JPH0743407A - Characteristic value measuring method for electronic circuit - Google Patents

Abstract

PURPOSE:To measure characteristic values of a number of electronic circuits efficiently. CONSTITUTION:In the application of a binary search method, a subject is taken as assembly 101 of electronic circuits which have a range (range of forecasting characteristic values)allowing the obtaining of characteristic values of a plurality of electronic circuits and measurement by the binary search method is performed by a comparison judgment means 104 using the same comparison reference value (value determined within the characteristic value forecasting range) for the individual electronic circuits belonging to the assembly. The plurality of electronic circuits belonging to the assembly are separated into two assemblies having different characteristic value forecasting ranges to store by a subject memory means 106 based on the results of the judgment in comparison with the comparison reference value. Thereafter, the measurement by the binary search method is repeated in the same way for the electronic circuit belonging to the assembly separated and when the characteristic value forecasting range in the respective assemblies are held below a fixed value, the characteristic values of the electronic circuits belonging to the respective assemblies are determined as fixed value in the characteristic value forecasting range.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for measuring characteristic values of electronic circuits, and more particularly to a method for efficiently measuring characteristic values of many electronic circuits.

[0002]

2. Description of the Related Art A method called a binary search method is known as a method for measuring the characteristic value of an electronic circuit. This method is characterized in that the absolute value of the desired characteristic value is not directly measured, but is indirectly measured by relative comparison measurement. In particular, when the characteristic value to be obtained can be predicted to some extent in advance and the magnitude relationship with a predetermined value can be easily determined, it is excellent in that efficient measurement can be realized. The binary search method will be described below with reference to FIG.

First, let α be the magnitude of the desired characteristic value,
Predicted range of this characteristic value (minimum value a ₁ , maximum value b
₁ ) (Fig. 5 (a)). Next, the central value of the minimum value a ₁ and the maximum value b ₁ (hereinafter referred to as the comparison reference value) is M
₁ (= (a ₁ + b ₁ ) / 2), and this comparison reference value M ₁
And the value α to be obtained are compared and judged (see FIG. 5).
(B)). As a result of this comparison judgment, the desired value α is in two ranges, that is, (between the minimum value a ₁ and the comparison reference value M ₁ ).
It becomes clear which range is between or (between the comparison reference value M ₁ and the maximum value b ₁ ).

If the desired value α is the comparison reference value M₁ than
If it is larger, α is the comparison reference value M ₁ And the maximum value b₁ Between
Will be in the range. Therefore, next to the value α
As the prediction range, the minimum value a₂ , Maximum value b₂ To
New comparison reference value M in this range₂ ((A₂ 
+ B₂ ) / 2) is set (FIG. 5 (c)). At this time,
In the case shown in FIG. 5B, a₂ = M₁ , B₂ =
b₁ Becomes

Next, as in the case of the first range (minimum value a ₁ and minimum value b ₁ ), the comparison reference value M ₂ and the desired value α are compared and judged, and Prediction range (minimum value a ₃ , maximum value b ₃ ) is defined. If the desired value α is smaller than the comparison reference value M ₂ , a ₃ =
a ₂ and b ₃ = M ₂ (corresponding to the case of FIG. 5D).

In the following, the prediction range of the value α to be obtained by the same method is gradually narrowed so as to be divided into two. Then, if the prediction range (| b _i −a _i | (i = 1, 2, ...)) becomes sufficiently small, the desired value α can be set in a sufficiently accurate range, and approximately α = It is calculated as a _i (or b _i ). FIG. 6 is a general flowchart of the above method. The expected range of the value α to be obtained is set, the magnitude relationship of the value α to be obtained is compared and judged with respect to the comparison reference value, which is the center point of the expected range, and the previous expected range is divided into two according to the result. The expected range is reset, and thereafter, the above steps are repeated until the expected range becomes equal to or less than a predetermined value (minimum resolution t), and the minimum or maximum value of the expected range at that time is obtained. This is an approximate value of the value α. Expected range is (1/2) ^n-1
Since it converges at a ratio of (n is the number of repetitions), it is possible to obtain an approximate value of the value α to be obtained with a relatively small number of times.

An application example of the method for measuring the characteristic value of an electronic component using the binary search method will be described below. FIG. 7 is a block diagram showing a measuring means for measuring the access time of the memory device. The access time is an important characteristic value of the memory device, and is defined as the time after the clock pulse is applied to the memory device until the output is determined.

