Model creation method and program

Abstract

To provide a model creation method for modeling a relationship between an internal voltage and a drain current of a transistor.SOLUTION: A model creation process includes an internal voltage calculation step S106 of calculating internal voltages Vgsi and Vdsi by using a gate voltage measurement value for expressing a measurement value of a gate voltage VG of a transistor, a drain voltage measurement value for expressing a measurement value of a drain voltage VD of a transistor, and a middle value ID_middle of a drain current calculation value for expressing a drain current calculated by a current source model, a current calculation step S107 of calculating a drain current calculation value by the current source model by using an internal voltage, and an update step S109 for updating a parameter of the current source model by using an error between the drain current calculation value, and a drain current measurement value for expressing a measurement value of a drain current ID(n) of a transistor. Until parameters of the current source model converge, a first loop process including the current calculation step and the update step is repeated.SELECTED DRAWING: Figure 5

Description

[Technical Field]

[0001] This disclosure relates to a model creation method and program. [Background technology]

[0002] Transistors are modeled for circuit design and other purposes. For example, Non-Patent Document 1 discloses a technique for creating a model of a GaN (gallium nitride) FET (field effect transistor). [Prior art documents] [Non-patent literature]

[0003] [Non-Patent Document 1] GaN FET transistor model extraction / IVCAD, Internet<URL:https: / / www.youtube.com / watch?v=Ihbg60DqRlU&t=300s> [Overview of the Initiative] [Problems that the invention aims to solve]

[0004] To create a transistor model, it is necessary to model the relationship between the internal voltage and the drain current, but this modeling can be difficult with conventional techniques.

[0005] This disclosure aims to provide a model and program that can model the relationship between the internal voltage and drain current of a transistor. [Means for solving the problem]

[0006] The model creation method of this disclosure is a model creation method for creating a current source model that represents the relationship between the internal voltage and drain current of a transistor, and the computer performs the following steps: an internal voltage calculation step in which the internal voltage is calculated using a gate voltage measurement value representing a measured value of the gate voltage of the transistor, a drain voltage measurement value representing a measured value of the drain voltage of the transistor, and an intermediate value of a drain current calculation value representing the drain current calculated by the current source model; a current calculation step in which the drain current calculation value is calculated by the current source model using the internal voltage; and an update step in which the parameters of the current source model are updated using the error between the drain current calculation value and a drain current measurement value representing a measured value of the drain current of the transistor, wherein the computer repeats a first loop process including the current calculation step and the update step until the parameters of the current source model converge.

[0007] This disclosure can be implemented not only as a model creation method that performs such characteristic steps, but also as a model creation device equipped with a processing unit that performs such characteristic steps, or as a program for causing a computer to perform such characteristic steps. Furthermore, the entire or a part of the model creation device can be implemented as a semiconductor integrated circuit, or as a model creation system including the model creation device. [Effects of the Invention]

[0008] According to this disclosure, the relationship between the internal voltage and drain current of a transistor can be modeled. [Brief explanation of the drawing]

[0009] [Figure 1] Figure 1 shows an example of an equivalent circuit of a transistor used in the large-signal model. [Figure 2] Figure 2 schematically shows a conventional method for creating a current source model. [Figure 3]Figure 3 shows an example of the hardware configuration of the model creation device according to the first embodiment. [Figure 4] Figure 4 shows an example of the functional configuration of the model creation device according to the first embodiment. [Figure 5] Figure 5 is a flowchart showing an example of the current source model creation process according to the first embodiment. [Figure 6] Figure 6 schematically shows how the drain current calculated using the current source model converges. [Figure 7] Figure 7 shows an example of a current collapse. [Figure 8] Figure 8 shows an example of the functional configuration of a model creation device according to the second embodiment. [Figure 9] Figure 9 is a flowchart showing an example of the current source model creation process according to the second embodiment. [Figure 10] Figure 10 is a flowchart showing an example of the weight calculation process according to the second embodiment. [Figure 11] Figure 11 shows an example of load line calculation. [Figure 12] Figure 12 shows an example of a region definition. [Figure 13] Figure 13 shows an example of calculating distance for weight calculation. [Modes for carrying out the invention]

[0010] [Description of Embodiments in this Disclosure] First, embodiments of this disclosure will be listed and described.

[0011] [1] A model creation method according to one aspect of the present disclosure is a model creation method for creating a current source model representing the relationship between the internal voltage and drain current of a transistor, the method comprising: an internal voltage calculation step of calculating the internal voltage using a gate voltage measurement value representing a measured value of the gate voltage of the transistor, a drain voltage measurement value representing a measured value of the drain voltage of the transistor, and an intermediate value of a drain current calculation value representing the drain current calculated by the current source model; a current calculation step of calculating a drain current calculation value using the current source model with respect to the internal voltage; and an update step of updating the parameters of the current source model using the error between the drain current calculation value and a drain current measurement value representing a measured value of the drain current of the transistor, wherein the computer repeats a first loop process including the current calculation step and the update step until the parameters of the current source model converge.

