JPH11296561A - Worst-case model parameter generation method and apparatus - Google Patents

Abstract

(57) [Summary] [Problem] Regarding element characteristics of interest, from conditions such as worst values and allowable limits of independent elements, and from conditions of independent elements corresponding to model parameters indicating specified element characteristic values (guaranteed values) A worst case model parameter generation method and apparatus are provided. A data input unit (1) inputs a reference value of an independent element which influences a variation of a model parameter, a value specifying a distribution of the independent element, a desired value of a characteristic of a target element, and a relational expression between a model parameter and an independent element. Is done. In the data processing unit 2,
The sensitivity, worst point, model parameters, and the like of the independent element to the element of interest are calculated. Based on the sensitivity data obtained by the sensitivity calculation, the reference value of the independent element, the desired value of the element characteristics, or the allowable limit value of the independent element, the existence probability of the model parameter matched the specified situation in the area of the constant independent element. The value of the independent element that takes the minimum value or the maximum value of the element characteristics is obtained.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method and an apparatus for supporting generation of a model parameter set used in a circuit simulation performed particularly in electronic circuit design and control of a process based on the model parameter.

[0002]

2. Description of the Related Art In recent years, due to the design of high-frequency circuits and the reduction of the power supply voltage, circuit designs with small margins for device characteristics have been made. In such a situation, fluctuations in device characteristics affect circuit reliability and, consequently, product yield. Therefore, worst case analysis is important in designing a highly reliable circuit and realizing a circuit product with a high yield.

Conventionally, as model parameters for worst-case analysis of analog circuits, independent factors such as process conditions (for example, various manufacturing factors that affect the characteristics of element circuits such as a certain temperature during the process and the dose of impurities). There is a parameter set (corner model) corresponding to the combination condition of the maximum value and the minimum value of the fluctuation factor).

FIG. 7 is a conceptual diagram showing the existence probability of the corner model in the independent element space. ΔP1 on the horizontal axis indicates a normalized shift amount from the center value of a certain independent element, and ΔP2 on the vertical axis indicates the normalized shift amount from the center value of another certain independent element. . The existence probability (only one independent element is 3
σ shift). The dashed line
The combination when the maximum value and the minimum value of each independent element are set indicates that a worst case model parameter such as a cross is generated in each quadrant.

[0005] From FIG. 7, if the maximum value and the minimum value of a certain independent element, that is, the so-called permissible limit value (for example, ± 3σ) are applied to each independent element, if the existence probability of the model parameter is lower than the assumed existence probability of the model parameter Model parameters are generated. In other words, there is a problem in that a parameter set under conditions that are stricter than the set existence probability of the independent element is generated.

Further, since a combination of the maximum value and the minimum value of the independent elements is used, the number n of the independent elements to be varied is
2 ⁿ parameters are generated. For this reason, when n increases, the number of generated parameters also increases rapidly, and there is a problem that analysis time and scale increase.

On the other hand, as another method of generating a parameter set, a method of obtaining a parameter set by setting a maximum value and a minimum value for one of the independent elements (bounda
ry model (boundary model) is also proposed.

FIG. 8 is a conceptual diagram showing the existence probabilities in the independent element space regarding the boundary model. ΔP1 on the horizontal axis indicates a normalized shift amount from the center value of a certain independent element, and ΔP2 on the vertical axis indicates the normalized shift amount from the center value of another certain independent element. .
The circle line represents a region having the same existence probability as in FIG.
When the maximum value and the minimum value of the existence probability of one independent element are set, the central value (zero deviation value) is set as the existence probability of the other independent element. As a result, the worst-case model parameters that ride on the line of the circle as indicated by the crosses, that is, match the assumed existence probability, are generated.

According to the above method, the number of parameters is suppressed to 2n for the number n of independent elements to be varied. However, there is a problem that a parameter having a worst value is not always generated in a region of the set existence probability of an independent element for a certain element characteristic of interest. That is, as shown in FIG. 6, when considering other element characteristics, for example, attention in the Z-axis direction, the parameters marked with x on the ΔP1 axis and ΔP2 axis are not necessarily worst values, Is generated.

