Computer system and method for optimizing temperature adjustment mechanism of ring oscillator
By using computer systems and methods, and employing techniques such as similarity search, clustering algorithms, or regression models, the optimal temperature coefficient set for ring oscillators can be quickly determined. This solves the problem of frequency instability with temperature changes and improves the frequency stability of ring oscillators and the overall stability of integrated circuits.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- NUVOTON
- Filing Date
- 2025-05-26
- Publication Date
- 2026-06-05
AI Technical Summary
Existing technologies make it difficult to efficiently find the optimal set of temperature coefficients to control the frequency variation with temperature within an acceptable range when mass-producing ring oscillators, resulting in unstable frequency signals.
Using a computer system and method, the RC characteristic dataset is stored in a storage unit. Based on the RC characteristic dataset and the target characteristics, the processing unit uses similarity search, clustering algorithm, classification algorithm or regression model to quickly determine the optimal temperature coefficient group of the target RC.
The optimal temperature coefficient set for the target RC circuit is found efficiently, which improves the frequency stability of the ring oscillator and the overall stability of the integrated circuit, while reducing the number of measurements and resource consumption.
Smart Images
Figure CN122159836A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to a temperature adjustment mechanism for a ring oscillator, and more particularly to a method for optimizing the temperature adjustment mechanism of a ring oscillator. Background Technology
[0002] A ring oscillator (RC) is a closed-loop circuit consisting of an odd number of inverters. Signals can be passed back and forth between these inverters and continuously reversed, thus creating oscillations. Therefore, ring oscillators (RC) are frequently used as the clock signal source for integrated circuits (ICs) or microcontroller units (MCUs).
[0003] Because the frequency of a ring oscillator (RC) varies with temperature (also known as "temperature drift"), when using a ring oscillator as a clock signal source, the frequency variation of the RC with temperature should be controlled within an acceptable range (e.g., +2%) to provide a clock signal of a specific and stable frequency for the IC or MCU.
[0004] Generally, the frequency variation of an RC circuit with temperature can be controlled using a temperature adjustment mechanism. This mechanism can be a combination of two temperature coefficients: a positive temperature trim (P trim) and a negative temperature trim (N trim). By adjusting different values of the positive and negative temperature coefficients (P and N values), the frequency variation of the RC with different temperatures can be determined. Then, from these many combinations of P and N values, the (P, N) combination with the smallest temperature-frequency variation can be identified as the optimal temperature coefficient set. This ensures that the RC used as the clock signal source uses the optimal temperature coefficient set. In other words, an optimal temperature coefficient set should be found for each RC circuit to provide a clock signal of a specific frequency.
[0005] Current methods for finding the optimal temperature coefficient set include: adjusting different positive temperature coefficient values (P value) and negative temperature coefficient values (N value); heating the entire IC in a temperature chamber at different (P, N) values, measuring and recording the change in the IC's RC frequency with temperature; and selecting the (P, N) value with the smallest change from the numerous records as the optimal temperature coefficient set (P). best N bestHowever, during mass production of ICs, it is impractical to measure the frequency change with temperature for each IC under different (P, N) RC conditions.
[0006] Therefore, a computer system and method are needed to optimize the temperature adjustment mechanism of a ring oscillator, which can solve the above problems. Summary of the Invention
[0007] This disclosed embodiment provides a computer system for optimizing the temperature adjustment mechanism of a ring oscillator (RC), comprising a storage unit and a processing unit. The storage unit stores an RC characteristic dataset, including one or more characteristics of each of a plurality of RCs and an optimal temperature coefficient set for each RC. The characteristics of each RC include the temperature-frequency relationship of that RC. The processing unit reads a program from the storage unit to receive one or more target characteristics of a target RC, and, based on the RC characteristic dataset and the target characteristics, determines a target temperature coefficient set for the target RC. The target characteristics include the temperature-frequency relationship of the target RC.
[0008] This disclosure provides a method for optimizing the temperature adjustment mechanism of a ring oscillator (RC), implemented by a computer system. The method includes receiving one or more target characteristics of a target RC, and determining a target temperature coefficient set for the target RC based on an RC characteristic dataset and the target characteristics. The target characteristics include the temperature-frequency relationship of the target RC. The RC characteristic dataset includes one or more characteristics of each of a plurality of RCs and an optimal temperature coefficient set for each RC. The characteristics of each RC include the temperature-frequency relationship of that RC.