[0012] After calculating the internal voltage, the calculation of the drain current value and parameter updates are repeated until the parameters of the current source model converge. Therefore, the internal voltage does not change during the repeated calculation of the drain current value and parameter updates. This allows the parameters of the current source model to converge.

[0013] [2] The computer repeats the second loop process, which includes the internal voltage calculation step and the first loop process, in [1] until the drain current calculation value converges. This allows the parameters of the current source model to be updated to fit the drain voltage measurement.

[0014] [3] In [2], the intermediate value is the value between the drain current calculation value calculated in the second loop processing in the nth iteration and the drain current calculation value calculated in the second loop processing in the (n-1)th iteration. This ensures the convergence of the drain current calculation value.

[0015] [4] Steps [1] to [3] include a weight calculation step in which the drain voltage measurement value and the drain current measurement value are used to calculate a weight for the drain current measurement value, and the error multiplied by the weight is used to determine whether or not the parameters of the current source model have converged.

[0016] [5] In [4], the weight calculation step calculates the load line of the large-signal model using the operating point of the large-signal operation of the transistor, the drain voltage measurement, and the drain current measurement, and calculates the weights for the drain current measurement such that the closer the drain current measurement is to the load line, the higher the weight of the drain current measurement, and the further away it is from the load line, the lower the weight of the drain current measurement.

[0017] [6] A program according to one aspect of the present disclosure causes a computer for creating a current source model representing the relationship between the internal voltage and drain current of a transistor to perform an internal voltage calculation step of calculating the internal voltage using a gate voltage measurement representing a measured value of the gate voltage of the transistor, a drain voltage measurement representing a measured value of the drain voltage of the transistor, and an intermediate value of a drain current calculation value representing the drain current calculated by the current source model; a current calculation step of calculating a drain current calculation value using the current source model with respect to the internal voltage; and an update step of updating the parameters of the current source model using the error between the drain current calculation value and a drain current measurement representing a measured value of the drain current of the transistor, and to repeat a first loop process including the current calculation step and the update step until the parameters of the current source model converge.

[0018] After calculating the internal voltage, the calculation of the drain current value and parameter updates are repeated until the parameters of the current source model converge. Therefore, the internal voltage does not change during the repeated calculation of the drain current value and parameter updates. This allows the parameters of the current source model to converge.

[0019] [Details of Embodiments of the Present Disclosure] Hereinafter, the first and second embodiments of the present disclosure will be described in detail, but the present disclosure is not limited thereto. In this specification and the drawings, substantially the same components may be denoted by the same reference numerals, and redundant description may be omitted.

[0020] The first and second embodiments of the present disclosure relate to a modeling device that models the relationship between an internal voltage and a drain current in order to create a transistor model (which is also called a large-signal model). Note that at least a part of each of the embodiments described below may be arbitrarily combined.

[0021] (Transistor Model) A transistor model (a large-signal model of a transistor) is a model obtained by modeling a circuit as shown in FIG. 1. FIG. 1 shows an equivalent circuit of a transistor used for a large-signal model.

[0022] In FIG. 1, G represents a gate, D represents a drain, S represents a source, R represents a resistor, L represents an inductance, I represents a current, V represents a voltage, and C represents a capacitance. Also, with respect to the subscripts given to R, L, I, V, and C, G and g represent the gate, D and d represent the drain, s represents the source, and i represents intrinsic.

[0023] The internal voltage V gsi and V dsi and the drain current I D need to be accurately modeled. As represented by the broken line 1000 in FIG. 1, this is because the drain current I D is determined by the internal voltages V gsi and V dsi .

[0024] That is, it is necessary to accurately obtain a function f such that I D = f(V gsi , V dsi ). Hereinafter, the model represented by this function f will be referred to as a current source model.

[0025] (Conventional method for creating current source models) Built-in voltage V gsi and V dsi It is known that this can be calculated as follows:

[0026] V gsi =V G -I D ×R s V dsi =V D -I D ×(R s +R d ) In general, R s , R s , R d It is known or easily measurable.

[0027] Conventional methods for creating current source models include, given V G Measured value, V D Measurement values ​​and I D Using the measured values ​​and the above calculation formula, I D =f(V gsi ,V dsi ) and V gsi and V dsi Calculate and I D The parameters of function f are repeatedly updated to minimize the error with the measured value. This is schematically shown in Figure 2. Figure 2 shows the parameter updates at a certain nth iteration of a conventional method for creating a current source model.

[0028] As shown in Figure 2, in the conventional example, the built-in voltage V calculated in the previous iteration gsi (n) and V dsi (n) From drain current I D (n) =f(V gsi (n) ,V dsi (n) ) calculates the internal voltage V used in the next iteration. gsi (n+1) and Vdsi (n+1) Calculate I D (n) and I D The error between the measured value and the function's parameters is used to update the function's parameters to minimize this error.