[0010]

In the process development, a range of variation in device characteristics due to process variation is set as a guaranteed value. In general, a circuit designer designs at least the guaranteed value so that the circuit operates normally.

In this case, the process variation is reflected, and
It is effective to execute a circuit simulation using model parameters whose element characteristics indicate a guaranteed value. For this reason, it is important to generate an accurate worst case model parameter set.

In the prior art, a corner model generates a worst-case parameter set by a combination of minimum and maximum values of independent elements. For this reason, there is a problem that a parameter having a stricter condition than actual conditions is generated. Further, the number of parameters increases rapidly when the number of independent elements increases, which is not preferable in terms of analysis time and cost of analysis means.

Further, the boundary model has a problem that the worst value is not always taken for the element characteristic of interest. Therefore, the worst case
With the parameter set generation method, it is difficult to generate accurate worst-case model parameters that lead to high reliability and yield improvement of circuit products.

The present invention has been made in consideration of the above circumstances, and has as its object to show the maximum value or the minimum value of the element characteristic of interest in an independent element region in which the existence probability of a model parameter is constant in consideration of a process variation. Provided is a method for determining model parameters, and an apparatus therefor. Furthermore, in other words, a worst-case model parameter generation method and apparatus having a function of presenting a condition of an independent element corresponding to a model parameter indicating a specified element characteristic value (for example, a guaranteed value) effective in process development Is to provide.

[0015]

According to a first aspect of the present invention, there is provided a worst-case model parameter generation method, wherein at least a reference value of one or more independent elements which influences a variation of a model parameter used in a circuit simulation; A data designating step of obtaining a value specifying a distribution, a desired value of a target element characteristic, and a relational expression between the independent element and a model parameter; a sensitivity calculating step in which a sensitivity of the independent element to the target element characteristic is calculated; Using the sensitivity obtained by the sensitivity calculation and the data obtained from the acquisition unit, the minimum value or the maximum value of the element characteristic of interest is matched with the desired value in the independent element region where the existence probability of the model parameter is substantially constant. A worst-point calculation step of obtaining the value of the independent element for calculating the worst-point value. Characterized by comprising a model parameter calculation step of calculating a model parameter value of the unit from the equation.

According to a second aspect of the worst case model parameter generation method, at least a reference value of one or more independent elements which influences a variation of a model parameter used in circuit simulation, a value designating a distribution of the independent elements, Tolerable limit value of the normalized distribution of the independent elements, and a data acquisition step of obtaining a relational expression between the independent elements and model parameters, and a sensitivity calculation step in which the sensitivity of the independent elements to the element characteristics of interest is calculated. Using the sensitivity obtained by the sensitivity calculation and the data obtained from the data acquisition means, the element characteristic of interest has a minimum value or a maximum value in an independent element region having a model parameter existence probability at an allowable limit value of the independent element distribution. A worst point calculation step for obtaining the value of the independent element indicating the value, and Wherein the values of the independent element and from the relationship to and a model parameter calculation step of calculating a model parameter.

According to a third aspect of the worst case model parameter generation method, at least a reference value of one or more independent elements which influences a variation of a model parameter used in circuit simulation, a value designating a distribution of the independent elements, A data acquisition step of obtaining a normalized permissible limit value of the distribution of the independent elements, an independent element designation input for designating one of the independent elements, a desired value of the element characteristic of interest, and a relational expression between the independent elements and model parameters And a sensitivity calculation step in which the sensitivity of the independent element to the element of interest is calculated; and an allowable limit value of the distribution of the independent element using the sensitivity obtained by the sensitivity calculation and the data obtained from the data acquisition means. The minimum or maximum value of the element characteristic of interest is the desired value of the element characteristic of interest in the area of the independent element having the existence probability of the model parameter in As shown, the worst point calculation step of adjusting the distribution of the independent elements specified by the independent element specification input to obtain the value of the independent element, and the independent element specification input corresponding to the worst point obtained by the worst point calculation Acquiring a distribution of the designated independent element.

