1. Building a cosmic ray detector
Samuel Haste
73000470
School of Physics and Astronomy
The University of Manchester
Fourth year MPhys Project Report
May 2012
This experiment was performed in collaboration with Lawrence Fsadni.
Abstract
A cosmic ray detector was designed and built to show the path of cosmic ray muons in the
form of a spark chamber. The spark chamber consists of 15 stainless steel plates, two scintillators
with avalanche photodiodes attached and is placed in a glass bell jar filled with a Helium/Neon gas
mixture.

2. 1. Introduction
Cosmic ray detectors have been used in science research since the 1960s and less so in recent
decades. Their use has gradually changed and they are often used as an easy way to visualise
cosmic rays for the expansion of science outreach; to this end one complete spark chamber was
built. The original plan was to build two spark chambers but order enough components for three
with the two chambers to be displayed in the Schuster building and the Museum of Science and
Industry (MoSI) to increase science outreach to the public. The third set of components was ordered
to act as spare parts should anything happen to the components for the first two chambers during the
building and testing stages. One full set of components has been created which are in the complete
spark chamber. Due to longer than expected lead times for components, the second and third sets
were not produced together with the first set and are ready to be made when more chambers want to
be built.
Two groups are working on this project. My lab partner and I focussed on the design of the
spark chamber, the electronics and its construction while the other pair has focussed so far on the
use and adaptation of avalanche photodiodes and scintillators.
2. Theory
2.1. Cosmic rays and Muon production
Cosmic rays are high energy particles that travel through the vacuum of space. They come
from sources such as mechanisms within stars and can have energies in excess of 1020
eV which is
much greater than the maximum energies achievable with Earth based accelerators. For this reason
they are still an active area of research and can still provide ways of conducting very high energy
studies. They are primarily made up of protons and helium nuclei with a small portion ~1% made
up of heavier nuclei.
As the high energy cosmic rays collide with the upper atmosphere a cascade of particles is
produced, known as an air shower (Fig 1). High energy cosmic rays produce pions and neutrons
when they collide with particles in the atmosphere. The charged pions then decay to muons and
neutrinos with the muons being the particles that are detected at sea level in the spark chamber.
Muons have a lifetime of 2.2 µs which even at the speed of light by conventional dynamics
gives the distance travelled in its lifetime as 2200  10-9
s  3  108
m s-1
= 659.5 m. This is not a
great enough distance for the muon to reach sea level in its lifetime to be detected. This suggests
muons produced by cosmic rays must be affected by special relativity and show evidence that time
dilation occurs in their reference frame as observed from the Earth. This means that they must have
energies much greater than their mass 105.7 MeV / c2
so that they can travel near the speed of light,
which is also backed up by the fact that muons need 2-3 GeV of energy to penetrate down to sea
level.

3. Figure 1. A diagram showing a high energy cosmic ray proton colliding with an atmospheric molecule producing a
shower of pions and neutrons. The charged pions then decay into a muon and its associated neutrino, the muon being
the particle detected in a spark chamber.[1]
At sea level the expected flux of muons ~1 cm-2
s-1
. This flux is at a sensible level for us to
count and produce a spark rate that is interesting to the viewer.
2.2. Spark Chambers
The chosen design, after research, for the particle detector was a spark chamber. They have
one of the most pleasing visual effects and can be used to explain cosmic rays, particle physics and
can be used as in §2.1. to lead into discussion about relativity if that level of understanding is
sought by an individual looking at the exhibit. The type of spark chamber chosen consists of plates.
Some are constructed with wire for finer measurement but plates are easier to install, maintain and
look at.
Spark chambers consist of plates, scintillators, photomultiplier tubes or other photon detector
and a chamber filled with a gas mixture. The output signals from the photon detectors trigger the
spark to be fired. A schematic is shown in figure 2.
When a muon enters the gas it ionises the molecules and leaves behind it a trail of ions. As the
muon travels quickly and the trigger pulse follows shortly after, the ion trail will not have time to
diffuse or regain electrons and so will remain as a distinct line. The muon first hits the scintillator,
discussed in detail in §3.1. where it causes photons to be released and these are detected by the
chosen type of detector, avalanche photo diodes in this case. The muon then carries on through the
gas passing through all the plates leaving the ion trail. Then it travels through the second (bottom)
scintillator and causes photons to be released in this one and a signal to be sent by the detector
exactly as in the top scintillator.
The signals from each of the detectors then travel to a coincidence circuit where if they
coincide a signal is sent out to the trigger unit. The trigger unit causes a large voltage to be put
across the plates with a short rise time; enough to break down the gas along the path of least
resistance. The path of least resistance is the ion trail left by the muon. The spark formed therefore
jumps between plates via the ion trail and reveals the path taken by the cosmic ray muon. This is the
direct visualisation of cosmic ray muons.

4. Figure 2. A schematic showing the main components of a spark chamber. The dotted line shows the path of the muon
though the chamber travelling though the scintillators, plates and gas mixture.[2]
3. Equipment
3.1. Scintillator
Scintillators produce photons when energetic particles pass through them. The energetic
particles collide with the atoms of the scintillator and produce excited states. These excited states
then decay via photon emission to lower energy states. The photons emitted reflect within the
scintillator until they are absorbed by an attached detector or the surroundings as shown in figure 3.
Avalanche photodiode and amplifier chipset.
Avalanche photodiode and amplifier chipset.

