Transcript PULSE COUNTING SYSTEMS The signal chain shown in Fig. 17.7

Pulse Processing
Chapter No. 17
Radiation Detection and Measurements,
Glenn T. Knoll,
Third edition (2000), John Willey .
1
2
PULSE COUNTING SYSTEMS
The signal chain shown in Fig. 17.7 represents a basic measurement
scheme in which only the number or rate of pulses from a radiation
detector are to be recorded. The tail pulse output of the preamplifier
typically has an amplitude of a few tens or hundreds of millivolts
and is too small to be counted directly. Furthermore, the pile-up of these
long pulses at high rates could cause stability problems. Therefore, the
next step is normally to process the pulses through a linear amplifier.
Here a voltage gain of 1000 or more caIl be provided so
that the shaped linear pulse at its output can easily cover a span of 0-10 V. The shaping requirements in a simple counting system are usually not
severe, and only at relatively high counting rates must one pay close
attention to the specific method of shaping chosen.
The interrelationship between the various signal pulses at different points
in a typical pulse processing system is illustrated in Fig. 17.8. The current
pulse from the detector are integrated on a long time constant circuit in
the preamplifier, producing a tail pulse output· The linear amplifier then
shapes this pulse to a much shorter width, and increases its amplitude by
a factor given by its gain.
3
4
5
A. Integral Discriminator
In order to count the pulses reliably, the shaped linear pulses must be
converted into logic:pulses. The integral discriminator is the simplest unit
that can be used for this conversion and consists of a device that produces a
logic output pulse only if the linear input pulse amplitude exceeds a set
discrimination level. If the input pulse amplitude is below the discrimination
level, no output appears. This selection process is illustrated in Fig. 17.9a.
Unless specifically designed otherwise, the logic pulse is normally produced
shortly after, the leading edge of the linear pulse crosses the discrimination
level. This leading edge timing is compared with other schemes of generating
the logic pulse in the later section on time pick-off methods. The
discrimination level is normally adjustable by a front-panel control. In many
counting situations. the level is set just above the system noise so that the
maximum sensitivity for counting detector pulses of all sizes is realized.
Other situations may call for a higher discrimination level to count selectively
only events above a set minimum size. For example, much of the background
may be limited to relatively low pulse amplitudes so that some finite
discrimination level may greatly enhance the signal-to-background counting
ratio. Integral discriminators must be designed to accept shaped linear input
pulses of a specific amplitUde span (usually 0-10 V positive). The stability
and linearity of the discriminator adjustment are usually adequate for routine
applications but may become important specifications for demanding
situations.
6
7
Differenrential Discriminator (Single-Channel Analyzer)
Another linear-to-Iogic converter in widespread use involves two independent
discrimination levels. As illustrated in Fig. 17.9b, a differential discriminator
or single-channel analyzer (SCA) produces a logic output pulse only if the input
linear pulse amplitude lies between the two levels. The action of the unit is
therefore to select a band of amplitudes or in which the input amplitude must
fall in order to produce an output pulse.
In some uni1ts lower-level discriminator (LLD) and upper-level discriminator
(ULD) are independently adjustable from front-panel controls. In others, the
lower level is labeled the E level, the window width or difference between
levels is labeled E and can be varied separately without affecting the E level.
In counting systems, the SCA can serve to select only a limited range of
amplitudie from all those generated by the detector.
A common example is one in which the window is set to correspond only to
those events in the detector that deposit the full energy of incident radiation.
In this way, one type or energy of radiation often can be measured selectively
in the presence of other radiations.
In normal. SCAs the time of appearence of the.logic pulse is not closely
coupled to actual event timing, and use of these logic pulses in timing
measurements will often give imprecise results. If one of the time pick-off
methods is incorporated into the SCA design, the logic pulse can be much more
closely correlated with this actual event time. Modules with this feature are
often called timing SCAs and are widely used in coincidence applications or 8
other timing measurements.