According to a first aspect of the worst case model parameter generation device, at least a reference value of one or more independent elements which influences a variation of a model parameter used in a circuit simulation, a value designating a distribution of the independent elements, Data acquisition means for obtaining a desired value of the characteristic of the element of interest and a relational expression between the independent element and the model parameter; a sensitivity calculation in which the sensitivity of the independent element to the element of interest is calculated; and a sensitivity obtained by the sensitivity calculation. Using the data obtained from the obtaining means, in the independent element region where the existence probability of the model parameter is substantially constant, set the value of the independent element for matching the minimum value or the maximum value of the element characteristic of interest to the desired value. The worst point calculation to be obtained, and the value of the independent element obtained in the worst point calculation and the relational expression Model parameter calculation for calculating a Dell parameters, and data processing means for performing each process, characterized by comprising an output means for outputting the worst parameter set of said independent elements according to the data processing means.

According to a second aspect of the worst case model parameter generation device, at least a reference value of one or more independent elements which influences a variation of a model parameter used in circuit simulation, a value designating a distribution of the independent elements, Data acquisition means for obtaining a normalized permissible limit value of the distribution of the independent element and a relational expression between the independent element and a model parameter; a sensitivity calculation for calculating the sensitivity of the independent element to the element of interest; and the sensitivity calculation. Using the sensitivity obtained by the above and the data obtained from the data acquisition means, the element characteristic of interest sets the minimum value or the maximum value in the area of the independent element having the model parameter existence probability at the allowable limit value of the distribution of the independent element. Worst-point calculation for determining the value of the independent element shown, and the independent element obtained by the worst-point calculation. Data processing means for performing each processing of a model parameter calculation for calculating a model parameter from the value of the relational expression and an output means for outputting a worst parameter set of the independent element corresponding to the data processing means. It is characterized by.

According to a third aspect of the worst case model parameter generation device, at least a reference value of one or more independent elements which influences a variation of a model parameter used in circuit simulation, a value designating a distribution of the independent elements, Data obtaining means for obtaining a normalized allowable limit value of the distribution of the independent elements, an independent element designation input for designating one of the independent elements, a desired value, and a relational expression between the independent elements and the model parameters; A sensitivity calculation in which the sensitivity of the independent element to the characteristic is calculated; a probability of existence of a model parameter at an allowable limit value of the distribution of the independent element using the sensitivity obtained by the sensitivity calculation and the data obtained from the data acquisition means; In the area of the independent element having Worst point calculation for adjusting the distribution of the independent elements specified by the element specification input to obtain the value of the independent element, and the independent element specified by the independent element specification input corresponding to the worst point obtained by the worst point calculation A data processing unit for performing each process of obtaining a distribution, and an output unit for outputting a worst parameter set of the independent element according to the data processing unit.

In the first to third inventions, the output step or the output means may correspond to a model parameter generated so as to indicate a desired element characteristic value in addition to the output of the worst parameter set. And outputs a value of each independent element, a shift amount from a reference value, or a normalized value of the shift amount.

In the present invention, the worst-case model parameters for the element characteristics of interest are generated under conditions such as the worst value of the element characteristics and the allowable limit value of the independent element.
Therefore, an appropriate worst case model parameter can be obtained for the element characteristics of interest. Further, the number of worst case model parameter sets is determined by the number of target element characteristics.

[0023]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 is a conceptual diagram showing a worst case model parameter generation device according to each embodiment of the present invention. The worst case model parameter generation device according to the first embodiment will be described with reference to FIG. The data input unit 1 provides a reference value of an independent element that influences a variation of a model parameter, and a value (for example, σ: a parameter of a process step or monitor information reflecting the parameter) of the independent element. Standard deviation), a desired value of the element characteristic of interest, and a relational expression (ResponseSurface Model: RSM expression) between the model parameter and the independent element are input. As an input method, in addition to typing from a keyboard or the like, create data in an electronic file in advance,
A method of loading the file can be considered.

