NRE 3112

Laboratory 1
Electronics Pulse Analysis, SCA, MCA
Alec Herbert, James Padgett, Michael Reilly, Tera Sparks
Team Nukeass
(Dated: September 15, 2016)
The common components of a larger radiation detection apparatus are independently studied for
their purpose and effect on events simulated by a pulser, providing a useful baseline for further
study in this field.

I.

INTRODUCTION

Radiation measurement and detection is an essential

aspect of the nuclear industry with respect to both reactor physics, particle accelerators, medical industry, and
many other facets of physics. Discerning not only the
source, but the types of radiation present is essential.
There are many ways to measure high energy radiation,
depending on which properties of the radiation source
are sought. However, although the energy per event is
very high on the atomic scale, it is not large on a human
one, and thus many steps should be taken to amplify and
shape the events to facilitate detecting, counting, and
further classification of other properties, such as energy
deposited, and signal shape.
In this experiment, the team of engineers conducted
a survey of common laboratory equipment employed in
the detection and classification of radiation sources. The
goals of this experiment are to determine a useful baseline for the behavior and effect of the equipment studied
in a controlled environment, so that its usage is well understood during analysis of random processes. This goal
is being performed on a well controlled pulser waveform,
rather than a random process, such as a radiation source,
because it is predictable and controllable, whereas a random process would introduce a number of variables and
behaviors that are not necessary or desired for this sort
of experiment.
II.

BACKGROUND

The equipment profiled include a preamplifier, linear

amplifier, delay line amplifier, a Single Channel Analyzer
(SCA) and a Multichannel Analyzer (MCA). To assist in
the analysis an oscilloscope, a pulse generator, a counter
and timer were used.
A preamplifier can come in several types, including
charge sensitive and voltage sensitive. In a charge sensitive preamplifier the initial pulse comes from the movement of charges generating a current that is then converted to a voltage in the preamplifier. The final voltage is linearly related to the initial charge pulse. Other
preamplifier types do not preserve this information to
the same degree. Moreover, voltage sensitive preamplifiers only function in a limited temperature range due to

the relationship between temperature and impedance. [1]

Therefore, the most common type of preamplifier in use
for radiation detection is of the charge sensitive type. [2]
After the amplitude of the signal is boosted by the
preamplifier, it is often desirable to shape the signal into
a more uniform profile for ease of counting. There are two
major types of amplifiers: linear and delay line. A linear
amplifier attempts to retain some of the characteristics of
the original signal while eliminating the long exponential
tail associated with the output of the preamplifier. [1] In
some situations it may be desirable to use a delay line
amplifier instead to preserve the rise time of the initial
pulse.
An SCA attempts to count the number of events within
a specific voltage band. An SCA can be made using
the difference between two Integral Discriminators and
is therefore sometimes called a Differential Discriminator. [2]. On the other hand an MCA will categorize every
event and place it into one of several different buckets,
essentially running several hundred or thousand SCAs in
parallel. [2]

III.
A.

PART I
Procedures

In order to understand the impact of each piece of

equipment, each one was added individually so that comparisons could be made between how the system worked
with and without equipment. Comparisons were made
by comparing pulse height, pulse width, and signal frequency, and a qualitative look at the shape of the pulse.
The initial settings of the system included an attenuation
factor of 2x, polarity set to negative, pulse height vernier
to 10.0. The oscilloscope was then set to use external
triggering, and to trigger on the falling edge.
Initially, there was only the pulse generator connected
to the oscilloscope, as shown in Figure 1. [3]
The preamplifier was connected so that gain and pulse
shape could be compared. Two separate channels existed
so that comparisons could be made with and without the
preamplifier.
A linear amplifier was connected to further define the
pulse. The shaping was set to 2 s. Additionally, the

FIG. 1. A schematic depicting the pulse generator connected

to an oscilloscope.

FIG. 3. A schematic depicting both a linear amplifier and the

preamplifier connected to the oscilloscope.

