You are on page 1of 98

INTERCALIBRATION OF PHOTOMULTIPLIER TUBE

TEST BENCHESTHROUGH PRECISION TESTING OF


THIER INTERNAL PHOTODIODES
Barry S. Spurlock
The University of Texas at Arlington
Abstract
The tile calorimeters of the ATLAS detector of the Large Hadron Collider at
CERN will require over ten thousand photomultiplier tubes.
These
photomultiplier tubes will be subjected to rigorous testing in order to ensure
that they fulfill their operational requirements before installation in the
detector. Testing such a large volume of photo-multiplier tubes on schedule
has necessitated the construction of a number of test benches. These tests are
performed by comparing the performance of photomultiplier tubes to that of a
large-area photodiode. Intercalibration of these test benches is required to
ensure that the test results are consistent at all test sites. This will be
accomplished through precise testing of the photodiodes used in each of the
test benches. This necessitated the design and construction of a unit capable
of performing the precision testing of these large-area photodiodes.

TABLE OF CONTENTS
Chapter 1. Introduction............................................................................................. 4
Chapter 2. Prototype Design Objectives .................................................................. 4
2.1. Main Chassis ............................................................................................... 5
2.2. Central Shaft................................................................................................ 5
2.3. Source Cap .................................................................................................. 5
2.3.1. Filter Holder ....................................................................................... 6
2.4. LAP Cap...................................................................................................... 7
2.5. Power Supply and Power Circuitry ............................................................. 7
2.6. Connections, Controls, and Displays .......................................................... 7
Chapter 3. Prototype Electrical Systems.................................................................... 9
3.1. Power System Design.................................................................................. 10
3.1.1. Power System Connections, Switches, and Controls ......................... 11
3.1.2. Internal Power Supply ........................................................................ 11
3.1.3. Voltage Regulators ............................................................................. 11
3.2. Input Circuit Design .................................................................................... 11
3.2.1. Light Emitting Diode (LED) .............................................................. 11
3.2.2. LED Intensity Control and LAP Size Selection Control.................... 13
3.3. Output Circuit Design ................................................................................. 13
3.3.1. Large Area Photodiodes (LAPs) ........................................................ 13
3.3.2. Amplifier Circuit ................................................................................ 14
3.3.3. Voltmeter/Display .............................................................................. 15
3.4. Circuit Analysis........................................................................................... 15
3.4.1. Analysis of Output Circuit ................................................................. 17
3.4.2. Analysis of LED/LAP Interaction ...................................................... 18
3.4.3. Analysis of Input Circuit .................................................................... 19
3.4.4. Conclusions Based on Numerical Analysis........................................ 19
Chapter 4. Viability Testing ..................................................................................... 19
4.1. Experimental Set Up ................................................................................... 19
4.2. Data from Viability Tests ............................................................................ 20
4.3. Conclusions from Viability Tests................................................................ 20
Chapter 5. Prototype Construction ........................................................................... 22
5.1. Design of Mechanical Components ............................................................ 22
5.1.1. LAP Cap............................................................................................. 22
5.1.2. LED Cap............................................................................................. 24
5.1.3. Central Shaft....................................................................................... 25
5.1.4. Main Chassis ...................................................................................... 28
5.2. Construction of Mechanical Components ................................................... 28
5.3. Selection and Purchase of Electrical Components ...................................... 28
5.4. Assembly of the Prototype .......................................................................... 28
5.5. Troubleshooting .......................................................................................... 29
5.5.1. Readout Fluctuation ........................................................................... 29
2

5.5.2. Low Output ........................................................................................ 29


5.5.3. Internal Capacitance ........................................................................... 29
5.5.4. LAP Connection................................................................................. 30
5.5.5. Temperature Gauge ............................................................................ 30
Chapter 6. Testing the Prototype.............................................................................. 31
6.1. Testing the Light Tight Chamber ................................................................ 31
6.2. Temperature Testing.................................................................................... 31
6.3. Repetition of Photodiode Installation.......................................................... 31
6.3.1. Data from Repetition Tests ................................................................ 31
6.3.2. Data Analysis ..................................................................................... 35
6.4. Testing Operation with 220VAC Input Power............................................ 43
Chapter 7. Conclusions ............................................................................................ 43
Appendix A. Parts List ............................................................................................. 48
Appendix B. Specifications for Electronic Components ......................................... 49
Appendix C. Dimensional Outlines for Mechanical Components........................... 55
Bibliography.............................................................................................................. 98

1. Introduction
The ATLAS detector calorimeters will use over ten thousand photomultiplier tubes
(PMTs). Due to the large number of PMTs, these will be tested at a number of locations
using identical test benches, constructed for this purpose. It is important that the results
from different PMT testing stations be comparable at the same level. To this end, a
method of intercalibrating the test stands has been developed. The various test benches
assess the quality of PMTs by comparing their response when stimulated by an
assortment of light emissions to the response of a central photodiode. Consequently, we
can intercalibrate the different test stands by testing their internal LAPs. Though no
specific requirements on the accuracy of this intercalibration were specified in the
technical design report of the ATLAS tile calorimeter, it offers a general statement that
the calibration of PMTs during operation should be ~1%. Thus, our goal will be to
achieve similar accuracy.
There are two possible ways of achieving our goal. Either we bring the LAPs to one place
for testing, or we ship the test equipment to each different location. Having the versatility
to operate in either of these ways is preferable and easily obtainable. This requires that
our unit must be compact enough to be easily shipped. Our prototype will operate solely
in DC mode in which the LAP is excited by a continuous, steady level of light. This will
be the most simple to construct, allowing testing to commence as soon as possible. Our
basic design will allow for increased capabilities to be added to future versions. One
improvement is the inclusion of a pulse mode, where the LAP is excited by a short burst
of light. Another possibility is the addition of an intermittent mode, where the photodiode
is excited by the emissions of a radioactive source. The emissions of a radioactive source
are more consistent that other sources, and this could lead to more accurate
measurements. We could also include a control port, which will allow an outside
computer to control all of these modes.
In this thesis, we will discuss the design and construction of a unit to test LAPs in DC
mode. We will present how this design achieves our goals. We will then present the
results of extensive studies of the performance of the test unit.

