SPICE model simulation system and simulation method

Abstract

The present invention discloses a SPICE model simulation system, comprising: a model import module, which is used to import a model file; a model parsing module, which is used to determine the type of the imported model file, and parse the model file type through a model card simulator lexical analyzer to obtain model parameters; a measured data import module, which is used to import device measured data; a simulation condition input module, which is used to input device physical characteristics and device simulation conditions; a simulation optimization module, which is used to output simulation results after performing simulation optimization on model files, model parameters, device measured data, device physical characteristics and device simulation conditions. The present invention also discloses a SPICE model simulation method. The present invention can reduce the technical requirements of operators and can quickly and accurately obtain simulation results.

Description

Technical Field

[0001] The present invention relates to the field of integrated circuit manufacturing, and in particular to a SPICE model simulation system. The present invention also relates to a SPICE model simulation method. Background Art

[0002] In the current R&D and manufacturing process of VLSI, efficient and accurate SPICE models are the key to the circuit from functional design to implementation. The developed SPICE models must be simulated with the help of various commercial EDA simulation software such as HSPICE, Spectre, Eldo, etc. Unless it is a complex circuit-level simulation, the general transistor-level electrical characteristics simulation using commercial EDA simulation software is often overkill. In addition, the use of these EDA software must also be familiar with the use of each software and the syntax of the simulation netlist, etc., and professional SPICE model engineers or circuit design engineers are required to implement complex simulation processes.

[0003] The core codes of MOSFET transistors in the current EDA simulation software in the industry are based on the fully open-source BSIM SPICE simulation algorithm (developed in C language) developed by the Berkeley Short-channel IGFET Model Group (http: / / bsim.berkeley.edu / ) of the Department of Electrical Engineering and Computer Sciences (EECS) at the University of California, Berkeley, and combined with the characteristics of their own simulation software and a series of licenses. With the international standards and the continuous optimization of internal process algorithms, the threshold of existing SPICE model simulation technology has gradually increased, and the technical requirements for semiconductor engineering are also getting higher and higher, which is not conducive to the development of engineering projects. Summary of the invention

[0004] A series of simplified concepts are introduced in the Summary of the Invention, which are all simplifications of the prior art in the field, and will be further described in detail in the Detailed Description of the Invention. The Summary of the Invention of the present invention does not mean to attempt to define the key features and essential technical features of the claimed technical solution, nor does it mean to attempt to determine the scope of protection of the claimed technical solution.

[0005] The technical problem to be solved by the present invention is to provide a SPICE model simulation system which can reduce the technical requirements of operators and can quickly and accurately obtain simulation results by only inputting model files, measured data and simulation conditions.

[0006] Another technical problem to be solved by the present invention is to provide a SPICE model simulation method that can reduce the technical requirements of operators and can quickly and accurately obtain simulation results by only inputting model files, measured data and simulation conditions.

[0007] In order to solve the above technical problems, the present invention provides a SPICE model simulation system, comprising:

[0008] Model import module, which is used to import model files;

[0009] A model parsing module is used to determine the type of the imported model file and parse the model file type through the lexical analyzer of the model card simulator to obtain model parameters;

[0010] A measured data import module, which is used to import the measured data of the device;

[0011] A simulation condition input module, which is used to input device physical characteristics and device simulation conditions;

[0012] By loading batch OP (Operation Point, working point simulation) file input files, the simulation values ​​of the same type of devices at the same voltage, different sizes, and temperatures can be realized. In addition, the path, name, type, and batch file information of the loaded model can be included.

[0013] The simulation optimization module is used to output simulation results after performing simulation optimization on the model file, model parameters, device measured data, device physical characteristics and device simulation conditions.

[0014] Optionally, the SPICE model simulation system is further improved, and the imported model file includes a model library file and / or a single model parameter file.

[0015] The model library file includes multiple MOS models and different process angle models;

[0016] Model parsing module To realize automatic parsing of model files, the present invention develops a model card simulator lexical analyzer combined with a GUI interface that can quickly load a single model file and the entire model library file and automatically parse to read model card parameters.

[0017] The automatic lexical analyzer is divided into a simple lexical analyzer and an efficient lexical analyzer. The simple lexical analyzer is specifically for the Compact model, and the efficient lexical analyzer is for the complex Sub-Circuit model.