FIG. 2. A schematic depicting both a preamplifier and a pulse

generator connected to an oscilloscope.

polarity was set to positive and the unipolar output was

measured.
By using pulse height controls, it was qualitatively
shown how a pulse can be saturated.
Within the circuit, the delay line amplifier was added
to maintain the rise time of the original pulse. The outputs of the preamplifier with and without the delay line
amplifier were also compared. Moreover, several different
values of for the Integ setting on the delay line amplifier
were compared.

FIG. 4. A schematic depicting both a delay line amplifier and

the preamplifier connected to the oscilloscope.

3
B.

Results

The output on the oscilloscope with no components,

other than wires, between the pulse generator the output
shape is shown in Figure 5.
The following values were measured for the output of
the pulser:

TABLE I: Pulse Generator Measurements

Pulse Height
21 mV
Pulse Width (FWHM)
60 s
Signal Frequency
990.1 Hz

Vout
Vin
5.84 mV
G=
14.2 mV
G = 0.4113
GdB = 20 log G
GdB = 7.72 dB
G=

TABLE II: Preamplifier Measurements

Pulse Height
5.84 mV
Pulse Width (FWHM)
17 s
Signal Frequency
990.1 Hz
Gain
7.72 dB

The next comparison was between the output of the

preamplifier alone versus the output from a preamplifier
and a linear amplifier.

FIG. 5. Output of the pulser set to 1kHz with no further

processing.
FIG. 7. The output of the linear amplifier (the larger waveform) compared with the output of the preamplifier (smaller).

The preamplifier was connected to channel two and its

output waveform was isolated, and it is shown in Figure 6. The gain between the pulse generator and the
preamplifier was then calculated:

A comparison of these two outputs can be seen in Figure 7. The output of the amplifier is of higher amplitude
and is relatively more narrow as compared to the preamplifier waveform.
TABLE III: Linear Amplifier vs Preamplifier Measurements
Property
Preamplifier Linear Amplifier
Pulse Height
5.28 mV
176 mV
Pulse Width (FWHM)
20 s
4.8 s
Signal Frequency
990.1 Hz
990.1 Hz
Gain
30.46 dB

Vout
Vin
176 mV
G=
5.28 mV
G = 33.33
GdB = 20 log G
GdB = 30.46 dB
G=

FIG. 6. The output of the preamplifier when the waveform

in 5 is fed through it. Of interest, the base voltage appears
somewhat inconsistent.

4
An attempt was made to investigate what would occur
if the amplifier were to become saturated. In order to isolate the edge of this behavior, some adjustments needed
to be made to the settings of the pulser and amplifier.
The adjustments to the pulser were necessary because
the amplifier lacked the resolution in its upper ranges to
exactly isolate the proper amount of gain otherwise.

son between the output of the preamplifier and the delay

line amplifier can be viewed in Figure 9.

FIG. 10. Detail view of the delay line amplifier waveform

using a Integ value of 0.25 s.

The following properties were measured for the output

waveform (visible in Figure 10):
FIG. 8. A saturated waveform obtained from turning the linear amplifier very high. The timescale of this figure is 10.0 s,
and the voltage scale is 2.00 V, with 0 V located half a mark
below the lowest major mark. (It is marked by the arrow on
the lower right.)

The pulse height vernier was set to 7.02 during this

experiment, while the amplifier was set to 3000 on the
course adjustment and 13.0 on the fine adjustment.
These settings produced the waveform visible in Figure 8.
The pulse is visibly clipped, and its maximum amplitude
was measured at 11.6 V.
Next, the delay line amplifier was observed. Initially,
the Integ parameter was set to 0.25 s, and the oscilloscope was connected to the unipolar output, as shown in
Figure 4.

FIG. 9. Output of the delay line amplifier (bottom waveform,

channel two) vs the output from the preamplifier (top waveform, channel one). Note that for clarity these waveforms
are offset from each other and have differing vertical scaling.
Review the legend for details.