5. Detector and
amplifier.
Figure 3. A diagram showing how a muon enters a scintillator and collides with multiple molecules within the
scintillator. The yellow lines represent photons and the path they take with a number of them ending up at the detector.
A signal is then produced by the detector if photons are detected.[3]
There is a selection of materials available to use as the scintillator. A plastic scintillator was
provided for the first investigations and plastic was chosen as the scintillator material to use in this
project. Plastic is easier to cut and obtain, easier to clean and maintain all meaning plastic will make
the best choice for the brief of this project.
Photons can be lost from the scintillator to the surroundings meaning fewer photons are
detected. To help reduce the loss of photons all scintillators have been prepared by cleaning their
surfaces, covering them in silver foil as smooth as is possible and then wrapping the entire
scintillator and detector set up in black insulating tape as seen in figure 4. The foil increases the
number of photons reflected adding to the number that will be reflected into the detector. The black
tape ensures that no photons can get into the detector from the surroundings so that only photons
produced in the scintillator should be detected, reducing noise.
Figure 4. Two pictures showing the original scintillator and detector being covered in foil and tape. As can be seen the
scintillator has a piece of Perspex attached to it so that the detector, in this case a photodiode, has a larger area to fix to
meaning it has a larger area to collect photons over. This scintillator and photodiode set up was used in the preliminary
stages of this project and as mentioned in §3.3. was not used in the final chamber.
3.2. Wavelength shifters
The sides of the scintillator are relatively large and the surface area of the detectors used is
small. This would lead to a large loss of photons that are not detected. In order to combat this
problem, wavelength shifters were fitted to opposite edges of the scintillators (figure 5) and the

6. detectors on the small ends of these so that any light incident on the side of the scintillator is
channelled to the detector, not just light incident on the detector itself had it been connected to the
scintillator directly. The wavelength shifter also changes the wavelength of the light emitted by the
scintillator. The scintillator emits light at 425 nm which the wavelength shifter changes to 625 nm.
The quantum efficiency of the detectors used peaks at 625 nm, thus converting a larger number of
photons into signal and giving a more frequent, reliable output. Wavelength shifters improve the
detection rate in two ways making them important parts of the design. The end of the wavelength
shifter that does not have a detector attached is coated in silver to aid the reflection of photons.[5]
Figure 5. A photo of the scintillator with wavelength shifters attached. A small air gap is present as that aids the photon
transmission.
3.3. Photodiodes
Photodiodes (PDs) are a small piece of equipment that can detect incoming photons. They can
be easily attached to scintillators and amplifiers making them suitable for use in a self-contained
spark chamber.
Photodiodes are photodetectors that turn incident photons into an electric current. They
operate using a PIN junction rather than a p-n junction used in classical photovoltaic cells. PIN
junctions have a faster response time and are better for use in photodetectors the p-n junctions.
Within the photodiode the PIN junction is reversed biased meaning the PIN diode does not allow
current to flow.[4]
There is a noise level of ‘dark current’ which covers all current produced when
there are no incident photons. One source of dark current is background radiation, meaning a small
current does flow in reality. Dark current needs to be minimised so that the signal to noise ration of
the photodiode is increased. The photodiode is powered by the amp chip set it is attached to.
When an incident photon hit the ‘I’ region of the diode it causes an electron/electron-hole pair
to be formed. The reverse bias causes the electron-hole to be accelerated out of the ‘I’ region and
causes a current to be produced. The produced current is therefore proportional to the number of
photons.

7. 3.4. Avalanche Photodiodes
Avalanche photodiodes (APDs) work in the same way but at a higher bias voltage. This
means that as the electron-hole is accelerated it gains more kinetic energy. As the electron-hole
collides with molecules in the scintillator it causes more electron/electron-hole pairs to be produced.
This chain reaction or avalanche effect increases the charge from an incident photon and creates a
much larger output signal than in a normal photodiode in the same way as photomultiplier tubes
work appendix A. Avalanche photo diodes were worked on by the other group[5]
and so far have
been successful at detecting cosmic ray muons and are working well in coincidence with the PMT
set up.
3.5. Photomultiplier tube
The testing of the APDs involved comparison to a known cosmic ray flux. For this purpose
photomultiplier tubes (PMTs) attached to scintillator paddles were used. PMTs are sensitive to very
low numbers of photons and are able to detect down to single photon levels. This makes them ideal
for spark chambers and on large scintillators where a large portion of photons may be lost to the
surroundings. They are however big and would not fit inside the chamber of a spark chamber easily,
hence their use only as a testing set up. They are also relatively heavy and require large voltages to
operate. Spark chambers that were researched all used PMTs as the chosen detectors and they were
situated outside the spark chambers above and below the gassed cavity. As the design for this spark
chamber requires it to be free standing and self-contained PMTs do not fit the project brief.
For information on how PMTs work, see appendix A.
3.6. Amplifier
Each APD was plugged into and amplifier chip board which included a discriminator. These
were worked on and designed by the other group[5]
. Figure 6 shows an APD attached to an
amplifier.
Figure 6. A photo of the APD (far left) attached to the amplifier chip. The 5 V and 40 V inputs can be seen along with
the analogue and digital outputs.
4. Photodiode testing
This section is Semester 1 work, but is given as background to the research and for
information to future students.
4.1. Tests of the scintillator, photodiode and amplifier chipset
To test the photodiode it was attached to the scintillator (figure 4) and placed between the two
larger scintillator paddles which are attached to photomultiplier tubes (PMTs). The signal from the
photodiodes is passed into an amplifier chip board. Once the output signal has been amplified it is
then sent to a coincidence unit.
The first test was to link the PD output to an oscilloscope to try and observe a signal. The PD
output was very noisy. Then a test with the PD and amplifier setup within a metal box was