2. Prototype Design Objectives


The highest priority is reproducibility. We must be able to create the exact test conditions
every time we perform a test on a LAP. This includes ensuring the precise separation and
alignment between the LAPs and the light source, as well as keeping the intensity and
orientation of the source exact. To this end, our parts will be rectangular rather than
circular to allow for greater precision in machining.

Another major concern is the elimination of crosstalk. In other words, we must ensure
that electrical signals directed to the source will have no effect upon the LAP's output
signal. The impact of noise is also an issue, and care must be taken to reduce noise
wherever possible. This is especially important in those locations where our signal to
noise ratio is low. Since the signal coming from the LAP is very small, we must be
particularly concerned with minimizing noise prior to amplification of this signal. Finally,
as with all electrical systems, the appropriate measures of electrical protection are needed
to minimize the possibility of damage due to shorts and surges.
2.1. Main Chassis
Unlike the central shaft, the outer casing is not required to be light tight. It must be
durable enough to withstand rough handling during shipping. The main chassis needs to
provide a stable base for our testing, and it must have ample room to place controls and
displays. It should be composed of a conducting metal, which can be grounded. This will
provide shielding against the influence of external electromagnetic fields.
2.2. Central Shaft
The central shaft is where the light from our source will propagate to the LAP under
testing. It must be light tight, meaning that it no outside light can penetrate into its
interior. At one end, it must connect to the source cap, which will hold the light source.
At the other end, it will connect to the LAP cap, which will hold the LAP to be tested.
This design of changeable source and LAP caps will allow us to use this unit with a
variety of input and output sources. Care must be taken to ensure that the connections
between the central shaft and the caps do not allow outside light to enter.
2.3. Source Cap
The source cap will hold our light source, and it must be removable to allow replacement
of that source. By making it interchangeable, we can perform a variety of tests using
different light sources with minimal adjustment to the basic unit. Our design will only
include a light emitting diode (LED) as a light source, but having the ability to upgrade to
allow operation with a pulsed or radioactive source will provide a more complete test of
the LAP characteristics. The cap must lock into position with the central shaft to ensure
that the spacing between the source and the LAP is identical in every test. As mentioned
earlier, this connection must also be light tight. The design will allow for the addition of a
filter holder, but this may not be necessary. A general layout for our test unit is shown in
figure 1.

Figure 1. Layout of the LAP Test Unit.


2.3.1. Filter Holder
Our design will be compatible with an easy-to-construct 3-filter holder. This will allow us
the option of using a combination of filters including wavelength selection, diffusion,
and/or attenuation. The DC prototype will first be tested without any filters. If it is
determined that filters are necessary, then the filter holder will be installed. The required
filters will be installed permanently during the design and testing stages since
replacement of these filters could lead to an additional source of error. If a mixer is
required, it will be connected to the filter holder.
6

2.4. LAP Cap


The detachable LAP cap must allow us to easily exchange LAPs. It must also lock into
position on the central shaft in order to ensure definite positioning of the LAP, and create
a light tight connection. The LAP will be held in precise position by a frame, which is
secured to the LAP cap. Since testing will be performed upon photodiodes of various
sizes, this will require a different frame for each size and shape. Each type of LAP will
also have slightly different pin positioning. This requires that we adopt an electrical
connection that can adapt to each type of LAP. An extra LAP cap will be constructed in
which the LAP is permanently attached. This will serve as an additional control.
2.5. Power Supply and Power Circuitry
There are a variety of supply voltages available from the PMT test benches themselves,
but variations from one bench to another could have an effect upon our readings. Instead,
our test unit will have its own internal power supply. Supplies are available that can be
powered from both 110VAC and 220VAC inputs. This will allow us to perform tests
anywhere there is a wall socket. Separate voltage regulators will provide power to the
LED and the LAP amplifier in order to minimize crosstalk. This is a cost-effective way to
supply the required voltages and it also helps to provide a degree of surge protection. The
meter used to read and display the LAP output signal will require its own independent
power source. Consequently, we will include one or more batteries to power the meter.
Most of the circuitry can be placed anywhere inside the main casing, but the LAP
amplifier should be placed as close as possible to the LAP itself within the LAP end cap.
This will allow us to amplify our signal before it becomes degraded by noise.
2.6. Connections, Controls, and Displays
The controls and displays are listed below. Many are visible in the perspective drawing,
figure 2.
1. Main Power Switch: This switch connects or disconnects the unit from the
external source of power.
2. DC Test Switch: This switch supplies power to the source LED.
3. LED Intensity Control: This adjusts the input voltage supplied to the LED in
conjunction with the LAP size selection control, and thus controls the
amount of light emitted. It will have five settings allowing us to test the
LAP response to a range of light intensities.
4. LAP Size Selection Control: This switch scales the LED intensity so that we
can obtain readings at optimal levels for each of the three sizes of LAPs
that will be tested.
5. Power Indicator Lights: These verify that the power supply is functioning.
6. Meter Display: This displays the voltage of our amplified LAP signal.
7

7. Meter Scale Switch: This adjusts the maximum voltage displayed by the meter,
thereby maximizing the precision of each reading.
8. Power Input Socket: This will be used to connect the test unit to external
power, namely a wall outlet.
9. Source Selection Switch: This adjusts the internal settings to accept either
110VAC or 220VAC as external power.

Figure 2. Perspective Diagram of Test Unit.

3. Prototype Electrical Systems


The circuitry of the prototype can be broken down into three separate component systems.
The first will be called the power system, and it will include the power supply and its
controls/displays. The second part will be called the input circuit, and it will contain all
of the components needed to produce the illumination required for the test (up to and
including the LED). The third part will be called the output circuit and it will contain
all of the components required to collect the light and display our measurement (including
the LAP and the display). The output circuit is essentially a light meter. The general
operation is shown by the flow chart shown in figure 3.

Figure 3. Operational Flow Chart.

