JPH03200200A - Frequency analyzing system - Google Patents

Abstract

PURPOSE:To allow the frequency analysis up to the high-frequency region of >=1/2 a sampling speed by adding a mechanism for slightly changing the speed of the sampling speed of an input signal and a mechanism for comparing the results of the frequency analysis before and after the change. CONSTITUTION:A sampling speed changing means 3 changes the sampling speed of a sampling means 1 by a slight value DELTAf and a peak comparing means 4 detects the frequency difference between the two peaks of the equal value in the output spectra of a frequency analyzing means 2 as a corresponding difference if the difference has a relation of integer times of DELTAf. The sampling speed changing means 3 and the peak comparing means 2 are provided in such a manner and discrimination is made as to which of the frequency bands folded back in the output of the frequency analyzing means 2 the peaks from the frequency difference of both peaks having the corresponding relation belongs, by which the analysis of the frequency component of the input signal of the sampling means 1 down to the high-frequency component exceeding 1/2 the sampling speed is executed. The frequency analysis down to the high-frequency component of >=1/2 the sampling frequency is executed in this way.

Description

[Detailed Description of the Invention] [Summary] Regarding a frequency analysis method for analyzing frequency components of an analog input signal, the present invention provides a frequency analysis method that enables frequency analysis up to a high frequency region of 1/2 or more of the sampling rate. With the aim of
A sampling rate changing means for changing f, and a peak comparing means for detecting as corresponding peaks where the frequency difference between two peaks of equal value in the output spectrum of the frequency analyzing means is an integral multiple of Δf are provided. By determining which frequency band the peak belongs to in which the peak is aliased in the output of the frequency analysis means from the frequency difference between the corresponding peaks, the frequency component of the input signal of the sampling means is reduced to 1/1 of the sampling rate.
It is configured by making it possible to analyze up to high frequency components exceeding 2.

[Industrial Application Field] The present invention relates to a frequency analysis method for analyzing frequency components of an analog input signal, and particularly relates to a frequency analysis method for analyzing frequency components of an analog input signal.
This invention relates to a frequency analysis method that makes it possible to analyze up to high frequency components of /2 or higher.

In digital frequency analyzer, analog 111
The input signal is frequency-analyzed by expanding the sampled output into the frequency domain.

In this case, it is desired that frequency analysis can be performed by reproducing the original analog signal up to a high frequency component of 1/2 or more of the sampling rate in the frequency analysis result.

[Prior Art] In a digital frequency analyzer, input analyzer r:j
Frequency analysis of the input signal is performed by expanding the signal obtained by analog-to-digital (AD) conversion of the input signal at a constant speed into the frequency domain. At this time, the output of the frequency analyzer is Since high-frequency components of 1/2 or more appear folded back, it becomes difficult to interpret the analysis results.

Therefore, conventionally, a low-pass filter called an anti-aliasing filter is inserted before the AD conversion device, a cutoff frequency of 1/2 or less of the sampling rate is set, and high-frequency components are removed before frequency analysis is performed. In this way, the influence of aliased frequency components is eliminated.

[Problem to be solved by the invention]

In the conventional frequency analysis method described above, the upper limit of the frequency that can be analyzed is limited by the sampling rate of the AD converter, and it is not possible to perform frequency analysis up to high frequency components of 1/2 or more of the sampling rate. .

Further, anti-aliasing filters are generally commercially priced, and simple AD conversion devices often do not include such filters.

The present invention aims to solve the problems of the prior art, and adds a mechanism for changing the sampling rate of the input signal at the minimum speed, and a mechanism for comparing the frequency analysis results before and after the change. The purpose of this invention is to provide a frequency analysis method that enables frequency analysis up to a high frequency region of 1/2 or more of the sampling rate.

