at one of the reference photomultipliers In that condition we assume
that the photocathode current has reached the full collection efficiency
plateau value.

Photocathode currents of the tested PMTs is so measured

Then logic of the command and control electronic switch between
photocathode mode and anode mode. In this anode mode, we interpose a
filter of known attenuation (this nominal value is around 1000 but should be
6
measured during the bench auto-test period) to keep the corresponding
current at an acceptable level of few A.

We deduce photomultipliers current amplification gains from the ratio of two
modes currents, taking into account the filter attenuation.

Measurements will be repeated every 50 Volts over the whole voltage range
(500V to 800V)
Fitting the obtained curves, we get the nominal high voltage corresponding to a
gain of 105 and the  of each photomultiplier.
Selection-rejection specification agreed with Hamamatsu in the contract is that
the nominal amplification (105) value must be reached in a voltage range
between a minimum value of 600 V and a maximum value of 800 V
Short term stability
The tube short term stability will be measured, recording the anode sensitivity
variation as a function of the time. The amplification is set to its nominal value
(105) with the ATLAS nominal voltage repartition. A DC light source, giving a
typical anode current of 3 A, will be monitored by a photodiode to correct from
light source unstability.
First, we put each photomultiplier under its own nominal high voltage and let
them without light for 10 hours.
Then, we alternate cycles of 3 hours with and without light for a total time
measurement of 3x5 hours. At regular intervals, we measure photomultipliers
currents and central photodiode one
During the measurement, the high voltage should be applied to the tested
tubes.
The measurement robustness is guaranteed by the temperature regulation
system.
The drift will be determined as the maximum of the variation (in %) of the anode
sensitivity between the beginning and the end of the 3 illumination periods
Selection-rejection specification agreed with Hamamatsu in the contract is that
the maximum of this short term drift has to be at maximum 1.5%, and the
averaged over the whole sub-batch less than 0.8%
STEP 1 schedule
As mentioned above, two peoples are needed for one hour for the
photomultipliers installation. After the bench auto-test, one of them has to check
7
if all tests have been passed successfully and so if the automatic acquisition
had started: this operation is a matter of few minutes. After that, everything is
automatic.
STEP 1 measurements of one batch will last two days: the succession of
operation is shown in following table. So 3 batches can be measured in one
week if batch installations are done on Monday, Wednesday, and Friday. To
complete the measurements of an Hamamatsu delivery batch, it will take 5
weeks if no major problems
Hour
To
To + 1
To + 2
To + 3
To + 4
To + 5
To + 6
To + 7
To + 8
To + 9
To + 10
To + 11
To + 12
To + 13
To + 14
To + 15
To + 16
To + 17
To + 18
To + 19
To + 20
To + 21
To + 22
To + 23
First Day
Batch installation
Bench auto-test
Second day
HV on and no light
Temperature stabilisation
Channel intercalibration
HV on and no light
Short term Stability
Dark current
Collection efficiency
Quantum efficiency
Current amplification
HV on and no light
Spare
STEP2 measurements
Here photomultipliers are coupled with their own dividers, same as in ATLAS,
and a pulsed light is used. For linearity measurements, the background is
simulated adding the DC light. Results on linearity, dark current could be used
to choose the PMT implantation in the Tilecal.
Prelude to PMTs measurements
This is the only part of the all STEP 2 which needs human actions. It seems that
it could take one hour with two people involved :
 First, one has to remove the grid from the box, remove the photomultipliers
from the grid, and replace them in the stocking area.
8



