Frequency response correction method for front-end analog channel of broadband digital oscilloscope

Abstract

The invention discloses a frequency response correction method for a front-end analog channel of a broadband digital oscilloscope, and the method comprises the steps: constructing an ideal fast edge signal and an expected channel frequency response characteristic in a digital domain; the output of the ideal fast edge signal after passing through the expected frequency response model is used as the expected response of the fast edge signal to be compared with the fast edge response output of the actual front-end simulation channel in time domain parameter, so as to obtain each parameter deviation; and correcting the amplitude-frequency characteristic and the phase-frequency characteristic of the frequency response Hc (ej omega) of the compensation filter by using deviation correction indication signals corresponding to the parameter deviations, so that an expensive high-sampling-rate, high-bandwidth and high-precision sampling oscilloscope does not need to be used for sampling to complete the correction of the front-end channel frequency response of the broadband oscilloscope. Besides, a frequency sampling method is used for uniformly designing compensation filters of amplitude-frequency response and phase-frequency response, and finally all the compensation filters are converted into FIR compensation, so that the method is efficient and simple to implement, high in compensation precision and applicable to occasions where channel amplitude-frequency characteristics and phase-frequency characteristics are very unsatisfactory.

Description

Technical Field

[0001] The invention belongs to the technical field of digital oscilloscope correction, and more specifically, relates to a frequency response correction method for a front-end analog channel of a broadband digital oscilloscope. Background Art

[0002] Digital oscilloscopes are the most commonly used instruments in general test and measurement instruments. With the continuous increase in sampling rate and bandwidth, the frequency response characteristics of the front-end analog channels have an increasingly greater impact on the measurement results in test scenarios such as high-speed signals and broadband signals. The frequency response characteristics of the front-end analog channels can be divided into amplitude-frequency characteristics and phase-frequency characteristics, which respectively reflect the change in amplitude and phase after the input signals of different frequencies pass through the front-end analog channels (the change in phase corresponds to the delay after the corresponding signal passes through the channel). Ideally, it is hoped that the amplitude-frequency characteristics of the front-end analog channels will present a flat characteristic within the band, while the phase-frequency characteristics will present a linear phase characteristic within the band. In practice, due to non-ideal characteristics such as hardware components, the frequency response characteristics of the front-end analog channels will present certain non-ideal characteristics. Before leaving the factory, broadband digital oscilloscopes must undergo strict frequency response calibration, i.e., frequency response calibration, to ensure the measurement accuracy during subsequent use. The purpose of frequency response calibration is to make the frequency response of the front-end analog channels after calibration close to the ideal characteristics or the characteristics expected by the user.

[0003] The frequency response correction of the analog channel at the front end of a broadband digital oscilloscope requires a fast edge signal as its correction signal, and the bandwidth of the fast edge signal itself must be greater than the bandwidth of the oscilloscope to be corrected. Since the fast edge signal is also generated and output by a fast edge generator, the fast edge signal itself is also a non-ideal fast edge signal. Therefore, a sampling device with "ideal frequency response characteristics" is required to sample the fast edge signal to obtain the frequency spectrum characteristics of the fast edge signal itself. In the prior art, a sampling oscilloscope with a higher sampling rate and higher bandwidth than the broadband digital oscilloscope to be corrected is used as a reference device. The frequency response characteristics of the sampling oscilloscope itself are considered to be "ideal". Therefore, it is used to sample the fast edge signal, and the sampled data is used as the standard data representing the correction signal, so that the "real" frequency spectrum characteristics of the fast edge signal can be obtained.

[0004] However, the aforementioned existing frequency response correction method for the front-end analog channel has two shortcomings: on the one hand, the sampling oscilloscope itself is expensive; on the other hand, even for a sampling oscilloscope, its channel frequency response is not absolutely ideal and still has certain non-ideal characteristics. In this case, using its sampling data as reference data will bring unavoidable errors to the final correction.