3.1. Power System Design


The power system will deliver electric power to every part of the unit with one exception;
the battery used as an independent supply to the digital panel meter will be included in the
output circuit. We intend to draw our external power from a standard wall outlet (either
110VAC or 220VAC), and deliver this to our internal power supply. This will be done
through a standard (IEC) computer cable and socket. The input will be grounded to the
main chassis, and fuses will be used to protect the power supply and other components.
The internal power supply will then provide 15VDC for operation of the components of
the tester. LEDs will be used as power indicators, with one accepting power from the
+15VDC source and the other accepting power from the -15VDC source.
The power supply output will then be further split into three voltage regulators. The first
will be called the input voltage regulator, and it will provide the voltage used in the input
circuit (+5VDC). The other two will provide the voltages needed for our amplifier circuit
(+12VDC and 12VDC). The power system will also include the power switch and the
power indicator light(s). A schematic is shown in figure 4.

Figure 4. Schematic Diagram of Power System.


10

3.1.1. Power System Connections, Switches, and Controls


A simple rocker switch will be used as our power switch. Two inexpensive LEDs will act
as indicators, one for the positive voltage output of the power supply and one for the
negative. To draw power from a wall socket, an ideal choice for the power input socket is
an IEC female power connector. This would enable us to use the standard power cords
used to power computers. These cords should be readily available if some problem should
arise. A source selector switch will be required to choose between 110VAC and
220VAC. This switch must also make the desired changes to the input connections of the
power supply (so that it can accept the appropriate input voltage), and divert the input
current through the correct fuse. The DC test switch will be placed prior to the voltage
regulator used for the input circuit. A toggle switch seems a very functional choice.
3.1.2. Internal Power Supply
Our internal power supply must take the input power, 110/220V@60Hz, and provide
power needed for our voltage regulators. All of our voltage regulators can be powered by
a supply that can provide both +15VDC and 15VDC. It was decided to use a linear
supply in order to minimize noise. The supply chosen was an IHAD15-0.4 from
International Power. It is small and relatively light, with good regulation. One concern
about this supply was its low power capacity (it can only produce 0.4A through either
channel), but estimations of our needs indicate that this is more than adequate.
3.1.3. Voltage Regulators
The input voltage regulator will take +15VDC input, and supply 5VDC to the LED via
the test selection switches (and resistors). The output voltage regulators will provide
+12VDC and 12VDC to the amplifier circuit. All are inexpensive and readily available.
The input voltage regulator will be connected to the DC test switch.
3.2. Input Circuit Design
The purpose of the input circuit is to produce the illumination needed to conduct our test.
It will consist of the LED, the LED intensity control, and the LAP size selection control.
The input circuit is shown in figure 5. This figure also includes the input voltage
regulator, which is part of the power system, in order to show the connections.
3.2.1. Light Emitting Diode
The LED being used is an NSPB310, which was in our possession at the beginning of this
project. The light emitted from the LED has a spectrum that can be approximated by a
Gaussian distribution with a peak at 465nm and a width of 30nm. The peak frequency is
11

close enough to 480nm to suit our needs, with a nominal intensity of half a candela. The
specifications for this LED can be found in appendix B.

Figure 5. Schematic Diagram of Input Circuit.


We can approximate the LED intensity with the equation shown. The leading coefficient,
I0, will be dependent upon the voltage applied to the LED:

12

In order to determine the total power emitted by the LED we must then integrate over all
possible wavelengths:

The fraction of this luminous power that reaches the photodiode will be determined by
the geometry of our experimental layout.
3.2.2. LED Intensity Control and LAP Size Selection Control
The function of both of these controls is to determine the amount of current sent to the
LED. Our goal is to produce a LAP output voltage near 5VDC for the high setting of the
LED intensity control and a LAP output voltage near 0.1VDC for the low setting of the
LED intensity control. It should produce these results for each size of LAP being tested.
This is complicated by the LEDs non-linear relationship between the input current and
intensity. In order to produce the desired results, we will combine the LED intensity
control with the LAP size selection control in the same configuration as a digital
multiplexer. Each of the fifteen possible settings of these two controls will divert the
current through a resistor unique to that setting. The resistor chosen will restrict the flow
of current to the LED to produce the required light intensity. The values of the resistors
were determined by experiment using a control photodiode. The original design neglected
to include the diodes in series with each resistor. Failure to include these allowed the
current to flow back through many different paths, increasing the current delivered to the
LED.
3.3. Output Circuit Design
The purpose of the output circuit is to measure the light produced by the input circuit and
display the results. It will consist of the photodiode, the photodiode amplifier circuit, and
the meter/display. It is essentially a powerful light meter. A schematic of the output
circuit is shown in figure 6.
3.3.1. Large Area Photodiodes
The photodiodes to be tested will be Hamamatsu model numbers S3590-03, S2744-03,
and S6337-01. Their active surface areas are 100 sq. mm, 200 sq. mm, and 400 sq. mm,
respectively. These photodiodes have shunt resistances which are greater than 1M (as
we will see, an exact figure is unnecessary), and generate an output current of 0.26A per
watt of incident light at 480nm. This is a slight approximation since our chosen LED will
emit light centered on 465nm. The equivalent circuit for these photodiodes can be seen in
figure 8.
13

Figure 6. Schematic Diagram of Output Circuit.


3.3.2. Amplifier Circuit
The amplifier must take a very small current from the photodiode, and from it generate a
significant voltage that can be measured by the voltmeter. The logical choice for an
amplifier circuit is to use the same circuit that is used in the PMT test stand. This circuit
is shown in figure 7. The op amps used are TL083s with an input resistance of 1012 and
a differential voltage gain of 2x105. For DC operation, we can safely ignore the feedback
capacitor of the first op amp, but the 4M feedback resistor will prove very significant.
The second op amp is merely a voltage follower used to isolate the amplifier from the
output. This is a low noise circuit, but noise can still affect our system prior to
amplification (where the signal is very small). Thus it is important to keep the distance
between the LAP and the amplifier as small as possible. An equivalent for this circuit can

14

be seen in figure 8, which shows the combination of amplifier and photodiode


equivalents.

Figure 7. Amplifier Circuit.

Figure 8. Equivalent Circuit for Output Circuit.