[0029] Here, the built-in voltage V gsi (n+1) When calculating V G Measured value and drain current I D (n) Using the above formula V gsi =V G -I D ×R s It is calculated by V. dsi (n+1) When calculating V D Measured value and drain current I D (n) Using the above formula V dsi =V D -I D ×(R s +R d ) is calculated by this. Therefore, in the conventional example, the drain current I D (n) Each time the value of changes, the built-in voltage V gsi (n+1) and V dsi (n+1) The value of will also change. Therefore, depending on the transistor being modeled (for example, a gallium nitride high electron mobility transistor (GaN HEMT)), the parameters of the function f may not converge, making modeling impossible, or requiring manual parameter adjustment for modeling.

[0030] (First embodiment) Below, we will describe a model creation device 10 that can accurately create a current source model f for the conventional example described above. Note that the model creation device 10 may be configured as an element of a model creation system that includes the model creation device 10.

[0031] First, the hardware configuration of the model creation device 10 according to this embodiment will be described with reference to Figure 3. As shown in Figure 3, the model creation device 10 according to this embodiment has the hardware configuration of a general computer or computer system, and includes, for example, an input device 101, a display device 102, an external I / F 103, a communication I / F 104, a processor 105, and a memory device 106. Furthermore, each of these hardware components is connected to each other so as to be able to communicate via a bus 107.

[0032] The input device 101 is, for example, a keyboard, mouse, touch panel, or physical button. The display device 102 is, for example, a display or display panel. Note that the model creation device 10 does not necessarily have to include at least one of the input device 101 and the display device 102.

[0033] External I / F 103 is an interface with the recording medium 103a. The model creation device 10 can read from and write to the recording medium 103a via the external I / F 103. Examples of recording medium 103a include CD (Compact Disc), DVD (Digital Versatile Disc), SD memory card (Secure Digital memory card), and USB (Universal Serial Bus) memory card.

[0034] The communication interface 104 is an interface for connecting the model creation device 10 to a communication network. The processor 105 is a type of computing device such as a CPU (Central Processing Unit) or a GPU (Graphics Processing Unit). The memory device 106 is a type of storage device such as an HDD (Hard Disk Drive), SSD (Solid State Drive), RAM (Random Access Memory), ROM (Read Only Memory), or flash memory.

[0035] The model creation device 10 according to this embodiment can realize various processes, including the current source model creation process described later, by having the hardware configuration shown in Figure 3. Note that the hardware configuration shown in Figure 3 is just an example, and the model creation device 10 may, for example, have multiple processors 105, multiple memory devices 106, or various other hardware not shown (for example, semiconductor integrated circuits other than CPUs and GPUs).

[0036] Next, the functional configuration of the model creation device 10 according to this embodiment will be described with reference to Figure 4. As shown in Figure 4, the model creation device 10 according to this embodiment has a model creation processing unit 201 and a storage unit 202. The model creation processing unit 201 is realized, for example, by causing a processor 105 to execute one or more computer programs stored in a memory device 106. The storage unit 202 is realized, for example, by a memory device 106 such as an HDD or SSD. The one or more computer programs that realize the model creation processing unit 201 may be stored in a recording medium 103a, for example, or downloaded from a predetermined server device via a communication I / F 104.

[0037] The model creation processing unit 201 uses V stored in the storage unit 202 G Measured value, V D Measurement values ​​and I D A current source model f is created using the measured values. That is, the model creation processing unit 201 performs V G Measured value, V D Measurement values ​​and I D The parameters of the current source model f are learned using the measured values ​​as training data. Here, the model creation processing unit 201 receives the drain current I D A current calculation unit 211 calculates the current, and the built-in voltage V gsi and V dsi The system includes an internal voltage calculation unit 212 that calculates the current source model f, and a parameter update unit 213 that updates the parameters of the current source model f. Note that any function f that has the parameters to be learned can be used as the function f representing the current source model.

[0038] The memory unit 202 stores the V G measurement values, V D measurement values, and I D measurement values. Also, the memory unit 202 stores the parameters of the current source model f.

[0039] Here, the V G measurement values, V D measurement values, and I D measurement values are the pre-measured gate voltage V G measurement value, drain voltage V D measurement value, and drain current I D measurement value, respectively. Hereinafter, with the index representing the measurement time being t, at t = 1, ···, T, the V G measurement value V G '(t), V D measurement value V D '(t), and I D measurement value I D '(t) are assumed to be those given to the model creation device 10. At this time, a T-dimensional vector having each V G measurement value V G '(t) as an element is defined as V G ' such that V G ' = (V G '(1), ···, V G '(T)). Similarly, a T-dimensional vector having each V D measurement value V D '(t) as an element is defined as V D ' such that V D ' = (V D '(1), ···, V D '(T)). Similarly, a T-dimensional vector having each I D measurement value I D '(t) as an element is defined as I D ' such that I D ' = (I D '(1), ···, I D '(T)). Note that when there is no misunderstanding, the T-dimensional vectors V G ', V D ', and I D ' are also the V G measurement values, VD Measurement values ​​and I D It is sometimes referred to as a measured value.