[0005] In the US invention patent US7050918B2, entitled "Digital Group Delay Compensator", which was authorized and published on May 23, 2006, a digital group delay (Digital Group Delay compensation method is used to compensate the frequency response of broadband digital oscilloscope: a high-precision sampling oscilloscope is used to sample the correction signal, and the sampled signal of the front-end analog channel of the oscilloscope to be corrected is deconvolved to obtain the pulse response of the channel, and then the ideal fast edge signal is used to convolve with the pulse response to obtain the output, and the group delay characteristics of the output are compensated, and the pulse characteristics of the compensated output time domain signal (such as overshoot, preshoot, rise time and other parameters) are calculated, and these parameters are further used to adjust the coefficients of the compensation filter to form a closed loop. The core of the method is to repeatedly correct the group delay characteristics of the front-end analog channel pulse response until the various parameters of the output fast edge signal meet the indicators. The group delay compensation filter adopts the IIR all-pass filter method. When the real group delay characteristics of the channel show more fluctuations and larger fluctuation characteristics, due to the limitation of the IIR all-pass filter design method, this method is difficult to converge and the corresponding compensation filter coefficients cannot be obtained, which leads to the failure of the algorithm. Summary of the invention

[0006] The purpose of the present invention is to overcome the shortcomings of the prior art and provide a frequency response correction method for the front-end analog channel of a broadband digital oscilloscope, so as to avoid the high cost problem caused by the high sampling rate and high bandwidth high-precision sampling oscilloscope being used as a reference device. At the same time, the problem that when the IIR all-pass filter compensates for the group delay characteristics of the front-end channel, when the group delay characteristics have more fluctuations and more violent fluctuations (with more "pits"), the filter is difficult to converge, so as to avoid algorithm failure.

[0007] To achieve the above-mentioned object of the invention, the frequency response correction method of the front-end analog channel of a broadband digital oscilloscope of the present invention is characterized by comprising the following steps:

[0008] (1) Obtain the initial compensation frequency response

[0009] A vector network analyzer is used to measure the frequency response of the analog channel of the broadband digital oscilloscope front end. The measurement result is recorded as H(e jω ), then take its reciprocal 1 / H(e jω ) as the compensation frequency response, that is, the frequency response of the compensation filter H c (e jω ), where ω is the angular frequency;

[0010] (2) Calculate the compensation filter coefficients

[0011] Using the frequency sampling method, the frequency response H c (e jω ) Generate compensation filter coefficients h c (n), output to the frequency response compensation module, where n is the sampling point number;

[0012] (3) Compensation filtering of input data

[0013] A fast edge signal x(t) is outputted by a fast edge signal generator and inputted into the broadband digital oscilloscope to be calibrated, where t is the time. After passing through the front-end analog channel, the signal enters the ADC, and the ADC outputs a sampling sequence x(n). The appropriate oscilloscope time base, voltage gear and trigger mode are set so that only one fast edge is displayed in the center of the screen of the broadband digital oscilloscope to be calibrated. The sampling sequence x(n) enters the averaging processing module. Under the control of the trigger signal, the averaging processing module averages several frames of data of the sampling sequence x(n), and then outputs the averaged data to the frequency response compensation module. The frequency response compensation module receives the compensation filter coefficient h. c (n), perform compensation filtering on the input data and output the data sequence y(n);

[0014] (4) Time domain parameter deviation extraction

[0015] First, an ideal fast-edge signal sequence s(n) is generated based on the front-end analog channel bandwidth and ADC sampling rate, and a desired frequency response H is generated. d (e jω ), the frequency response H d (e jω ) The filter corresponding to the impulse response filters the fast edge signal sequence s(n) and outputs the fast edge signal sequence q(n);

[0016] Then, by comparing the time domain parameters of the data sequence y(n) and the fast edge signal sequence q(n), the deviation of each parameter is obtained and judged. If the deviation of each parameter is within the set range, the closed-loop correction is terminated, and the compensation filter coefficient h of the channel frequency response compensation is c (n) is no longer updated, and the compensation filter coefficient h c (n) stored for frequency response compensation of the front-end analog channel when the broadband digital oscilloscope is officially used. If the deviation of each parameter is not completely within the set range, the corresponding deviation correction indication signal is output to the compensation frequency response correction module;

[0017] (5) Compensation frequency response correction

[0018] The compensation frequency response correction module corrects the compensation frequency response, i.e. the frequency response H of the compensation filter, according to the input deviation correction indication signal. c (ejω )’s amplitude-frequency characteristic and phase-frequency characteristic are corrected, and then the process returns to step (2).