3.3.3. Voltmeter/Display
A number of suitable digital panel meters are available commercially. These include both
the logic and the display and have high accuracy. The meter we have selected is powered
by 9VDC and is accurate to 0.05%. Specifications can be found in appendix B.
3.4. Circuit Analysis
In this analysis, we will omit for greater clarity. All values will assume SI standards for
units. Thus, currents will be in amperes, all voltages will be in volts, and all resistances
will be in ohms. The complete circuit is shown in figure 9.

15

Figure 9. Schematic Diagram of Complete Circuit.


16

3.4.1. Analysis of Output Circuit


For analysis purposes, the only components of the output circuit that concern us are the
LAP and the DC amplifier circuit. An equivalent circuit is shown in figure 8. Our
starting point is the current equation at node A:
IL = ID + IR + IO IF
We will consider each of the currents separately. The photodiode specifications indicate
that we can treat IL as a current source which is dependent upon the power of the incident
light, PI (in watts), that strikes the surface of the photodiode:
IL = 0.26PI
We will assume that the current through the diode obeys the standard equation for diodes
at room temperature. The equation relates the diodes current, ID, to the voltage across it,
VP, and a constant called the saturation current, IS:
ID = IS{exp[VP/26mV]-1}
The resistor current, IR, the op amp input current, IO, and the feedback current, IF, are
given by Ohms law:
IR = VP/R
IO = VP/1012
IF = (VO VP) / RF = (2x105VP VP) / RF ~ 2x105VP/ RF = VP/20
Our current equation can now be simplified. Our first simplification comes from noting
that IR and IO are insignificant compared to IF (Since R is on the order of M and the
input resistance of the op amp is 1012). Next, we can simplify our diode current. Since
we expect an output voltage on the order of 1V, VP must be on the order of 10V.
Consequently, VP/26mV is much less than 1, and we can expand the exponential:
exp[VP/26mV] ~ 1 + VP/26mV
ID = IS{exp[VP/26mV]-1} ~ ISVP/26mV
When we consider that IS is usually on the order of nanoamperes, we find that ID is also
insignificant compared to IF. Thus we obtain a simple current equation:

17

IL = IF
Now we substitute the values of these currents:
0.26PI = VP/20
Finally, we obtain VO in terms of PI:
VO = 2x105VP = 2x1055.2PI = 1.04x106PI
3.4.2. Analysis of LED/LAP Interaction
The power of light incident on the photodiode (PI) is dependent upon the intensity (I0) of
the LED, the area (A) of the photodiode, and the distance (d) between the LED and the
photodiode. The intensity of the LED will be given in candelas (cd). To keep things
simple, we will use the approximation that the distance between the LED and every part
of the photodiode is the same. We will use the result shown earlier in section 3.2.1 for the
combined intensity emitted by the LED, which relates the total power emitted by the
LED, ITotal, to the power emitted at the peak frequency, I0:

We can determine the power that reaches the LAP from simple geometry:

Here, A is the active area of the LAP, and d is the distance between LED and LAP.
This ignores associated with adding filters in the light's path. It also ignores additional
losses that might be caused by the directivity of the LED where the power is not
uniformly distributed across the face of the LAP.
We can now combine this result with the formula from section 3.4.1 to obtain a value for
VO:

3.4.3. Analysis of Input Circuit

18

At this point, all we need is to determine the LED intensity (I0) in terms of the voltage
applied to the LED. Unfortunately, this relationship is non-linear. In addition, the
specifications we want are given as graphs rather than equations or equivalent
components, and this further complicates matters. A quick order of magnitude estimate
will shed some light on the situation. Assuming that A and d2 are of the same magnitude
simplifies things dramatically. We discover that to produce an output voltage on the order
of 1VDC, we need an intensity of 1mcd (millicandela).
In order to produce the low light levels required, the current supplied to the LED will be
very small. This region of operation is not shown in the product specifications. This limits
what may be accomplished with calculations, and forces us to rely upon experimentation
to obtain the desired operating voltages and currents.
3.4.4. Conclusions Based on Numerical Analysis
The primary factors that will determine the output for a given LED are the intensity of the
LED and the LED/LAP separation. The LED intensity will be controlled through the
choice of resistors in our circuit, and these values will be refined through further
experimentation. The LED will be operating at very low current levels to produce the
values needed for our tests. The LED/LAP separation will be a fixed parameter, and once
it is built into the design it will not change. Thus, the choice for this separation is a very
important element that we need to determine through viability tests.

4. Viability Testing
4.1. Experimental Set Up
Our goal at this stage was to ensure that the LED would operate normally in the region we
expect it function, and to ensure that there were no major surprises waiting for us. To this
end we set up an extremely crude version of our apparatus. A variable resistor was placed
in series with a standard power supply in order to create a variable power supply. This
was used to supply power to our LED, while a meter was attached to measure the voltage
across the LED. This measurement is what we have called the input voltage. The S359003 LAP, which has an active surface that is 10mm on a side, was attached to an amplifier
circuit. This pair was then laid in the light tight box facing the LED. A second power
supply was used to power the amplifier, and its output was run to a second voltage meter.
This voltage is what we have called the output voltage. The LAP/LED separation (d) was
set using a ruler. The alignment of the two components was made by visual inspection
while the components lay on their sides in the box. The LAP and LED were not secured
in these positions.
4.2. Data from Viability Tests
19

We took measurements for separations of 5cm, 7.5cm, 10cm, and 12.5cm. For each
distance the input voltage was varied from 2.4VDC (the lowest our setup would allow),
and increased until our output voltage reached the level of amplifier saturation (around
9.7VDC). The results from this test are shown in table 1, and this data is also presented as
a plot in figure 10.