[0040] Next, the process for creating the current source model according to this embodiment will be explained with reference to Figure 5. This current source model creation process includes a small loop (the loop from steps S107 to S109) that fixes the internal voltage and updates the parameters, and a large loop (the loop from steps S105 to S111) that changes the drain current using the bisection method and recalculates the internal voltage. In the following, it will be assumed that the parameters of the current source model f have been initialized using known methods, etc.

[0041] The built-in voltage calculation unit 212 of the model creation processing unit 201 calculates the V stored in the storage unit 202 G Measured value, V D Measurement values ​​and I D Using the measured values, the built-in voltage V gsi and V dsi The built-in voltage V is calculated (step S101). Specifically, the built-in voltage calculation unit 212 calculates the built-in voltage V as follows: gsi and V dsi Calculate.

[0042] V gsi =V G '-R s I D ' V dsi =V D '-(R s +R d )I D ' Note V gsi and V dsi Both are T-dimensional vectors, and V gsi =(V gsi (1), ···,V gsi (T)), V dsi =(V dsi (1), ···,V dsi Note that it is expressed as (T). Also, R s and R d It should also be noted that all of these are scalars. That is, for each t=1,···,T, Vgsi (t) = V G '(t)-R s I D '(t), V dsi (t) = V D '(t)-(R s +R d )I D It is (t).

[0043] The parameter update unit 213 of the model creation processing unit 201 initializes n, an index that increases by 1 each time the large loop is executed, to 1 (step S102).

[0044] The current calculation unit 211 of the model creation processing unit 201 calculates the built-in voltage V calculated in step S101 above. gsi and V dsi Using this, the drain current I when n=1 D (1) The current calculation unit 211 calculates (step S103). Specifically, the current calculation unit 211 calculates I D (1) =f(V gsi ,V dsi ) due to drain current I D (1) Calculate I. D (1) is a T-dimensional vector, and I D (1) =(I D (1) (1), ···, I D (1) (T))=(f(V gsi (1), V dsi (1)),···,f(V gsi (T),V dsi Please note that (T))) is expressed as follows: That is, for each t=1,···,T, I D (1) (t) = f(V gsi (tPAV) dsi (t))

[0045] The current calculation unit 211 of the model creation processing unit 201 stores the intermediate value of the drain current in a variable I D_middle I D(1) Set / 2 (step S104). Specifically, the current calculation unit 211 sets I D_middle =(I D_middle (1), ···, I D_middle (T))=(I D (1) (1) / 2,···,I D (1) Set it to (T) / 2).

[0046] The parameter update unit 213 of the model creation processing unit 201 adds 1 to n (step S105).

[0047] The built-in voltage calculation unit 212 of the model creation processing unit 201 is I D_middle And, V G Measured values ​​and V D Using the measured value, the built-in voltage V gsi and V dsi The built-in voltage V is calculated (step S106). Specifically, the built-in voltage calculation unit 212 calculates the built-in voltage V as follows: gsi and V dsi Calculate.

[0048] V gsi =V G '-R s I D_middle V dsi =V D '-(R s +R d )I D_middle That is, for each t=1,...,T, V gsi (t) = V G '(t)-R s I D_middle (t), V dsi (t) = V D '(t)-(R s +R d )I D_middle Let (t) be the case.

[0049] The current calculation unit 211 of the model creation processing unit 201 calculates the built-in voltage V calculated in step S106 above. gsi and V dsi Using this, drain current ID (n) The current calculation unit 211 calculates (step S107). Specifically, the current calculation unit 211 calculates I D (n) =f(V gsi ,V dsi ) due to drain current I D (n) Calculate I. D (n) =(I D (n) (1), ···, I D (n) (T))=(f(V gsi (1), V dsi (1)),···,f(V gsi (T),V dsi Please note that (T))) is expressed as follows: That is, for each t=1,···,T, I D (n) (t) = f(V gsi (tPAV) dsi (t))

[0050] The parameter update unit 213 of the model creation processing unit 201 determines whether or not the parameters of the current source model f have converged (step S108). The parameter update unit 213 determines, for example, I D (n) and I D If the error with ' falls below a predetermined threshold th1, it is determined that the parameters have converged; otherwise, it is determined that the parameters have not converged. Here, I D (n) and I D Various error functions (error function) E can be used, but for example, the mean squared error shown below can be used.

[0051]

number

[0052] In step S108 above, if it is determined that the parameters of the current source model f have not converged, the parameter update unit 213 of the model creation processing unit 201 will D (n) and I D Using the error, the parameters of the current source model f are updated to minimize that error (step S109). Various optimization methods can be used to update the parameters of the current source model f, but for example, the parameters of the current source model f can be updated using Newton's method.