[0009] The computer system and method disclosed herein for optimizing the temperature adjustment mechanism of a ring oscillator (RC) can find the optimal temperature coefficient set for a target RC based on only a limited amount of data in the RC characteristic dataset. Furthermore, compared to the traditional approach of measuring all possible temperature coefficient sets for a target RC to find the optimal set, the embodiments disclosed herein, through data comparison or machine learning, can more efficiently find the optimal temperature coefficient set for the target RC, enabling the target RC to have optimal temperature drift performance at the output frequency, thereby improving the stability of IC operation. Attached Figure Description
[0010] This disclosure will be better understood from the following description of exemplary embodiments in conjunction with the accompanying diagrams. Furthermore, it should be understood that the execution order of the blocks in the flowcharts of this disclosure may be changed, and / or certain blocks may be altered, deleted, or merged.
[0011] Figure 1The present invention provides a system architecture diagram of a computer system according to an embodiment of the present invention, wherein the processing unit of the computer system executes a method for optimizing a temperature adjustment mechanism for a ring oscillator.
[0012] Figure 2 A data flow diagram illustrating the steps for determining the target temperature coefficient group of a target RC according to an embodiment of the present invention.
[0013] Figure 3 A data flow diagram illustrating the steps for determining the target temperature coefficient group of a target RC according to an embodiment of the present invention.
[0014] Figure 4 A data flow diagram illustrating the steps for determining the target temperature coefficient group of a target RC according to an embodiment of the present invention.
[0015] Symbol Explanation
[0016] 10: Computer System
[0017] S10: Method
[0018] 11: Storage unit
[0019] 12: Processing Unit
[0020] 13: RC Feature Dataset
[0021] 14: Target Characteristics
[0022] S101, S102: Step 15: Target Temperature Coefficient Group
[0023] 21: Similarity Search Algorithm
[0024] 22: Most similar to RC
[0025] 31: Clustering Algorithm
[0026] 32, 34: RC category
[0027] 33: Classification Algorithm
[0028] 41: Regression Model Detailed Implementation
[0029] The following description illustrates various embodiments of the present invention, but is not intended to limit the scope of the invention. The actual scope of the invention is defined by the claims.
[0030] In the embodiments listed below, the same or similar elements or components will be represented by the same reference numerals.
[0031] The serial numbers in this specification and the claims, such as "first," "second," etc., are for ease of explanation only and there is no sequential relationship between them.
[0032] The descriptions of embodiments of the apparatus or system in this specification also apply to embodiments of the method, and vice versa.
[0033] Figure 1 This is a system architecture diagram of a computer system 10 for optimizing the temperature adjustment mechanism of a ring oscillator according to an embodiment of the present invention. Figure 1 As shown, the computer system 10 includes a storage unit 11 and a processing unit 12.
[0034] Computer system 10 can be any computer system or processing device with computing power, such as a personal computer (e.g., a desktop computer or a notebook computer), a server computer, or a mobile device (e.g., a tablet computer or a smartphone), but this disclosure does not limit it.
[0035] Storage unit 11 may include any device containing non-volatile memory (such as read-only memory, electrically erasable programmable read-only memory (EEPROM), flash memory, or non-volatile random access memory (NVRAM)) such as hard disk (HDD), solid-state drive (SSD), or optical disk, but this disclosure is not limited thereto.
[0036] Processing unit 12 may include any one or more general-purpose or special-purpose processors and combinations thereof for executing instructions, such as a central processing unit (CPU) and / or a graphics processing unit (GPU). Processing unit 12 may further include volatile memory such as dynamic random access memory (DRAM) and / or static random access memory (SRAM), but this disclosure is not limited thereto.
[0037] like Figure 1 As shown, storage unit 11 stores RC characteristic dataset 13. RC characteristic dataset 13 may contain one or more characteristics of each RC and the optimal temperature coefficient set of each RC. The characteristics of each RC may also include its temperature-frequency relationship.
[0038] The following table, , is provided as an example of an RC characteristic dataset. As shown in , an RC characteristic dataset 13 may contain at least an RC identifier, a characteristic, and an optimal temperature coefficient group (P). best N best The data consists of three interrelated fields: temperature-frequency relationship (TFR), temperature coefficient, and frequency coefficient. Each characteristic may at least include the temperature-frequency relationship of the corresponding RC, i.e., the temperature-frequency change of the RC. Each column in Table 1 represents one data point in the RC characteristic dataset 13. In this example, the RC characteristic dataset 13 records a total of K data points (however, this disclosure does not limit the number of data points in the RC characteristic dataset 13), and each data point represents the temperature-frequency relationship and optimal temperature coefficient group (P) for each RC. best N best For example, RC1 has a temperature-frequency relationship 1 and an optimal temperature coefficient group (0, 0), RC2 has a temperature-frequency relationship 2 and an optimal temperature coefficient group (3, 0), and so on. Each RC has its optimal temperature coefficient group (P). best N best The temperature frequency relationship under these conditions exhibits the best temperature drift performance (i.e., minimum temperature drift).