In FIG. 7, the prediction range setting means 202 sets a range of access time predicted in advance for the memory device 201, and the comparison reference value setting means 203 sets the center point of the access time range according to the data of this access time range. The comparison reference value (measurement time (M) in this case) is set. A clock pulse is applied to the memory device, its output is input to the comparison / determination circuit 204, its output is measured, and if the output is confirmed within the above measurement time (M), "pa
If it is not determined, it is judged as “fail.” This judgment corresponds to the comparison judgment of the magnitude relationship between the access time desired by the memory device and the measurement time (M),
Such a determination can be easily made in a short time by observing the operation state of the memory device.

The result of the comparison / decision circuit 204 ("pass")
Or, according to "fail"), the prediction range setting means 202 sets again a new access time range obtained by dividing the previous access time range, and thereafter performs the same measurement, and the access time range has a predetermined value (minimum resolution t). Repeat until converged to. The access time to be finally obtained is approximately obtained by using the minimum value or the maximum value of the converged access time range or the measured time when the access time converges.

If the variation in access time of the memory device is 50%, the measurement by the above binary search method is repeated four times (convergence rate (1/2) ³ = 1/8). ), A measurement accuracy of about 6% is obtained, and highly accurate measurement can be realized in a short time measurement.

[0011]

As described above, since the measurement of the characteristic value of the electronic circuit by the binary search method can realize the accurate measurement in a short time, even when the characteristic value of a large number of electronic components is measured. Many are used. However, even when such characteristic values of a plurality of electronic components are measured by the binary search method, the variations in the characteristic values of the individual electronic components are irregular, and thus all the measurements must be performed individually. Normally, the range of expected variations in the characteristic values of a plurality of electronic components is set as the initial expected range in the binary search method, so if all individual measurement accuracy is the same,
The number of repetitions n is also set to be the same in all individual measurements. Therefore, assuming that one measurement time is T, it takes (N × n × T) time to measure N electronic components. Therefore, in the case of measuring the characteristic values of a large number of memory integrated circuits, there is a problem that the total measurement time becomes considerably long even if the measurement time for each time is short.

The present invention solves the above-mentioned conventional problems, and an object of the present invention is to provide a method for efficiently measuring the characteristic values of a large number of electronic components.

[0013]

In order to achieve this object, an electronic circuit characteristic value measuring method according to the present invention is characterized in that when a conventional binary search method is applied, the object of measurement is the characteristic of a plurality of electronic circuits. Possible value range (Characteristic value prediction range)
And the measurement by the binary search method using the same comparison reference value (a predetermined value within the characteristic value prediction range) for each electronic circuit belonging to the collection. Then, based on the result of the comparison judgment with the comparison reference value, the plurality of electronic circuits belonging to the assembly are separated into two assemblies each having a different characteristic value prediction range, and then separated into each assembly. Repeat the measurement by the binary search method in the same way for the electronic circuit belonging to the assembly,
When the characteristic value prediction range in each aggregate converges to a certain value or less, the characteristic value of the electronic circuit belonging to each aggregate is determined as a constant value within the characteristic value prediction range. .

[0014]

According to the method of the present invention, one characteristic value prediction range including a range of variation of characteristic values of a plurality of electronic circuits predicted in advance is set, and a plurality of electronic circuits included in the characteristic value prediction range are set. As an aggregate, and for each electronic circuit belonging to this aggregate, by comparing and determining the magnitude relationship of the characteristic values by the binary search method using the same comparison reference value, the aggregate of the electronic circuit, A conventional dichotomous search method for individual electronic circuits by sequentially "convoluting" a set of electronic circuits having new (divided) characteristic value prediction ranges narrower than the characteristic value prediction range. At
It is possible to eliminate the requirement that measurement of all electronic circuits must start from the initial characteristic value prediction range,
It is possible to perform efficient measurement without performing measurement in an unnecessary range, and it is possible to significantly reduce the measurement time.

[0015]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 shows a flow chart of a characteristic value measuring method for a plurality of electronic circuits according to an embodiment of the present invention. First, the magnitude of the characteristic value of each of the n electronic circuits to be obtained α = (α ₁ , α ₂ , α ₃ , ..., α _n ) and the prediction in all the electronic circuits in consideration of the variation in all the electronic circuits A set P having the range of characteristic values (minimum value a, maximum value b) as the elements of the set is defined. Let M be the center point of the minimum value a and the maximum value b (hereinafter referred to as the comparison reference value).