[0053] If step S109 above is executed, the model creation processing unit 201 returns to step S107. This results in the built-in voltage V gsi and V dsi With the parameters fixed, the small loop (steps S107 to S109) is repeatedly executed until the parameters of the current source model f converge.

[0054] On the other hand, if it is determined in step S108 above that the parameters of the current source model f have converged, the parameter update unit 213 of the model creation processing unit 201 will perform I D (n) It is determined whether or not the result has converged (step S110). The parameter update unit 213, for example, I D (n) and I D (n-1) When the difference between the two becomes less than or equal to a predetermined threshold th2, I D (n) Determine that it has converged, otherwise I D (n) We just need to determine if it has not converged. Here, I D (n) and I D (n-1) Various methods can be used to determine the difference, but for example, the mean absolute error shown below can be used.

[0055]

number

[0056] In step S110 described above, D (n) If it is determined that the process has not converged, the current calculation unit 211 of the model creation processing unit 201 will D_middle (I D (n) +I D (n-1) Set ) / 2 (step S111). That is, the current calculation unit 211 sets I D (n) and I D (n-1) I is the midpoint between the two. D_middle Set to I D_middle =(I D_middle (1), ···, I D_middle (T))=((I D (n) (1) + I D (n-1) (1)) / 2,···,(I D (n) (T)+I D (n-1) Set (T)) / 2). Note that following the dichotomy, I D_middle I D (n) and I D (n-1) We set an intermediate value between these, but it is not limited to this; for example, I D (n) and I D (n-1) A value obtained by dividing the distance between the two values ​​by a predetermined ratio may be set.

[0057] If step S111 above is executed, the model creation processing unit 201 returns to step S105.D_middle Using the built-in voltage V gsi and V dsi The calculation is performed again, and the small loop is executed once more.

[0058] On the other hand, in step S110 above, D (n) If it is determined that the process has converged, the model creation processing unit 201 terminates the current source model creation process. This means that a current source model f with the learned parameters has been created.

[0059] As described above, the model creation device 10 according to this embodiment creates a current source model f using an algorithm that includes a small loop in which the internal voltage is fixed and the parameters are updated, and a large loop in which the drain current is changed by the bisection method and the internal voltage is recalculated. In this algorithm, since the internal voltage remains fixed and does not change during the execution of the small loop, it is possible to prevent the parameters of the current source model f from diverging instead of converging.

[0060] Furthermore, in the large loop, the drain current value I is calculated using the bisection method (or a similar method). D_middle Calculate the drain current value I D_middle Using the built-in voltage V gsi and V dsi To recalculate, I D (n) =f(V gsi ,V dsi The convergence of ) is also guaranteed. For example, if the vertical axis is I D , the horizontal axis is V D I D -V D On a plane, D Measurement value I D ' and the calculated value I calculated using the current source model f D (n) The diagram is shown in Figure 6. As shown in Figure 6, as n increases, the bisection method gives I D (n) and I D (n-1) As the difference becomes smaller, D (n)The convergence of is guaranteed, and its convergence value is I D Measurement value I D It can be seen that it fits accurately.

[0061] Therefore, according to the model creation device 10 of this embodiment, a highly accurate current source model f can be created without requiring, for example, manual parameter adjustment.

[0062] (Second Embodiment) Gallium nitride high electron mobility transistors (GaN HEMTs) can experience a problem called current collapse. Current collapse occurs when the drain voltage (V) is relatively large. D (This is called stress.) When the drain current and drain voltage are measured immediately after applying this to the GaN HEMT, I D -V D I on a plane D -V D This refers to the phenomenon of waveform distortion. D -V D Waveform and I of GaN HEMT immediately after stress application D -V D The waveform is shown in Figure 7. As shown in Figure 7, a current collapse occurs immediately after stress is applied, and I D -V D You can see that the waveform is distorted.

[0063] When modeling transistors that can experience current collapse, such as GaN HEMTs, the effect of current collapse is as follows: D -V D When the waveform is distorted, the parameters of function f may not converge, making it impossible to model (or even if modeled, with low accuracy). Therefore, in conventional examples, manual parameter adjustment was often necessary for modeling. On the other hand, it is often not necessary to create a current source model that fits all measured values; a current source model that fits measured values ​​on and around a certain load line is often sufficient for circuit design purposes.

[0064] Therefore, the following describes a model creation device 10 that can accurately create a current source model f by modeling transistors that can generate current collapses and weighting the measurements on and around the load line to fit them.

[0065] In this embodiment, we will mainly describe the differences from the first embodiment, and will omit the explanation of components that are the same as those in the first embodiment (that is, for matters not described in this embodiment, the first embodiment can be applied as is).