[0019] The object of the invention of the present invention is achieved in this way:

[0020] The frequency response correction method of the front-end analog channel of the broadband digital oscilloscope of the present invention does not need to use an expensive high-sampling rate, high-bandwidth, high-precision sampling oscilloscope for sampling to complete the correction of the front-end channel frequency response of the broadband oscilloscope. Instead, an ideal fast-edge signal and an expected channel frequency response characteristic are directly constructed in the digital domain according to the channel bandwidth and sampling rate characteristics of the broadband digital oscilloscope to be corrected. The output of the ideal fast-edge signal after the expected frequency response model is used as the expected response of the fast-edge signal and the fast-edge response output of the actual front-end analog channel is compared with the time domain parameters to obtain the deviations of various parameters (overshoot, preshoot, rise time, fall time deviation, etc.), and the deviation correction indication signals corresponding to these parameter deviations are used to compensate the frequency response H of the filter. c (e jω ) to correct the amplitude-frequency characteristic and phase-frequency characteristic of the channel, forming a closed-loop feedback system. When the deviation of each parameter is less than the set threshold, the compensated channel frequency response meets the expected requirement, and the closed-loop correction is terminated. In addition, the present invention uses a frequency sampling method to uniformly design compensation filters for amplitude-frequency response and phase-frequency response, and finally converts all of them into FIR compensation, achieving high efficiency and simplicity, and high compensation accuracy, and can be applied to occasions where the channel amplitude-frequency characteristic and phase-frequency characteristic are very undesirable. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 It is a flow chart of a specific implementation method of the frequency response correction method of the front-end analog channel of the broadband digital oscilloscope of the present invention;

[0022] Figure 2 It is a signal processing schematic diagram of a specific implementation method of the frequency response correction method of the front-end analog channel of the broadband digital oscilloscope of the present invention;

[0023] Figure 3 is an example of the amplitude-frequency characteristic in the desired frequency response;

[0024] Figure 4 is a schematic diagram of the structure of the deviation extraction module;

[0025] Figure 5 is a schematic diagram of phase characteristic rotation operation;

[0026] Figure 6 These are two example diagrams of different group delay characteristics, where (a) is a fixed value group delay, and (b) is a relatively large delay corresponding to the high frequency part;

[0027] Figure 7: is a frequency response correction example diagram, where (a) is an ideal fast edge signal, (b) is a fast edge signal to be corrected, and (c) is a fast edge signal after compensation, i.e. correction;

[0028] Figure 8 This is a diagram of the normal usage scenario of the oscilloscope after calibration. DETAILED DESCRIPTION

[0029] The specific implementation of the present invention is described below in conjunction with the accompanying drawings so that those skilled in the art can better understand the present invention. It should be noted that in the following description, when the detailed description of known functions and designs may dilute the main content of the present invention, these descriptions will be omitted here.

[0030] Figure 1 , 2 They are respectively a flow chart and a signal processing schematic diagram of a specific implementation of a method for correcting the frequency response of a front-end analog channel of a broadband digital oscilloscope of the present invention.

[0031] In this embodiment, if Figure 1 As shown, the frequency response correction method of the front-end analog channel of the broadband digital oscilloscope of the present invention comprises the following steps:

[0032] Step S1: Obtaining the initial compensation frequency response

[0033] A vector network analyzer is used to measure the frequency response of the analog channel of the broadband digital oscilloscope front end. The measurement result is recorded as H(e jω ), then take its reciprocal 1 / H(e jω ) as the compensation frequency response, that is, the frequency response of the compensation filter H c (e jω ).