[0018] The simple lexical analyzer uses the dictionary method in program programming to establish a binary structure {parameter name, parameter value} for each parameter, so that the model card parameters can be easily read during the simulator operation. For example, the form shown in Table 1 below;

[0019]

[0020]

[0021] Table 1

[0022] The efficient lexical analyzer is a function parser developed for the increasing number of mathematical functions used in the current SPICE model to meet the fitting accuracy. It performs function judgment by recognizing specific characters after conversion into ASCII codes. It can automatically parse out operations such as MIN (minimum value), MAX (maximum value), INT (rounding) functions, PWR, LOG, SQR, and a judgment operation.

[0023] Optionally, the SPICE model simulation system is further improved, and the device measured data at least includes MOSFET voltage-current scanning data.

[0024] By loading the MOSFET voltage-current scan data, the voltage-current data is imported and the corresponding current value is output based on the simulation of each voltage value. And multiple data files can be loaded at the same time using the visual interface to perform simulation calculations at the same time, which can realize the simultaneous display of scan data and simulation, and can select multiple data for rapid simulation at the same time.

[0025] Optionally, the SPICE model simulation system is further improved, and the simulation optimization module performs simulation optimization including at least:

[0026] 1) Eliminate the internal voltage iteration of the device caused by parasitic resistance caused by wiring;

[0027] 2) Threshold voltage search iteration;

[0028] 3) Statistical mismatch model, generate mismatch model simulation random numbers.

[0029] Through the above optimization process and the preset output parameter definition, simulation and basic electrical characteristics output are achieved. The output content includes saturation current under DC, four-terminal port current such as gate / source / drain / substrate, threshold voltage, simulation based on Monte Carlo mismatch model, and also includes AC characteristics such as gate oxide capacitance, channel capacitance, 1 / f flicker noise at different frequencies and other commonly used simulation results output.

[0030] Optionally, the SPICE model simulation system is further improved, and the simulation optimization module performs simulation optimization 1) including the following steps;

[0031] 1.1) Calculate the source-drain current under external applied voltage;

[0032] 1.2) Calculate the voltage due to the wire and silicon oxide contact resistance (Rsc) and source / drain diffusion resistance (Rs,diff) using Ohm's law;

[0033] 1.3) Subtract the parasitic voltage from the initial voltage and substitute it as the new initial voltage to calculate the new current;

[0034] 1.4) Repeat the above steps 1.1) to 1.3) until the deviation between the initial voltage and the last applied initial voltage is less than the first threshold.

[0035] Optionally, the SPICE model simulation system is further improved, and the simulation optimization module performs simulation optimization 2) using the following steps;

[0036] 2.1) Set the minimum voltage, maximum voltage and average voltage, and calculate the simulated current value under the average voltage;

[0037] 2.2) The simulation current value is obtained based on the binary scanning voltage cycle iteration calculation, and the simulation current is compared with the target current value until the error between the simulation current and the target current is less than the second threshold value or the number of iterations of the simulation current is greater than the third threshold value, then the iteration is terminated and the average voltage at this time is used as the threshold voltage.

[0038] Optionally, the SPICE model simulation system is further improved, and the simulation optimization module performs simulation optimization 3) using the following steps;

[0039] 3.1) Use the random number function Rnd() to generate random number variables x1 and x2 between 0 and 1;

[0040] 3.2) Use formula (1) to generate a set of random numbers that conform to the normal distribution;

[0041]

[0042] 3.3) The two devices call the random number at intervals according to the number of simulations to implement the mismatch model simulation.

[0043] The present invention provides a SPICE model simulation method, comprising the following steps:

[0044] S1, import model file;

[0045] S2, determine the type of the imported model file, and parse the model file type through the model card simulator lexical analyzer to obtain model parameters;

[0046] S3, import device measured data;

[0047] S4, input device physical characteristics and device simulation conditions;

[0048] S5, outputting simulation results after performing simulation optimization on the model file, model parameters, device measured data, device physical characteristics and device simulation conditions.

[0049] Optionally, the SPICE model simulation method is further improved, and the imported model file includes a model library file and / or a single model parameter file.