[0039]
[0040]
[0041]
[0042] In one embodiment, the optimal temperature coefficient group (P) for each RC best N best Multiple temperature-frequency relationships can be measured by adjusting the temperature coefficient group (P, N), and the temperature-frequency relationship with the minimum temperature drift can be selected. The temperature-frequency relationship refers to the change of the frequency of the corresponding RC circuit with temperature, and can be represented by a linear function, a nonlinear function, a statistical chart, or a mapping table. Furthermore, the aforementioned temperature test chamber can be used to heat the IC to obtain the frequency of each RC circuit at different temperatures, thus obtaining the temperature-frequency relationship. However, this disclosure does not limit the specific data collection methods for the RC characteristic dataset.
[0043] like Figure 1 As shown, the processing unit 12 executes a method S10 for optimizing the temperature adjustment mechanism of the ring oscillator, which may include steps S101 and S102.
[0044] In step S101, the processing unit 12 receives one or more target characteristics 14 of a target RC. As previously described, the target characteristic 14 includes at least the temperature-frequency relationship of the target RC.
[0045] In step S102, the processing unit 12 determines the target temperature coefficient group 15 of the target RC based on the RC characteristic dataset 13 and the target characteristic 14. Figure 1 The middle is represented as (P) target N target The RC feature dataset 13 may be loaded from storage unit 11 into the random access memory of processing unit 12 to execute step S102.
[0046] In one embodiment of the present invention, the temperature-frequency relationship of the target RC can be obtained by adjusting the temperature of the target RC and recording the corresponding frequencies of the target RC at multiple temperatures. For example, the implementation of temperature adjustment may involve the use of a cold / hot plate, a temperature chamber, an infrared heater, a laser heater, or other similar tools, but this disclosure is not limited thereto. The temperature of the target RC can be known through a temperature sensor inside the target IC, but this disclosure is not limited thereto.
[0047] In one embodiment of the present invention, without needing to adjust the ambient temperature of the target RC, the temperature of the target RC can be regulated by adjusting the electrical parameters of the target IC, such as voltage or current. In one implementation, the voltage and / or the current generated during operation of the target IC can be adjusted according to an industry-standard formula to regulate the temperature of the target RC by means of the power consumption and package heat dissipation of the target IC. The aforementioned formula is as follows:
[0048] Temperature rise = Voltage × Current × Package temperature coefficient
[0049] It should be noted that the characteristics in target characteristic 14 and RC characteristic dataset 13 should use the same consistent representation method, such as linear functions, nonlinear functions, statistical charts or mapping tables.
[0050] In one embodiment of the present invention, the characteristics in the RC characteristic dataset 13 include not only the temperature-frequency relationship but also the temperature-power consumption relationship. Correspondingly, the target characteristic 14 includes not only the temperature-frequency relationship of the target RC but also the temperature-power consumption relationship of the target RC. Table 2 is provided below as an example of the RC characteristic dataset in this embodiment.
[0051]
[0052] RC Feature one Feature 2 <![CDATA[(P best ,N best )]]> 1 Temperature-frequency relationship 1 Temperature-power relationship 1 (0，0) 2 Temperature-frequency relationship 2 Temperature-power relationship 2 (1，0) … … … … K Temperature-frequency relationship K Temperature-power relationship K (0，3)
[0053] In this example, RC characteristic dataset 13 records a total of K data points (however, this disclosure does not limit the number of data points in RC characteristic dataset 13). Each data point represents the temperature-frequency relationship, temperature-power consumption relationship, and optimal temperature coefficient group (P) corresponding to each RC. best N best For example, RC1 has temperature-frequency relationship 1, temperature-power consumption relationship 1, and optimal temperature coefficient group (0, 0); RC2 has temperature-frequency relationship 2, temperature-power consumption relationship 2, and optimal temperature coefficient group (1, 0)... and so on. Each RC has its optimal temperature coefficient group (P... best N best The temperature-frequency relationship and temperature-power consumption relationship under these conditions exhibit the best temperature drift performance.
[0054] Figure 2 To implement according to an embodiment of the present invention Figure 1 The data flow diagram for step S102 is shown. In this embodiment, step S102 may involve selecting the most similar RC22 from the RC feature dataset 13 based on the target feature 14 using a similarity search algorithm. Then, the optimal temperature coefficient group most similar to RC22 is obtained as the target temperature coefficient group 15.
