Time domain test method suitable for signal integrity test

Abstract

The invention discloses a time domain testing method suitable for signal integrity testing, and belongs to the technical field of testing. The method comprises the following steps: providing a plurality of frequency domain windowing functions; allowing a user to input a relation value between a certain length of custom window function coefficient and pulse width or rise time; obtaining accurate step rise time by using an interpolation or extrapolation method, and expanding the upper limit and the lower limit of the step rise time by using an extraction algorithm; calculating a direct current expansion point; a normalization method is provided to allow a user to select whether to normalize the step response result. A frequency domain data truncation time window function can be selected and can be input by a user; the interpolation precision is improved by using a pre-stored value; expanding a step rise time upper limit by using an extraction algorithm; a user-defined and automatic normalization method is provided, the problem of low-frequency jitter in impedance test is avoided, and a more effective solution is provided for various high-frequency cable tests.

Description

Technical Field

[0001] The invention belongs to the field of testing technology, and in particular relates to a time domain testing method suitable for signal integrity testing. Background Art

[0002] With the increase of digital communication rates, cable testing has entered the high-frequency, high-speed, and high-efficiency testing stage. The test items cover frequency domain indicators such as loss and crosstalk, as well as time domain indicators such as impedance, differential delay, and eye diagrams. Traditionally, time domain reflectometers are used to test indicators such as impedance and delay. Modern vector network analyzers convert frequency domain data into time domain data through mathematical algorithms to obtain the above test functions. Time domain impedance testing requires the simulation of a low-pass step response function to be implemented. In engineering implementation, it is impossible to use an ideal step response function, which gives rise to the concept of step response rise time. A large number of test cables with different structures and indicators require different step rise times when testing indicators such as rise time decay and delay difference. The original vector network analyzer only provides one transformation method, the Kaiser window, and the upper and lower limits of the transformation time are fixed, which can no longer meet the cable testing needs.

[0003] In response to the delay, impedance, offset and other testing requirements of various high-speed data connection cables, combiners, high-speed differential circuits, high-frequency network transformers and other passive components, domestic vector network analyzers provide time domain testing and advanced time domain testing functions. The step response rise time and pulse impulse width are only based on linear interpolation of several points of the Kaiser window function frequency response, resulting in inaccurate rise time or impulse width calculated for different Kaiser windows and different β parameters. There is only one frequency response mode of the Kaiser window, which cannot meet the testing needs of some users when testing indicators such as rise time decay, skew, and impedance.

[0004] At present, domestic vector network analyzers cannot change the step time characteristics, the upper and lower limits of the step time are fixed in a narrow band and cannot be expanded, the impedance normalization function is not provided, and the low-pass expansion point algorithm is fixed. These problems cannot meet the personalized needs of a wide variety of cable tests and have certain limitations. Summary of the invention

[0005] In view of the above technical problems existing in the prior art, the present invention proposes a time domain testing method suitable for signal integrity testing, which has a reasonable design, overcomes the shortcomings of the prior art, and has good effects.

[0006] In order to achieve the above object, the present invention adopts the following technical solution:

[0007] A time domain testing method suitable for signal integrity testing comprises the following steps:

[0008] Step 1: Provide a variety of frequency domain windowing functions;

[0009] Step 2: Allow the user to input the relationship between the custom window function coefficient of a certain length and the pulse width or rise time; use the interpolation or extrapolation method to obtain the accurate step rise time, and use the extraction algorithm to expand the upper and lower limits of the step rise time;

[0010] Step 3: Calculate the DC expansion point;

[0011] Expand the calculation points to the high-frequency range, and then expand from the high-frequency range to the low-frequency range;

[0012] Step 4: Provide a normalization method to allow the user to choose whether to normalize the step response result.

[0013] Preferably, in step 1, the multiple frequency domain windowing functions include Gaussian, exponential, Chebyshev, Hanning, Hamming and Blackman functions.

[0014] Preferably, in step 2, a spline interpolation algorithm is used to obtain any point in the interval, and a decimation algorithm is used to obtain values ​​exceeding the upper limit of the interval.