Even when compared with the linear amplifier, the delay line amplifier has a very short duration. A compari-

TABLE IV: Delay Line Amplifier vs Preamplifier Measurements

Property
Preamplifier Delay Line Amplifier
Pulse Height
5.12 mV
90.4 mV
Pulse Width (FWHM)
16.0 s
960.0 ns
Signal Frequency
990.1 Hz
990.1 Hz
Gain
24.94 dB

Vout
Vin
90.4 mV
G=
5.12 mV
G = 17.66
GdB = 20 log G
GdB = 24.94 dB
G=

FIG. 11. Detail view of the delay line amplifier waveform

using a Integ value of 0.1 s.

Then the experiment was repeated for the other two

Integ parameters, 0.1 s (Figure 11) and 0.04 s (Figure 12). But while the pulses were shaped slightly differently, the measured characteristic parameters remain
more or less identical.

FIG. 12. Detail view of the delay line amplifier waveform

using a Integ value of 0.04 s.
TABLE V: Delay Line Amplifier Integ
Property
0.25 s
Pulse Height
90.4 mV
Pulse Width (FWHM) 960.0 ns
Signal Frequency
990.1 Hz

IV.

Parameter Comparison
0.1 s 0.04 s
98.4 mV 96.0 mV
920.0 ns 920.0 ns
990.1 Hz 990.1 Hz

The counting system was setup as shown in Figure 13.

The unipolar output from the linear amplifier was connected to the first oscilloscope channel via a T-Connector
and the positive output from the SCA was connected to
the second oscilloscope channel.
The SCA was setup to accept input events from 0 V to
5 V, using a lower threshold of 0 V and a E of 5 V. Then
the amplitude of the pulse generator signal was varied to
test its dependence on signal character.
Next, the amplitude of the input to the SCA was manually set to 2 V, as measured by the oscilloscope, and the
number of SCA events was counted using the counter and
timer over the course of 10 s. The expected count of this
measurement is given by the number of events in 10 s,
which is given by:

10 s 990.1 Hz = 9901
990.1 Hz being the measured frequency of the input
signal.

PART II

FIG. 14. Waveform of the linear amplifier (top waveform,

channel one) vs the output of the SCA (bottom waveform,
channel two). The waveforms have been separated and have
different vertical scaling for clarity.

The number of counts was measured as 9901 events,

for an error off of the expected measurement of 0%. The
height and width of the SCA output was found to not
be dependent upon the pulser input, provided that the
input was in the correct voltage range upon reaching the
analyzer. The nature of the SCA output was a very sharp
negative voltage as seen in Figure 14.
The voltage was then adjusted up to 4 V and another
count was taken over 10 s. The second count was measured as 9946 events, for an error rate of 0.452%.

FIG. 13. A schematic depicting the linear amplifier connected

to channel one of the oscilloscope and the SCA connected to
channel two.



 Expected Observed 


Error = 

Observed


 9901 9946 

Error = 

9946
Error = 0.452%

6
Then the voltage was adjusted up a third time to 6 V,
and another count was taken over 10 s. Because the generated pulse was set to 6 V, it is out of range of the threshold, and so the expected value for counts is 0.
Instead, 171 events were measured over 10 seconds, for
an error rate of 100%, which is the only possible error
rate for an expected value of 0 events. The existence
of events in this area are likely due to line noise being
picked up as an event by the detector, but this is merely
conjecture.
Next, the lower threshold on the SCA was set to 4 V
and the E was set to 2 V. Another count was taken
with these settings over 10 s and a total of 0 events were
detected. Its difficult to say the expected count in this
situation, because the value to measure lay right on the
discontinuity of the counting function.
Next, the E was adjusted until the SCA began counting events from the 6 V signal. This value was found to be
approximately 2.4 V. A 10 s and 30 s signal was counted
by the SCA. The counts for these were 6235 and 18624
respectively.
TABLE VI: SCA Event Counting
Input Voltage Lower Threshold E Duration Counts
2V
0V 5V
10 s
9901
4V
0V 5V
10 s
9946
6V
0V 5V
10 s
171
6V
4V 2V
10 s
0
6V
4 V 2.4 V
10 s
6235
6V
4 V 2.4 V
30 s 18624

V.