8. conducted, again with high levels of noise. Testing with a radioactive source was carried out so as
the see if any signal could be seen at all and it was not. This meant that the initial scintillator or the
PD/amplifier board was damaged.
After finding the minimum incident photons needed to trigger a signal by using a pulsed laser
it was calculated that incoming muons would not have deposit enough energy to trigger a detection
in the photodiodes. This means that the photodiodes are not suitable for the detection of cosmic ray
muons in this scintillator and so cannot be used in this project.
For detailed calculations and results see appendix B.
5. Designs
Once it was established that incoming muons do not have enough energy to trigger the
photodiode, the next stage of the Gantt chart was to design the spark chamber, its set up,
construction and submit drawings to the drawing office to be drawn in CAD so that the workshop
can produce the components. The main criteria to be considered in terms of the design are: that is
has to look good as it’s for display, it has to be easy to repair and source spares and there should be
the option at a later date to install photo-chips in order to photograph sparks.
Chambers already built by Birmingham and Cambridge were studied to understand how
people had approached spark chamber construction before.
5.1. Birmingham design
The Birmingham design consisted of many individual modules that interconnected to build
the chamber. Having this modular construction means that there is no overall larger gassed chamber
and that repair would be relatively easy as individual modules can be replaced. Between each
module there was a potential difference of 4.8 kV which is a high voltage even for spark chamber
standards when added across multiple modules. The modules were made of Perspex with
aluminium sheets contained, both of which are materials easy to source and manipulate. Figure 7.
The pros of this design are that it should be easier to repair due to its modular nature and that
the plates cover a large surface area thus more chance of a muon passing through per time. However
this design had a constant gas flow and was not left free standing and sealed by itself. The design
for this project needs to be a free standing chamber that does not need to be permanently gassed.
The scintillators were also connected to PMTs which seems to be the standard photodetector used
which is not going to be used in this project. The Birmingham design has quite closely placed plates
and as they are big any photo device would need to be placed away from the chamber.
Figure 7. A photograph of the Birmingham spark chamber showing the modular nature of the detector along with the
black scintillator paddles at the top and bottom which were connected to PMTs.[8]

9. 5.2. Cambridge design
The Cambridge design uses sheets, but not encased in Perspex modules. The plates are
supported at the corners by metal rods and are held apart by plastic insulating cylinders. The
construction of the box is square and made of metal with a Perspex sheet covering one side for
viewing. The unit is self-contained with the scintillators inside the metal sections at the top and
bottom of the chamber, however as a unit it is not very aesthetically pleasing. Figure 8.
This model was taken to science fairs to expand science outreach. But when on display was
encased in a display board such that only the Perspex viewing side was visible. This made it appear
very nice to look at but encasing it would not be a viable option at MoSI.
The chamber was a sealed standalone unit, but it was not left for extended periods of time
without gas replacement. Again for this chamber the issue of difficult photography arises.
Figure 8. A photograph showing the Cambridge spark chamber out of its display stand. This is the way the spark
chamber will be displayed in this project and as can be seen, the Cambridge design is not particularly engaging to look
at.[6]
5.3. Princeton design
This design was the basis of the project. It was the first design shown and has the aesthetics
that this project requires. Figure 9. It uses a glass bell jar rather than a metal or Perspex box which
makes it more pleasant to look at. It appears to be standalone, however the scintillator paddles are
situated outside of the glass bell jar above and below and the PMTs are out of the picture. It is also
constantly gasses and is held down by the springs and blue straps that can be seen in the picture,
which adds to obscure the view.
Figure 7. A photograph showing the Princeton spark chamber. It shows the glass bell jar, the external scintillators and
the spring/strap mechanism for making and air tight seal around the rim of the bell jar.[9]

10. Taking into account the designs from Cambridge, Birmingham and Princeton design models
were created. The images of the designs are discussed below in the evolution of the concept
followed by the evolution of the individual components.
5.4. Overall spark chamber concept
At first the idea was to build the chamber based around the Cambridge model, so square in
nature with square plates and an easy to construct shell. After consultation with supervisors, it was
decided that a glass bell jar should be used as in the Princeton design. The advantages of using a
bell jar are that there is only one edge that needs to be sealed around the base, they can be
purchased rather than having to be constructed and they are nice to look at which is important for
the display quality.
As the design of the chamber was round, it was decided to try and keep things as circular as
possible including the plates because circular plates in a bell jar will cover a greater surface area
than square plates will.
Unlike all the other chambers, the scintillators also need to be inside the chamber with
photodiodes rather than situated outside with PMTs. Also the chamber needs to be sealed
completely and not have a constant flush of gasses into it.
5.5. Towers
As in the Princeton design, the chamber is supported by towers. Originally one tower was
proposed at the back of the chamber but the use of three towers was settled upon. The use of three
towers gives symmetry, stability and enough space for all wires and components to be concealed.
The towers have removable back sections that allow easy access for installation and repair of
components. (Appendix D). These are held on by M2 screws.
The towers have recesses where plates fit and have a hole where a tab on the plate will slide
through to the back side of the tower. The inner edge of the tower is flat to fit snugly to the flat edge
of the plates. The tab will create further stability for the towers and an easy mounting point for
electronic connections. The holes are only placed where corresponding tabs would be located on
plates. As can be seen in figure 10, holes alternate down the tower with larger holes at the top and
bottom for scintillator boxes. The one of the other towers has the opposite alternating holes while
the third tower only had two holes, one at the top and one at the bottom for connection into the
scintillator boxes.
Figure 10. An image produced by the drawing office of the high voltage tower next to a photo of the grounded tower
and a photo of the high voltage tower with plates and scintillator boxes installed. The two large holes for the scintillator
box can be seen as can the alternating holes that the plates fit into.