[0015] Preferably, the extraction algorithm is as follows:

[0016] Step S1: Start;

[0017] Step S2: Determine whether the user input value is within the pre-stored value range;

[0018] If: the judgment result is that the user input value is within the pre-stored value range, execute step S3;

[0019] Or if the judgment result is that the user input value is not within the pre-stored value range, then execute step S4;

[0020] Step S3: Determine whether the user input value falls within the pre-stored value;

[0021] If: the judgment result is that the user input value falls on the pre-stored value, then the pre-stored value is taken;

[0022] Or if the result of the determination is that the user input value does not fall within the pre-stored value, then step S6 is executed;

[0023] Step S4: Determine whether the user input value exceeds the lower limit of the pre-stored value or the upper limit of the pre-stored value;

[0024] If: the judgment result is that the user input value exceeds the lower limit of the pre-stored value, then the lower limit of the pre-stored value is taken;

[0025] Or if the judgment result is that the user input value exceeds the upper limit of the pre-stored value, then execute step S5;

[0026] Step S5: Calculate the proportional coefficient r, r = user set value / pre-stored value upper limit, and use the window function order / r as the new window function order;

[0027] Step S6: taking two pre-stored values ​​greater than and less than the user input value for linear interpolation;

[0028] Step S7: End.

[0029] Preferably, step 3 specifically includes the following steps:

[0030] Step 3.1: Correct low-frequency data using a quadratic curve fitting algorithm;

[0031] Step 3.2: Use low-frequency data to extrapolate and obtain the value of the DC expansion point;

[0032] Step 3.3: Compare the calculated DC expansion point value with the previous value. If it is within the error range, the calculation ends. Otherwise, the difference between the two calculation results is displayed.

[0033] Preferably, in step 4, the normalization method includes a user-specified method and an automatic method; wherein, in the user-specified method, the user specifies the normalization position, normalization time and normalization value, and divides the data below the normalization position by the normalization value for normalization; the automatic method refers to using the value of the 90% position of the rise time as the normalization point, and dividing the data at the remaining positions by the value of the 90% position for normalization.

[0034] Beneficial technical effects brought by the present invention:

[0035] The frequency domain data truncation time window function of the present invention can be selected and input by the user; the interpolation accuracy is improved by using pre-stored values; the upper limit of the step rise time is expanded by using an extraction algorithm; and user-defined and automatic normalization methods are provided to avoid the low-frequency jitter problem of impedance testing. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 Flowchart for pulse width / rise time calculation. DETAILED DESCRIPTION

[0037] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments:

[0038] The vector network analyzer is a test system based on the frequency domain, and the time domain transformation function is implemented based on the IFFT algorithm. The frequency domain data of the vector network analyzer can be regarded as a sampling function with a frequency interval of △ω:

[0039]

[0040] In reality, the frequency domain test data is a finite length sequence, and the value of n cannot be infinite. The value of n and the value of the frequency step △ω are determined by the number of measurement points and the sweep range of the vector network analyzer. The sampling of finite length frequency domain data can be considered as a sequence with a sampling period of 1 / △ω. When converted to time domain data through IFT, a truncation effect will occur. Truncation is the product of frequency domain data and a rectangular window function, and the truncation point of the frequency domain data is usually not close to zero, resulting in a sidelobe effect during the inverse transformation, and large sidelobes will obscure the real signal at that point, resulting in erroneous measurement results. The solution is to add different window functions during truncation to reduce the truncation effect, and windowing will expand the impulse width or step response rise time. The impulse broadening characteristics of the window function make the time domain transformation result of the ideal open-circuit response after the frequency domain sampling data is windowed not zero at t=0, and the window function needs to be normalized. The normalized frequency domain truncation windowing time domain transformation result is:

[0041]

[0042] Where n is the count, N is half of the order of the window function, ω0 is the normalized value of the window function, ω() is the window function, and F() is the frequency domain sampling function.

[0043] The windowed time domain transform suppresses the sidelobe response while also causing the width of the corresponding unit impulse to be widened. The window function of the current vector network analyzer uses the Kaiser window function, which balances the sidelobe suppression and the width expansion of the unit impulse response by adjusting its β parameter.

[0044] The time domain transformation digital excitation of the vector network analyzer includes impulse and step. The normal working mode of the time domain reflectometer is step DC excitation, which is also the test requirement for cable test skew, rise time decay and other indicators. The step response is the integral result of the time domain impulse response. It requires that the time domain is transformed into a low-pass mode and the frequency domain data to be transformed is in a harmonic relationship and extended to zero. Because of the windowing, the step rise time will inevitably be larger than the ideal value. Different cable manufacturers require different rise times. The present invention provides multiple options and allows users to customize input, while using algorithms to expand the upper and lower limits.