[0050] Optionally, the SPICE model simulation method is further improved, and the model card simulator lexical analyzer includes a simple lexical analyzer and an efficient lexical analyzer;

[0051] Simple lexical analyzer, which is used for Compact model, uses dictionary method to build binary structure for each parameter and read it into model card parameter

[0052] An efficient lexical analyzer is used for complex Sub-Circuit models. It recognizes specific characters in the model file to make function judgments and can automatically parse and obtain operation algorithms and judgment operations.

[0053] Optionally, the SPICE model simulation method is further improved, and the device measured data at least includes MOSFET voltage-current scanning data.

[0054] Optionally, the SPICE model simulation method is further improved, and when step S5 is implemented, the simulation optimization at least includes:

[0055] 1) Eliminate the internal voltage iteration of the device caused by parasitic resistance caused by wiring;

[0056] 2) Threshold voltage search iteration;

[0057] 3) Statistical mismatch model, generate mismatch model simulation random numbers.

[0058] Optionally, the SPICE model simulation method is further improved, and the simulation optimization 1) includes the following steps:

[0059] 1.1) Calculate the source-drain current under external applied voltage;

[0060] 1.2) Calculate the voltage due to the wire and silicon oxide contact resistance (Rsc) and source / drain diffusion resistance (Rs,diff) using Ohm's law;

[0061] 1.3) Subtract the parasitic voltage from the initial voltage and substitute it as the new initial voltage to calculate the new current;

[0062] 1.4) Repeat the above steps 1.1) to 1.3) until the deviation between the initial voltage and the last applied initial voltage is less than the first threshold.

[0063] Optionally, the SPICE model simulation method is further improved to perform simulation optimization 2) using the following steps;

[0064] 2.1) Set the minimum voltage, maximum voltage and average voltage, and calculate the simulated current value under the average voltage;

[0065] 2.2) The simulation current value is obtained based on the binary scanning voltage cycle iteration calculation, and the simulation current is compared with the target current value until the error between the simulation current and the target current is less than the second threshold value or the number of iterations of the simulation current is greater than the third threshold value, then the iteration is terminated and the average voltage at this time is used as the threshold voltage.

[0066] Optionally, the SPICE model simulation method is further improved to perform simulation optimization 3) using the following steps;

[0067] 3.1) Use the random number function Rnd() to generate random number variables x1 and x2 between 0 and 1;

[0068] 3.2) Use formula (1) to generate a set of random numbers that conform to the normal distribution;

[0069]

[0070] 3.3) The two devices call the random number at intervals according to the number of simulations to implement the mismatch model simulation.

[0071] The BSIM4 model is a MOSFET transistor model standard recognized by the industry and developed by the UC Berkley BSIM group. It has been widely used in technology nodes from 0.35um to 20nm. The present invention can develop a simulation system based on a user visual GUI interface using the VB programming language based on the core algorithm of BSIM4.5 developed by UC Berkley. Through a few simple steps, it can realize multiple functions from model library analysis, electrical simulation to parameter adjustment, simulation output, etc., and can continuously expand functions according to actual needs.

[0072] The core algorithm of the SPICE model standard BSIM4.5 formulated by the UC Berkley BSIM group is implemented using C language. The mainstream commercial EDA simulation software in the industry directly adopts its code and combines its own algorithm and interface. To use these commercial simulation software to realize the simulation of SPICE models, the simulation netlist must be written in accordance with the grammatical rules of their respective software, requiring users to have a strong professional background; the present invention realizes multiple functions such as model library parsing, electrical simulation and parameter adjustment under the GUI user interface through code conversion from C language to VB. Users only need to input the GUI interface to complete the commonly used simulation without strong professional requirements, and all model formulas and parameters are in accordance with the standards provided by the UC Berkley BSIM group. The simulation accuracy is completely consistent with the commercial EDA simulation software, and the demand for simulation output of more electrical characteristics can be further expanded and realized by improving the code. The present invention is compared with the results of commercial EDA simulation software, and the simulation calculation accuracy is completely consistent, and due to the use of simulation optimization, a faster simulation speed can be achieved, realizing the transistor-level fast simulation of the SPICE model. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] The drawings of the present invention are intended to illustrate the general characteristics of the methods, structures and / or materials used in specific exemplary embodiments of the present invention, and to supplement the description in the specification. However, the drawings of the present invention are schematic diagrams not drawn to scale, and thus may not accurately reflect the precise structure or performance characteristics of any given embodiment, and the drawings of the present invention should not be interpreted as limiting or restricting the range of values ​​or properties covered by the exemplary embodiments of the present invention. The present invention is further described in detail below in conjunction with the drawings and specific embodiments:

[0074] Figure 1 This is a diagram of the workflow of a simple lexical analyzer for the Compact model.