[Means for Solving the Problems] As shown in FIG. 1, the present invention has a 4F sampling means 1 that samples an analog signal input at a constant rate, and a sampling output that is 1/2 of the sampling rate. In a frequency analysis mechanism having a frequency analysis means 2 developed in a frequency region, a sampling rate change means 3 and peak comparison means 5 to 10 are provided, and this peak is frequency analyzed based on the frequency difference between both peaks in a corresponding relationship. By determining which frequency band ζ is folded back to the output of means 2, the sumbri
It is possible to analyze the 11-stage human power (RJ frequency component) up to high frequency components exceeding 1/2 of the sampling rate.

Here, the sampling speed changing means 3 changes the sampling speed in the sampling means 1 by a minute value Δf, and the peak comparing means 5 to 10 change the sampling speed between two peaks of equal value in the output spectrum of the frequency analyzing means 2. Those in which the frequency difference is an integer multiple of Δf are detected as corresponding ones.

[Effect]

Generally, sampling data is subjected to Fourier transform (FFT)
If you analyze the frequency by
It is known that two or more frequency components cause an error called aliasing.

Figures 3 (a) and (b) explain aliasing, where (a) shows the case where the target signal frequency is 1/2 or less of the sampling rate, and (b) shows the case where the target signal frequency is 1/2 or less of the sampling rate. This shows a case where the speed is 1/2 or more.

In the case of Fig. 3(a), the sampling interval ΔT is less than 1/2 of the signal period, so the digital signal waveform (shown by the dotted line) connecting the AD conversion values shown by ○ is the original signal shown by the solid line. Similar to analog signal waveform.

On the other hand, in the case of Fig. 3 (1)), the sampling interval Δ'
Since F is more than 1/2 of the period No. 13, AD indicated by ○
The digital signal waveform (shown by the dotted line) that connects the converted values is
The waveform is different from the original analog signal waveform shown by the solid line.

Figures 4 (a) to (C) illustrate aliasing errors on frequency, where (a) shows the frequency components of the input analog signal, and (t)) shows f, ,X ( f...
1/ΔT) as the sampling frequency and shows the folded signal components in each frequency band when AD conversion is performed, and (C) shows the frequency analysis output.

In Fig. 4(a), the input analog signals include ■～■
It is shown that there is a peak frequency component at each point shown in . Such a signal is sampled at a sampling rate r,,X (f
,,X=1/ΔT) When AD variation ta is performed, Figure 4 (
As shown in b), signals existing in each frequency band that is an integer multiple of r□,/2 are folded back and appear in the frequency analysis output.

Therefore, in the output signal from the frequency analyzer, the output signals of each band shown in FIG. 4(b) are added, and the output signals shown in FIG.
C).

Now, if we slightly change the sampling speed by Δf, we get
The frequency of each folded output changes.

Figures 5 (a) to (e) illustrate changes in frequency data when the sampling rate is slightly changed, where (a) shows the input analog signal and sampling rate, and (b) shows the input analog signal and sampling rate. Folded signal output for each frequency band when AD conversion is performed at the sampling rate f□, (C1 is the frequency analysis output when AD conversion is performed at the sampling rate f six, (d) is the sampling rate (f,, , 10Δf), the folded signal output of each frequency band when AD conversion is performed, (e) is the sampling rate (f,
It shows the frequency analysis output when AD conversion is performed in r).

In FIG. 5(a), it is assumed that the input analog signal has a peak frequency component at each point, not indicated by ■■. Such a signal is sampled at a sampling rate f saw (A,,X=1
/ΔT), as shown in FIG. 5(b), aliasing signal outputs ■, ■ are generated for each frequency band that is an integer multiple of (,,X/2).The output signal from the frequency analyzer Now, the output signals of each band shown in FIG. 5(b) are added, resulting in the result shown in FIG. 5(C).

Now, if we slightly change the sampling speed by Δf, we get
The frequency of each folded output changes, and folded signal outputs ■' and ■' are generated for each frequency band that is an integer multiple of (Il, As for the signal, the output signals of each band shown in FIG. 5(d) are added to form the signal shown in FIG. 5(e).