P = {α ₁ , α ₂ , α ₃ , ..., α _n , (a, b)} M = (a + b) / 2 Next, the comparison reference value M and the desired characteristic value (α ₁ , α ₂ ,
α ₃ , ..., α _n ) are compared and judged.
As a result of this comparison determination, it is determined whether the n electronic circuits are in two ranges, that is, between (minimum value a and comparison reference value M) or (between comparison reference value M and maximum value b). Prove. Therefore, as the next characteristic value prediction range of the set of values to be obtained, a set having a larger characteristic value than the comparison reference value M is set as a set P ₀ including the maximum value (b) and the minimum value (M), and the comparison reference value M A set having a smaller characteristic value than the above is set as a set P ₁ including the maximum value (M) and the minimum value (a).

P₀ = {Α₁ ，… ， Α_k, (M, b)} P_l= {Α_{k + 1} ，… ， Α_l, (A, M)} First, the set P₀ Is a new target for comparison
As M₀ To set. M₀ = (M + b) / 2 Next, for example, the comparison reference value M₀ And the desired characteristic value (α
₁，… ， Α_k) Is compared and judged. This comparison
As a result, the set P₀ There are two more electronic circuits included in
Range, ie (comparison reference value M and comparison reference value M₀ Between)
Or (Comparison reference value M₀ And the maximum value b)
It turns out to be in range. Comparison reference value M ₀ Than the characteristic value
The maximum value (b) and the minimum value (M₀ ) Including
Combine P₂ And the comparison reference value M₀ Set with smaller characteristic value than
The maximum value (M₀ ) And the minimum value (M), the set P₃ Tosu
It

P ₂ = {α ₁ , ..., α _l , (M, b)} P ₃ = {α _{l + 1} , ..., α _k , (a, M)} The above method is used to predict the characteristic value. If the prediction range becomes sufficiently small by repeating the set with a narrow range, the desired value is determined with a sufficiently high accuracy range, and the characteristic values α _i of all the electronic circuits included in the set in the prediction range are set. Approximately, α _i = M (or a, b) is obtained.

FIG. 2 is a block diagram of a measuring apparatus when the measuring method according to the embodiment of the present invention is applied to measuring the access time of a plurality of memory devices. The predicted range setting means 102 sets a range of access time predicted in advance for the plurality of memory devices 101, and the comparison reference value setting means 103 is used by using the data of the access time range.
The measurement time (M) which is the center point of the access time range is set with. A clock pulse is applied to the plurality of memory devices 101, the outputs thereof are input to the comparison / judgment circuit 104, the outputs thereof are measured, and if the outputs are confirmed within the above measurement time (M), it must be confirmed as "pass". For example, the judgment "fail" is made.

The result of the comparison / decision circuit 104 ("pass")
Alternatively, according to "fail"), the prediction range setting means 102 sets again a new access time range obtained by dividing the previous access time range into two. At this time, the measurement target storage means 10
6, the result of the comparison / determination circuit 104 is "pas".
Each identification code of the electronic circuit assembly that has become "s" and each identification code of the electronic circuit assembly that has become "fail" are temporarily stored with the values of the predicted range of each electronic circuit attached, and sequentially. By reading the stored identification code of the aggregate and repeating the above-described measurement method for each aggregate from which the identification code is read, the comparative determination can be performed for all the aggregates.

The above method is repeated until the access time range converges to a predetermined value (determined by the convergence determination means 105). The access time of each memory device 101 to be finally obtained is approximately obtained by using the minimum value or the maximum value of the converged access time range or the measurement time at the time of convergence. The effect of this embodiment will be specifically described in comparison with the case of the conventional measurement performed on individual devices.

FIG. 3 shows a measurement example in which the access times of four memory devices are obtained according to the above-described measurement method, which will be described with reference to FIGS. 2 and 3. First, the prediction range setting means 102 sets a range in which the measured values of access times of all the memory devices 101 are always included (minimum value 75 ns, maximum value 83 ns) (FIG. 3A).

Next, the comparison reference value setting means 103 sets the measurement time (M ₁ ) to a minimum value of 75 ns and a maximum value of 83 n.
The position is set to 1/2 of s (79 ns), and the comparison / determination means 104 is used to determine pass / fail for all the memory devices 101 (FIG. 3B). As a result, the memory devices P, Q and R become "pass" and the memory device S becomes "fail". Where "pas
The identification code of the memory device P, Q, R that has been "s", the minimum value of 75 ns and the maximum value of 79 ns (hereinafter referred to as set A), and the identification code of the "failed" memory device S and the minimum value. 79 ns and a maximum value of 83 ns (hereinafter, referred to as set B) are stored in the measurement target storage unit 106.