One can then take the new batch, associate each photomultiplier with its
own divider, mount them in the grid, and replace the grid in the box.
After having introduced the serial number of the new batch of
photomultipliers and of the corresponding dividers in the controlling
computer,
one can start the bench auto-test.
Bench auto-test is done by comparison with expected values and lasts about
two hours.
Tested systems are:
 light (and therefore preamplifiers),
 programmable voltages, LECROY HV4032,
 filter-wheels (movements, and attenuation of the filters).
Temperature stabilisation is achieved as in STEP 1
Dark current
The photomultipliers and the divider dark current will be measured at room
temperature defined as 25°C ± 3°C.
Dark current will be measured over the whole voltage range from 500V to 900V.
Pulsed light amplification
The high voltage is varying from 500V to 800V. For each value, 10000 events
are acquired with the pulsed light only. The amplification is deduced from the
width of the spectrum. Points taken too close to the pedestal or to the end of the
range are automatically removed. Fitting the remaining points, we get the
nominal high voltage corresponding to a gain of 10 5 and the  of each
photomultiplier in pulsed mode.
Linearity without background
We put each photomultiplier under its own nominal high voltage obtained in 3-2.
Using the filter wheel, we change the amount of light received. Taking as a
reference the pulsed photodiode inside the photomultipliers box, we obtain ten
measurements corresponding to the different positions of the filter wheel.
The non-linearity over the full dynamic range should not exceed 2%
Linearity with background
Same as above except that we first adjust the DC light at the desired
background value using one of the reference photomultipliers. Possible values
are 0.1, 1, and 2 A.
STEP 2 schedule
As mentioned above, two peoples are needed for one hour for the
photomultipliers installation. After the bench auto-test, one of them has to check
if all tests have been passed successfully and so if the automatic acquisition
had started: this operation is a matter of few minutes.
9
After that, everything is automatic.
STEP 2 measurements of one batch will last one day: the succession of
operation is shown in the following table. So 5 batches can be measured in one
week if batch installations are done on Monday, Tuesday, Wednesday,
Thursday and Friday. To complete the measurements of an Hamamatsu
delivery batch, it will take 3 weeks.
Hour
To
To + 1
To + 2
To + 3
To + 4
To + 5
To + 6
To + 7
To + 8
To + 9
To + 10
To + 11
To + 12
To + 13
To + 14
To + 15
To + 16
To + 17
To + 18
To + 19
To + 20
To + 21
To + 22
To + 23
First Day
Batch installation
Bench auto-test
Temperature stabilisation
HV on and no light
Dark current
Pulsed light amplification
Linearity without background
Linearity with background
Spare
10
Quantity : 10140 photomultipliers
Unit
105
Nominal Tilecal amplification
High voltage at nominal gain
Requirement
Minimum
600
V
Maximum
800
V
2
2
%
Non-linearity for pulses of 20 ns and 50 mA currents
Dark current (DC800) at room Maximum
temperature and 800 Volts
Averaged
Dark current (DC900) at room Maximum
temperature and 900 Volts
DC900/DC800
250
8
nA
pA
nA
 10
%
Minimum at 480 nm
18
15
Averaged at 520 nm
 12.5
%
HV between cathode and first dynode for 90%
(Cathode linearity)
Rise time for delta-like pulses
 50
V
 2.5
ns
Response time FWHM to delta-like pulses
8
ns
Drift
1.5
%
Short term average
 0.8
%
Long term
 2
+ 10, - 30
%
< ±10
%
< 60
mm
250
mm2
Quantum efficiency
Averaged at 480 nm
Short term maximum
Ageing (relative output for integrated Q = 100 C): variation
respect the initial response
Photocathode non-uniformity
Length (including pins)
Useful photocathode area
80
Stability with temperature (variation of nominal gain per ºC
~ 0.25
between 20 and 30 ºC)
Insensitivity to neutron and gamma fluxes
Magnetic sensitivity (relative output for 20 Gauss)
%
%
%
%
Table 1 : Summary of the Tilecal requirements
11
Control
Anode sensitivity measured with ATLAS voltage
repartition at 800 Volts.
Photocathode sensitivity measured in the same
condition that the anode sensitivity
480 nm quantum efficiency
Dark current measured with ATLAS nominal
repartition at 800 and 900 Volts
Current amplification voltage dependence with ATLAS
voltage repartition and 480 nm DC light
Photocathode linearity with 480 nm DC light
Short term stability at nominal amplification
Amplification voltage dependance with standart divider
Dark current voltage dependance with standart divider
Linearity with divider
Made by
Hamamatsu
Made by
Tilecal
X
X
X
X
X
X
X
X
X
X
Table 2 : quality control measurements summary
12
Item
1
2
3
Approved
Rejected
Anode sensitivity at 800 V with
nominal voltage repartition
Photocathode sensitivity
measured in the same condition
that anode sensitivity
480 nm Quantum efficiency
>15% (individual)
5
480 nm Quantum efficiency
<Sub-Batch> > 18%
6
G=10^5 Voltage Point
(650 < HV < 800 Volts)
7
Dark Current (800 V, 25 °C)
< 2 nA (individual),
8
Dark Current (800 V,25 °C)
<Sub-Batch> < 250 pA
9
Dark Current (900 V, 25 °C)
< 8 nA (individual),
10
(DC900/DC800) < 10
11
Non-Linearity < 2%
(20 ns pulses and 50 mA)
13
ATLAS
Value
Visual Inspection
4
12
Producer
Value
Short-Term Drift < 1.5 %
(300 pC, 1KHz)
<average> < 0.8%
90% Photocathode to dynodes
Photocatode linearity
< 50 Volts
Table 3 : PMT quality sheet
13