JPS601578A - Tester for integrated circuit - Google Patents

Abstract

PURPOSE:To test an integrated circuit easily and quickly with free control of the weight or the like of a test pattern by controlling a random number distribution converter for controlling the random number responding to a random number generator with an external binary digit. CONSTITUTION:The distribution of a uniform random number of a fundamental input from a uniform random number generator 10 is controlled with a random number distribution converter 12 according to binary digits alpha and beta from an external unit 11. With the value alpha, the right and left movement of the probability distribution varies with a conversion function unit 13 of an equipment 12 to determine the right and left balance with respect to the weight center while with the value beta, the geometry of the distribution changes through a conversion function unit 14 to control the ratio of magnitude between the center peak of the probability distribution and both ends thereof. A test pattern is outputted from a test pattern generator 15 responding to the equipment 12 to test a 2 input AND gate with the probability of 3/4 for the appearance of 0, for instance, by a pattern of the corresponding appearance probability thereby enable the testing of an integrated circuit easily and quickly.

Description

DETAILED DESCRIPTION OF THE INVENTION [Technical Field to which the Invention Pertains] The present invention relates to an integrated circuit testing device applicable to testing LSIs and the like.

[Prior art and its problems]

Conventionally, various test pattern generation devices have been devised, but among random number input generation devices, devices that generate test patterns by inputting generated random numbers as they are or by uniformly converting them have been mainly proposed. For example, if the :l easy matching number of 24 human-powered circuits is AD converted and generated as a test pattern, 24 - sets of 11" or 'O" will be
It is formed as one test turn. Therefore, the number of TL 1 // in the - set (hereinafter abbreviated as "1 weight")
is plotted on the horizontal axis, and the probability of occurrence corresponding to that 1 turn is plotted in 1 vertical column.The distribution of 3000 sets of test turns is shown in the 11th threshold using a uniform random number generator.

From the figure, it can be seen that as long as the force is a uniform random number, its distribution becomes a binomial distribution.

Also, the uniform transformation, ie the appearance of V″1″*0/l6;'
If the +f ratio is simply changed, the distribution shown in Figure 1 is nothing but the distribution shifted horizontally to the east (1 degree). However, in the above method, the 'l' distribution becomes a binomial distribution. Therefore, when testing by generating as few (1) test patterns as possible, even if you want to generate patterns with various weights, the patterns limited to the probability portion XU will not occur (1). ; Many 1.).

For example, if there are 20 10'' and 4 ``1'' test turns, the probability is approximately 0.03%, that is, they will appear 3 times in 10,000 badans. Only.

Even when further uniform transformation is applied, the same thing can be said for the 'weights' in the opposite direction that are translated in parallel, and this tendency becomes more pronounced as the number of inputs increases.For example, AND
The conventional method cannot be applied at all when the structure is composed of a gate and an OR gate and the ratios thereof are different.

[Purpose of the invention]

The present invention takes this point into consideration and aims to facilitate testing of integrated circuits by freely controlling the weights of test patterns using external data such as circuit data.

[Summary of the invention]

The present invention provides a random number distribution conversion device that controls the distribution of random numbers generated after a uniform random number generator, and further includes a test pattern generator that outputs a specific logical value with a predetermined probability according to this random number distribution. According to the present invention, the external binary values control the movement of the appearance probability distribution of the logical values in the weight axis direction and the change in shape thereof. [Effects of the Invention] According to the present invention, the external binary values By fully controlling the amount of movement and changes in the distribution shape, it becomes possible to freely set the ``1'' probability distribution, and the test time for integrated circuits is greatly shortened.

For example, for an integrated circuit that includes an AND gate 1 and an OR gate y at a predetermined ratio, the ratio of the appearance probabilities of the edges on the axis is calculated according to the ratio of the two, and the shift in the weight axis direction is performed. The quantity is set to achieve a balance according to the ratio, and then by changing the distribution shape from a binomial distribution to a concave shape, the size of the appearance probability is set at each end, and this is extremely short. It is possible to generate a test pattern that completes the test in time.

[Embodiments of the invention]

Hereinafter, one embodiment of the present invention will be described in detail using the drawings.

To explain the outline of the device, first, a conversion function is prepared, and this is defined as k G t (xi. Here, specifically, Q
1 (xl-tan 2X ・・・・・・・・・・・・
・Set as (1).

Here, a uniform random number between -1 and 1 is generated and transformed according to equation 11, and if this is 'tP, that is, the generated random number is fo, then P = tan-γ.・・・・・・・・・・・・
・(2) becomes. This is the standard value.