[0055] In a practical example, the similarity search algorithm 21 may involve calculating the similarity between each RC characteristic and the target characteristic 14 in the RC characteristic dataset 13, and selecting the one with the highest similarity to obtain the most similar RC and the corresponding optimal temperature coefficient set. The similarity may be evaluated using Euclidean distance, Manhattan distance, cosine similarity, or other measures, but this disclosure is not limited to this.
[0056] Figure 3 To implement according to an embodiment of the present invention Figure 1 The data flow diagram for step S102 is shown. In this embodiment, step S102 may involve using a clustering algorithm 31 to determine multiple RC categories 32 of the RC in the RC characteristic dataset 13, and the optimal temperature coefficient group corresponding to these RC categories 32. Then, based on the target characteristic 14, a classification algorithm 33 is used to determine which of the RC categories 32 the target RC belongs to, and the optimal temperature coefficient group corresponding to the RC category is obtained as the target temperature coefficient group.
[0057] In one embodiment, after using the clustering algorithm, the RC characteristic dataset 13 can further include two data fields: RC category and optimal temperature coefficient group (P). class N class As shown in Table 3, the RC characteristic dataset 13 here records K temperature-frequency relationships and optimal temperature coefficient groups (P) corresponding to RC values. best N best ), and the M types of RC corresponding to these K RC and the optimal temperature coefficient group (P) class N class ), where K>M. The optimal temperature coefficient group (P) corresponding to the RC category to which the target RC belongs. class N class If the target RC is determined to belong to RC category 1 by the classification algorithm, then the optimal temperature coefficient group (0, 0) corresponding to RC category 1 will be used as the target temperature coefficient group 15; if the target RC is determined to belong to RC category 2 by the classification algorithm, then the optimal temperature coefficient group (2, 0) corresponding to RC category 2 will be used as the target temperature coefficient group 15; and so on.
[0058]
[0059] RC Feature one <![CDATA[(P best ,N best )]]> RC category <![CDATA[(P class ,N class )]]> 1 Temperature-frequency relationship 1 (0，0) one (0，0) 2 Temperature-frequency relationship 2 (2，0) two (2，0) 3 Temperature-frequency relationship 3 (0，1) one (0，0) … … … … … K Temperature-frequency relationship K (0，2) M (0，3)
[0060] In one embodiment, the clustering algorithm 31 may be, for example, K-means, density-based spatial clustering of applications with noise (DBSCAN), or spectral clustering, but this disclosure is not limited thereto.
[0061] In one embodiment, the classification algorithm 33 may be a machine learning classification model established using the features and RC categories in Table 3, such as: the nearest centroid classifier (also known as the Rocchio classifier), the k-nearest neighbors algorithm (k-NN), the decision tree algorithm, the support vector machine algorithm (SVM), or the neural network classification model (NN), but this disclosure is not limited to this.
[0062] Figure 4 To implement according to an embodiment of the present invention Figure 1The data flow diagram for step S102 is shown. In this embodiment, step S102 may involve establishing a regression model 41 based on the RC characteristic dataset 13. Then, the target characteristic 14 is input into the regression model 41, and the target temperature coefficient group 15 output by the regression model is directly obtained.
[0063] Specifically, regression model 41 can be based on the characteristics of each RC and the corresponding optimal temperature coefficient set (P) of the RC characteristic dataset 13. best N best Machine learning regression models established by this disclosure include, but are not limited to, linear regression, decision tree regression, support vector regression (SVR), or neural network regression models.
[0064] The computer system and method disclosed herein for optimizing the temperature adjustment mechanism of a ring oscillator (RC) can find the optimal temperature coefficient set for a target RC based on only a limited amount of data in the RC characteristic dataset. Furthermore, compared to the traditional approach of measuring all possible temperature coefficient sets for a target RC to find the optimal set, the embodiments disclosed herein, through data comparison or machine learning, can more efficiently find the optimal temperature coefficient set for the target RC, enabling the target RC to have optimal temperature drift performance at the output frequency, thereby improving the stability of IC operation.
[0065] The preceding paragraphs describe various forms. Clearly, the teachings herein can be implemented in multiple ways, and any particular architecture or functionality disclosed in the examples is merely representative. Based on the teachings herein, it should be understood in the art that the individual forms disclosed herein can be implemented independently, or that two or more forms can be implemented in combination.
[0066] Although the present disclosure has been described above with reference to embodiments, it is not intended to limit the present disclosure. Those skilled in the art can make some modifications and refinements without departing from the spirit and scope of the present disclosure. Therefore, the scope of protection of the invention shall be determined by the scope defined in the appended claims.