Table 1. Data from Viability Test


This same data was used to make a second plot, which is shown in figure 10. A few
additional measurements were made at 10cm separation to complete this chart. On this
graph the output voltage is plotted against the inverse square of the distance of separation.
Our expectation was that this would produce nice linear results. Instead we found a
distinct curvature to the lines. We believe that this curvature is a direct result of effects
occurring at the edges of the LAP.
In the analysis we assumed that the distance from the LED to every point on the LAP
surface was the same, and the validity of this assumption grows weaker as the separation
shortens. Another effect is caused by the directivity of the LED. As the separation grows
smaller, the area of the LAP begins to extend outside the region most strongly illuminated
by the LED. This causes the voltage to become nonlinear as the separation grows smaller.
4.3. Conclusions from Viability Testing
The viability tests provided useful insights. First and foremost, it has shown that the basic
concept is sound. The photodiode/amplifier combination is very sensitive and requires
little illumination to produce useful results. This opens the possibility of using a weaker
(and hopefully less expensive LED). As the separation nears 5cm, we begin to
concentrate power upon the center of the LAP, resulting in weaker testing of the edges
and corners of the LAP. We can compensate for this effect with a diffusion filter (or a
20

mixer) or by changing to an LED with a broader illumination. Of course, this effect also
disappears by increasing the LED/LAP separation. We have decided to operate at a
separation of 5cm. At this range the separation is large enough that fringe effects are still
minimal. It is also large enough to allow for the insertion of a filter holder and filters. At
this range, the LED is strong enough that any losses in such filters will pose no serious
problem.

Figure 10. Voltage Gain with LAP/LED Separation.

21

Figure 11. Output Voltage vs. LAP/LED Separation.

5. Prototype Construction
5.1. Design of Mechanical Components
The design of the mechanical components began with the LAP and proceeded outward.
The LAP cap was designed around the photodiode. Next, the LED cap was made to
match up with the LAP Cap, and then the central shaft was designed to hold both caps.
Finally, the main chassis was designed around the central shaft.
5.1.1. LAP Cap
Checking the dimensional outlines of the various LAPs gave us the minimum size
required for the LAP frames, which are designed to hold the LAPs in precise position. In
order to accommodate three different sizes of photodiodes we need three different frames.
The three frames were made with the same outer dimensions and placement of retaining
22

screw holes, making them interchangeable. The mounting surface was recessed to allow
the LAP frames to be positioned correctly.
Making our cap function with any of the three sizes of LAPs created another problem.
The placement of the pins is different for each size of photodiode. The different pin
placements were confined to a small enough space that using separate connections for
each LAP size would have been problematic. Instead we elected to make the connections
by hand, after the LAP and frame had already been attached to the cap. Since the
separation between the two pins was the same for each size, a two-pin connector was
used to make the connection.
A second layer, to which the amplifier would be attached, was placed a short distance
behind the first. Placing the amplifier as close to the LAP as possible will minimize the
effects of noise. Any noise entering the system prior to amplification would be amplified
along with our signal. Despite this, enough space was kept in between so that we could
make our manual connections.
Both layers are held in place inside a rectangular tube. This tube would fit snugly inside
the central shaft. The overlap of end caps and central shaft would help to eliminate any
possibility of light passing in between and reaching the central shaft. Rails were allowed
to extend along opposite sides with the intention of having them slide into grooves inside
the central shaft. This would help us to precisely position the LAP cap.
A retaining pin was designed to pass through both tubes, the LAP cap, and the central
shaft, as another aid to LAP cap positioning. A plate was installed inside the LAP cap to
isolate the retaining pin holes from that end of the light tight chamber.
A second pair of holes were added, one to allow us to bring in power to the amplifier and
one to take our signal out. We elected to make external connections so that we can ensure
good electrical connections. Our initial intention was to fill these holes after our signal
and power lines were passed through, but it turns out that these holes do not degrade the
light-tight nature of our central shaft.
A plate was placed over the rear of the cap to seal it, and a rim was left extended around it
to provide something to grip when removing the cap. The interior surface was painted flat
black to minimize reflection. Finally, decals were added to identify the lines leaving the
LAP cap and ensure its proper orientation. A scale drawing of the LAP cap is shown in
figure 12.

23

Figure 12. Scale Drawing of the LAP Cap.

5.1.2. LED Cap


The LED cap was designed to use the same rectangular tube shape as the LAP cap. In this
manner, it would slide into the same grooves from the other side of the shaft. It was left
shorter since there was less need for space. A single plate acts as a mounting surface for
the LED, and it will also act as a mounting surface for a filter holder should one be added.
The LED mount securely holds the LED in place. The line that carries power to the LED
mount passes through a hole in the side. As with the LAP cap, a retaining pin holds the
LED cap in place, and a plate covers the back side of the LED cap with a rim left as a
grip. In order to minimize reflections, the interior surface was painted flat black. Decals
were added to identify the line leaving the LED cap and to ensure that it is oriented
correctly. A scale drawing of the LED cap is shown in figure 13.
24

Figure 13. Scale Drawing of the LED Cap.


5.1.3. Central Shaft
The central shaft was designed as a rectangular tube, large enough so that both the LED
and the LAP caps would fit inside. Grooves were positioned to align with the rails on the
caps, allowing the caps to be precisely positioned. Inside the shaft, a rectangular stop
was placed on opposite sides. This would catch both the LED and the LAP cap, and
ensure that they were appropriately separated. Extensions were left on either end to allow
the retaining pins of each cap to pass through. A scale drawing of the central shaft is
shown in figure 14. A similar drawing showing how the LED and LAP caps connect is
shown in figure 15.

25

Figure 14. Scale Drawing of the Central Shaft.


26

Figure 15. Scale Drawing of the Central Shaft with Caps Attached.
27

5.1.4. Main Chassis


The primary consideration for the size of the main chassis was the size of the components
that needed to be within it. The width was designed to ensure that the installed LAP and
LED caps would extend no further than the edge of the main chassis. The height was
chosen to leave extra space below the shaft to allow wires to pass beneath. The depth was
chosen to allow room for the power supply behind the central shaft and room for the
controls on the front panel. The front panel was designed to allow installation of all of the
controls and displays, with the exception of the input power selection switch, which is
located on the back panel. The power input and fuses were also placed on the back panel.
Holes were also added to the back panel to vent any excess heat generated by the power
supply. Decals were added to identify the electrical connections on the surface, as well as
identifying all of the controls and displays, and rubber feet were added to the bottom
surface to keep the metal box from destroying the surfaces upon which it sits. We named
the prototype LAPLACE, which stands for large area photodiode luminance analysis
calibration equipment. A perspective drawing made to scale from the dimensional
outlines for the components was shown in figure 2.
5.2. Construction of Mechanical Components
The LED and LAP caps, the main chassis, and the central shaft were all machined to a
tolerance of one thousandth of a inch in the machine shop of the UTA physics department
by Jimmy Hanhart utilizing a Trump B6EC. Wallace Lutes and I performed additional
machining.
5.3. Selection and Purchase of Electrical Components
Major electrical components, [and by this I mean the expensive components] were
selected by comparison shopping. Minor components were added to these orders to keep
shipping costs to a minimum. A list of the electrical components and the suppliers of
those components are shown in appendix A.
5.4. Assembly of the Prototype
I assembled the prototype in the UTA high energy physics electronics lab. Armen
Vartepatian assisted me by soldering the most difficult connections. Onder Anilturk gave
me additional support by performing some soldering while keeping me company in the
lab late at night.