Next, the set B stored in the measurement target storage means 106 is newly read as a measurement target (the order of reading the sets A and B is arbitrary). The measurement reference time setting means 103 measures the measurement time (M ₂ ) at a position 81 which is a half of the stored minimum value 79 ns and maximum value 83 ns.
set to ns, and the comparison / determination means 104 passes / f
The judgment of ail is made (FIG. 3 (c)). As a result, the memory device S becomes "pass".

From the above result, the memory device S which has become "pass", the minimum value of 79 ns and the maximum value of 81 ns (hereinafter, referred to as a set C) are stored by the measuring object storage means 106. Since there is no memory device that has become “fail” by the determination unit 104, the measurement is continued for the set C. Next, the measurement time (M ₃ ) is set by the comparison reference value setting means 103.
Position 80n which is 1/2 of the minimum value 79ns and the maximum value 81ns
s, and the comparison / determination means 104 passes / fa
The judgment of il is performed (FIG. 3 (d)). As a result, the memory device S becomes "fail". Here, the memory device S, the minimum value is 80 ns, and the maximum value is 81 ns.
Is stored in the measurement object storage means 106. Since the convergence width of the measurement value prediction range is set to 1 ns, the maximum value 81
ns is defined as the measured value of the memory device S (minimum value 80
(ns may be defined as a characteristic value) Next, a set of measurement values, which is stored in the measurement object storage means 106 and whose measurement values are not yet determined, is read as a measurement object, and the measurement is continued. Here, the set A is read as the measurement target. The comparison reference value setting means 103 sets the measurement time (M ₄ ) to a minimum value of 75 ns and a maximum value of 79 ns.
Of the device P, Q, R
On the other hand, the comparison / determination means 104 determines the pass / fail (FIG. 3E). As a result, the memory device P
Is "pass", and memory devices Q and R are "fail"
Becomes Here, the “pass” memory device P
And a minimum value of 75 ns and a maximum value of 77 ns
The memory devices Q and R that have been “failed” (hereinafter, referred to as a set D) and 77 ns as a minimum value and 79 ns as a maximum value (hereinafter, referred to as a set E) are stored by the measurement target storage means.

Next, the set E is newly set as a measurement target.
Measurement time (M ₅ ) by the comparison reference value setting means 103
Of the stored minimum value 77 ns and maximum value 79 ns
The position / 1/2 is set to 78 ns, and the comparison / determination unit 104 determines pass / fail (FIG. 3 (f)). As a result, the memory devices Q and R become "fail".
Since the measured value prediction range has converged to 1 ns or less, the maximum value 79 ns is determined as the measured value of the memory devices Q and R.

Finally, the measurement is continued with the set D stored in the measurement object storage means 106 as the measurement object.
Measurement time (M ₆ ) by the comparison reference value setting means 103
Of the stored minimum value 75 ns and maximum value 77 ns
The position of / 2 is set to 76 ns, and the comparison / determination unit 104 determines the pass / fail (FIG. 3 (g)). As a result, the memory device P becomes "pass". Here, the memory device P, the minimum value of 75 ns, and the maximum value of 76 ns are stored by the measurement target storage means 106. However, since the measured value prediction range has converged to 1 ns or less, the maximum value of 76 ns is stored in the memory device P. Confirm as the measured value.

With the above, the measured values of the access time of all the memory devices 101 are fixed. The effects of this embodiment will be specifically described below in comparison with the case of single measurement. Consider the case where the number of memory devices is set to 4 and the minimum width of the measurement value prediction range is set to 1 ns, and the measurement value of the access time is measured. Since the minimum width of the measurement value prediction range is 1 ns and the measurement value prediction range is 8 ns, at least three (2 ³ = 8) binary searches are required to measure the access time of one memory device. is necessary. Therefore, twelve binary searches are required to measure the access time of four memory devices. On the other hand, in the present embodiment, the access time of 4 memory devices was measured by 6 binary searches. In this embodiment, the maximum number of binary searches occurs when the access time of each memory device is present every 2 ns, which is the minimum width of the measurement value prediction range. In the search, the difference between the comparison reference value and the minimum value is 1 ns, and at the same time, the difference between the comparison reference value and the maximum value is 1 ns.) At this time, the access time of the memory device can be measured by seven binary searches.
In other words, the effect increases as the number of memory devices to be measured increases. FIG. 4 is a graph showing the relationship between the number of measurement devices and the number of convergences (which is determined by the measurement value prediction range and the minimum value of the measurement value prediction range) based on this embodiment. By using this algorithm, the measurement time can be significantly shortened by setting the expected range of the characteristic values of a plurality of memory devices for every several aggregates. As described above, it can be seen that the larger the number of measurements, the shorter the measurement time and the more efficient the measurement.