[0075] Figure 2 It is a schematic diagram of the workflow of an efficient lexical analyzer for complex Sub-Circuit models.

[0076] Figure 3 It is a flow chart of the SPICE model simulation method of the present invention.

[0077] Figure 4 It is a schematic diagram of an equivalent circuit of the calculation process of the present invention. DETAILED DESCRIPTION

[0078] The following describes the implementation methods of the present invention through specific specific embodiments, and those skilled in the art can fully understand other advantages and technical effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through different specific implementation methods, and the details in this specification can also be applied based on different viewpoints, and various modifications or changes can be made without deviating from the overall design concept of the invention. It should be noted that the following embodiments and the features in the embodiments can be combined with each other in the absence of conflict. The following exemplary embodiments of the present invention can be implemented in a variety of different forms and should not be interpreted as being limited to the specific embodiments described herein. It should be understood that these embodiments are provided to make the disclosure of the present invention thorough and complete, and to fully convey the technical solutions of these exemplary specific embodiments to those skilled in the art.

[0079] First embodiment;

[0080] The present invention provides a SPICE model simulation system, comprising:

[0081] A model import module, which is used to import model files, including model library files and / or single model parameter files;

[0082] A model parsing module is used to determine the type of the imported model file and parse the model file type through the lexical analyzer of the model card simulator to obtain model parameters;

[0083] A measured data import module, which is used to import the measured data of the device;

[0084] A simulation condition input module, which is used to input device physical characteristics and device simulation conditions;

[0085] The simulation optimization module is used to output simulation results after performing simulation optimization on the model file, model parameters, device measured data, device physical characteristics and device simulation conditions.

[0086] Second embodiment;

[0087] The present invention provides a SPICE model simulation system, comprising:

[0088] A model import module, which is used to import model files, including model library files and / or single model parameter files;

[0089] The model parsing module is used to determine the type of the imported model file and parse the model file type through the model card simulator lexical analyzer to obtain the model parameters; the model card simulator lexical analyzer includes a simple lexical analyzer and an efficient lexical analyzer; the simple lexical analyzer is used for the Compact model and uses the dictionary method to establish a binary structure for each parameter and read it into the model card parameter;

[0090] The simple lexical analyzer uses the dictionary method in program programming to establish a binary structure {parameter name, parameter value} for each parameter, which makes it easy to read the model card parameters during the simulator operation. The Corner parameters are read in the model card, and each Corner SPICE model can be simulated, which can solve most of the Compact Model text reading and parsing work. The main functions include at least:

[0091] a) Filter spaces;

[0092] b) Identification identifier: {parameter name, parameter value};

[0093] c) Recognize numbers (1|2|3|4|5|6|7|8|9|0);

[0094] d) Identify typical operators and separators, such as +, -, *, / , (), etc.;

[0095] e) Have error handling capabilities and be able to detect invalid characters in the programming language, etc.

[0096] The workflow of a simple lexical analyzer for the Compact model is as follows Figure 1 shown.

[0097] An efficient lexical analyzer is used for complex Sub-Circuit models. It recognizes specific characters in the model file to make function judgments and can automatically parse and obtain operation algorithms and judgment operations.

[0098] The efficient lexical analyzer is a function parser developed for the increasing number of mathematical functions used in the current SPICE model to meet the fitting accuracy. It can automatically parse out functions such as MIN (minimum value), MAX (maximum value), INT (rounding) functions, PWR, LOG, SQR and other operations as well as a judgment operation by recognizing characters converted to ASCII codes. For example, the PWR power function conversion is used as an example.