11. For the top of each tower there is a cap for completeness and holes in the base of the tower for
fixing the towers down to the inner base plate. The towers and tower caps are made of Delrin as it
has a high breakdown voltage and can be machined to the required accuracies.
The plates that were manufactured had slight curves on them (in §5.10.) and so there was a lot
of resistance when inserting a few of the into the towers. A region for future improvement may be
to increase the cut out height by a couple of mm so that the spacing of the plates is not changed but
so that slight curvature does not hinder inserting plates into the towers. This assumes that curved
plates do not affect the overall operation of the spark chamber (in §5.10.).
5.6. Glass Bell Jar
A glass bell jar, as in the Princeton design (Figure 9), is the chosen vacuum chamber. Bell
jars only have one sealing edge that needs to be kept air tight and as such there should be a much
lower rate of leakage. They are also aesthetically appealing to look at, fitting in well with the brief
of the project.
Bell jars come in standard sizes up to 12”, 301 mm inner diameter (ID) and 315 mm outer
diameter (OD) and this is the size of bell jar used. Three were supplied by Applied Vacuum, at a
height of 480 mm with a 10 mm flange for clamping. The bell jar flange is measured as the width
across the bottom of the flange not the top edge which was assumed at time of ordering. The bell
jars, being made to order and out of glass are not perfectly circular with one constant radius. As
discussed in §5.9. this had multiple effects on the clamp system design.
During the testing process the bell jar got damaged. A section was taken out of the flange (at
position 16, see §5.9. for what this means) but did not damage the flat edge on the bottom of the
bell jar, figure 11. After testing the bell jar under vacuum (with a blast shield) it was found that the
damage did not affect the bell jar’s strength to any observable degree so the project continued to use
the same bell jar.
Figure 11. A photo of the damage to the bell jar. As can be seen a section of the flange has been damaged, but the flat
edge under the sides of the bell jar are undamaged and an air tight seal is still made. The chip is conveniently located
such that should the numbers on the bell jar come off over time, the correct orientation can still be achieved. The ‘peak’
of the chip lines up with the holes for clamp 16.

12. 5.7. Gas Mixture
Following research into the gas mixtures chosen by other spark chamber projects the ratio of
70/30 Ne/He was decided upon. The ranges of per cent He used by groups ranged from 30% down
to 5% all with satisfactory sparking results. It was decided to go for the upper end of this range as if
He leakage did occur from the system, the amount of He would decrease as the amount of Ne
stayed the same, thus reducing the per cent of He in the chamber. If this occurred then the ratio of
the gasses would tend towards 95/5 Ne/He which is known to still work. Using this ratio of gasses
helps reduce the adverse effects of any Helium leakage.
The canister of gas purchased is 50 l at 200 bar, giving an amount of gas equal to ~11,000 l at
900 mbar. The chamber has a volume of ~45 l and so almost 250 refills of the chamber can occur
from this one canister. This is not a reason to waste gas in the future, as the gas mixture does not
come cheap. Preserve the gas whenever possible and try not to fill it for the sake of it.
5.8. Base Plates
The base plate has many functions within this project. It makes a seal with the bell jar, it has
the sealed connections for all the electronics and for signals enter/leave the chamber and it supports
the towers and weight of the spark plates. In order to hide cables and allow easier fixing of the
towers to the base plate it was decided to have a double base plate set up. This consists of an inner
and outer base plate made from Aluminium.
The outer base plate is the one that creates the air tight seal with the bell jar and allows
connections through from the outside into the chamber. The inner base plate is the attaching point
for the towers and creates a space underneath for wires and connectors to be stored.
Figure 12. The first design of the outer base plate produced by the drawing office. Seven signal and three high voltage
connections can be seen along with the O-ring groove and bolt holes for the clamp. The picture on the right is the final
set up of the inner base plate. As can be seen, the high voltage and Fischer connectors have swapped places and only
three connectors in total are present. It can also be seen the number of holes drilled for the clamping mechanism has
increased too.
The design for the outer base plate did not change much over the course of the project, what
changed significantly was the number, type and size of each connector meaning the layout and
dimensions of connection holes has evolved. Originally one high voltage and one signal connecter
were going to be placed under each tower leaving the centre clear just for a KF25 tube to enter the
chamber for the vacuuming and filling of gasses. The designs of the towers changed and their
functions developed so the number of connectors increased as each tower has a specific role. This

13. led to a more deliberately specified layout of holes, as can be seen in figure 12. The holes in the
centre and bottom left were to be signal connectors (Fischer) and the three at the top, which were all
to be under one tower, are the high voltage connectors. When the project was put together, it was
decided that there was no point adding connectors for the sake of it. The positions of the high
voltage and Fischer connectors was changed along with the number of connectors. The final
chamber has one high voltage and two 16-way Fischer connectors as can be seen in figure 12.
The outer base plate also has a groove cut into it for the placement of an O-ring to aid the
sealing of the bell jar to the base plate as well as twenty M8 holes for tightening the clamp
mechanism that is to be used.
The inner base plate is designed so that it fits snuggly into the recess of the outer base plate
and it has holes for connecting the towers as well as lager holes allowing cables to pass into the
towers. Originally it was designed to be solid so that all connections and cables were completely
concealed but due to the size of the high voltage connectors the central section of the plate was
removed. This does not affect the stability of the towers or inner base plate and due to the
scintillator box it does not reveal much of the under section to the public as can be seen in yellow in
appendix D and in figure 13.
Figure 13. A photo of the scintillator box situated in the towers and fixed to the inner base plate. As can be seen the cut
out section of the base plate cannot be seen from this angle and so the general public will not be able to the wires that
run under it.
5.9. Clamp
The chamber is run at 900 mbar (~90% of atmospheric pressure) so that the force from the
atmosphere creates a seal between the bell jar and O-ring. It was decided to aid this force by adding
an aluminium clamping ring. This means that should any leakage into the container occur, no loss
of seal strength would occur. Due to the nature of glass bell jars and the fact that the ones produced
had no single constant radius a single clamping ring could not be produced for the bell jar. The idea
to have a series of individual clamps that fit the radii of different parts of the flange was settled
upon. This lead to the design as can be seen in figure 14.