PART III

The MCA was connected in the place of the SCA in the

previous section, the major difference is that the MCA
does not have an output to the oscilloscope, instead its
output is the computer. This configuration is shown in
Figure 15.
The computer interpreted the output of the MCA via a
program named Gamma Acquisition and Analysis, which
needed to be configured to accept input from the MCA,
by setting the source as the external detector.
The program was instructed to used 4096 channels
with a live time of 30 s. Due to the aforementioned issues
with linear amplifier resolution, the value of the highest
channel, V0 , was impossible to obtain. Instead, a separate formula was developed and used to obtain the lower
and upper thresholds.
After the signal was isolated near channel 2000, we
measured the pulse height at this value, which was found
to be 4.92 V. A full 30 s capture was ran to establish a
baseline and determine the region of interest. The mode
channel was 2007, and the region of interest was determined to be channels 2005-2008, which is a surprisingly
tight range. The number of events was found to be 34165.
Channel 2007 was taken to correspond with the pulse
height of 4.92 V. Then the same assumptions that the

FIG. 15. A schematic depicting the linear amplifier connected

to the MCA. Only the linear amplifier is connected to the
oscilloscope.

original formula used was employed, namely that the differences in energy between channels is approximately linear, and that channel 0 corresponds to 0 V.

2005
V2007
2007
2008
V2007
=
2007

Tlow =
Thigh

Which gives values of Tlow = 4.915 V and Thigh =

4.9225 V, from which E can be obtained.

E = Thigh Tlow 0.01 V

Next, the SCA was used to test these findings. A 30 s
out was setup using the lower threshold and E calculated in the previous section. No events were measured.
Because it was established that the voltage readings on
the SCA were not properly calibrated previously, with
the guidance of the TA we widened the SCAs voltage
window.
Using a lower threshold of 4.88 V and E of 0.8 V a
second 30 s count was ran, and a value of 29698 counts
was obtained.

7
VI.
A.

DISCUSSION
Questions Posed

Why are the preamplifier, linear amplifier, and

delay line amplifier important in detection systems?
Because, although the energy and particles emitted by
nuclear decay have very high energies on the atomic scale,
and can easily ionize multiple atoms, they are still very
small compared to human sized scales, and thus make
only very events that have to be distinguished from noise
due to thermal systems, background radiation, static interactions and so on.
These devices allow the signal to be separated from
the noise and shaped into a form that is convenient to
count and analyze. Further discussion of these devices is
located in the Background section II.
If the input signals are random pulses from a
radiation detector, what would you expect from
the oscilloscope?
If, instead of pulse generator, a radiation detector was
connected that was measuring counts from either the
background or a continuous spectrum source, then there
would be a number of important differences.
First, there would not be a well defined frequency of
events. There would be a statistical frequency based on
the mean time between events, which is based on the particle flux. The events would come at random, following
a Poisson process.[4]
Because these events are unfiltered, there is a possibility that there may be some pileup based on the number
of events and the detectors dead time and behavior (i.e.,
paralyzable vs non-paralyzable.)[2]
What applications can the SCA be used for?
SCAs are most useful for isolating any event that has a
discrete energy spectrum. These are usually only rays
or characteristic x-ray events, as, although particles
have discrete energies, they have such a short mean free
path, and will likely deposit its energy partially before
reaching the detector. However, other discrete processes
exist, such at Internal Conversion or Auger electrons.
(But not, for instance, particles.)[1]
When using the MCA with a radioactive
source, what features might you expect if the
pulse height spectrum corresponds to the energy
deposition in the detector volume?
This depends on the characteristics of the radioactive
source. It depends on the energy of the decay, the type
of decay, whether or not the spectrum of the decay is
continuous or discrete, and so on.
For particles or other isoenergetic photon sources,
there is also a dependence on energy, because of the different types of interactions available. For instance, if the
energy of the photon is less than 1.02 MeV then youll
find a photopeak, a backscatter peak, peaks for characteristic X-Rays based on the source and the detector, a
bowl shaped area of partial energy deposition via Comp-

ton Scattering, not to mention background sources, side

reactions, contamination, and so on. On the other hand,
if the photon energy is sufficient for pair production then
you can also end up peaks smaller than the photopeak
by 511 keV or 1.02 MeV, if one or both photons produced
by the positron annihilating with an electron escape the
detector.[5]
For different types of particles its a different story.
For beta particles, presumably there would be the standard energy curve of a beta particle, with a peak about
one third of its maximum energy. However, there would
also be energies associated with Bremsstrahlung radiation, etc.

