JP2637036B2 - Trigger device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a trigger device used for activating a fault recorder, a transient recorder, and the like, and more particularly to a trigger device for detecting and triggering a discrete phenomenon such as particles and radiation. Things.

[0002]

2. Description of the Related Art FIG. 4 shows a schematic configuration of an apparatus for recording the progress of a time series phenomenon such as a transient recorder.
First, the input signal I is supplied to the trigger circuit 11 and the delay circuit 12.
, And the input signal I
Is determined whether or not satisfies a predetermined condition. When the trigger circuit 11 determines that the predetermined condition is satisfied, the recorder 13 is activated. As a result, the input signal I that arrives at the recorder 13 with a time delay after the delay circuit 12 is stored in the recorder 13, and the state of the input signal I at that time can be analyzed later.

A signal used as a trigger may be an analog signal such as a voltage signal, but when connected to a radiation detector or the like, a random digital signal in the natural world is targeted.

For example, a transient recorder is disposed around a linac, a nuclear reactor, a cyclotron accelerator, or the like to detect and record when an environmental radiation level or a count level becomes higher than an allowable count level. May be used. In such an apparatus, it is common to perform counting for a certain time by a radiation detector such as a GM tube, to detect when the value exceeds a certain value as a burst, and to trigger based on this. I have.

[0005]

However, if a burst is detected based on the measurement of such a fixed time, the following problem occurs. That is, when a random count is measured, if the count period is long and a sufficient number of counts can be obtained as a population, a normal distribution is exhibited. But,
When the measurement period is short, it is known that the measurement follows a Poisson distribution. The approximate expression in the Poisson distribution is as follows.

[0006]

(Equation 1)

Here, λ is the average number of input pulses per unit time, and x is the actual number (the number of pulses actually input). Now, when the measurement time is long enough and one condition is input stochastically during a certain measurement period (this is also called statistical probability), how the counting will be performed according to the Poisson distribution Substitutes 1 for λ and sets x to 0
It can be found by solving equation (1) while changing the value in the range of 10.
The results are as shown in Table 1 below.

[0008]

[Table 1]

From Table 1, it can be seen that the probability that the count is 0, that is, the possibility of x = 0, is approximately 36.8% under the condition that one is input stochastically. At the same time, under the condition that one is input stochastically, the probability that the count indicates 9;
It turns out that the possibility of x = 9 is 1 / 1,000,000.
In addition, the probability that the count indicates 10, x = 10, is not completely zero. Therefore, for example, in burst detection,
Even when the trigger level, which is the threshold value, is set to the count value x = 9 to 10, the condition (λ = 1) where one is input stochastically is detected and the burst is determined. It indicates that there is a possibility to happen.

Generally, the count P0 is equal to the trigger level Pt.
h, that is, whether P0> Pth or P0 ≧ Pth is compared to determine the burst in many cases. The integrated expected values (expected value integration) and the expected values in each case of x = 1 to 10 where λ = 1 to 10 are summarized below, which are used when determining a burst based on the comparison of P0 ≧ Pth. It is. In Table 2, for example, when λ = 1, Pth = 2 is set.
It can be seen that the probability of reaching this trigger level is about 92%, which is the sum of the expected values of x = 0, 1, 2.

[0011]

[Table 2]

In such a burst detector, for example, suppose that Pth is set to 9 under the condition of λ = 1 while allowing a one-millionth malfunction. Then, it is assumed that an abnormality has occurred in the accelerator and a state in which five are input stochastically, that is, a state of λ = 5 is reached. In this case, the possibility of reaching x = 9 which is Pth from the expected value integration is almost 97
%. Therefore, it is not possible to detect an increase with a probability of less than 3%. Thus, in the simple digital comparison circuit as described above, it is impossible to reliably detect a burst with a small number of counts. Of course, it is possible to treat as a normal distribution by lengthening the measurement time and increasing the count value, but the detection speed is reduced, and it is difficult to function as a trigger device used in a transient recorder or the like. Becomes

In view of the above problems, it is an object of the present invention to provide a trigger device which has a high detection speed and is capable of detecting a burst while reliably preventing a malfunction. .

