A fast estimation method of long link S parameters based on Python data processing

Abstract

The present invention discloses a method for quickly estimating the S parameters of a long link based on Python data processing, comprising: step 1, extracting a reference long link full-band S parameter model S ref ; Step 2, extract the low-frequency S parameter model S of the long link to be measured original ; Step 3, based on the properties of different types of S parameters, refer to the full-band S parameter model S of a typical long link ref , for the low frequency S parameter source file S original Correction is performed; step 4, the corrected S parameter model is used for link circuit simulation to obtain time domain results, thereby completing the long link signal integrity assessment. The present invention extracts the low-frequency S parameters of the link to be tested and corrects them based on the properties of the S parameters to obtain a full-band S parameter model that can be used for long link signal integrity estimation. The method of directly estimating the full-band S parameter file from the low-frequency S parameter model greatly improves the efficiency of signal integrity simulation analysis and reduces the time cost of high-speed digital circuit performance evaluation.

Description

Technical Field

[0001] The present invention relates to a method for quickly estimating S parameters of a long link based on Python data processing, and belongs to the field of high-speed digital circuit signal integrity simulation analysis and Python data processing. Background Art

[0002] As the data transmission rate of high-speed digital circuit interfaces continues to increase, the design of electronic systems is no longer a simple connection with wires. Signal integrity issues such as signal reflection and crosstalk, loss in transmission lines, and synchronous switching noise have become major problems that circuit design engineers cannot avoid.

[0003] The main object of SI analysis is the signal transmission line. The transmission line on the PCB is different from the wires on the traditional circuit. The transmission line consists of two branches, namely the signal path and the reference path (return path), which also leads to various signal integrity issues.

[0004] Transmission lines on a PCB are constructed from conductors on the signal layer and the power or ground layer, with an insulating medium between them. Ideally, the conductor's resistance would be zero, while the insulating medium's resistance would be infinite. However, such a conductor and insulating medium do not exist in reality. A conductor's resistance, no matter how small, exists; as long as it has resistance, it will distribute a certain voltage. Similarly, the insulating medium's resistance, no matter how large, is finite; as long as its resistance is finite, it will distribute a certain current. Therefore, transmission line losses are unavoidable. In the era of low-speed circuits, due to the high voltages across the chips within the circuit, even if the voltage was distributed by the conductor, it would not significantly affect the logic waveform transmitted by the signal line. However, as the operating frequencies of digital integrated circuits increase, signal rise times decrease, and chip voltages continue to decrease, these losses have become more significant.

[0005] An ideal transmission line is a uniform transmission line. Although impedance exists—the sum of the obstacles to the signal caused by resistance, capacitance, and inductance during transmission—it remains constant, thus unaffecting signal transmission along the line. However, impedance variations inevitably occur at the junctions between circuit components, at vias, or where other factors necessitate changes in transmission line width. These impedance variations can lead to reflections of signal currents—a crucial factor affecting link signal integrity.

[0006] For high-speed digital circuits, the ideal transmission line consists of a signal link and one or two corresponding reference planes. However, in reality, any two conductors separated by an insulating medium will constitute a signal transmission path. Part of the signal energy is lost through these non-ideal paths, forming crosstalk between links.

[0007] The signal integrity issues of these high-speed digital circuits can all be measured using the link's S-parameters. During signal integrity analysis, electromagnetic field simulation software is typically used to extract the link's S-parameters. These S-parameters serve as an important basis for link signal integrity assessment and as an equivalent link model for subsequent circuit simulation. However, for long links, especially those operating at high frequencies, extracting S-parameters is time-consuming. Furthermore, as a key model for subsequent simulations, they cannot be performed simultaneously with circuit simulation, significantly reducing the efficiency of signal integrity simulation analysis. Summary of the Invention

[0008] The purpose of the present invention is to propose a fast S parameter estimation method based on Python data processing to address the technical defect that the extraction of S parameters of simple structure long links in high-speed digital circuits takes a long time.