[0066] First, the functional configuration of the model creation device 10 according to this embodiment will be explained with reference to Figure 8. As shown in Figure 8, the model creation processing unit 201 of the model creation device 10 according to this embodiment has each I D Measurement value I D Weight ω for (t) t (However, 0≦ω t Furthermore, a weight calculation unit 214 is included that calculates weight ω when determining whether the parameters of the current source model f have converged. t The error is calculated taking this into consideration.

[0067] Next, the current source model creation process according to this embodiment will be described with reference to Figure 9. Steps S202 to S208 of this current source model creation process are the same as steps S101 to S107 of the current source model creation process according to the first embodiment, and steps S210 to S212 are the same as steps S109 to S111, so their explanations will be omitted.

[0068] The weight calculation unit 214 of the model creation processing unit 201 is V D Measurement values ​​and I D Using the measured values, each I D Measurement value I D Weight ω for (t)t (However, 0≦ω t Calculate the weights ω (≤ 100) (step S201). Details of the process in this step will be described later, and below these weights ω t We will continue the explanation assuming that these have been calculated. Furthermore, as will be discussed later, these weights ω t The value will be high for measurements taken on and around the load line, and low for measurements taken in areas far from the load line, and in areas close to the operating point with low power or low efficiency.

[0069] Following step S208, the parameter update unit 213 of the model creation processing unit 201 determines whether or not the parameters of the current source model f have converged (step S209). At this time, the parameter update unit 213 according to this embodiment determines I D (n) and I D The weight ω represents the error with ' t Using an error E that takes this into account, for example, if the error E becomes less than or equal to a predetermined threshold th3, it is determined that the parameters have converged; otherwise, it is determined that the parameters have not converged. Here, weight ω t As an error E that takes this into account, for example, the following errors can be used.

[0070]

number

[0071] This means I D (n) (t) and I D Of the differences with (t), the weight ω tFor differences corresponding to low values ​​of t, the impact on parameter updates can be minimized. Therefore, even when modeling transistors where current collapse can occur, it is possible to prevent the parameters from diverging instead of converging, and to create a current source model f that accurately fits the measured values ​​on and around the load line.

[0072] Next, the weight calculation process in step S201 described above will be explained with reference to Figure 10.

[0073] First, the weight calculation unit 214 calculates the given operating point and V D Measured value V D 'and I D Measurement value I D Using ' and, I D -V D The load line on the plane is calculated (step S301). Here, the operating point is determined by the person who uses the transistor model to design the circuit (for example, the person who receives the transistors and designs and manufactures the circuit, etc.). D and I D This is the value of Q0. Below, the operating point is Q0 = (I D_set ,V D_set As an example of load line calculation, we will explain with reference to Figure 11. Note that each measurement point in Figure 11 is I D Measurement value I D '(t) and V D Measured value V D (t) and the plotted points (I D '(t),V D It refers to (t).

[0074] The weight calculation unit 214 calculates the load line according to the following Steps 1-1 to 1-5.

[0075] Step 1-1: First, the weight calculation unit 214 calculates θ = arccos(1-I D ' / (I D '-I D_setCalculate ). Note that θ is a T-dimensional vector and can be expressed as θ=(θ(1),···,θ(T)). Also, I D_set Note that is a scalar. That is, for each t=1,···,T, θ(t)=arccos(1-I D '(t) / (I D '(t)-I D_set )) is.

[0076] Step 1-2: Next, the weight calculation unit 214 calculates the following output power P out Alternatively, calculate the efficiency (Eff).

[0077] P out =(I D '-I D_set )×(V D_set -V D ')×(θ-sinθ×cosθ) / 2π Eff = 0.5 × (1 - V) D ' / V D_set )×(θ-sinθ×cosθ) / sinθ Note P out Alternatively, Eff is a T-dimensional vector, and P out =(P out (1), ···,P out Note that (T) and Eff can be expressed as Eff=(Eff(1),···,Eff(T)). That is, for each t=1,···,T, P out (t)=(I D '(t)-I D_set )×(V D_set -V D '(t))×(θ(t)-sinθ(t)×cosθ(t)) / 2π, Eff(t)=0.5×(1-V D '(t) / V D_set )×(θ(t)-sinθ(t)×cosθ(t)) / sinθ(t).

[0078] Step 1-3: Next, the weight calculation unit 214 calculates the output power P in Step 1-2 above. out If you calculate P outThe measurement point corresponding to the element with the largest value among the elements is calculated. That is, t max =argmax(P out If (t) then, (I D '(t max ),V D '(t max )) calculate. Note that this measurement point (I D '(t max ),V D '(t max )) is also called the output matching point.

[0079] On the other hand, if the weight calculation unit 214 has calculated the efficiency Eff in Step 1-2 above, it calculates the measurement point corresponding to the element with the largest value among the elements of Eff. That is, t max If we set =argmax(Eff(t)), then (I D '(t max ),V D '(t max )) calculate. Note that this measurement point (I D '(t max ),V D '(t max )) is also called the efficiency matching point.