[0014]

In order to solve the above-mentioned problems, in the present invention, the result of comparison based on the Poisson distribution is further processed using a binomial distribution, and bursts are reliably detected while maintaining the detection speed. It is possible. That is, the trigger device according to the present invention includes a counting means capable of counting continuously for a predetermined time, and a flag setting means for setting a flag when the count value of the counting means is equal to or greater than a preset reference count value. A flag holding means for performing an evaluation based on the Poisson distribution and further holding a state of a plurality of flags set by the flag setting means; a flag counting means for counting the flags held in the flag holding means; An evaluation based on a binomial distribution is performed by a flag number comparing unit that compares the number of flags counted by the flag counting unit with a preset reference flag number.
Then, a determination means for outputting a trigger signal based on the result of the flag number comparison means is provided.

In such a trigger device, it is desirable to have reference count value setting means capable of variably setting the reference count value, and it is also effective to have reference flag number setting means capable of variably setting the reference flag number. .

[0016]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiments of the present invention will be described below based on actually used circuits. FIG. 1 shows a configuration of a trigger device according to an embodiment of the present invention. The present apparatus includes a counting circuit 2 capable of detecting and counting radiation for a predetermined time under the control of a reference clock 1, a count P measured by the counting circuit 2, and a trigger level Pth preset by a threshold setting unit 3. And a comparison circuit 4 for comparing
, An 8-bit shift register 5 for setting a flag when the count P ≧ Pth is satisfied, a flag check circuit 6 for checking the number of flags set in the shift register, and a flag counted by the flag check circuit 6 A flag comparing circuit 7 for comparing the number F with a reference flag number Fth preset by the reference flag setting section 8
And a signal transmission circuit 9 for outputting a burst signal based on the result of the flag comparison circuit 7.

The counting circuit 2 is based on the reference clock 1,
For a predetermined period, measure the number of detected radiation events,
It is output to the comparison circuit 4 as the count value P. Thereafter, the count is reset, measurement is performed again for a predetermined period, and this operation is repeated. The comparison circuit 4 compares the count P input from the counting circuit 2 with the trigger level Pth, and counts P ≧ Pth
Is established, a flag is set in the shift register 5 that is performing a synchronous operation with the same reference clock 1. The shift register 5 is an 8-bit shift register and incorporates the presence / absence of a flag according to the reference clock 1, so that the results of the comparison circuit 2 in eight counting periods can be accumulated retroactively. Then, the flag check circuit 6
Is the number of flags F recorded in the shift register 5.
Can be counted, so that the flag check circuit 6 can refer to the results of the past eight counting periods.

For example, suppose a condition that the statistical probability of the counted input pulse is 1 within a predetermined period. This condition is the average number of input pulses per unit time described above, and corresponds to the condition where λ is 1. Then, Pth is set to 4. Under this condition, the probability that the flag is set in the shift register 5 by the comparison circuit 4 is an expected value in which the number of actually input pulses is 4 or more,
= 0 to 3 is the probability of subtracting the integrated value of the expected value. That is, the probability F (Pth) that the flag is set in the register is as follows: F (Pth) = 1- (expected value integration of x = 3) = 1-0.981012 = 0.018988 (2) This corresponds to the hatched portion 21 of the Poisson distribution (λ = 1) shown in FIG.

When a flag is set with such a statistical probability,
Among the 8-bit shift registers 5, the number and the probability of the registers where the flag is set follow a binomial distribution. Therefore, the probability R (y) that a flag is set in y registers R is:

[0020]

(Equation 3)

Here, N is 8 since the shift register has 8 bits, and 0.018988 obtained by the equation (2) is used as F (Pth). The values of R (y) when the value of y is changed from 0 to 8 are summarized in Table 3 below.

[0022]

[Table 3]

As can be seen from Table 3, when y = 6 or more, R
(Y) becomes 0 to about 8 digits after the decimal point, and can be regarded as 0 for measurement. Therefore, the number of flags counted by the flag check circuit 6 is changed by 6 to the reference flag number Fth set by the reference flag number setting unit.
By performing the comparison in the flag comparison circuit 7 as described above, the statistical probability of the input pulse to be counted set to 1 is 1
Under the condition, that is, the condition corresponding to the condition of λ = 1, it is possible to completely prevent the trigger device from malfunctioning.