As a result of using an actual memory device, the measured value has a binomial distribution and has a small variation, so that the number of times of convergence can be further saved (about half of the theoretical calculation value).

[0030]

According to the characteristic value measuring method of the electronic circuit of the present invention, by comparing and judging the magnitude relation of the characteristic values by the binary search method using the same comparison reference value for each electronic circuit, The conventional electronic circuit is performed by "convoluting" the set of electronic circuits into a set of electronic circuits having a new characteristic value prediction range that is narrower than the characteristic value prediction range. In the binary search method, the requirement that measurement of all electronic circuits must start from the initial characteristic value prediction range is eliminated, and efficient measurement that does not perform measurement in an unnecessary range is possible, resulting in a large measurement time. Therefore, the characteristic values of a large number of electronic components can be efficiently measured.

[Brief description of drawings]

FIG. 1 is a flowchart of a plurality of electronic circuit characteristic values according to an embodiment of the present invention.

FIG. 2 is a block diagram of a plurality of electronic circuit characteristic values in one embodiment of the present invention.

FIG. 3 is an explanatory diagram of a method of measuring a plurality of electronic circuit characteristic values in one embodiment of the present invention.

FIG. 4 is a graph showing the relationship between the number of convergences and the number of measurements for explaining the effect of the invention in one embodiment of the present invention.

FIG. 5 is an explanatory diagram of a conventional binary search method.

FIG. 6 is a flowchart of a conventional binary search method.

FIG. 7 is a block diagram of a conventional binary search method.

[Explanation of symbols]

 101 memory device 102 prediction range setting means 103 comparison reference value setting means 104 comparison determination means 105 convergence determination means 106 measurement target storage means

Claims (3)

[Claims] 1. A method for measuring characteristic values of a plurality of electronic circuits by using a binary search method, the characteristic value prediction range having maximum and minimum possible characteristic values of the plurality of electronic circuits. And a plurality of electronic circuits belonging to the characteristic value prediction range,
A second step of making a comparison decision by a binary search method for the characteristic value of each electronic circuit by using a certain comparison reference value within the characteristic value prediction range; The electronic circuit is divided into a first group of electronic circuits having a characteristic value larger than the comparison reference value and a second group of electronic circuits having a characteristic value smaller than the comparison reference value. In the third step, for the electronic circuits belonging to the first aggregate, the comparison reference value is set to the minimum value, and the maximum value of the characteristic value prediction range in the previous comparison determination is set to the current comparison determination time. Of the characteristic value prediction range is set to the maximum value, and the comparison reference value is set to the maximum value for the electronic circuits belonging to the second aggregate, and the characteristic value prediction range at the time of the previous comparison determination is set to the minimum value. Value is the minimum value for this comparison And a fourth step of setting a characteristic prediction range, wherein the second, third, and fourth steps are performed on the characteristic value prediction range for electronic circuits belonging to the first and second aggregates, respectively. Is repeated until it converges to a certain value or less, and at a predetermined value within the characteristic value prediction range at the time of convergence,
A characteristic value measuring method for an electronic circuit, wherein the characteristic values of the electronic circuits respectively belonging to the first aggregate and the second aggregate are used. 2. The identification codes of the electronic circuits respectively belonging to the first aggregate and the second aggregate are temporarily stored in a storage device together with the values of the characteristic value prediction range, and the first aggregate and the second aggregate are stored. 2. The electronic circuit according to claim 1, wherein the identification code and the value of the characteristic value prediction range of the electronic circuit belonging to each aggregate are read from the storage device when the measurement is performed on the electronic circuit belonging to each body. Characteristic value measurement method. 3. The method for measuring a characteristic value of an electronic circuit according to claim 1, wherein the comparison reference value is set to the center value of the characteristic value prediction range.