The data processing unit 2 performs data processing such as sensitivity calculation of an independent element with respect to a target element characteristic, worst point calculation, model parameter calculation, and the like. In the sensitivity calculation routine, the reference value of the independent element, R
Circuit simulator 5 represented by SPICE from SM type
Is generated, and a method of calculating the sensitivity from a circuit simulation result using the model parameter file is considered.

In the worst point calculation routine, based on the sensitivity data obtained by the sensitivity calculation, the reference value of the independent element, and the desired value of the element characteristic, the minimum element characteristic is obtained in the area of the independent element where the model parameter existence probability is constant. The value (point) of the independent element whose value or maximum value matches the desired value is determined. This makes it possible to specify one worst-case point even when a plurality of independent elements affect the element characteristics of interest.

In the model parameter calculation routine, the value of the independent element obtained by the worst point calculation is R
The model parameters corresponding to the worst point are calculated by substituting into the SM formula.

The data output unit 3 outputs worst case model parameters obtained by worst point calculation, model parameter calculation, and the like. In addition, a value of an independent element corresponding to the parameter, a shift amount from a reference value or a normalized shift amount, a graph display of a relationship between a target element characteristic and an independent element, and the like are output.

Although a display is usually used for the output unit, the output is also performed to the data storage unit 4 at the same time. Obtaining these outputs allows the user to know what independent factors are affecting the generated model parameters, and how much margin the process has for the desired values of the device characteristics. Can be determined.

The data storage unit 4 stores not only the above output results but also intermediate data (for example, circuit simulation results, temporary model parameters, etc.) generated in sensitivity calculation, worst point calculation, and the like. An external storage device such as a hard disk is usually used for the data storage unit 4.

The worst case model parameter generation device according to the second embodiment will be described with reference to FIG. Data input unit 1 shown in the first embodiment
In, instead of inputting the desired value of the element characteristic, the allowable limit value of the normalized independent element distribution, for example, 3σ is input.

In the worst-point calculation routine, the allowable limit value is a minimum value in the region of the independent element having the model parameter existence probability at the allowable limit value together with the sensitivity data and the reference value of the independent element. Alternatively, it is used to calculate the value of the independent element indicating the maximum value. This makes it possible to specify one worst point.

The worst case model parameter generation device according to the third embodiment will be described with reference to FIG. The first embodiment incorporates the second embodiment. The value of the tolerance of the independent element is specified,
The region in which the element characteristic of interest indicates the minimum value or the maximum value is determined, but the desired value of the element characteristic is also specified and input, so that the independent element specified by the independent element specification input is matched to the desired value of the element characteristic. The distribution is adjusted. The distribution of the specified independent element obtained by the worst point calculation routine is output to the data output unit 3. At the same time, the data is output to the data storage unit 4.

FIG. 2 is a flowchart showing an example of a data processing flow showing a worst case model parameter generation method according to each embodiment of the present invention. Each step will be described with reference to this figure.

Step 1: In data input, a reference value of an independent element and a value designating a distribution, a desired value of a target element characteristic,
The RSM formula is input. Step 2: In sensitivity calculation, reference value of independent element, RSM
A model parameter is created using the equation, and the sensitivity of an independent element to a target element characteristic (for example, a DC amplification factor hFE in a bipolar transistor, a collector-emitter saturation voltage Vce = 5 V, a collector current Ic = 100 μA) is calculated using a circuit simulator. Ask for.

Step 3: In the worst point calculation,
The worst point is calculated using the following assumptions and formulas. As assumptions, the following two points (i) and (ii) are considered. (I) The elements that change the model parameters are independent and Gaussian distributions, and if normalized, the shape of a region having an equal distribution probability is an n-dimensional sphere (a is the radius of the sphere).

[0036]

(Equation 1) (Ii) The variation of the independent element is minute relative to the reference value of the independent element, and the effect on the element characteristics can be approximated by the product sum of the sensitivity and the shift amount of each independent element.

[0037]

(Equation 2) The formula for calculating the worst point is the following formula in the case of the first embodiment in which the desired value of the element characteristics is specified.

[0038]

(Equation 3) Also, in the case of the second embodiment in which the allowable limit value of the independent element is specified, if the allowable limit value is aσ (3 = 3 = a = 3), it can be expressed by the following equation.