28

5.5. Troubleshooting
A number of difficulties arose during the construction of this prototype. What follows is a
short description of each problem and how we resolved them.
5.5.1. Readout Fluctuation
The first problem encountered were small, random fluctuations in the readout. Initially it
was thought that the cause might be the absence of a load resistor. A one mega-ohm load
resistor was installed, but this did not solve the problem. We discovered that the problem
occurred in the power provided to the digital panel meter. Originally we had supplied
power to the digital panel meter via a dedicated voltage regulator, which received its
power from the units power supply. Unfortunately, the meter required an independent
source, completely isolated from the signals power supply. This was solved by adding a
9V battery with the sole purpose of providing power to the meter. The previous diagrams
include this modification.
5.5.2. Low Output
The next difficulty was that the meters readings were significantly different from our
expectations. The resistors selected by the intensity control and the LAP size selection
control were chosen to achieve the appropriate output levels for each setting. When we
tested photodiodes, our reading were less than half of what we had intended. We found
that the original multiplexer design allowed current to flow through more than one
resistor by flowing back through other resistors. This was solved by the installation of
diodes in series with each resistor in order to prevent these currents. These diodes are
shown in the previous diagrams. Addition of the diodes forced us to change the values of
all of these resistors. The output values still differ from our desired levels, and we believe
that the new resistor values are at fault. It was decided that correcting these resistors
would be postponed until a later date since the output range is adequate for our
measurements.
5.5.3. Internal Capacitance
Our next hurdle was a capacitance problem. When testing the stability of our
measurements with time we discovered that the output values would slowly fall. After
taking readings for two weeks we found that this decay was very close to an exponential
function with a time constant of about 6 days. This data is shown in figure 16.
The large time constant could only occur with a large value of resistance. Consequently,
we examined the circuit around the largest resistors. It was found that the connection
between the one mega-ohm load resistor and ground was loose. At this time we realized
that the amplifier circuit has a built-in ten kilo-ohm load resistor, making the one mega29

ohm load resistor redundant. We removed the redundant resistor, and this solved the
problem.

Figure 16. Exponential Decay of Output Voltage.


5.5.4. LAP Connection
The connector initially used to connect to the LAP was designed to slide onto the ends of
the pair of photodiode pins. We found that the fragile pins were bending during insertion
of the LAP. It was obvious that repeated bending would eventually break the pins. To
avoid this, new test leads were installed. The new connectors attach individually to each
of the pins, and these connections are made by hand. These new connections do not
damage the LAP pins.
5.5.5. Temperature Gauge
In order to determine if temperature would have a significant effect upon testing, the unit
was placed inside a refrigerator in order to cool it before measurements were taken. A
significant change in our readings was discovered. In order to compensate for these
temperature effects, a panel mount thermometer was purchased and installed on the unit.
Once the temperature dependence is determined, we can make corrections to the test data.

30

6. Testing the Prototype


6.1. Testing Light Tight Chamber
In order to ensure that the light-tight chamber would not allow any light to enter, the dark
current was measured while a 500W halogen light was directed at the unit. Measurements
were taken with the light at various orientations. In each case, there was no change in the
output voltage measurement. We can conclude that even the brightest of external lights
will have no effect upon our measurements.
6.2. Temperature Testing
We subjected the test unit to temperatures over a range from 9C to 30C (48F to 86F).
The low temperatures were obtained by placing the unit in a refrigerator. Two sets of data
were taken under these conditions. The data was taken with no illumination from the
LED in addition to all five intensity levels. To obtain the readings at the higher
temperatures, the unit was placed near a heat lamp. An additional set of data was taken
under these conditions. The three sets of data were mapped on the same graphs with one
graph for each intensity setting. An examination of these graphs, figures 17 through 22,
shows unusual behavior. It was hoped that we could make a linear approximation that
would reasonably fit this data, but the behavior precludes this simple type of correction.
Instead, we will use a table of multipliers generated from the data. We will use one
multiplier for each even value of temperature, and we will interpolate for decimal values.
We will use 22C (72F) as our base temperature, giving no correction (a multiplier of 1).
These multipliers are shown in table 2.
6.3. Repetition of Photodiode Installation
In this test, we took measurements for each of eight different LAPs. After measurements
were taken at each of the six possible illumination levels, the photodiode was removed
and replaced with a different LAP. This ensured that each photodiode was completely
reinstalled after each measurement. To vary the test conditions as much as possible,
measurements were taken in three separate sessions on three separate days in two
different locations. Nine measurements were taken for each photodiode.
6.3.1. Data from Repetition tests
The data from repetition testing is shown in tables 3 and 4. It has been sorted by
photodiode to allow comparison of the results. The column on the left identifies the
photodiode under test. Those that begin with S are the smallest photodiodes, S3590-03,
which are numbered consecutively. Those that begin with M are the intermediate sized
photodiodes, S2744-03. Those that begin with L are the largest of the LAPs, S6337-01.
31

Figure17. Output Voltage Temperature Dependence for Dark Current.

Figure 18. Output Voltage Temperature Dependence for Setting 1.

32

Figure 19. Output Voltage Temperature Dependence for Setting 2.

Figure 20. Output Voltage Temperature Dependence for Setting 3.

33

Figure 21. Output Voltage Temperature Dependence for Setting 4.