14. The clamps all have two holes, one for an M8 screw and another for a Stabilising pin. The
screw is to enable the tightening of the clamps and the pin is to stop them rotating while they are
being tightened.
The production of the clamps was outsourced to a separate company and to reduce the cost of
manufacture the clamp sections were produced in groups of 4 and 5 with the same radius. The
radius was measured at 20 points and the results grouped into similar ranges. The numbers that can
be seen in pen on the bell jar and base plate were to ensure the bell jar was always aligned correctly
to match with the clamps. Four little silicon tabs were fitted within the O-ring groove of the outer
baseplate for stability to stop the bell jar wobbling once the clamps were fitted.
To prevent the contact of the clamps on the glass flange of the bell jar and causing damage a
flat rubber ring was purchased as a buffer. The rubber ring also adds to the thickness that the clamps
have to work on and so increasing the tightening range of the clamps.
Figure 14. Top left is the drawing office design for the clamps with a cross section. Bottom left and bottom right show
clamp number 1 fitted with screw and pin mechanism. Top Right shows the underside of clamp 11 with its number
etched into it.
Due to the damage to the bell jar as discussed in §5.6. clamp number 16 must not be tightened in
case it causes further damage. If the spark chamber is for display, then clamp 16 may be placed in
the correct position to look complete but must not the tightened.
5.10. Plates
Creating the chamber using a bell jar lead immediately to the idea for using round plates
instead of square ones. Round plates cover a larger surface area and remove the four large corners
that can lead to edge sparking in square plate chambers. Edge sparking occurs when charge builds

15. up due to a point (or corner) creating an increased electric field which causes the breakdown of the
gasses in the chamber and the flow of charge. The current discharges through the edge sparks and
so does not spark between plates when a muon passes through.
To fix the plates in the towers each plate has three flat edges and on one of these flat edges is
located a circular tab. These tabs have holes in for fixing wires to the plate. The tabs slot into holes
in the towers as discussed §5.5. and can be seen in figure 10.
The plates were manufactured from stainless steel off site in order to speed up the building
process. They came to the workshop with the three flat edges and the workshop added the 5 mm
tabs with holes to the centre of the flat edges and rounded the edges of the plates to further reduce
the chance of edge sparking. From the manufacturer the plates were provided mirror polished as
these are one of the most aesthetic components in the chamber. The workshop ground and polished
the edges of the plates after they had been rounded again to increase their visual appeal.
When the plates arrived, only 15 of the 45 plates were flat. The other plates all had curves of
at least 1 mm on them. These are most likely no good for the spark chambers as Cambridge found
that the tolerance on plate separation was 1 mm maximum. Some of the plates were flattened by the
workshop, but even this did not produce satisfactory results. The curved plates also caused
problems fitting them into the towers as discussed above and in §5.5.
An area for future research would be to investigate whether the plates that are curved still
produce adequate sparks. One would assume that if some plates are more curved than others, the
inter-plate distance will vary and as such the sparks will only occur between plates that are close
together, always leaving some gaps never sparking.
5.11. Scintillator Box
In order to place the scintillators inside the chamber and have the necessary photodiodes,
amplifier chipsets and power supplies concealed it was decided to house them inside individual
boxes at the top and bottom of the towers, above and below the stack of plates. Figure 15.
The scintillator boxes are the same shape as the plates, having the same diameter and three
flat edges so that they fit into the towers uniformly with the plates. To access the inside of the
scintillator boxes three removable sections were designed, one between each pair of towers meaning
both scintillator boxes are the same design. The removable sections are held on with M2 screws.
To hold the scintillators in place so that they are both in line and have the greatest possible
overlapping surface area, small pegs were designed and manufactured. These are threaded pegs
with hexagonal tops (for easy screwing) and tapped holes can be drilled for them at points marked
out around the scintillator. This will lead to an accurate and snug fit. The reason for using threaded
pegs over glued pegs is so they can be removed on one side allowing the scintillator to slide out
should maintenance tasks need to be carried out. The pegs have not yet been fitted and the
scintillators were wrapped in light tight cloth giving them a snug fit within the box. Future work on
the scintillators can include getting the pegs fitted for a permanent position for the scintillators.

16. Figure 15. This is a photograph of one of the scintillator boxes. It has two of the three removable sections (black parts)
attached and one off for demonstration.
5.12. Handles
A method of removing the inner base plate, with towers and plates attached, from the outer
base plate was devised so that the wires and connectors underneath can be fitted and maintained
more easily. Different handle concepts were hypothesised such as premade ones that can be
removed or using metal wire as an easy to fit, removable handle. The drawing office had some
premade designs that fitted onto the inner base plate better than any premade handles that were
found in catalogues . The handles were made and were fitted to the inner base plate during
installation. Figure 16.
Figure 16. A photo showing two of the three handles on the inner base plate. Two people have to carry the inner base
plate at a time, it is heavy and easy to over balance.

17. 6. Electronics
Every design researched had the same basic components in similar places but with slight
variations. This design is taken from the basic concept of all the layouts but itself again is a
different variation. Outside the red box is the main circuit for sparking and inside the red box is the
clearing field circuit.
Figure 17. A schematic showing the layout of the circuit from the scintillators through to the trigger circuit that causes
the spark to be created along the ion trail. The grey box covers the components that will be inside the chamber and
everything that is outside of the grey box will be concealed under the spark chamber in a display cabinet type box.
6.1. Single resistor and capacitor set up
Other designs had one resistor in series with each individual plate and hence in parallel with
each other. Combining the resistors makes no difference to the operation of the system and means
that one resistor can be located outside the chamber, which is better in terms of the space
requirement. It is easier to test one resistor to see if it is broken than test or replace seven. R2 in
figure 17. This is the route that the charge will flow if a spark doesn’t occur when the system is
triggered. Resistor R1 reduces the current from the HT supply to ~1mA, this creates a time delay
between sparking and the capacitor becoming charged again. This ensures a more reliable spark due
to the low current acting to reduce the potential difference across the plates.
This design has one capacitor whereas other designs have one per high voltage plate again.
The understanding of the use of a capacitor is to act as a high pass filter so that the DC high voltage
current does not discharge via the plates and resistor to the ground. The understanding of the
capacitor acting in this way, after much debate, seems to be correct. The arguments against using
just a single capacitor is that if one plate sparks early, then all the charge would dissipate between
the one set of plates and only one spark would be seen. Multiple capacitors would prevent all the
charge dissipating through a single pair of plates. The circuit works with the single capacitor in the
main circuit, with sparks being seen on multiple plates not just a single plate, further backing up the
high pass filter idea. C1 in figure 17.
Future investigation can make use of the space inside the high voltage tower should the need
arise to have one capacitor per high voltage plate. The benefit this may have is that it could increase
the mean number of gaps that spark every time a muon goes through. The capacitors used on the
R1
R2
C2
C1