[0009] In order to achieve the above objectives, the following technical solutions are adopted:

[0010] The method for quickly estimating the S parameters of a long link based on Python data processing relies on systems including Ansys electromagnetic field simulation tools, general Python compilation tools, and Aurora automated simulation cloud platform;

[0011] The method for quickly estimating passive link S parameters based on Python data processing includes the following steps:

[0012] Step 1: Extract a full-band S parameter model of a typical long link as a reference model for rapid estimation of the S parameters of the link to be measured, denoted as S ref The full frequency band refers to the frequency band of 3-5 times the circuit operating frequency;

[0013] The characteristics of the long link include the same number of ports, link materials, and interface types as the link under test; the extracted S-parameter model has a large valid range that meets the S-parameter constraints of the link under test; and the simulation tool used for parameter extraction is Ansys electromagnetic field simulation software. The link materials include, but are not limited to, FR4, resin, fiberglass cloth, aluminum substrates, and other dielectric materials suitable for general simulation methods. The link interface types include, but are not limited to, DDR, PCIE, HDMI, MIPI, USB, DP, and other common interface types suitable for the simulation method used in this article.

[0014] Step 2: Extract the low-frequency S-parameter model file of the long link to be measured as the source file for rapid estimation, denoted as S original ; The low frequency refers to the frequency band below 1 GHz;

[0015] The simulation tool used for S-parameter extraction is Ansys electromagnetic field simulation software;

[0016] Step 3: Based on the properties of different types of S parameters, refer to the full-band S parameter model S of a typical long link ref , for the low frequency S parameter source file S original After correction, a full-band S-parameter model S is obtained, which can be used to evaluate the signal integrity of the long link to be tested. result ;

[0017] The software compilation environment used for model modification is Pvthon. The modified S parameters include but are not limited to S11, S12, and S13 (depending on the number of link ports), that is, all S parameters that characterize signal link reflection, loss, and crosstalk. For the convenience of description, the S parameters that characterize reflection are recorded as S according to the properties of each type of S parameter. nn , the S parameter characterizing the loss is recorded as S nn+1 , the S parameter characterizing the crosstalk is recorded as S mn ;

[0018] Since the S parameters extracted by the simulation software have convergence, that is, only the S parameter results in the middle frequency band are consistent with the actual situation, and engineering only has constraints on the S parameter values in the frequency band 3-5 times the circuit operating frequency, this method also only optimizes the effective value of the S parameter;

[0019] Step 3.1: S that represents the reflection nn Parameter correction

[0020] Among them, S nn The basis for parameter correction is that according to the telegraph equation, the relationship between the incident wave, reflected wave, voltage and current at the port of the multi-port network is as follows:

[0021]

[0022]

[0023] Among them, a n represents the incident wave at port n, b n represents the reflected wave at the n port, v n is the voltage at the n-port, i n is the current at the n-port, Z n is the impedance at the n-port, and * is the mathematical symbol for taking the conjugate. According to the definition of S parameters and reflection coefficient, we can get:

[0024]

[0025] Where Γ represents the reflection coefficient. For the S parameter matrix, the simulated voltage and current are both sinusoidal waves, so the voltage and current are periodic and have equal periods. Snn is also a periodic value. Therefore, we can use the low-frequency S parameter matrix S original S in nn The value is corrected and the periodic extension is performed to obtain the full-band S parameter model S result S nn ;

[0026] Step 3.2: S that represents the loss nn+1 Parameters are modified;

[0027] Among them, S nn+1 The basis for parameter correction is that for simple long links, factors such as line height, line width, dielectric constant and thickness that determine the properties of the transmission line remain almost unchanged, so the loss is proportional to the length, which can be obtained from the S of a typical long link. nn+1 The S of the long link to be measured is obtained in proportion to the parameter and the link length. nn+1 Parameter value;

[0028] In actual engineering, S parameters are generally converted into dB form, and the conversion formula is as follows:

[0029] dB(S ij )=20log 10 [Mag(S ij )]

[0030] Similarly, the length relationship is also converted in dB form:

[0031] dB(lenth_ratio)=20log 10 [Mag(lenth_ratio)]

[0032] Wherein, dB represents the value in dB form, and Mag represents the value in multiple form;

[0033] Step 3.3: S that characterizes crosstalk mn Parameters are modified;

[0034] Among them, S mn The basis for parameter correction is that for high-speed interfaces, adjacent data links remain parallel, so the non-ideal signal transmission channels they form are also very close, and the S parameter model of a typical long link can be used directly. mn value instead.

[0035] Step 4: Use the modified S-parameter model for link circuit simulation to obtain time domain results, thereby completing the long link signal integrity assessment;

[0036] The simulation tool used for link circuit simulation is the Aurora automated simulation cloud platform;

[0037] At this point, from step 1 to step 4, a fast estimation method based on the S-parameter properties of a long link is completed.