Figure 22. Output Voltage Temperature Dependence for Setting 5.

34

Table 2. Temperature Corrections


The next two columns show the session and temperature, respectively. Beside these are
six columns of data, one for each possible intensity setting and one with no illumination.
Each photodiode has nine rows of data beside it. The data in each row was taken
consecutively, and there was at least a half hour delay before the next set of
measurements were taken on that photodiode.
6.3.2. Data Analysis
Compilation of the data prior to temperature corrections shows less than desirable results.
This is shown in table 5. We expect the best results to come from the highest illumination
setting due to the fact that we will have the highest signal to noise ratio prior to
amplification. Consequently, we will focus our attention on measurements taken at the
highest setting. The high intensity data shows an average standard deviation of 2.12%,
which is larger than desired. Better results are expected when temperature corrections are
taken into account. Graphs of the high intensity data are shown in figures 23 and 24.
Although the graphs in figures 23 and 24 drift over a large range, the curves seem to
follow the same trends, rising and falling together. In addition, the data points for each
session, separated by the dashed lines, seems to be grouped together. These general trends
could be the result of some aspect of the environment. Temperature corrections were
included for the compilation of data in table 6. The high intensity data now shows an
average standard deviation of 1.89%. This is an improvement over the data taken without
correction, but still falls short of our goals. Graphs of the temperature-corrected data are
shown in figures 25 and 26.
35

Table 3. Data from Repetition Testing of the Small LAPs


36

Table 4. Data from Repetition Testing of the Large LAPs


37

Table 5. Compilation of Data without Temperature Corrections

Figure 23. High Intensity Measurements of Small LAPs.


38

Figure 24. High Intensity Measurements of Large LAPs.

Table 6. Compilation of Data after Temperature Corrections


39

Figure 25. Temperature Corrected Data for Small LAPs.

Figure 26. Temperature Corrected Data for Large LAPs.


40

The temperature correction seems to work well for the small photodiodes. Though the
first two points, the data from the first session, do not match up with the other data points,
they agree with each other. This may indicate that another environmental factor is
affecting our data. The temperature corrected data for the large photodiodes, however,
seems to be worse than the uncorrected data. This indicates that the temperature
correction, derived from and used successfully on the small photodiodes, cannot be
applied to the data for the medium and large LAPs.
It appears that the data taken in each session was more closely related than the overall
standard deviation. A compilation of the data by session is shown in tables 7 and 8. When
the data is separated out on a session to session basis, the high intensity data has an
average standard deviation of 0.64%. Temperature corrections were made, and this data is
shown in tables 9 and 10. The high intensity data then has a standard deviation of 0.69%.
It should be noted that the data improves significantly for the small photodiodes, and the
standard deviation is approximately 0.40%. The same correction applied to the larger
photodiodes makes the data worse. Once again, this indicates the need for a separate set
of temperature corrections for the larger photodiodes.
As suggested earlier, the correlation of the data from session to session indicates that
there may be additional environmental factors that are affecting our readings. The shift in
our data from one session to the next may be caused by a single factor, such as humidity
for example. It may be possible to make corrections similar to those made for
temperature.
One possible method of taking all possible environmental factors into account is to
simply compare the performance of the photodiode being tested with the performance of
a control photodiode. We can approximate this to a limited degree in the data already
taken by treating the first photodiode as our control. Since the temperature drifted over
the measurements, temperature corrections are included. This is shown in table 11.
The standard deviation of the ratio of the high intensity readings is less than 1.00% for
most of the photodiodes. It is curious that the standard deviation deteriorates as we move
down the chart. This may be a result of the delay between the testing of the photodiode
and the control. Since the diodes were tested in order, the further down the diode is listed,
the larger the delay between its test and that of the control.
To present this data in the form of a graph, we have taken the ratio of these data points
and scaled them by a factor of 3.20 in order to compare them easily with previous graphs.
These graphs are shown in figures 27 and 28.

41

Table 7. Compilation of Small LAP Data by Session without Temperature Corrections


Improvement in the small LAP data is clearly evident. In the case of the large LAPs, we
see some improvement, but the data is still very erratic. Once again, this indicates that we
need to consider the larger diodes separately. Figure 29 shows the ratio of the large
photodiode data to that of M1. This brings about a dramatic improvement in our data
taken from the other two medium sized diodes. The data from the large photodiode,
however, doesnt show much improvement.
From this data, we might estimate that simultaneous testing (or near simultaneous testing)
of the photodiodes and a control photodiode of the same size will produce standard
deviations near 0.50%.

42

Table 7. Compilation of Large LAP Data by Session without Temperature Corrections


6.4. Testing Operation with 220VAC Input Power
The unit was set up to accept 220VAC as its input power source and a few measurements
were taken to ensure that it was operating correctly under the higher voltage.

7. Conclusions
The test unit is functioning as designed, and by comparison with a control we exceed the
precision we had sought. In normal operation, straight measurement of the diode
response, it gives results good to 1.89%. This falls just short of the precision sought. By
comparing our measurements with those of a control we can give results good to
approximately 0.80%, which exceeds our expectations.
43

Table 9. Compilation of Small LAP Data by Session with Temperature Corrections


There are encouraging signs that the results from normal operation can be improved.
First, we saw that when the data is separated by session our results improved to within
approximately 0.60%. This may indicate that other environmental factors are affecting
our data. If these can be accounted for our results in normal operation should improve
significantly.
By increasing the high intensity output to a value closer to the maximum amplifier output,
perhaps outputs of 8V or so, will increase the signal to noise ratio prior to amplification.
This should create some improvement in out results.

44

Table 10. Compilation of Large LAP Data by Session with Temperature Corrections

Table 11. Average Ratio of Diode Measurements and Standard Deviation

45

Figure 27. Small LAP data adjusted by comparison with S1.

Figure 28. Large LAP data adjusted by comparison with S1.


46

Figure 29. Large LAP data adjusted by comparison with M1.