Claims
1. A computer system for optimizing the temperature adjustment mechanism of a ring oscillator (RC), characterized in that, include: A storage unit stores an RC characteristic dataset, which includes one or more characteristics of each of a plurality of RCs and an optimal temperature coefficient set for each RC, wherein the plurality of characteristics of each RC include a temperature-frequency relationship for that RC; and A processing unit reads a program from the storage unit for execution: Receive one or more target characteristics of a target RC, wherein the multiple target characteristics include the temperature-frequency relationship of the target RC; and Based on the RC characteristic dataset and the multiple target characteristics, a target temperature coefficient group for the target RC is determined.
2. The computer system as described in claim 1, characterized in that, The temperature-frequency relationship of the target RC was obtained by adjusting the temperature of the target RC and recording the corresponding frequencies of the target RC at multiple temperatures.
3. The computer system as described in claim 2, characterized in that, The processing unit adjusts the temperature of the target RC by adjusting an electrical parameter of the target integrated circuit where the target RC is located.
4. The computer system as described in claim 1, characterized in that, The plurality of characteristics of each RC in the RC characteristic dataset further include a temperature-power dissipation relationship for that RC; and The aforementioned multiple target characteristics further include the temperature-power consumption relationship of the target RC.
5. The computer system as described in claim 1, characterized in that, Based on the multiple target characteristics, the processing unit uses a similarity search algorithm to select the most similar RC from the RC characteristic dataset and obtains the best temperature coefficient group of the most similar RC as the target temperature coefficient group.
6. The computer system as described in claim 1, characterized in that, The processing unit further uses a grouping algorithm to determine multiple RC categories of the multiple RCs in the RC characteristic dataset, and the optimal temperature coefficient group corresponding to each RC category; and The processing unit further uses a classification algorithm based on the multiple target characteristics to determine which of the multiple RC categories the target RC belongs to, and obtains the optimal temperature coefficient group corresponding to the RC category as the target temperature coefficient group.
7. The computer system as described in claim 1, characterized in that, The processing unit further uses the RC feature dataset to build a regression model; and The processing unit further inputs the target characteristic into the regression model and obtains the target temperature coefficient set output by the regression model.
8. A method for optimizing the temperature adjustment mechanism of a ring oscillator RC, characterized in that, Implemented by a computer system, the method includes: Receive one or more target characteristics of a target RC, wherein the multiple target characteristics include a temperature-frequency relationship of the target RC; and Based on an RC characteristic dataset and the multiple target characteristics, a target temperature coefficient set for the target RC is determined; The RC characteristic dataset includes one or more characteristics of each RC and an optimal temperature coefficient set for each RC, wherein the multiple characteristics of each RC include a temperature-frequency relationship of that RC.
9. The method as described in claim 8, characterized in that, The temperature-frequency relationship of the target RC was obtained by adjusting the temperature of the target RC and recording the corresponding frequencies of the target RC at multiple temperatures.
10. The method as described in claim 9, characterized in that, This temperature regulation is achieved by adjusting an electrical parameter of the target integrated circuit where the target RC is located.
11. The method as described in claim 8, characterized in that, The plurality of characteristics of each RC in the RC characteristic dataset further include a temperature-power dissipation relationship for that RC; and The aforementioned multiple target characteristics further include the temperature-power consumption relationship of the target RC.
12. The method as described in claim 8, characterized in that, Based on the RC characteristic dataset and the multiple target characteristics, the target temperature coefficient set for the target RC is determined, further including: Based on the multiple target characteristics, a similarity search algorithm is used to select the most similar RC from the RC characteristic dataset, and the best temperature coefficient group of the most similar RC is obtained as the target temperature coefficient group.
13. The method as described in claim 8, characterized in that, Including: The RC characteristic dataset is determined by using a grouping algorithm to determine the multiple RC categories of the multiple RCs, and the optimal temperature coefficient group corresponding to each RC category; The determination of the target temperature coefficient group of the target RC based on the RC characteristic dataset and the multiple target characteristics further includes: using a classification algorithm based on the multiple target characteristics to determine which of the multiple RC categories the target RC belongs to, and obtaining the optimal temperature coefficient group corresponding to the RC category as the target temperature coefficient group.
14. The method as described in claim 8, characterized in that, Based on the RC characteristic dataset and the multiple target characteristics, the target temperature coefficient set for the target RC is determined, further including: Based on this RC characteristic dataset, a regression model is established; and Input the target characteristic into the regression model and obtain the target temperature coefficient set output by the regression model.