[0038] Beneficial effects of the present invention:

[0039] The present invention extracts the low-frequency S parameters of the link to be tested and corrects them based on the properties of the S parameters to obtain a full-band S-parameter model that can be used for long-link signal integrity estimation. The method of directly estimating the full-band S-parameter file from the low-frequency S-parameter model greatly improves the efficiency of signal integrity simulation analysis and reduces the time cost of high-speed digital circuit performance evaluation. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 is a flow chart of the rapid estimation method provided in the embodiments of the present application;

[0041] Figure 2 is a reference model S11 graph provided in the embodiments of the present application;

[0042] Figure 3 It is a simulation link diagram provided in the embodiment of the present application;

[0043] Figure 4 This is a graph of a single cycle S11 of the link to be tested provided in an embodiment of the present application;

[0044] Figure 5 This is a time domain eye diagram of the transmission line on the left side of the link to be tested provided in an embodiment of the present application;

[0045] Figure 6 This is the time domain eye diagram of the transmission line on the right side of the link to be tested provided in the embodiment of the present application. DETAILED DESCRIPTION

[0046] The following further illustrates and describes in detail the method for rapid estimation of passive link S parameters based on Python data processing according to the present invention in conjunction with the accompanying drawings and embodiments.

[0047] This embodiment describes in detail the estimation results of a method for rapid estimation of passive link S parameters based on Python data processing according to the present invention when it is specifically implemented, aiming to verify the estimation effect of a method for rapid estimation of passive link S parameters based on Python data processing.

[0048] The long link to be measured used in this embodiment is a four-port network. The S parameter matrix extracted using electromagnetic field simulation software is a 4×4 matrix, including 32 data points in amplitude and phase. Corrections are performed separately according to the type of S parameters. The operating frequency of the link is 1600 MHz, that is, the S parameter constraint range (3-5 times the operating frequency) is 4800 MHz-8000 MHz.

[0049] Figure 1 This is a flowchart of a method for fast S-parameter estimation of long links based on Python data processing. The specific implementation steps are as follows:

[0050] Step 1: Extract a full-band S-parameter model of a typical long link as a reference model for rapid S-parameter estimation of the link under test. The characteristics of the long link include the same number of ports, link material, and interface type as the link under test. The extracted S-parameter model has a large valid range that can meet the S-parameter constraint range of the link under test.

[0051] In this embodiment, the number of ports is 4, the link material is copper and FR4, the interface type is PCIE, and the S-parameter extraction frequency domain settings are shown in the following table:

[0052]

[0053] Figure 2 The S11 curve of the reference model shows that the effective range of the model is approximately 3500MHz-9400MHz, which meets the frequency constraint range of the link to be tested.

[0054] Step 2: Extract the low-frequency S-parameter model file of the long link to be measured as the source file for the fast estimation method;

[0055] In this embodiment, the number of ports is 4, the link material is copper and FR4, and the interface type is PCIE, which is consistent with the reference model. Figure 3 is a simulation link diagram of this embodiment, Figure 4 is a graph of S11 for a single cycle of this embodiment;

[0056] Step 3: Based on the properties of different types of S parameters and referring to the full-band S-parameter model of a typical long link, the low-frequency S-parameter source file is modified to obtain a full-band S-parameter model that can be used for signal integrity assessment of the long link to be tested.

[0057] In this embodiment, the link under test is a four-port network, that is, there are 16 S parameters of three types, and its S parameter matrix is as follows:

[0058]

[0059] Step 3.1, modifying the S parameters characterizing the reflection, which in this embodiment include S11, S22, S33, and S44;

[0060] Taking S11 as an example, the effective frequency band of 2600-3500MHz is taken as the corrected reference period, and it is extended backward by 6 periods to meet the constraint range of 3-5 times the operating frequency. That is, the corrected S parameter effective value of 2600-8900MHz is obtained.

[0061] Step 3.2, modifying the S parameters characterizing the loss, which in this embodiment include S12, S21, S34, and S43;

[0062] The reference link length in this embodiment is 84.7 mm, and the link length to be measured is 79.8 mm. Therefore, the amount to be corrected is as follows:

[0063]

[0064] Wherein, dB(fix) represents the amount to be corrected, lenth_ref represents the reference link length, and lenth-test represents the test link length. From this formula, we can see that the amount to be corrected for the loss-related S parameters is -0.51 dB.