It is possible that some of our error stems from output fluctuations from the source LED.
Increasing the LEDs output might shift it to a more stable level. By adding an attenuation
filter we can increase the LEDs output power while maintaining the amount of light
incident on the surface of the LAPs. This may also improve our results.
Making separate temperature corrections for each size of photodiode will improve the
testing of the larger photodiodes. Building a separate LAP cap with a permanently affixed
control LAP could reduce the error obtained for comparison testing.
By making all of these modifications, we should be able to improve our accuracy. It is
possible that this could reduce our error into the range of 0.40%0.50%.

47

Appendix A. Electronic Parts List


Part
AMP
BTH
BTT
CRD
DPM
F01
F02
F03
K02
K04
K05
K06
LD1
LD2
LD3
LP1
LP2
LP3
PS1
S01
S02
S03
S04
S05
S06
SKT
THM
VR1
VR2
VR3

Name
Amplifier
9V Battery Holder
9V Battery
Power Cord
Panel Meter
110VAC Fuse
220VAC Fuse
Fuse Holder (2)
Power Selector Knob
Intensity Control Knob
Area Selector Knob
Meter Scale Knob
Source LED
+15 Power Indicator
-15 Power Indicator
100mm2 LAP
200mm2 LAP
400mm2 LAP
Power Supply
Power Switch
Power Selector Switch
DC Test Switch
LED Intensity Control
LAP Area Selector
Meter Scale Switch
AC Power Inlet
Thermometer
Input V Regulator
+Output V Regulator
-Output V Regulator

Description
LAP output amplifier (UTA)
Radio Shack #27-326
Duracell 9V Battery
IEC-US Power Cord (Standard Computer Cord)
4 Digit LCD Digital Panel Meter (MPJA#7160ME)
0.50A Fuse, 5mm x 20mm
0.25A Fuse, 5mm x 20mm
Bussman HTB-62M for 5mm x 20mm fuses
Mouser #5164-1510
Mouser #5164-1500
Mouser #5164-1500
Mouser #5164-1510
Nichia #NSPB310A
Mouser #351-5004
Mouser #351-5004
Hamamatsu #S3590-03
Hamamatsu #S2744-03
Hamamatsu #S6337-01
International Power IHAD15-0.4 (MPJA#6638)
Rocker Switch On/Off (Mouser #107DS850S-22)
Rotary, 2 Pos/3 Poles, Non-Short (Mouser #10WA166)
Miniature Toggle Switch (Mouser #10TA420)
Rotary, 5Pos/1 Pole, Non-Short (Mouser #10WA164)
Rotary, 3 Pos/1 Pole, Non-Short (Mouser #10WA164)
Rotary, 3 Pos/3 Pole, Non-Short (Mouser #10WA166)
IEC Plug (Mouser #161-0707-7-187)
Taylor 9940 Panel Mount Thermometer
+5VDC Voltage Regulator (511-L7805CV)
+12VDC Voltage Regulator (511-L7812CV)
-12VDC Voltage Regulator (511-L7912CV)

48

Appendix B. Specifications
B.1. Large Area Photo Diode Specifications
B.1.1. S3590-03 and S2744-03 Data

49

B.1.2. S3590-03, S2744-03, and S6337-01 Dimensional Outlines

50

B.2. NSPB310 Light Emitting Diode

51

52

B.3. International Power IHAD15-0.4 Power Supply


Marlon P. Jones Associates SN: 6638
Input: 110/220 VAC 47-63Hz
Outputs:
#1: +15VDC@0.4A
#2: -15VDC@0.4A
+/- .05% Regulation
4.87x4.00x2.10

Open frame linear supply


Foldback current limit
Remote Sense
No Minimum Load
Weight: 2 lbs.
123.70mm x 101.60mm x 53.34mm

B.4. Digital Panel Meter


Marlon P. Jones Associates SN: 7160ME
4 digit LCD display
200.0mVDC input
0.05% accuracy
Adjustable decimal point
2 readings per second
Holes on board to add scale resistors
2 x 1 x

53

character height
9VDC powered
Input impedance > 100M
Auto polarity
Through panel mounting
WT: 1 lb.

B.5. Other Electronic Components

54

Appendix C. Dimensional Outlines

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

72

73

74

75

76

77

78

79

80

81

82

83

84

85

86

87

88

89

90

91

92

93

94

95

96

97

REFERENCES
1. P. Horowitz and W. Hill, The Art of Electronics, Cambridge University Press
(1980)
2. Mohammed S. Ghaussi, "Electronic Devices and Circuits: Discrete and Integrated",
Holt, Rinehart, and Winston (1985)
3. The manual of "Photomultipliers Data Handbook", Philips Corporation (1990)
4. M. Bosman et al., Development of a Scintillator Tile Sampling Hadron Calorimeter
with Longitudinal Tile Configuration, CERN/DRDC (1993)
5. ATLAS Technical Proposal, CERN/LHCC (1994)
6. E. Berger et al., Construction and Performance of an Iron-Scintillator Hadron
Calorimeter with Longitudinal Tile Configuration, CERN/LHCC (1994)
7. The manual of "Silicon Photodiodes and Charge Sensitive Amplifiers for Scintillation
Counting and High Energy Physics", Hamamatsu Corporation (1995)
8. The manual of "Photodiodes", Hamamatsu Corporation (1995)
9. Z. Ajaltouni et al., Response of the ATLAS Tile Calorimeter Prototype to Muons,
CERN/PPE (1996)
10. ATLAS Tilecal Collaboration, ATLAS Tile Calorimeter Technical Design Report,
CERN/LHCC (1996)
11. ATLAS Collaboration (Calorimetry and Data Acquisition), Results from a
Combined Test of an Electromagnetic Liquid Argon Calorimeter with a Hadronic
Scintillating Tile Calorimeter, CERN/PPE (1996)
12. ATLAS Tilecal Collaboration, Technical Specifications of the Test Bench for the
Test of the ATLAS Tile Calorimeter Photomultipliers, CERN/LHCC (1996)
13. V. Reece, ITC Submodule Assembly Handbook, CERN (1998)
14. ATLAS Collaboration (Calorimetry and Data Acquisition), Results from a new
Combined Test of an Electromagnetic Liquid Argon Calorimeter with a Hadronic
Scintillating Tile Calorimeter, CERN/PPE (2000)

98

You might also like