18. board are too big to fit into the space within the tower, however the capacitors used are rated to 15
kV as this was part of our contingency plan, should high voltages (>10 kV) need to be used.
Smaller capacitors of the same value can be found and soldered onto the tower PCBs.
6.2. Spark Gap
To trigger the plates to spark quickly, rather than just using a pulse from the trigger unit to
spark a cross the plates, the designs researched used spark gaps. The signal from the trigger unit
creates a small spark inside the spark gap that causes a breakdown across a larger gap, creating a
larger spark which then transfers charge across one side of the circuit with a small rise time. The
quick rise time of the sparked charge causes charge to also spark across the plates rather than
through the resistors attached to them. The sparks follow the path of least resistance which is the
path of ions left by the incident muon as discussed in §2.2.
Commercially available spark gaps were not a viable option as they either have too short a
lifetime, are too expensive or are only available from military suppliers which would require large
order sizes. Other projects have used car spark plugs as a basis for a spark gap. Due to information
from CERN we researched and chose to use a thyristor as detailed below.
6.3. Thyristor switch
Due to the reasons above, commercially available spark gaps were not suitable for this project
and rather than research into developing our own, advice was given from CERN. At CERN, spark
chambers have been built using thyristors. They are a self-contained electronic unit that can trigger
a large pulse from a low voltage (3 – 6 V) trigger. Unlike spark gaps, thyristors exhibit stable
switching characteristics, not dependant on temperature or age. Thyristors have a much longer
lifetime than a spark gap would also a bonus given the low level of maintenance required in the
brief. Thyristors are more reliable and have a much longer mean time between failures that spark
gaps, meaning sparks are seen more often.
Thyristors have a longer rise time than spark gaps do, but it was calculated that the increase in
rise time didn’t cause the ions in the gas to diffuse or recombine before a spark could be formed.
Unlike with spark gaps, the thyristor switching behaviour is not affected by the rise time or
amplitude of the trigger, meaning less specific criteria had to be considered in producing the
coincidence circuit. More information is available in the appendix C.
6.4. Clearing Field
The section of figure 17 in the red box is the added clearing field circuit. The clearing field is
used to remove the ions within the chamber after sparking or in the event of a muon going through
and no sparking. Clearing fields reduce the delay time between sparks as it remove ions faster than
the rate of just recombination, hence giving a greater time resolution. They however reduce the
spatial resolution as the clearing field causes the ions and electrons to migrate towards the plates,
meaning the ion trial becomes more defuse within the chamber and the sparks can occur over a
larger space range. As this project is not concerned with taking spatial measurements, at least not at
this stage, a reduction is spatial resolution is not an issue.
The clearing field is created by the two 5 MΩ resistors that create a potential divider. The
capacitor on the far right within the red box acts to decouple the plates from the high voltage supply
when triggered. The negative potential spike will follow the path of least resistance and could return
to the HT power supply through the 5 MΩ resistors if the capacitor wasn’t acting to decouple. A
switch was installed so that the clearing field can be turned off if so desired. For example if it
becomes a hindrance to different applications of the spark chamber.

19. The clearing field is set at 200 V in the current set up of the spark chamber. A future area of
research could be to vary the potential of the clearing field and observe the effect on the sparks
formed.
6.5. High Voltage Connectors
A hermetic high voltage connector (SHV connector) is used to allow the high voltage into
the chamber and onto the plates. Six spaces for connectors have been allowed in the base plate for
connections to the plates, however only one is drilled fully and used. The connectors are rated up to
10 kV as research showed spark chambers can operate in the 6 – 8 kV region. Plugs rated to 15 kV
were originally intended but none could be found that were hermetic and confirmed to work in
anything except a vacuum.
The first SHV connector used did not seal properly and it was assumed that the type we used
was not designed to be hermetic. When testing for leaks, ethanol was sprayed around and into
connectors and was seen to bubble through the central section of the SHV connectors. Having been
told the connectors are listed as hermetic we tested a second connector and it was fully sealed.
Cable rated to 15 kV is used to link the connectors in the base plate with the printed circuit boards
(PCBs) in the towers, §6.7. One cable travels from the central pin to the high voltage tower and a
second cable travels from an eyelet attached to the grounded section of the connector to the PCB in
the grounded tower.
6.6. Signal Connectors
Two Fischer signal connectors that are hermetically sealed will be used. Two holes have been
drilled for them in the base plate. They each have 16 connectors on them and this provides the
correct number of connections for the photodiodes, amplifiers and ground connections.
One connector carries the power for the amplifiers (5 V) and discriminators (40 V) as well as
a ground for each power supply. As each of the four photodiodes has its own amplifier and
discriminator, four cables for each of the four APDs gives 16 cables which fills one connector. The
second connector carries the signals from the amplifiers and discriminators to the outside of the
chamber onto the trigger circuit board. Each of these outputs also has a ground associated with it.
Two signals and two grounds for each of the four APDs, again gives 16 cables which is one full
connector. All the signals are carried by ribbon cable which is soldered onto the Fischer connectors
as in figure 18.
Figure 18 Photos showing the soldering of the ribbon cable to the Fischer socket (left) and to the plug (three photos on
the right) The cable with the red stripe is number one and the circular pattern on the Fischer information sheets is
followed in numerical order.