[0099] +inv_sa1ref='1e-21 / pwr(((sa1_ref+0.1*l)*scale_m),3)'

[0100] Take the right side of the above formula 1e-21 / pwr(((sa1_ref+0.1*l)*scale_m),3) to search for the pwr keyword and record the location of the keyword. Find the nearest "(". We know that if the number of "(" and ")" in any expression is equal, the expression is complete. This method can be used to extract

[0101] pwr(((sa1_ref+0.1*l)*scale_m),3) converts "," to "^", and finally gets the expression 1e-21 / (((sa1_ref+0.1*l)*scale_m)^3), which enables the computer to read and calculate normally after the unknown quantity is introduced. However, some expressions have the phenomenon of pwr function nested in pwr function, so we need to iterate continuously through this method until there is no pwr function in the function.

[0102] In expressions, you will often encounter basic functions that are not included in VB programming, such as MIN and MAX. The solution process is as follows: Figure 2 As shown in the figure, an efficient lexical analyzer workflow for complex Sub-Circuit sub-circuit models.

[0103] +SEL_SA=MAX(MIN(SA,SA_MAX),SA_MIN)

[0104] Take the right side of the above formula MAX(MIN(SA,SA_MAX),SA_MIN), first substitute SA, SA_MAX and SA_MIN to get MAX(MIN(4.0E-7,1.0E-5),1.0E-7), then search for the MAX function to distinguish which is larger, MIN(4.0E-7,1.0E-5) and 1.0E-7, then parse MIN(4.0E-7,1.0E-5) to get 4.0E-7, then rearrange the formula to get MAX(4.0E-7,1.0E-7), and use the MAX function to get the final +SEL_SA=4.0E-7. This is a relatively simple function parsing process. This method can be used to infer the parsing of other functions and better improve the versatility of the model card lexical analyzer.

[0105] The measured data import module is used to import the measured data of the device; the measured data of the device at least includes the voltage-current scanning data of MOSFET. In addition, the basic information such as device type, size, scale, temperature, etc. is intuitively imported; the path, name, type and batch file information of the loaded model are also imported.

[0106] A simulation condition input module, which is used to input device physical characteristics and device simulation conditions;

[0107] By loading batch OP (Operation Point, working point simulation) file input files, the simulation values ​​of the same type of devices at the same voltage, different sizes, and temperatures can be realized. In addition, the path, name, type, and batch file information of the loaded model can be included.

[0108] The simulation optimization module is used to output simulation results after performing simulation optimization on the model file, model parameters, device measured data, device physical characteristics and device simulation conditions; the simulation optimization module performs simulation optimization including at least:

[0109] 1) Eliminate the internal voltage iteration of the device caused by parasitic resistance caused by wiring;

[0110] For a SPICE model simulation software, simulation accuracy and speed are two important factors. As far as transistor simulation accuracy is concerned, the more accurate the actual voltage applied to the inside of the transistor device is, the more accurate the simulation value will be. Because the transistor device structure produced through a complex semiconductor manufacturing process is composed of a variety of different materials, the externally applied voltage will be divided into a part of the voltage (such as the voltage of the transistor) before it is given to the inside of the device through the four ports of the transistor due to the influence of parasitic factors such as the connection and silicon oxide contact resistance (Rsc), source / drain diffusion resistance (Rs,diff), etc. Figure 4 ), if the original voltage value is still used as the input condition, it will cause errors. Currently, commercial software generally adopts the node analysis method based on Kirchhoff's Current Law (KCL), which realizes complex circuit analysis by solving the matrix equations composed of each node voltage and iterating the node analysis. Its advantages are high simulation accuracy and can cope with extremely complex circuit structure DC analysis, AC analysis, and transient analysis; but its disadvantages are long simulation time, complex algorithm, and a large number of control statements are required to handle unexpected situations in the program.

[0111] This patent is mainly used to realize transistor device-level simulation, and introduces the endpoint parasitic resistance voltage iteration method to achieve the same accuracy and faster speed. Figure 4 As shown, first calculate the source-drain current under the external applied voltage (as the initial voltage); then calculate the tiny voltage caused by the connection and silicon oxide contact resistance (Rsc) and the source / drain diffusion resistance (Rs,diff) through Ohm's law; then use the initial voltage minus the parasitic voltage as the new initial voltage to re-substitute into the calculation of the new current; then use the obtained current to calculate the new parasitic voltage and also obtain the new applied voltage as the initial voltage. When the deviation of the new initial voltage calculated through continuous iterations and the last applied initial voltage is less than a certain error range, generally set to 1uV (1E-6V), the cycle ends, and the source-drain current at this time is recorded as the precise current value after eliminating parasitic resistance such as connection.