9
C. Scalers or Counters
As the final step in a counting system, the logic pulses must be accumulated
and their number recorded over a fixed period of time. The device used for
this purpose may be a simple digital register that is incremented by one count
each time a logic pulse is presented to its input.
such devices are sometimes called scalers as a historic anomaly that dates
from the time when digital registers of reasonable size were not widely
available. Then it was common to use a scaling circuit to divide the input pulse
repetition rate by a fixed factor such as 100 or 1000 so that the rate would
be low enough to be directly recorded by an electromechanical register. These
systems have been replaced by all-electronic digital registers. The scaling
function persists only in the sense that an overflow output pulse is often
provided when the maximum content of the register is exceeded (usually no
less than HP or 106 counts). We henceforth refer to such units as counters·
because that term more adequately describes the actual function.
Counters are commonly operated in one of two modes: preset time and
preset count. In the preset time mode, the counting period is controlled by an
internal or external timer. This timer may be built as part of a common chassis
with the counter, or separate timers can be obtained as individual modules.
In the preset count mode, the counter will accumulate pulses until a specified
total has been achieved, at which point the counting period is terminated. If
the period of time over which these counts have been accumulated can be
recorded independently, the counting rate can be determined.
The preset count mode has the advantage that a given statistical precision10can
Blind counters do not provide a visual display but, instead, can generate a
coded logic readout of the register content when triggered by an external
command. Because blind counters are less expensive than the display type,
they have found favor in large-scale systems in which many independent
counts must simultaneously be accumulated. In that event, the blind
counters are often part of a CAMAC system in which the interrogation and
readout take place over the dataway.
A printing counter is one in which an interface has been provided to
generate the proper readout signals to drive a conventional line printer or
other device. Other features found in some counters include an internal
input gate that can be controlled by a gate pulse supplied to the unit, or a
built-in integral discriminator to eliminate any noise that may appear along
with the input pulses. Other specifications of importance are the minimum
time separation between the leading edge of two logic pulses in order that
they be counted as separate events (the pulse pair resolving time) and the
maximum counting rate at which the counter may be driven.
11
Timers
The function of a timer is simply to start and stop the accumulation
period for an electronic counter or other recording device. Obviously
its most important property is the precision to which the time interval
is controlled. Two general methods of control are commonly
encountered. The simplest and least expensive method is to base the
timing interval on the frequency of the alternating current of the
power line to which the unit is connected. The precision of the timing
is therefore determined solely by the accuracy and stability of the
power line frequency. Utility companies usually do a good job of
controlling the accuracy of the power line frequency when integrated
over a day or more in order to maintain the accuracy of clocks also
synchronized to the line frequency. On the other hand, the frequency
may wander substantially over short periods of time, and timers based
on·power line synchronization thus may give rise to substantial timing
interval errors if the interval is less than a few hours. In order to
guarantee better accuracy, timers based on internal crystal-controlled
circuits are preferred for more exacting measurements. The most
important specification in this case then becomes the stability of the
timing frequency to changes in temperature.
12
I.. Counting Rate Meter
In some situations it is advantageous to have a visual indication of the
rate at which pulses are being counted in the system. counter in which
the experimenter visually observes the rate at which counts are
accumulated. Because of the random spacing between nuclear events,
small changes in counting rate are difficult to observe in this way,
particularly at low counting rates.
A counting rate meter provides a more direct means of indicating the
rate at which pulses are being accumulated. In its most common form,
a rate meter can be represented by the diode pump circuit outlined in
FIg. 17.10. The output stage of the logic device driving the rate meter
is represented by the voltage generator and series output impedance
Rf . Each logic pulse, as it enters the circuit, deposits a small fixed
amount of charge on the storage capacitor Cr- This capacitance is also
continuously discharged by a current that flows through the resistance
R. If the rate of arrival of logic pulses is constant, an equilibrium
will eventually be established in which the rate of charge deposition on
the capacitor is just equal to the rate of its discharge through the
resistance. Equilibrium is reached after several values of the time
constant of the circuit have elapsed following an increase or
decrease in the rate. This time constant is given simply by the product
of the capacitance Ct and the parallel resistance R.
13
If the conditions shown in the figure are met, the average voltage appearing
at the out“ put of the circuit is
v = iR = QrR = CFrR (17.2)
where r is the average rate at which pulses are supplied to the circuit, and
Q is the charge deposited per pulse given by the product of the coupling
capacitor Cf and the pulse ampli“ tude V. This output voltage is therefore
proportional to the rate of arrival of the input logic pulses.