20. With the Fischer socket and plug taking the signals through the base plate, it was decided not
to have them connect directly to the main circuit board outside or the amplifiers inside. Instead 16
way plugs were used so that when the inner base plate is lifted it can be disconnected without the
need to remove the Fischer connector and thus preventing damage to the seal. The 16 way plugs are
clamped to the end of the ribbon cable Figure 19 and coax is used from the connector to the
amplifier boards. The coax is grouped in units of four for each connector. Each APD has its own set
of four in each connector. The four way plugs have the arrangement in figures 20 and 21 so that a
uniform system is kept.
Figure 19. This photo shows the 16-way plug being clamped onto the ribbon cable. The arrow on the top left
indicates the position of wire number one.
Each APD set up has a digital output which is from the discriminator. This feeds into the trigger
circuit. Each APD also has an analogue output that was used at first to check the shape and size of
the signals from the amplifiers. Due to interference because of the ribbon cable, the shape of the
analogue signals was distorted. This lead to the analogue signal cables all being capped with a 50 Ω
resistor inside the chamber rather than outside. This is to improve the peaks of the signals and to get
more signals through the discriminator.
Once the 16-way plugs had been clamped, they were checked to ensure there was no cross
connection. The second plug that was made had cross connections due to the angle the ribbon cable
was in the plug. This had to be cut off and started again.
Figure 20. A diagram showing the layout of the sixteen pins on one of the connectors and the four way plug that
has the wires from a single APD/amplifier set up. ‘A’ stands for analogue signal, ‘D’ stands for digital signal and ‘G’ is
the grounded connection relative to the connection above it. The view of the four way plug is from the end where the
crimps are plugged, so the ‘back side’ of the plugs.
A D
G G
A D
G G
A D
G G
A D
G G
A D
G G
Pin 1
Pin 16

21. Figure 21. A diagram showing the layout of the sixteen pins on one of the connectors and the four way plug that has the
wires from a single APD/amplifier set up. ‘5’ stands for the 5 V power supply, ‘40’ stands the 40 V power supply and
‘G’ is the grounded connection relative to the connection above it. The view of the four way plug is from the end where
the crimps are plugged, so the ‘back side’ of the plugs.
6.7. Tower circuit boards
In order to connect all the plates to the same high voltage and ground without the need for
15 cables routed through the towers, it was decided to use a printed circuit board (PCB) in the both
the high voltage and ground towers and connect the plates with short wires from a common ground
strip. The wires were soldered onto the PCB and the screwed into the tapped hole on the plate tabs
as in figure 22. The tension in the cables is enough to hold the PCBs inside the towers and the tower
backs fit around the PCBs to conceal them.
Figure 22. A photo showing the wires screwed into the plates and a close up of the soldering onto the tower PCB.
6.8. Trigger Circuit Board
In order to hold all the chips and resistors in a secure place a main PCB was created. As can
be seen in figure 23 it holds the logic circuit (§6.9.) and the thyristor. The trigger PCB is where the
digital signals from the discriminators go and is where the high voltage pulse originates. The
spacing on the board reflects the need to have high voltage tracks separated. The boards were
5 40
G G
5 40
G G
5 40
G G
5 40
G G
5 40
G G
Pin 16
Pin 1
5 40
G G

22. manufactured off site but soldered in the lab. They have the components labelled and different
settings indicated, such as the choice between triggering from 2/4 or 3/4 signals. The two chips are
standard AND and OR gate chips which make up the coincidence circuit. There are added resistors
for each of the four input signals such that a broken APD with wandering output will be tied to
ground and will not cause phantom triggering. A high voltage shield is in the process of being
ordered and should have arrived by the 20th
of May in order to shield uses from the voltages on the
trigger PCB. The trigger circuit was tested with generated inputs before real input were fed to it.
The test came back positive and so the circuit was confirmed to work.
The future investigations on the size of the clearing field will rely on changing the size of the
resistors with the clearing field section of the PCB. No opening of the camber is needed which
saves, time and gas mixture.
Figure 23. The top photo shows the PCB during testing with all the components including the thyristor in place. The
bottom photo shows the circuit with all the soldered components attached. The labels can be seen clearly along with the
size of the capacitors, which are so large because they are rated to 15 kV as discussed above.

23. 6.9. Logic / Coincidence Circuit
When the signals come from the APDs the coincidence circuit checks the number that come at
the same time. At first a magnitude comparator was researched as the coincidence unit, but these
take four inputs and give them all a binary value increasing by column. So if the triggering value is
set to 4 then if just the fourth signal fires it is in the 8 value column and so will cause the system to
trigger. This method of value comparison would not produce sensible results and so was dropped.
The system in use comprises of a series of OR and AND gates that take in the four input signals.
Originally the circuit was built to trigger off any three signals or off one signal from each APD and
was later adapted to be able to trigger off any two signals as there was a spare OR gate on the chip.
Triggering off any two could lead to problems if only the APDs on the top scintillator send a signal,
but the spark chamber can deal with a false trigger. The benefit is that if a muon goes through the
top scintillator and not the bottom one due to its angle, a partial track with still be sparked along and
will show that some muons come from greater angles away from the vertical. Future study into the
rate of muons that go through one scintillator and not the other could be conducted.
The layout of AND and OR gates can be seen in figure 24 along with pathways which have
the inputs concerned next to them.
Figure 24. These show the original circuit diagram and also the improved circuit diagram that shows the extra OR gate
on the right hand end that allows triggering from any two signals.
7. Testing
7.1. Helium Testing
In order to test the connectors for not only air tightness but He leakage, the base plate and
bell jar were taken to the 4th
floor nuclear lab. Here the bell jar was evacuated and connected to a
mass spectrometer. The theory is that if the jar has leaks or that the connectors are not sealed
properly then when He is sprayed around the bell jar, some of it will get inside the chamber and
show up on the mass spectrometer. As He is the hardest gas to contain, if it is not detected then the
rate of leakage of all gasses can be assumed to be very low (if not negligible).
When the mass spec was turned on a few peaks were seen which are explained by molecules
such as water which produced a peak at mass 18 and Nitrogen from the air which produced a peak
at mass 28. Figure 25. When testing with He began nothing was detected after a short period of time
which proved no He was getting into the system. This gave a huge confidence in the quality of the
achieved seal created.