[0080] In the following, the output matching point or efficiency matching point is defined as Q1=(I D '(t max ),V D '(t max )) Note that whether to calculate the output matching point or the efficiency matching point is predetermined by, for example, the person who designs the circuit using a transistor model.

[0081] Step 1-4: Next, the weight calculation unit 214 calculates the following points Q2, Q3, Q4, and Q5.

[0082] A straight line passing through Q0 and Q1, and V D axis (that is, I D Intersection point Q2 with the line (=0). Q1 to V D The perpendicular line drawn from the axis, and V D Intersection point Q3 with the axis Q0 to VD The perpendicular line drawn from the axis, and V D Intersection point Q4 with the axis Let H be the distance between Q3 and Q4, V D Q5 is a point on the axis located at a distance H from Q4, and is different from Q3. Note that point Q5 mentioned above is also called the maximum operating point on the off side.

[0083] Step 1-5: The weight calculation unit 214 then uses a line (a polyline in the example shown in Figure 11) composed of line segments passing through Q1, Q0, and Q2, and line segments passing through Q2 and Q5 as the load line. Note that the line segments passing through Q1, Q0, and Q2 are I D Since the value of is greater than or equal to 0, it is also called the on-state load line. On the other hand, the line segment passing through Q2 and Q5 is I D Because its value is always 0, it is also called the off-state load line.

[0084] Following step S301, the weight calculation unit 214 calculates each measurement point (I D '(t),V D Weight ω for (t) t Calculate the weight ω t to I D Measurement value I D '(t) is the weight (step S302). Here, the weight calculation unit 214 is I D -V D After defining regions on a plane, the weight of each measurement point is calculated based on the region to which it belongs and its distance to the load line.

[0085] First, I D -V D The definition of the region on the plane will be explained with reference to Figure 12. The weight calculation unit 214 performs the following steps 2-1 to 2-7: D -V D Define a region on a plane. Here, in the following, I D -V D In a plane, all measurement points {(I D '(t),V D Let W be the smallest region that contains all measurement points {(ID '(t),V D The region containing '(t)|t=1,···,T} and having the smallest area corresponds to region W. Note that the output matching point or efficiency matching point is a point on the boundary of region W.

[0086] Step 2-1: The weight calculation unit 214 is a straight line parallel to the ON state load line, and I D -V D Calculate the line a1 that passes through the origin O of the plane.

[0087] Step 2-2: The weight calculation unit 214 calculates a line a2 that is parallel to the on-state load line and passes midway between line a1 and the on-state load line. In other words, the weight calculation unit 214 calculates a line a2 that is parallel to the on-state load line and passes midway between the perpendicular line drawn from the on-state load line to line a1.

[0088] Step 2-3: The weight calculation unit 214 is I D Calculate a3, which is a straight line parallel to the axis and passes through the operating point Q0.

[0089] Step 2-4: The weight calculation unit 214 is V D Calculate a4, which is a straight line parallel to the axis and passes through the operating point Q0.

[0090] Step 2-5: The weight calculation unit 214 is V D A straight line parallel to the axis, and also a line a4 and V D The line a5 passing midway between the axis is calculated. In other words, the weight calculation unit 214 calculates V D A straight line parallel to the axis, and from line a4 to V D Calculate the line a5 that passes through the midpoint of the perpendicular line drawn to the axis.

[0091] Step 2-6: The weight calculation unit 214 is I D Calculate the line a6 that is parallel to the axis and passes through Q2.

[0092] Step 2-7: The weight calculation unit 214 defines regions W1, W2, W3, and W4 as follows.

[0093] The boundary of region W, the line a2, and V D The region enclosed by the axis, line a3, and load line is defined as W1. The region enclosed by the boundary of region W, the load line, and the line a3 is called W2. The boundary of region W, line a5, line a6, and V D The area enclosed by the axis is W3 Of the regions W, the regions other than regions W1, W2, and W3 are designated as W4. Next, each measurement point (I D '(t),V D The measurement point (I) depends on the region to which (t) belongs and the distance to the load line. D '(t),V D The weight ω of (t) t The calculation of will be explained with reference to Figure 13. The weight calculation unit 214 calculates each measurement point (I) according to the following Steps 3-1 to 3-7. D '(t),V D The weight ω of (t) t Calculate.

[0094] Step 3-1: The weight calculation unit 214 calculates the distance Δ1, which represents the length of the perpendicular line drawn from the ON state load line to the straight line a2.

[0095] Step 3-2: The weight calculation unit 214 calculates each measurement point (I D '(t),V D For each (t)∈W1, calculate Δ2(t), which represents the length of the perpendicular line drawn from the measurement point to the on-state load line.

[0096] Step 3-3: The weight calculation unit 214 calculates each measurement point (I D '(t),V D For each (t)∈W2, calculate the distance Δ3(t) which represents the length of the perpendicular line that passes through the measurement point and is drawn from the on-state load line to the straight line a3.