[0039]

(Equation 4)

Also, the allowable limit value of the independent element and the desired value of the element characteristics are specified, and the independent element k for adjusting the distribution is specified.
Is specified in the third embodiment, the allowable limit value is a
σ, and the scale for the distribution of independent elements k is b,
The following equation is obtained.

[0041]

(Equation 5)

In the above case, the resulting distribution of the independent elements for distribution adjustment will be bΔpkσ. Step 4: In the model parameter calculation, the model parameter is calculated by substituting the value of the worst point (the value obtained by adding Δpi obtained by the above equation to the reference value) obtained by the worst point calculation into the RSM equation.

FIG. 3 shows an example of a form for inputting a fixed value to an independent element. The form displays a list of independent elements, and includes a column for inputting a reference value or a fixed value and a column for inputting a standard deviation σ. To the right is a button that toggles whether the independent element has a fixed value. The user can input a desired value in the reference value / fixed value column of the independent element for which a fixed value is to be set, and press the switch button to make the setting. The button to the right of the switching button has a function of returning the value in the reference value / unique value column to the default reference value.

The effect of giving fixed values to independent elements is that model parameters can be generated under more limited conditions. As an example, the boundary model can be obtained by using all but one independent element as reference values.

FIG. 4 is an example of a form for describing the element characteristic calculation procedure. The element characteristics of interest vary depending on the type of circuit, and it is difficult to prepare all the element characteristics in advance. Therefore, by allowing user definition in the element characteristic calculation procedure, it is possible to cope with the above problem.

For example, the SIM (“FILE_A” in FIG. 4)
DATA) means that a simulation is performed on the FILE_A netlist, and the result is stored in DATA. GET_Y (100E-6 DATA) indicates that the value of Y from which the value of X becomes 100E-6 is obtained from DATA. Means to return.

FIG. 5 is a conceptual diagram illustrating the existence probability of the independent element in the space. In the two-dimensional case as shown in FIG.
According to the above assumption, a region in the independent element space in which the distribution of independent elements (probability of existence of model parameters) is equal is represented by a circle centered on the reference value (center value).

Here, in the corner model shown in FIG. 7, in the space of the two-dimensional independent element, the point marked with X corresponds to the corner model. Therefore, if an assumed allowable limit value (for example, 3σ) is applied to each independent element, an existence probability lower than the existence probability of the assumed model parameter, that is, a too strict model parameter is generated.

Further, it can be seen that the boundary model shown in FIG. 8 generates model parameters matching the assumed model parameter existence probability in the two-dimensional independent element space. However, considering the element characteristics that are focused in another dimension, for example, the Z-axis direction, the model parameters that indicate the minimum (maximum) element characteristics are not always generated.

FIG. 6 is a conceptual diagram for explaining a model parameter set according to the present invention in association with a space of independent elements. In the present invention, by using the sensitivity of the independent element that has calculated the effect on the element characteristics of interest, when the element characteristics of interest are assigned to another dimension (for example, the Z-axis direction in this case), the characteristics of the element characteristics such as It is possible to generate a minimum or maximum model parameter. The mark “x” corresponds to the boundary model.

As described above, in the conventional corner model and boundary model, the model parameter corresponding to the minimum or maximum value in the region of the independent element where the existence probability of the model parameter is constant is generated for the element characteristic of interest. Not done. However, according to the present invention, an appropriate worst case model parameter can be generated by using the sensitivity. Furthermore, since the number of worst cases depends on the element characteristics of interest, the number of model parameters does not increase more than necessary. This means that the circuit simulation can be performed efficiently.

Further, since model parameters for which the guaranteed value of the process is the worst case can be easily generated, the circuit operation verification at the guaranteed limit of the process can be appropriately performed. Further, a function for setting a fixed value to an independent element and a function for defining an element characteristic calculation procedure can easily generate a worst-case model parameter desired by a user.

From the viewpoint of process development, it is possible to quantitatively know how much the distribution of the independent elements should be controlled in order to match the variation in the element characteristics to the target. This has a significant effect on the early launch of the process.