[0112] This method has a simple algorithm, the calculation accuracy is almost the same as that of commercial software, and the simulation speed is extremely fast. It is a core optimization algorithm in this SPICE model simulation system.

[0113] 2) Threshold voltage search iteration;

[0114] The threshold voltage refers to setting a low voltage (usually 0.1V or 0.05V) at the drain end of the transistor device, and then applying a scanning voltage from low to high at the gate end, while measuring the source and drain currents. When the source and drain currents reach a set target value, the gate voltage at this time is recorded as the threshold voltage of the transistor (also called the turn-on voltage).

[0115] In current commercial EDA simulation software, to simulate the precise threshold voltage of transistor devices, it is often necessary to scan the voltage from low to high according to a certain voltage step. The smaller the scanning voltage step, the more accurate the threshold voltage is, but the longer it takes. This patent adopts an iterative method based on binary scanning voltage, that is, a set of minimum voltage, maximum voltage and average voltage (generally the working voltage of the device) is set at the beginning, and the simulation current under the average voltage is compared with the target value. If the current is less than the target value, the average voltage is taken as the new minimum voltage; if the current is greater than the target current, the average voltage is taken as the new maximum voltage. Through continuous cyclic iteration calculation, when the error between the simulation current and the target current under the average voltage is less than a set value (for example, 1mV) or the number of cycles is greater than a certain number (for example, 32 times), the iteration is ended, and the voltage at this time is the threshold voltage. This approach is equivalent to folding a piece of paper in half continuously, and we can get the result we want after a limited number of iterations (generally not more than 10). Compared with dozens or even hundreds of iterations of commercial EDA simulation software, the speed advantage is obvious, and it is also an indispensable core optimization algorithm in this patent.

[0116] 3) Statistical mismatch model, generate mismatch model simulation random numbers.

[0117] The present invention adopts the normal distribution to generate random numbers when used for statistical mismatch model simulation. The specific method is to first use the random number function Rnd() to generate random number variables x1 and x2 between 0 and 1, and then use the formula A set of random numbers that conform to the normal distribution can be generated. Since mismatch simulation is generally used to characterize the performance difference between two identical devices due to local variations in the manufacturing process, random numbers are generated in advance according to this method and stored in the memory before simulation. During simulation, these random numbers are directly called from the memory one by one according to the number of simulations for the two devices to achieve fast mismatch model simulation.

[0118] Third embodiment;

[0119] like Figure 3 As shown, the present invention provides a SPICE model simulation method, comprising the following steps:

[0120] S1, import model file;

[0121] S2, determine the type of the imported model file, and parse the model file type through the model card simulator lexical analyzer to obtain model parameters;

[0122] Through the lexical analysis of the model card simulator, if you select the process corner model for the model library file, including multiple process corner models such as standard, optimal, and worst, and select the device name, such as NMOS or PMOS

[0123] If it is a single model file, directly select the device name through the lexical analysis of the model card simulator;

[0124] The model card simulator lexical analyzer includes a simple lexical analyzer and an efficient lexical analyzer;

[0125] Simple lexical analyzer, which is used for Compact model, uses dictionary method to build binary structure for each parameter and read it into model card parameter

[0126] An efficient lexical analyzer is used for complex Sub-Circuit models. It recognizes specific characters in the model file to make function judgments and can automatically parse and obtain operation algorithms and judgment operations.

[0127] S3, import device measured data;

[0128] S4, input device physical characteristics and device simulation conditions;

[0129] Input device physical characteristics: MOS gate length, gate width, miniaturization ratio, layout size, etc.; device physical characteristics input methods include direct interface input, text file batch loading, and data file loading.

[0130] Input simulation conditions: temperature, voltage (gate, source, drain, substrate, etc.), threshold current, DC conditions are used to simulate current, AC conditions are used to simulate capacitance, DC plus frequency is used to simulate noise, etc.