[0065] Step 3.3, correcting the S parameters characterizing the crosstalk, which in this embodiment include S13, S31, S23, S32, S14, S41, S24, and S42;

[0066] Step 4: Use the modified S-parameter model for link circuit simulation to obtain time domain results. The following table shows the comparison of the main parameter results of the eye diagram. The general engineering requirement for signal integrity testing is that the error is less than 10%. It can be seen that the error of this embodiment meets this error requirement. Figure 5 、 Figure 6 They respectively represent the eye diagram results of the two transmission lines of the link to be tested.

[0067]

[0068] The above description is only a preferred embodiment of the present invention, and the present invention should not be limited to the contents disclosed in the embodiment and the accompanying drawings. Any equivalent or modification completed without departing from the spirit disclosed in the present invention shall fall within the scope of protection of the present invention.

Claims

1. A method for quickly estimating long link S parameters based on Python data processing, comprising a reference long link full-band model, a low-frequency model of the long link to be measured, and a model correction system; characterized by: The method comprises the following steps: Step 1: Extract the reference long link full-band S-parameter model ; Step 2: Extract the low-frequency S-parameter model of the long link to be tested ; Step 3: Based on the properties of different types of S parameters, refer to the full-band S parameter model of a typical long link , for low frequency S parameter source files Make corrections; This includes all S parameters that characterize signal link reflection, loss, and crosstalk. The specific number of S parameters that need to be corrected depends on the number of long link ports. Among them, the S parameter that characterizes the signal link reflection is recorded as , the S parameter that characterizes the signal link loss is recorded as , the S parameter characterizing the signal link crosstalk is recorded as ; Step 3.1: Characterize the S parameters of reflection based on periodicity Make corrections; Step 3.2: S parameters characterizing loss based on proportional relationships Make corrections; Step 3.3: S-parameters for characterizing crosstalk based on structural relationships Make corrections; Step 4: Use the modified S-parameter model for link circuit simulation to obtain time domain results, thereby completing the long link signal integrity assessment.

2. The method for fast estimating long link S parameters based on Python data processing according to claim 1, characterized in that: In step 1, the conditions of the reference link include the following: the number of ports, link material, and interface type are the same as those of the link to be tested. The extracted S-parameter reference model has a large effective range that can meet the S-parameter constraint range of the link to be tested, that is, a frequency band of 3-5 times the operating frequency.

3. The method for fast estimating long link S parameters based on Python data processing according to claim 2, characterized in that: The link is made of FR4, resin, glass fiber cloth or aluminum substrate.

4. The method for fast estimating long link S parameters based on Python data processing according to claim 2, characterized in that: The link interface type is DDR, PCIE, HDMI, MIPI, USB or DP.

5. The method for fast estimating long link S parameters based on Python data processing according to claim 1, characterized in that: In step 2, the conditions that the low-frequency S-parameter model of the link to be tested meets include at least one complete cycle within the valid interval, so that the parameters of the extended S-parameter model are complete.

6. The method for fast estimating long link S parameters based on Python data processing according to claim 1, characterized in that: In step 3, the types of S parameters corrected using Python data processing include three types: S parameters characterizing reflection, S parameters characterizing loss, and S parameters characterizing crosstalk.

7. The method for quickly estimating long link S parameters based on Python data processing according to claim 1, characterized in that: Use the numpy data processing library function in Python to modify the parameters.

8. The method for fast estimating long link S parameters based on Python data processing according to claim 1, characterized in that: In step 3.1, when modifying the S parameters that characterize the reflection, the basis is the periodicity of this type of S parameters in frequency. in The value is corrected, that is, the period is extended to obtain the full-band S parameter model of .

9. The method for fast estimating long link S parameters based on Python data processing according to claim 1, characterized in that: In step 3.2, when correcting the S parameters that characterize the loss, the basis is the proportional relationship between this type of S parameter and the link length, that is, the correction is made based on the fact that the loss is proportional to the length.

10. The method for fast estimating long link S parameters based on Python data processing according to claim 1, characterized in that: In step 3.3, when correcting the S parameters that characterize crosstalk, the basis is the correlation between this type of S parameters and the link structure, and the typical long link S parameter model is directly used. in value instead.

Images