[0034] In this embodiment, if Figure 2 As shown, the compensation frequency response initialization module 1 is used to initialize the frequency response H of the compensation filter. c (e jω ), and then output to the compensation frequency response correction module 2. Since no parameter deviation is obtained at this time, the compensation frequency response correction module 2 directly converts the frequency response H c (e jω ) is output to the compensation filter coefficient calculation module 3.

[0035] Step S2: Calculate compensation filter coefficients

[0036] In the compensation filter coefficient calculation module 3, the frequency response H is calculated by using the frequency sampling method. c (e jω ) Generate compensation filter coefficients h c(n), and output to the frequency response compensation module 4. Specifically: c (e jω ) performs an inverse FFT (IFFT) operation to obtain the corresponding compensation filter impulse response, and performs a windowing process on it to obtain the final compensation filter coefficient h c (n).

[0037] Step S3: Compensation filtering of input data

[0038] A fast edge signal x(t) is outputted by the fast edge signal generator 5 and inputted into the broadband digital oscilloscope to be calibrated, wherein t is time. After passing through the front-end analog channel 6, the signal enters the ADC 7. The ADC 7 outputs the sampling sequence x(n). The appropriate oscilloscope time base, voltage gear and trigger mode are set so that only one fast edge is displayed in the center of the screen of the broadband digital oscilloscope to be calibrated. The sampling sequence x(n) enters the averaging processing module 8. Under the control of the trigger signal, the averaging processing module 8 averages several frames of data, such as 8192 frames of data, on the sampling sequence x(n), and then outputs the averaged data to the frequency response compensation module 4. The frequency response compensation module 4 receives the compensation filter coefficient h. c (n), perform compensation filtering on the input data and output the data sequence y(n).

[0039] The present invention uses a frequency sampling method to uniformly design compensation filters for amplitude-frequency response and phase-frequency response, and finally converts all of them into FIR compensation. The reason why the phase-frequency characteristic compensation in the present invention does not use an IIR all-pass filter is that the IIR all-pass filter is suitable for compensating for relatively regular group delay characteristics, but is not suitable for compensating for group delay characteristics with more fluctuations or more drastic fluctuations.

[0040] Step S4: Time domain parameter deviation extraction

[0041] First, an ideal fast-edge signal sequence s(n) is generated based on the front-end analog channel bandwidth and ADC sampling rate, and a desired frequency response H is generated. d (e jω ), the frequency response H d (e jω ) The filter 9 corresponding to the pulse response filters the fast edge signal sequence s(n) and outputs the fast edge signal sequence q(n).

[0042] Fast edge signals can be generated like this (taking the rising edge as an example):

[0043]

[0044] Among them, TD represents the delay of the rising edge, T s is the fast edge signal sampling rate, Indicates rounding, ω0=2π×P / trise , P is a constant related to the hardware circuit, t rise It represents the rise time, for example, 20 ps (picoseconds), and this value is determined according to the front-end analog channel bandwidth of the actual wideband digital oscilloscope to be calibrated.

[0045] Desired frequency response: can be pre-generated and stored in the machine. According to the signal bandwidth and sampling rate, various frequency response characteristic combinations can be pre-generated. For example, the amplitude-frequency characteristics of various shapes can be generated. Figure 3 As shown, the amplitude-frequency characteristic can be Butterworth type (butter), Chebyshev type I (cheby1), Chebyshev type II (cheby2), Bessel type (besself), etc., or other characteristics; the phase-frequency characteristic can be linear phase, or other phase characteristics. Here, the desired frequency response can be selected from the existing characteristic curves, or it can be redefined by the user.

[0046] Then, in the deviation extraction module 10, by comparing the time domain parameters of the data sequence y(n) and the fast edge signal sequence q(n), such as preshoot, overshoot, rise time / fall time and other time domain parameters, the deviation of each parameter is obtained and judged. If the deviation of each parameter is within the set range, the closed-loop correction is terminated, and the compensation filter coefficient h of the channel frequency response compensation is c (n) is no longer updated, and the compensation filter coefficient h c (n) is stored for frequency response compensation of the front-end analog channel when the broadband digital oscilloscope is officially used. If the deviation of each parameter is not completely within the set range, the corresponding deviation correction indication signal is output to the compensation frequency response correction module 2.