[0131] S5, outputting simulation results after performing simulation optimization on the model file, model parameters, device measured data, device physical characteristics and device simulation conditions, wherein the simulation optimization at least includes:

[0132] 1) Eliminating the internal voltage iteration of the device caused by the parasitic resistance of the connection, including the following steps;

[0133] 1.1) Calculate the source-drain current under external applied voltage;

[0134] 1.2) Calculate the voltage due to the wire and silicon oxide contact resistance (Rsc) and source / drain diffusion resistance (Rs,diff) using Ohm's law;

[0135] 1.3) Subtract the parasitic voltage from the initial voltage and substitute it as the new initial voltage to calculate the new current;

[0136] 1.4) Repeating the above steps 1.1) to 1.3) until the deviation between the initial voltage and the last applied initial voltage is less than a first threshold;

[0137] 2) Threshold voltage search iteration, perform simulation optimization 2) The following steps are adopted;

[0138] 2.1) Set the minimum voltage, maximum voltage and average voltage, and calculate the simulated current value under the average voltage;

[0139] 2.2) The simulation current value is obtained based on the binary scanning voltage cycle iteration calculation, and the simulation current is compared with the target current value until the error between the simulation current and the target current is less than the second threshold value or the number of iterations of the simulation current is greater than the third threshold value, then the iteration is terminated and the average voltage at this time is used as the threshold voltage.

[0140] 3) Count the mismatch model, generate random numbers for mismatch model simulation, and perform simulation optimization 3) The following steps are adopted;

[0141] 3.1) Use the random number function Rnd() to generate random number variables x1 and x2 between 0 and 1;

[0142] 3.2) Use formula (1) to generate a set of random numbers that conform to the normal distribution;

[0143]

[0144] 3.3) The two devices call the random number at intervals according to the number of simulations to implement the mismatch model simulation.

[0145] The output simulation result method is determined based on the device physical characteristics input method;

[0146] If the physical property input method is GUI interface input, the simulation result output method is GUI interface output;

[0147] If the physical characteristics input method is text file batch loading, the simulation result output method is text output;

[0148] If the physical characteristics input method is to load the data file, the output simulation result method is to display the data and simulation image;

[0149] If the corresponding simulation results do not meet the design requirements, the model parameters can be fine-tuned and the simulation optimization can be re-executed.

[0150] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as those commonly understood by those skilled in the art to which the present invention belongs. It will also be understood that, unless expressly defined herein, terms such as those defined in general dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant art, and not interpreted in an ideal or overly formal sense.

[0151] The present invention has been described in detail above through specific implementation modes and embodiments, but these do not constitute limitations of the present invention. Without departing from the principles of the present invention, those skilled in the art may also make many variations and improvements, which should also be regarded as the protection scope of the present invention.

Claims

1. A SPICE model simulation system, characterized in that: include: Model import module, which is used to import model files; A model parsing module, which is used to determine the type of the imported model file and parse the model file type through the model card simulator lexical analyzer to obtain model parameters; the model card simulator lexical analyzer includes a simple lexical analyzer and an efficient lexical analyzer; Simple lexical analyzer, which is used for the Compact model, uses the dictionary method to establish a binary structure {parameter name, parameter value} for each parameter and reads it into the model card parameter; An efficient lexical analyzer, which is used for complex sub-circuit models. It can automatically parse and obtain operation algorithms and judgment operations by identifying specific characters in the model file for function judgment; A measured data import module, which is used to import the measured data of the device; A simulation condition input module, which is used to input device physical characteristics and device simulation conditions; The simulation optimization module is used to output simulation results after performing simulation optimization on the model file, model parameters, device measured data, device physical characteristics and device simulation conditions, including: 1) Eliminate the internal voltage iteration of the device caused by parasitic resistance of the connection, including: 1.1) Calculate the source-drain current under external applied voltage; 1.2) Calculate the voltage due to the wiring and silicon oxide contact resistance and source / drain diffusion resistance using Ohm's law; 1.3) Subtract the parasitic voltage from the initial voltage and substitute it as the new initial voltage to calculate the new current; 1.4) Repeating the above steps 1.1) to 1.3) until the deviation between the initial voltage and the last applied initial voltage is less than a first threshold; 2) Threshold voltage search iteration, including: 2.1) setting the minimum voltage, the maximum voltage and the average voltage, and calculating the simulation current value under the average voltage; 2.2) The simulated current value is obtained by iterative calculation based on the binary scanning voltage cycle, and the simulated current value under the average voltage is compared with the target current value. If the simulated current value is less than the target current value, the average voltage is taken as the new minimum voltage; if the simulated current value is greater than the target current value, the average voltage is taken as the new maximum voltage, until the error between the simulated current value and the target current value is less than the second threshold value, or the number of iterative calculations of the simulated current value is greater than the third threshold value, then the iteration is terminated, and the average voltage at this time is taken as the threshold voltage; 3) Statistical mismatch model, generating mismatch model simulation random numbers, including: 3.1) Use the random number function Rnd() to generate random number variables x1 and x2 between 0 and 1; 3.2) Use formula (1) to generate a set of random numbers that conform to the normal distribution; 3.3) According to the number of simulations, two identical devices call the random numbers from the memory one by one to realize the mismatch model simulation; Mismatch simulation is used to characterize the performance differences between two identical devices due to local variations in the manufacturing process.