[0054]

As described above, according to the present invention,
Appropriate worst-case model parameters can be obtained for the element characteristics of interest. Since the number of worst-case model parameter sets is determined by the number of element characteristics of interest, it is possible to efficiently perform circuit verification with a small number of parameter sets. That is, the worst-case model parameters for the element characteristics of interest can be generated in accordance with conditions such as the worst element characteristics specified by the user and the allowable limit value of the independent element. For this reason, it is possible to perform circuit verification that matches the intention of the circuit designer. Further, since the control range of the independent element for clearing the process verification value can be obtained quantitatively, it is possible to provide a worst-case model parameter generation method and apparatus that promote process development.

[Brief description of the drawings]

FIG. 1 shows a worst case according to each embodiment of the present invention.
FIG. 2 is a conceptual diagram showing a model parameter generation device.

FIG. 2 shows a worst case according to each embodiment of the present invention.
5 is a flowchart illustrating an example of a data processing flow illustrating a model parameter generation method.

FIG. 3 is a display diagram showing an example of a form for inputting a fixed value to an independent element.

FIG. 4 is a display diagram showing an example of a form describing an element characteristic calculation procedure.

FIG. 5 is a conceptual diagram illustrating the existence probability of an independent element in a space.

FIG. 6 is a conceptual diagram illustrating a model parameter set according to the present invention in association with a space of independent elements.

FIG. 7 is a conceptual diagram showing the existence probabilities of the corner model in the independent element space.

FIG. 8 is a conceptual diagram showing an existence probability in an independent element space relating to a boundary model.

[Explanation of symbols]

 REFERENCE SIGNS LIST 1 data input unit 2 data processing unit 3 data output unit 4 data storage unit 5 circuit simulator

Claims (12)