[0097] Step3-4: The weight calculation unit 214 calculates, for each measurement point (I D '(t), V D '(t)) ∈ W2, the distance Δ4(t) representing the length of the perpendicular line drawn from the measurement point to the straight line a3, respectively.

[0098] Step3-5: The weight calculation unit 214 calculates the distance Δ5 representing the length of the perpendicular line drawn from the V D axis to the straight line a5.

[0099] Step3-6: The weight calculation unit 214 calculates, for each measurement point (I D '(t), V D '(t)) ∈ W3, the distance Δ6(t) representing the length of the perpendicular line drawn from the measurement point to the straight line a5, respectively.

[0100] Step3-7: The weight calculation unit 214 calculates the weight ω D '(t), V D '(t)) of each measurement point as follows. t

[0101] · When (I D '(t), V D '(t)) ∈ W1 Set c1(t) = (Δ2(t) / Δ1) × π / 2, and ω t = 100 × cos(c1(t)) · When (I D '(t), V D '(t)) ∈ W2 Set c2(t) = (Δ4(t) / Δ3(t)) × π / 2, and ω t = 100 × cos(c2(t)) · When (I D '(t), V D '(t)) ∈ W3 Set c3(t) = (Δ6(t) / Δ5) × π / 2, and ω t = 100 × cos(c3(t)) · When (I D '(t), V D '(t)) ∈ W4 ω t = 0 As a result, for each measurement point (I D '(t), V D '(t)) ∈ W, its weight ω t is obtained. In the above calculation method, a weight with a large value is obtained for the measured values on and around the load line, the value of the weight decreases as the distance from the load line increases, and the value of the weight is set to 0 in a region separated by a certain degree or more or in a specific region. Therefore, it becomes possible to update the parameters of the current source model f so as to fit the measured values on and around the load line.

[0102] As described above, the first and second embodiments of the present disclosure have been described in detail. However, the present disclosure is not limited to these specific embodiments, and various modifications and changes are possible within the scope described in the claims.

Description of Reference Numerals

[0103] 10: Model creation device 101: Input device 102: Display device 103: External I / F 103a: Recording medium 104: Communication I / F 105: Processor 106: Memory device 107: Bus 201: Model creation processing unit 202: Storage unit 211: Current calculation unit 212: Built-in voltage calculation unit 213: Parameter update unit 214: Weight calculation unit

Claims

1. A method for creating a model that generates a current source model representing the relationship between the internal voltage and drain current of a transistor, An internal voltage calculation step that calculates the internal voltage using a gate voltage measurement value representing the measured gate voltage of the transistor, a drain voltage measurement value representing the measured drain voltage of the transistor, and an intermediate value of the drain current calculation value representing the drain current calculated by the current source model, A current calculation step in which the drain current is calculated using the built-in voltage and the current source model, An update step to update the parameters of the current source model using the error between the calculated drain current value and the measured drain current value representing the measured drain current of the transistor, It includes, The computer repeats a first loop process, which includes the current calculation step and the update step, until the parameters of the current source model converge; and further repeats a second loop process, which includes the built-in voltage calculation step and the first loop process, until the drain current calculation value converges. The aforementioned intermediate value is, A model creation method in which the calculated drain current value is an intermediate value between the drain current calculated in the second loop processing in the nth iteration and the drain current calculated in the second loop processing in the (n-1)th iteration.

2. The weight calculation step includes calculating a weight for the drain current measurement using the drain voltage measurement and the drain current measurement, The model creation method according to claim 1, wherein the error multiplied by the weights is used to determine whether or not the parameters of the current source model have converged.

3. The aforementioned weight calculation step is: Using the operating point of the transistor's large-signal operation, the measured drain voltage, and the measured drain current, the load line for the large-signal operation is calculated. The model creation method according to claim 2, wherein the weight of the drain current measurement value is calculated such that the closer the drain current measurement value is to the load line, the higher the weight of the drain current measurement value is, and the further away the drain current measurement value is, the lower the weight of the drain current measurement value is.

4. A computer is used to create a current source model that represents the relationship between the internal voltage and drain current of a transistor. An internal voltage calculation step that calculates the internal voltage using a gate voltage measurement value representing the measured gate voltage of the transistor, a drain voltage measurement value representing the measured drain voltage of the transistor, and an intermediate value of the drain current calculation value representing the drain current calculated by the current source model, A current calculation step in which the drain current is calculated using the built-in voltage and the current source model, An update step to update the parameters of the current source model using the error between the calculated drain current value and the measured drain current value representing the measured drain current of the transistor, Make it run, The first loop process, which includes the current calculation step and the update step, is repeated until the parameters of the current source model converge, and the second loop process, which includes the built-in voltage calculation step and the first loop process, is repeated until the drain current calculation value converges. The aforementioned intermediate value is, A program that provides a value intermediate between the drain current calculated in the second loop process during the nth iteration and the drain current calculated in the second loop process during the (n-1)th iteration.

Images