[0045] The frequency response of the window function is the reason for the broadening of the rise time of the time domain step response. Usually, users are not clear about what coefficients should be selected for the window of the time domain transformation of the vector network analyzer, but the time resolution that the user's test device needs to achieve can be determined. Different window functions have different broadening factors. The present invention provides a variety of commonly used window functions including Gaussian, exponential, Chebyshev, Hanning, Hamming, Blackman, etc., and corresponds between the coefficients of the window function and the pulse width and step rise time. The user inputs the pulse width or rise time, and the system automatically calculates the accurate coefficients, thereby generating a window function that meets the width or rise time requirements.

[0046] In order to more accurately correspond to the relationship between the coefficient and the rise time, the present invention pre-stores the rise time and pulse width of some commonly used coefficients. When the rise time or pulse width set by the user is within the range of the pre-stored value, first determine whether it falls on the pre-stored value. If so, take the pre-stored value, otherwise use the linear interpolation algorithm to calculate the corresponding coefficient; when the user adjusts the window function coefficient, the same algorithm is used to calculate the corresponding pulse width or rise time. When the time set by the user exceeds the lower limit of the pre-stored value, take the lower limit of the pre-stored value, because the pre-stored value saves the minimum coefficient value of the window function; and for window functions such as the Caesar window and the Gaussian window, due to the uncertain upper limit of the window coefficient, it cannot be implemented by the pre-storage method. Of course, it can be implemented by the extrapolation method, but the interpolation algorithm during extrapolation has a greater impact on the interpolation result. When the user-set value is not much different from the upper limit, pchip or spline is used as the extrapolation algorithm. At this time, the extrapolation result is more in line with expectations, and an algorithm selection interface is provided to increase flexibility.

[0047] When the user-set value exceeds 0.5 times the upper limit, the use of the interpolation method will inevitably lead to a situation where the user-set value becomes less accurate the farther it is from the pre-stored value. Therefore, the present invention provides an extraction algorithm, which utilizes the proportional relationship between different orders of the window function under the same coefficient. When the user-set value exceeds the upper limit of the pre-stored value, the proportional coefficient r is calculated, r=ceil(user-set value / upper limit of the pre-stored value), and the window function order / r is used as the new window function order. At this time, the pulse width or rise time of the new window function will be r times that of the original window function. This method can extend the pulse width or rise time by any r times.

[0048] The present invention allows the user to input the relationship between the custom window function coefficient of a certain length and the pulse width or rise time. At this time, the system uses the spline interpolation algorithm to obtain any point in the interval, and adopts the extraction algorithm to implement the value exceeding the upper limit of the interval.

[0049] Since the pulse width and rise time are widened and the response of DC and low-frequency components is amplified when the frequency domain data is windowed and amplitude normalized, small jitter of low-frequency data will also cause the time domain step response to jump up and down. During impedance testing, it manifests as the impedance at the starting position cannot be fixed. However, in a physical sense, the calibrated front-end termination impedance of the device under test should be a stable system impedance. The present invention provides a normalization method to avoid jitter problems during impedance or delay testing.

[0050] The DC component of the low-pass time domain transformation of the vector network analyzer is calculated by using an algorithm based on the low-frequency component expansion. The previous vector network analyzer used the first 2 or 3 points for linear extrapolation. However, the test accuracy of the low-frequency test points of the vector network analyzer is often poor, and there is a large jitter phenomenon, so the extrapolation result is usually not ideal. The present invention takes into account the low-frequency characteristics of the vector network analyzer, expands the calculation points to a high-frequency range with good dynamics and relative stability, and expands from the high-frequency range to the low-frequency range. First, the low-frequency data is corrected using a quadratic curve fitting algorithm, and then the low-frequency data is extrapolated to obtain the value of the DC expansion point. The calculated value of the DC expansion point is compared with the previous value. If it is within the error range, the calculation ends, otherwise the difference between the two calculation results is displayed.

[0051] The present invention allows the user to choose whether to normalize the step response result. Normalization can avoid the low-frequency jitter problem mentioned above. The normalization method includes a user-specified method and an automatic method. In the user-specified method, the user specifies the normalization position (time) and the normalization value. At this time, the data below the normalization position is divided by the normalization value for normalization; in the automatic method, the system obtains the value of the 90% position of the rise time, and the data of the remaining positions are divided by the value of the 90% position for normalization.