As described above, in the trigger device of this embodiment, random natural phenomena measured by the radiation measuring device or the like from the counting circuit 2 and the comparing circuit 4 are processed based on the Poisson distribution, and the result is further shifted by the shift register. 5 and the flag check circuit 6 can perform processing based on the binomial distribution. As a result, it is understood that a random natural phenomenon occurring at a predetermined statistical probability can be detected without malfunction. Of course, even with a conventional device that performs a trigger based only on the Poisson distribution, the probability of malfunction can be reduced to 1 / 100,000 or 1 / 10,000 by increasing Pth sufficiently. Thus, at first glance, this difference may be negotiable small.
However, if a computer that performs a calculation once per microsecond is normally present and the probability of the miscalculation is one millionth or one tenth of a million, the result of the calculation is one second or one second.
A malfunction occurs once every 0 seconds. Therefore, there is a great difference between the fact that the probability of malfunction is very small and that the probability of malfunction is zero.

As described above, by using the present apparatus,
The measurement time is short, and it is possible to detect a burst while preventing a malfunction while following a natural phenomenon.

As described above, each count rate (λ)
F (Pth) = 4, the number of reference flags Fth
Table 4 below shows the result of calculating the expected value at = 6.

[0027]

[Table 4]

As can be seen from this table, a high probability is exhibited when the input count is 5 or more. The purpose of the present device is to reliably detect the transition to the steady state or the transient state as soon as possible, and furthermore, malfunction in the steady state is not allowed. Therefore, as described above, when the input count is 5 or more, the burst can be reliably detected by adjusting the measurement time, Pth, Fth, and the number of memories.

In the trigger device of this embodiment, the reference flag number Fth can be freely set by using the reference flag setting section 8, and the trigger level Pth
Can also be freely set by the threshold setting unit 3. When measuring natural decrease, the average number of input pulses per unit time of the Poisson distribution can be freely selected by changing the measurement time. Therefore, in such a trigger device, first, it is desirable that the number of reference flags be about 80% of the number of measurements that can be stored in the shift register, and in this example, it is desirable that the number be 6 for an 8-bit register. Then, the trigger level Pth, which is the threshold value of the Poisson distribution, may be set to a value at which the number of repetitions y in the binomial distribution is considered to be 0 at the previously set Fth.
Furthermore, by setting the measurement time to a time that is the average number of input pulses at which the trigger level Pth can be measured in the measurement of the Poisson distribution, it is possible to detect a random natural phenomenon without malfunction by the trigger device described above. Become.

[0030]

As described above, the trigger device according to the present invention further processes the comparison result of the Poisson distribution for random natural phenomena using the binomial distribution, and has a short detection time. Malfunction can be prevented, and burst can be detected reliably. Therefore, in a device in which it is difficult to secure a large count value to be detected based on an accurate normal distribution from a measurement time, an accurate and high-speed trigger can be realized instead of the conventional trigger device including only the Poisson distribution. can do.

[0031]

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of a trigger device according to an embodiment of the present invention.

FIG. 2 is a graph showing a Poisson distribution.

FIG. 3 is a graph showing a binomial distribution.

FIG. 4 is a block diagram showing an outline of functions of a transient recorder.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 ... Reference clock 2 ... Counter circuit 3 ... Threshold setting circuit 4 ... Comparison circuit 5 ... Shift register 6 ... Flag check circuit 7 ... Signal output circuit 11 ... Trigger Circuit 12 ・ ・ ・ Delay circuit 13 ・ ・ ・ Recorder

Claims (3)

(57) [Claims] 1. A counting means capable of counting continuously for a predetermined time, a flag setting means for setting a flag when the count value of the counting means is equal to or larger than a preset reference count value, and a flag setting means. Flag holding means capable of holding the states of a plurality of set flags, flag counting means for counting the flags held in the flag holding means, the number of flags counted by the flag counting means and a predetermined reference A trigger device comprising: a flag number comparing unit that compares the number of flags; and a determining unit that outputs a trigger signal based on a result of the flag number comparing unit. 2. The trigger device according to claim 1, further comprising a reference count value setting unit that can set the reference count value. 3. The trigger device according to claim 1, further comprising a reference flag number setting unit capable of setting the reference flag number.