B.

Sources of Error and Oddities

Strangely, the gain from the preamplifier stage was less

than one. The presumed cause of this was discovered
nearly at the end of the laboratory procedures. The attenuation factor on the pulse generator was not set to 2x
as expected, instead the toggle switches were all thrown
exactly opposite as instructed, giving us a different, unknown attenuation factor to what was expected.
The base voltage of the preamplifier was very inconsistent, moreover it seemed to either have a very long
period or be aperiodic. This behavior did not seem to affect the performance of the amplifiers; however, it is odd.
It was conjectured that perhaps the voltage floor fluctuation was caused by induction from a 60 Hz AC power
line, but due to time constraints and apparent lack of
effect on results, it was not investigated further.
The output waveforms were somewhat noisier than expected. This may be caused by high impedance lines or
BNC connectors, or an impedance mismatch between the
line and the source or the oscilloscope, or other sources
of electrical and/or thermal noise.
The pulse height from the preamp seemed to be somewhat dependent on what equipment was connected. This
is possibly an artifact of a small sample size, or it could
have something to do with voltage drops due to TConnectors, or the resistance of other equipment changing the current and therefore the voltage drop across the
preamplifier.
The frequency was extremely consistent when measured. On one hand this isnt terribly surprising given
the waveforms being measured had very precise peaks
and a uniform methodology was used to measure the period. However, some variation is still expected. Perhaps
this is an artifact of the resolution of the oscilloscope,
but the exact same value was also obtained from the first
SCA count. So it may just be that the pulse generator is
extremely consistent.
The second and third SCA counts were both overmeasured; it is conjectured that these events can be
caused by noise; however, if thats the case why isnt
the first measurement affected? Coincidence? Further
experiments are required to explore this further.

8
The linear amplifier pulse was clipped at 11.6 V, which
is quite close to a very common voltage used by electronics circuits: 12 V. It is unknown to the authors if there is
a causal relationship between these two values, or if its
merely a coincidence.

The Integ parameter did not effect the FWHM as

much as expected. The shape of the signal did change;
however, it did not change uniformly, (E.g., 0.25 s
was more like 0.04 s than 0.10 s.) Presumably, this
parameter is meant to control how swiftly the waveform
is reversed, but if that is the case, then it would be
expected that the pulse width would be getting smaller
when turning the Integ down. Instead the pulse width
seems more or less invariant with respect to Integ.

[1] M. Reilly, Nre3112 class notes, (2016).

[2] G. Knoll, Radiation detection and measurement (John Wiley, Hoboken, N.J, 2010) pp. 111, 122, 639, 642, 705.
[3] A. E. (Assumed), Laboratory 1 assignment sheet,
(2016).
[4] A. Erickson, Lecture 4 - statistics of nuclear detection,

VII.

CONCLUSION

The purpose of these procedures was testing and discovering the limitations of the equipment, prior to using
it for productive work in a less controlled setting, as well
as understanding typical values of clean waveforms for
the different types of amplifiers.
Several interesting limitations and surprising behaviors were discovered that require bearing in mind, such
as the basic noise floor from the wiring and impedance
effects, the near invariance of the FWHM of the delay
line amplifier when Integ was adjusted, the saturation of
the linear amplifier and the lack of resolution for the amplifier in higher ranges, SCA false positives and boundary
conditions, and so on.
Time permitting, it would be interesting to revisit
some of the mysterious behaviors mentioned in the discussion, such as how stable the frequency seemed, and
the strange behavior of the base voltage from the preamplifier output.
Team Nukeass looks forward to applying this new
knowledge to further labs and expects this information
gleaned from this experiment to be valuable.

(2016).
[5] I. Rittersdorf, Gamma ray spectroscopy, http:
//www-personal.umich.edu/~ianrit/gammaspec.pdf
(2007), [Online; accessed 15-September-2016].