[0052] A time domain test method suitable for signal integrity testing, such as Figure 1 As shown, the implementation steps are as follows:

[0053] Step 1: To meet the step rise time characteristics requirements of different users, multiple frequency domain windowing types such as Kaiser window, Gaussian window, Hanning, and Hamming are provided. At the same time, interpolation or extrapolation and decimation algorithms are used to obtain accurate step rise time and expand the rise time limit;

[0054] Step 2: Allow user interface to allow user to input window function;

[0055] Step 3: In order to reduce the impact of low-frequency jitter on the test results, the DC extension point is calculated again after fitting the low-frequency data;

[0056] Step 4: To avoid up and down jitter in impedance or delay test results, a normalized test method is provided, allowing the user to specify or the system to automatically normalize.

[0057] Of course, the above description is not a limitation of the present invention, and the present invention is not limited to the above examples. Changes, modifications, additions or substitutions made by technicians in this technical field within the essential scope of the present invention should also fall within the protection scope of the present invention.

Claims

1. A time domain testing method suitable for signal integrity testing, characterized in that: The steps include: Step 1: Provide a variety of frequency domain windowing functions; Step 2: Allow the user to input the relationship between the custom window function coefficient of a certain length and the pulse width or rise time; use the interpolation or extrapolation method to obtain the accurate step rise time, and use the extraction algorithm to expand the upper and lower limits of the step rise time; Step 3: Calculate the DC expansion point; Expand the calculation points to the high-frequency range, and then expand from the high-frequency range to the low-frequency range; Step 4: Provide a normalization method to allow the user to choose whether to normalize the step response result.

2. The time domain testing method suitable for signal integrity testing according to claim 1, characterized in that: In step 1, a variety of frequency domain windowing functions include Gaussian, exponential, Chebyshev, Hanning, Hamming and Blackman functions.

3. The time domain testing method suitable for signal integrity testing according to claim 1, characterized in that: In step 2, the spline interpolation algorithm is used to obtain any point in the interval, and the extraction algorithm is used to obtain the value exceeding the upper limit of the interval.

4. The time domain testing method suitable for signal integrity testing according to claim 3, characterized in that: The extraction algorithm is as follows: Step S1: Start; Step S2: Determine whether the user input value is within the pre-stored value range; If: the judgment result is that the user input value is within the pre-stored value range, execute step S3; Or if the judgment result is that the user input value is not within the pre-stored value range, then execute step S4; Step S3: Determine whether the user input value falls within the pre-stored value; If: the judgment result is that the user input value falls on the pre-stored value, then the pre-stored value is taken; Or if the result of the determination is that the user input value does not fall within the pre-stored value, then step S6 is executed; Step S4: Determine whether the user input value exceeds the lower limit of the pre-stored value or the upper limit of the pre-stored value; If: the judgment result is that the user input value exceeds the lower limit of the pre-stored value, then the lower limit of the pre-stored value is taken; Or if the judgment result is that the user input value exceeds the upper limit of the pre-stored value, then execute step S5; Step S5: Calculate the proportional coefficient r, r = user set value / pre-stored value upper limit, and use the window function order / r as the new window function order; Step S6: taking two pre-stored values ​​greater than and less than the user input value for linear interpolation; Step S7: End.

5. The time domain testing method suitable for signal integrity testing according to claim 1, characterized in that: Step 3 specifically includes the following steps: Step 3.1: Correct low-frequency data using a quadratic curve fitting algorithm; Step 3.2: Use low-frequency data to extrapolate and obtain the value of the DC expansion point; Step 3.3: Compare the calculated DC expansion point value with the previous value. If it is within the error range, the calculation ends. Otherwise, the difference between the two calculation results is displayed.

6. The time domain testing method suitable for signal integrity testing according to claim 1, characterized in that: In step 4, the normalization method includes a user-specified method and an automatic method; wherein, in the user-specified method, the user specifies the normalization position, normalization time and normalization value, and divides the data below the normalization position by the normalization value for normalization; the automatic method refers to using the value of the 90% position of the rise time as the normalization point, and dividing the data at the remaining positions by the value of the 90% position for normalization.

Images