[0047] In this embodiment, if Figure 4 As shown, before comparing the time domain parameters of the data sequence y(n) and the fast edge signal sequence q(n), a certain proportion of interpolation processing is performed on the two sequences as needed to improve the calculation accuracy of the parameters. Specifically, Figure 4 As shown, in the deviation extraction module 10, an interpolation module 1001 is used to interpolate the data sequence y(n) and the fast edge signal sequence q(n), and then a time domain parameter calculation module 1002 is used to obtain the three time domain parameters of the pre-shoot, overshoot and rise time of the data sequence y(n) and the fast edge signal sequence q(n), and then in the deviation calculation module 1003, the time domain parameters of the data sequence y(n) and the fast edge signal sequence q(n) are compared to obtain the deviation of each parameter and judge. If the deviation of each parameter is within the set range, the closed-loop correction is terminated. If the deviation of each parameter is not completely within the set range, the corresponding deviation correction indication signal 1, 2, ... m is output to the compensation frequency response correction module 2.

[0048] Step S5: Compensation frequency response correction

[0049] The compensation frequency response correction module 2 corrects the compensation frequency response, i.e., the frequency response H of the compensation filter, according to the input deviation correction indication signal. c (e jω )’s amplitude-frequency characteristic and phase-frequency characteristic are corrected, and then the process returns to step S2.

[0050] For example, when it is detected that the rise time of the data sequence y(n) is greater than the rise time of the fast edge signal sequence q(n), the phase-frequency characteristic of the compensation frequency response can be corrected. Specifically, the phase-frequency characteristic can be rotated by a certain angle. For example, Figure 5 Assume that the initial phase delay curve is OA, and now we want to rotate the curve by angle θ, then the coordinate calculation formula of point B after rotation is:

[0051] x2=(x1-X)cosθ-(y1-Y)sinθ+X

[0052] y2=(x1-X)sinθ+(y1-Y)cosθ+Y

[0053] Figure 6 These are two example graphs of different group delay characteristics. Figure 6 The group delay characteristic shown in (a) is a fixed value group delay, that is, the delay of signals of different frequencies after passing through the system is the same, and its corresponding phase characteristic is a linear phase characteristic. This group delay characteristic is generally used when the input signal is a broadband signal. Figure 6 (b) This group delay characteristic has a relatively large delay corresponding to the high-frequency part. This characteristic is generally used in the analysis of pulse signals, such as eye diagram analysis. This group delay characteristic does not cause excessive pre-throw in the time domain waveform when the input is a high-speed pulse signal or a fast-edge signal. Therefore, for example, when it is detected that the pre-throw value of the data sequence y(n) is significantly greater than the pre-throw value of the fast-edge signal sequence q(n), the group delay characteristic and amplitude-frequency characteristic of the compensation frequency response can be corrected. Specifically, the group delay corresponding to the high-frequency part can be appropriately increased and the amplitude attenuation corresponding to the high-frequency part can be reduced. Examples include Figure 6 (b) as shown.

[0054] Figure 7 (a) is an ideal fast edge signal, Figure 7 (b) is a fast edge signal to be corrected (with a large pre-shoot) collected by ADC, which is corrected by the present invention. Finally, Figure 7 As shown in (c), a compensated or corrected fast edge signal (with almost no pre-shoot) that meets the user's expectations can be obtained.

[0055] Once the calibration is completed, the filter coefficients for channel frequency response compensation are no longer updated. Figure 8 As shown, the coefficient is stored and used to correct the frequency response of the front-end analog channel when the frequency response of the broadband digital oscilloscope is officially used.

[0056] The present invention does not require a special high-precision sampling oscilloscope, which greatly saves the cost required for correction. In addition, the present invention uses a frequency sampling method to uniformly compensate amplitude and phase, with higher compensation accuracy and a wider compensation range.