Furthermore, as a conversion function, G (xi = α+β tan −x...
・・・・・・・・・ [31 will be held regularly and prepared. (α
, β are control parameters described below) Then, a test pattern sequence is generated using uniform random numbers. All random numbers γ generated assuming that the number of elements in the test pattern sequence is 24, ~γ24-converted pattern element Q, ~
Q24, then Qk = α + β tan -; rk (k = 1, 2, -
...24)"--441.

Depending on (2), a test pattern is generated from the determined P and (4) determined Q, ~Q24. The method is defined as follows, where the generated test pattern is T, ~T24. By repeating steps (1) to (5) according to this method, test patterns are generated one after another.

Here, a supplementary explanation will be given regarding α and β that appeared in equation (3). kzuα, but to put it simply, this is the first
This is a parameter that controls the horizontal axis direction in the figure, that is, the parallel shift l1TtI to the left and right of the probability distribution at 1 weight'. The role of this parameter is to carry out the uniform change described above. Figure 2 shows that α as Figure 2 t・tβ=5.0
As α=1. O(at, α=-1,0 (bl). This is an example of changing. This determines the left and right balance with respect to the weight center ("IN appearance probability 12 times). Next, β Dirt is a parameter that changes the entire distribution shape. Figure 3 shows an example of changing β. The figure shows that when α = 0.0, β - 10,0 (at, β =
o, i tb+ As can be seen from FIG. 3, by completely changing the distribution shape, the magnitude relationship between the center (peak portion) and both ends of the probability distribution and the ratio thereof are controlled. FIG. 4 shows an example of the device. In FIG. 4, a uniform random number group γ is input as basic human input from a uniform random number generator IO. , γ1...γ1. and external data input circuit 1
The external parameters α and β input from 1 are random number distribution changes!
Input it into the device tTt12, and then first input γ. are converted and outputted by the conversion function device 13 of the formula fil type, and srl to γ24 are respectively converted by the inflection function device 14 of the formula (3) type. This output sequence P , Qt −Qt,・
. . , Q□F are further input to the test pattern generator 15, and the final test patterns T, ˜T, , ' are generated by conversion of the equation (5) type.

The entire flowchart described above is shown in FIG.

Furthermore, test patterns T1 to T24 are input to the circuit under test 16. PI, PI, . . . , PIn are its input terminals. If the circuit under test is an AND or OR gate, the following procedure may be used.

If you explain the two-man AND gate in advance, the input will be '00'.
The output is 0 for the 3 patterns of '', '01'', and 10''.
Therefore, the output of only the input 'll' becomes '1'. That is, the output ``0'' appears with a probability of 3/4. In the case of a two-man OR gate, conversely, only the input ``00'' results in an output 11''.

Therefore, if the circuit under test is mainly composed of AND system and OR system logic circuits, α is determined from the input pin ratio of AND system and OR system, and NND system and OR system for all input bins are determined.
Determine β from the system occupancy. And β=0 in Fig. 3.
If seven test patterns are generated with an 11' appearance probability distribution such as 1, the circuit under test can be efficiently tested and the test can be completed in an extremely short time.

It should be noted that the present invention is not limited to the above-mentioned embodiments, and various types of conversion functions may be used, for example. In addition, random numbers are not limited to those generated as a complete number of 1 from -'I,
The logic 'x'w may be generated at random time intervals, and the time intervals may be made to correspond to random numbers.

[Brief explanation of drawings]

In Figure 1, the horizontal axis is the weight, the vertical axis is the probability, and the pattern elements (11
``, total number of ``0'') is a diagram showing the ``weight''-probability distribution curve of the test pattern generated by 24 uniform random numbers. Figure 2 shows the ``weight'' - probability distribution curve of the test pattern generated by changing only α. Kyomi "-probability distribution song i'i?J k diagram, 3rd
The figure is a diagram in which only β is changed, FIG. 4 is a configuration diagram of an apparatus according to the present invention, and FIG. 5 is a flowchart diagram of the present invention. Representative Patent Attorney Noriyuki Chika (and 1 other person) Figure 8 → Meeting Figure 4

Claims (1)

[Claims] A random number generation device, a random number distribution conversion device that controls the distribution of the generated random numbers, and a test pattern generation device that outputs a specific logical value with a predetermined probability according to the random number distribution,
An integrated circuit testing device characterized in that the movement of the appearance probability distribution of the logical value in the weight axis direction and the shape change thereof are completely controlled by external binary values.