2. The SPICE model simulation system as claimed in claim 1, characterized in that: The imported model files include model library files and / or single model parameter files.

3. The SPICE model simulation system according to claim 1, characterized in that: The device measured data at least includes MOSFET voltage-current scanning data.

4. A SPICE model simulation method, characterized in that: The following steps are involved: S1, import model file; S2, determine the type of the imported model file, and parse the model file type through the model card simulator lexical analyzer to obtain model parameters; the model card simulator lexical analyzer includes a simple lexical analyzer and an efficient lexical analyzer; Simple lexical analyzer, which is used for the Compact model, uses the dictionary method to establish a binary structure {parameter name, parameter value} for each parameter and reads it into the model card parameter; An efficient lexical analyzer, which is used for complex sub-circuit models. It can automatically parse and obtain operation algorithms and judgment operations by identifying specific characters in the model file for function judgment; S3, import device measured data; S4, input device physical characteristics and device simulation conditions; S5, outputting simulation results after performing simulation optimization on the model file, model parameters, device measured data, device physical characteristics and device simulation conditions, including: 1) Eliminate the internal voltage iteration of the device caused by parasitic resistance of the connection, including: 1.1) Calculate the source-drain current under external applied voltage; 1.2) Calculate the voltage due to the wiring and silicon oxide contact resistance and source / drain diffusion resistance using Ohm's law; 1.3) Subtract the parasitic voltage from the initial voltage and substitute it as the new initial voltage to calculate the new current; 1.4) Repeating the above steps 1.1) to 1.3) until the deviation between the initial voltage and the last applied initial voltage is less than a first threshold; 2) Threshold voltage search iteration, including: 2.1) Set the minimum voltage, maximum voltage and average voltage, and calculate the simulated current value under the average voltage; 2.2) The simulated current value is obtained by iterative calculation based on the binary scanning voltage cycle, and the simulated current value under the average voltage is compared with the target current value. If the simulated current value is less than the target current value, the average voltage is taken as the new minimum voltage; if the simulated current value is greater than the target current value, the average voltage is taken as the new maximum voltage, until the error between the simulated current value and the target current value is less than the second threshold value, or the number of iterative calculations of the simulated current value is greater than the third threshold value, then the iteration is terminated, and the average voltage at this time is taken as the threshold voltage; 3) Statistical mismatch model, generating mismatch model simulation random numbers, including: 3.1) Use the random number function Rnd() to generate random number variables x1 and x2 between 0 and 1; 3.2) Use formula (1) to generate a set of random numbers that conform to the normal distribution; 3.3) According to the number of simulations, two identical devices call the random numbers from the memory one by one to realize the mismatch model simulation; Mismatch simulation is used to characterize the performance differences between two identical devices due to local variations in the manufacturing process.

5. The SPICE model simulation method as claimed in claim 4, characterized in that: The imported model files include model library files and / or single model parameter files.

6. The SPICE model simulation method according to claim 4, characterized in that: The device measured data at least includes MOSFET voltage-current scanning data.

Images