[0057] Although the above describes the illustrative specific embodiments of the present invention to facilitate those skilled in the art to understand the present invention, it should be clear that the present invention is not limited to the scope of the specific embodiments. For those of ordinary skill in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the attached claims, these changes are obvious, and all inventions and creations using the concept of the present invention are protected.

Claims

1. A method for correcting the frequency response of a front-end analog channel of a broadband digital oscilloscope, characterized in that: The following steps are involved: (1) Obtain the initial compensation frequency response A vector network analyzer is used to measure the frequency response of the analog channel of the broadband digital oscilloscope front end. The measurement result is recorded as H(e jω ), then take its reciprocal 1 / H(e jω ) as the compensation frequency response, that is, the frequency response of the compensation filter H c (e j ω ), where ω is the angular frequency; (2) Calculate the compensation filter coefficients Using the frequency sampling method, the frequency response H c (e jω ) Generate compensation filter coefficients h c (n), output to the frequency response compensation module, where n is the sampling point number; (3) Compensation filtering of input data A fast edge signal x(t) is outputted by a fast edge signal generator and inputted into the broadband digital oscilloscope to be calibrated, where t is the time. After passing through the front-end analog channel, the signal enters the ADC, and the ADC outputs a sampling sequence x(n). The appropriate oscilloscope time base, voltage gear and trigger mode are set so that only one fast edge is displayed in the center of the screen of the broadband digital oscilloscope to be calibrated. The sampling sequence x(n) enters the averaging processing module. Under the control of the trigger signal, the averaging processing module averages several frames of data of the sampling sequence x(n), and then outputs the averaged data to the frequency response compensation module. The frequency response compensation module receives the compensation filter coefficient h. c (n), perform compensation filtering on the input data and output the data sequence y(n); (4) Time domain parameter deviation extraction First, an ideal fast-edge signal sequence s(n) is generated based on the front-end analog channel bandwidth and ADC sampling rate, and a desired frequency response H is generated. d (e jω ), the frequency response H d (e jω ) The filter corresponding to the impulse response filters the fast edge signal sequence s(n) and outputs the fast edge signal sequence q(n); Then, by comparing the time domain parameters of the data sequence y(n) and the fast edge signal sequence q(n), the deviation of each parameter is obtained and judged. If the deviation of each parameter is within the set range, the closed-loop correction is terminated, and the compensation filter coefficient h of the channel frequency response compensation is c (n) is no longer updated, and the compensation filter coefficient h c (n) stored for frequency response compensation of the front-end analog channel when the broadband digital oscilloscope is officially used. If the deviation of each parameter is not completely within the set range, the corresponding deviation correction indication signal is output to the compensation frequency response correction module; (5) Compensation frequency response correction The compensation frequency response correction module corrects the compensation frequency response, i.e. the frequency response H of the compensation filter, according to the input deviation correction indication signal. c (e jω )’s amplitude-frequency characteristic and phase-frequency characteristic are corrected, and then the process returns to step (2).

2. The frequency response correction method of the front-end analog channel of a broadband digital oscilloscope according to claim 1, characterized in that: In step (4), in the deviation extraction module, an interpolation module is used to interpolate the data sequence y(n) and the fast edge signal sequence q(n), and then a time domain parameter calculation module is used to obtain the three time domain parameters of the pre-shoot, overshoot and rise time of the data sequence y(n) and the fast edge signal sequence q(n). Then, in the deviation calculation module, the time domain parameters of the data sequence y(n) and the fast edge signal sequence q(n) are compared to obtain the deviation of each parameter and judge. If the deviation of each parameter is within the set range, the closed-loop correction is terminated. If the deviation of each parameter is not completely within the set range, the corresponding deviation correction indication signal 1, 2, ... m is output to the compensation frequency response correction module.

3. The frequency response correction method of the front-end analog channel of a broadband digital oscilloscope according to claim 2, characterized in that: The fast edge signal sequence s(n) is a rising edge signal sequence: Among them, TD represents the delay of the rising edge, T s is the fast edge signal sampling rate, Indicates rounding, ω0=2π×P / t rise , P is a constant related to the hardware circuit, t rise Indicates the rise time.

Images