24. Helium can diffuse through glass itself, not just through gas in the seal. The leakage rate
over the surface area of the bell jar worked down to ~1 cm3
per year of Helium escaping. It would
be easy to detect a large leakage of Helium because the pressure will drop by 30% inside the
chamber as the partial pressure of He reduces. No such loss of pressure is observed at all.
Figure 25. A photograph of the mass spectrometer read out. The peaks at mass 18 and 28 can be seen. The peak at mass
2 is a statistical anomaly. Nothing is seen at mass 4 which is where He would show up were it leaking into the chamber.
7.2. Gas Leakage
When the chamber was first filled with the He/Ne gas mixture, the pressure took a few
minutes to settle in the same way the virtual leak took a few minutes to settle out. When the
chamber was filled it was left for 5 days so that leakage could be observed. What was recorded was
a graph of pressure fluctuations. These were hypothesised to be related to temperature change in the
lab, as the lab temperature had a relatively large range. Preliminary results gave approximately 3
mbar of pressure change per Kelvin. If the ideal gas law is used then the pressure change expected
can be calculated.
As V, n and R are constant the change in pressure due to temperature can be written as:
Taking the temperature as 298 K and the pressure of the gas mixture as 900 mbar, the final result
for pressure change gives:

25. Which is as observed in the investigation. The data were plotted on a graph to show this graphically
and the results can be seen in figure 26 which agrees with the calculated result. The correlation of
the data on the graph and the agreement with the ideal gas law gives good evidence that the pressure
changes observed are the result of temperature changes in the gas and not due to gas leakage in
either direction. The data was taken over 10 days and no downward or upward trend was seen with
time, suggesting no leakage in or out of the chamber. The little or no leakage means that the
chamber should be able to be run for long periods of time without needing to be refilled, even
without the clamps on as the clamps were not used in these tests.
Figure 26. A graph showing the pressure as a function of temperature in the chamber over a period of 10 days. As can
be seen the gradient is 3.25 mbar K-1
in agreement with the calculations above.
7.3. Final Build
When all the parts had been received and tested, construction took place. All the
components were assembled into the towers. Cables were soldered onto the PCBs in the towers and
connected to the SHV connector. The signal cables in their fours were connected to the 16 black
plugs and tested before powering to make sure they were still all isolated from each other. A note
for the future, when testing the 16-way signal cable ribbon connector, the grounded connections
(one for analogue and one for digital) are linked via the amplifier chips inside the scintillator boxes,
so do not panic when they set each other off during connection testing. Once everything had been
tested the high voltage power supply was connected to the circuit PCB.
Output signals were again verified and the discriminator voltage set to 40 V. The high
voltage power supply has a maximum voltage of 6.4 kV. It was started at 4.5 kV and a noise was
heard. The potential was put up to the full 6.4 kV and sparks were seen. Figure 27 and Figure 28.
The discriminator voltage was also increased, up to its maximum of 42 V. This was to produce
larger signals so more coincidences were registered per time. Future investigation could look at the
effects of decreasing high voltage potential and lowering the discriminator voltage. Also if a higher
potential power supply could be found that can produce enough current then an investigation into
how increasing potential affects the spark rate, size, brightness and if it possibly induces problems
within the system.

26. Figure 27. A photograph of the completed spark chamber just before it was switched on.
Figure 28. Two screen shots from a video of the spark chamber. As can be seen most but not all of the gaps had sparks
in them.

27. 8. Installation and Display
MoSI has joined with the science museum in London due to funding and so the display area
that was to be built where the chamber was to be displayed may not go ahead. The curator of MoSI
was still enthusiastic about the spark chamber at a meeting in March so the scope to install it is still
a real possibility. Following safety tests and more input into displays this could make an excellent
visitor attraction. A preliminary poster has been produced that could be used in a display about the
chamber in MoSI, schools or the Schuster building.
Work needs to be done to correctly enable easy installation of the chamber. A manual on
maintenance needs to be official produced along with general operating procedures. An area for
future development could be to install a timer or motion tracker to the circuit so that it does not
spark 24/7.
9. Discussion and Summary
The spark chamber works. As can be seen above, the spark chamber works and sparks can be
seen on multiple plates. This is the first spark chamber to be stand alone, to be able to be left to run
without constant gas flow and it is the first that does not use PMTs and as such is able to be fully
self-contained with only three output connector required. Now that the spark chamber has been
proven to work and can hold its pressure for significant periods of time work on installation and
safety testing can begin. The use of a single capacitor rather than one per plate has been shown to
work. Sections where future work can be conducted have a small paragraph at the end for
directional purposes.
Acknowledgements
I would like to thank Professor Stefan Soldner-Rembold, Dr David Urner, Dr Jo Pater, Mr
David Chorlton, Mr Michael Perry, Mr Julian Freestone and Mr Mervyn Posnett for all their time,
encouragement, knowledge and patience which has been invaluable and highly appreciated by me
and the rest of the group.
References
[1] http://en.wikipedia.org/wiki/File:Atmospheric_Collision.svg
[2] Adapted from: http://www.ep.ph.bham.ac.uk/general/outreach/SparkChamber/text4dhh.html
[3] Adapted from: http://www-f9.ijs.si/~rok/sola/praktikum4/mioni/muonexp.html
[4] Cox, J. 2002. Fundamentals of Linear Electronics: Integrated and Discrete, Delamr, USA.
[5] Lab reports by Jacob Stein and Ashley Timmons. University of Manchester, May 2012.
[6] http://www.hep.phy.cam.ac.uk/~lester/teaching/SparkChamber/SparkChamber.html
[7] Still image taken from: http://www.youtube.com/watch?v=F7SuyIsMv5c
[8] http://www.ep.ph.bham.ac.uk/index.php?page=general/outreach/SparkChamber/index