If the input pulses were regularly spaced in time, the output voltage would
have the appearance sketched in Fig. 17.l1a. Longer time constants result in
a more nearly constant signal, but the response to abrupt changes in rate
will be slower. The full-scale range of the meter is normally varied by
selecting the value of R with a front-panel control. Other ratemeter circuits
have been developed ll that provide an output proportional to the logarithm
of the count rate. These meters allow compression of the counting rate scale
so that several decades may be monitored without the inconvenience of
switching between scales.
When dealing with events from a radiation detector, the spacing between
pulses is irregular and fluctuations in the output voltage arise as a result of
the random variation in‘ spacing. The rate meter signal then has an
appearance typical of that sketched in Fig. 17.llb. The standard deviation
(J' of this signal can be defined as the square root of the variance of the
values derived by sampling the signal many times at random and may be
derived as follows.
The differential contribution to the output voltage produced by a rate r 14
during the time between t and t + dt is (QrICI) dt. Because C1 is
15
16
17
F. Deadtime in Counting Systems
An important consideration in many counting applications is the loss of events
due to the:! dead time of the system. For some detectors (notably the G-M
tUbe) the detector mechllinism itself limits the minimum interval between
events for which two distinct pulses can b~counted. More often, however, the
detector will be capable of producing pulses that are s~
arated by a time that is less than the dead time inherent in the operation of
an electroni£component in the signal chain, and therefore it will be this
component that determines the system dead time. In the simple counting
systems shown in Fig. 17.7, this limiting component is usually the integral
discriminator or the SCA Although there are many exceptions, thEl'
dead time of a discriminator or SCA is typically related to the width of the
linear pulse pre-! sented to its input and is characteristically a microsecond or
two larger than this width. The validity of the corrections for dead time losses
discussed in Chapter 4 depends oUlt the assumption that the dead time is
constant for all events. The inherent dead time of elec..";
tronic units can sometimes vary with the amplitude or shape of the input pulse,
so that steptt to set the dead time artificially are warranted in some critical
applications. In this approach,‘ the dead time is standardized by an element
such as a linear gate, which is held closed for a:
fixed period of time following each pulse. This time is chosen to be larger than
the dead time of any component in the system, so that accurate corrections
can be made, even under con~ ditions in which wide variations in pulse
18
amplitude are encountered. The dead time of this
19
20
21
10.8 COINCIDENCE-ANTICOINCIDENCE MEASUREMENTS
In radiation measurements it is desirable or necessary to discard the
pulses due to certain types of radiation and accept only the pulses from
a single type of particle or from a particle or particles coming from a
specific direction.
Here are two examples of such measurements:
1. Detection of pair-production events. When pair production occurs,
two 0.511-MeV gammas are emitted back-to-back. To insure that only
annihilation photon are counted, two detectors are placed 180" apart,
and only events that register simultaneously (coincident events) in both
detectors are recorded.
2. Detection of internal conversion electrons. Radioisotopes emitting
internal conversion (IC) electrons also emit gammas and X-rays. The
use of a single detector to count electrons will record not only IC
electrons but also Compton electrons produced in the detector by the
gammas. To eliminate the Compton electrons, one can utilize the X-rays
that are emitted simultaneously with the IC electrons. Thus, a second
detector is added for X-rays and the counting system is required to
record only events that are coincident in these two detectors. This
technique excludes the detection of Compton electrons.
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
For example, consider a coincidence
measurement involving two detectors (Fig.
10.42). The detector signals are fed into a
coincidence unit, which then is used to gate
the corresponding ADCs. The amplified
detector pulses that are coincident are
thus digitized by the ADCs, and the
information is stored in the memory of the
system. Any event that reaches the
memory is defined like a point in a twodimensional space. For example, if a pulse
from ADC1 has the value 65 (i.e., ADC
channel 65) and one from ADC2 has the
value 18, the event is registered as 6518
(assuming 100 channels are available for
each parameter). The measured data may
be stored in the computer, for subsequent
analysis, and may also be displayed on the
screen of the monitor for an immediate
preliminary assessment of the results. The
results of a dual-parameter system such as
that shown in Fig. 10.42.
39
40