[Claims] At least one of a reference value of one or more independent elements that influences a variation of a model parameter used in a circuit simulation, a value specifying a distribution of the independent elements, a desired value of a characteristic of a target element, and A data acquisition step of obtaining a relational expression of a model parameter; a sensitivity calculation step of calculating the sensitivity of the independent element with respect to the element of interest; and a sensitivity obtained by the sensitivity calculation and data obtained from the obtaining unit. A worst point calculating step of obtaining a value of the independent element for matching a minimum value or a maximum value of the element characteristic of interest to the desired value in a region of the independent element in which the existence probability of the model parameter is substantially constant; Model parameter calculation for calculating a model parameter from the value of the independent element obtained by calculation and the relational expression And generating a worst case model parameter. 2. A reference value of at least one independent element which influences a variation of a model parameter used for circuit simulation, a value specifying a distribution of the independent element, and a permissible limit of the normalized distribution of the independent element. A data acquisition step of obtaining a value and a relational expression between the independent element and the model parameter; a sensitivity calculation step of calculating the sensitivity of the independent element to the element of interest; and a sensitivity obtained by the sensitivity calculation and the data acquisition means. Worst-point calculation for obtaining the value of an independent element in which the element characteristic of interest shows the minimum value or the maximum value in the area of the independent element having the existence probability of the model parameter at the allowable limit value of the distribution of the independent element using the data obtained from Calculating a model parameter from the value of the independent element obtained in the worst point calculation and the relational expression. A worst-case model parameter generation method, comprising: 3. In the data acquisition step, at least a reference value of one or more independent elements that influences a variation of a model parameter used in a circuit simulation, a value specifying a distribution of the independent elements, A data acquisition step of obtaining an allowable limit value of the distribution of elements, an independent element designation input for designating one of the independent elements, a desired value of the element property of interest, and a relational expression between the independent element and the model parameter; A sensitivity calculating step in which the sensitivity of the independent element is calculated; and a probability of existence of a model parameter at an allowable limit value of the distribution of the independent element using the sensitivity obtained by the sensitivity calculation and the data obtained from the data acquisition unit. The independent element finger is set such that the minimum value or the maximum value of the element characteristic of interest indicates a desired value of the element characteristic of interest in the area of the independent element having A worst point calculation step of adjusting the distribution of the independent elements specified by the input to obtain the value of the independent element, and the distribution of the independent elements specified by the independent element specification input corresponding to the worst point obtained by the worst point calculation A worst-case model parameter generation method. 4. The method according to claim 1, wherein fixed values are specified for some of said independent elements.
The worst case model parameter generation method described in (1). 5. The worst-case model parameter generation method according to claim 1, wherein a procedure description that describes a calculation procedure for each element characteristic is used in the sensitivity calculation step. 6. A step of outputting a value of each independent element, a shift amount from a reference value, or a normalized value of the shift amount corresponding to a model parameter generated so as to indicate a desired element characteristic value. The worst-case model parameter generation method according to any one of claims 1 to 3, wherein: 7. At least a reference value of one or more independent elements that influences a variation of a model parameter used for circuit simulation, a value specifying a distribution of the independent elements, a desired value of a target element characteristic, and Data acquisition means for obtaining a relational expression of model parameters; sensitivity calculation for calculating the sensitivity of the independent element to the element of interest characteristic; and a model using the sensitivity obtained by the sensitivity calculation and data obtained from the acquisition means. The worst point calculation for obtaining the value of the independent element for matching the minimum value or the maximum value of the element characteristic of interest to the desired value in the area of the independent element where the existence probability of the parameter is substantially constant, and the worst point calculation. Calculation of a model parameter from the obtained independent element value and the relational expression. Data processing means and said data processing means worst-case model parameter generating apparatus characterized by comprising an output means for outputting the worst parameter set of said independent elements according to the performing. 8. A reference value of at least one independent element which influences a variation of a model parameter used in a circuit simulation, a value designating a distribution of the independent element, a permissible limit of the normalized distribution of the independent element. Data acquisition means for obtaining a value, a desired value of the element characteristic of interest, and a relational expression between the independent element and the model parameter; a sensitivity calculation for calculating the sensitivity of the independent element to the element characteristic of interest; a sensitivity obtained by the sensitivity calculation And the value of the independent element in which the element of interest shows the minimum value or the maximum value in the area of the independent element having the existence probability of the model parameter at the allowable limit value of the distribution of the independent element using the data obtained from the data acquisition means. Worst point calculation to obtain the model parameter from the value of the independent element obtained in the worst point calculation and the relational expression. Data processing means for performing each process of model parameter calculation for calculating data, and output means for outputting a worst parameter set of the independent element according to the data processing means. Model parameter generation device. 9. A reference value of at least one independent element which influences a variation of a model parameter used for circuit simulation, a value specifying a distribution of the independent element, a permissible limit of the normalized distribution of the independent element. Data acquisition means for obtaining a value, an independent element designation input for designating one of the independent elements, a desired value of the element property of interest, and a relational expression between the independent element and the model parameter; Attention is paid to an area of the independent element having the calculated sensitivity, the sensitivity obtained by the sensitivity calculation, and the existence probability of the model parameter at the allowable limit value of the distribution of the independent element using the data obtained from the data acquisition means. The distribution of the independent elements specified by the independent element specification input is set such that the minimum value or the maximum value of the element characteristics indicates the desired value of the element characteristic of interest. Data processing means for performing worst-point calculation for determining the value of an independent element by obtaining the value of the independent element, and obtaining the distribution of the independent element specified by the independent-element designation input corresponding to the worst point obtained by the worst-point calculation A worst case model parameter generation device, comprising: an output unit that outputs a worst parameter set of the independent element according to the data processing unit. 10. The worst-case model parameter generation device according to claim 7, wherein fixed values are specified for some of the independent elements. 11. The worst-case method according to claim 7, further comprising a procedure description unit for describing a calculation procedure for each element characteristic in the sensitivity calculation of the data processing unit. Model parameter generation device. 12. The output unit outputs, in addition to the output of the worst parameter set, a value of each independent element corresponding to a model parameter generated so as to indicate a desired element characteristic value, a shift amount from a reference value, 10. The worst-case model parameter generation device according to claim 7, wherein a normalized value of the shift amount is output.