Comparing Fig. 5 (C1, (e)), both peak signals are ■.

■' has the same frequency, but the folded signals ■ and ■ have different frequencies. The frequency change π in this case is equal to the amount of change Δr in the sampling frequency. Similarly, the fourth
In the case of the peak signal ■2■ in a higher frequency band as shown in Figure (1)), the frequency changes of the corresponding aliasing signals are 2Δf, 3Δf, and 3Δf, respectively.
...becomes.

Therefore, by checking the rate of change of each frequency between the original frequency component in the AD conversion result and the frequency component when the sampling rate is slightly changed, it is possible to know the frequency band in which each frequency component exists. from which its true frequency value can be estimated.

The present invention utilizes this fact to obtain frequency components higher than the sampling frequency.

[Embodiment] FIG. 2 shows an embodiment of the present invention, in which 11 is an AD converter, 12 is a frequency analyzer, 13 is an AD conversion clock generator, and 14 is an AD conversion clock switching section. , 15
16.17 is a frequency analysis data storage section, 18.19 is a spectrum peak detection section, 20 is a spectrum peak comparison section, and 21 is a spectrum display section.

In FIG. 2, an input analog signal is subjected to AD conversion by an AD converter 11 at a constant sampling rate.
Generates an output consisting of a digital signal. As such a sample value generating means, for boat fishing, a zambrar which samples the input signal at a constant speed may be used.

The frequency analyzer 12 generates an output signal obtained by expanding the discrete output signal in the time domain from the AD converter 11 into the frequency domain.

At this time, the AD conversion clock generator 13
Based on the control n of the AD conversion clock switching unit 14, the sampling frequency K at 1 is generated, but the frequency slightly changes from [□8 to “□8+Δf”.

The changeover switch 15 switches the frequency analyzer 1 in synchronization with the frequency switching control of the AD conversion clock switching unit 14.
The output of No. 2 is switched to the frequency analysis data recording section 16 and 17 for manual input.

The frequency analysis data storage units 16 and 17 store the input outputs of the frequency analyzer 12, respectively.

The spectral peak detection section 18.19 extracts each peak value and its frequency data from the data stored in the frequency analysis data storage section 16.17.

The spectral peak comparator 20 is configured to detect peaks of equal value in the outputs of the spectral peak detectors 18 and 19, and to check the difference in frequency between both peaks by an integral multiple of Δf.

In the spectrum display section 21, the outputs of the spectrum peak detection sections 18 and 19 are added and displayed, thereby facilitating the comparison operation in the spectrum peak comparison section 20.

Therefore, in the present invention, the correct position on the frequency of the frequency component that appears folded in the output of the frequency analyzer 12 can be easily estimated from the spectrum peak comparison result obtained in this manner.

〔Effect of the invention〕

As explained above, according to the present invention, the sampling frequency in the sampling means is By comparing the data of the frequency analysis results when making small changes, it is possible to know the correct position on the frequency of the frequency component that is folded back and appearing in the output of the frequency analysis results. On the other hand, it becomes possible to perform frequency analysis up to high frequency components of 1/2 or more of the sampling frequency.

[Brief explanation of drawings]

FIG. 1 is a diagram showing the basic configuration of the present invention, FIG. 2 is a diagram showing an embodiment of the present invention, FIGS. 3(a) and (b) are diagrams explaining aliasing, and FIG. al~(C) are diagrams illustrating aliasing errors in frequency, and Figures 5(a)~(
]) is a diagram illustrating a change in frequency data when the sampling rate is slightly changed. 1 is a +i sampling means, 2 is a frequency analysis means, 3 is a sampling rate changing means, and 4 is a peak comparison 1 river role. Figure 1 (3) Shows an example of perfecting 2. Figure 3 (a) 2. Figure 3 (a) 2. Figure 3 (a) 2. Figure 3 (a). Figure 4 Spectrum number (b) (d) (c) (e) Figure 5

Claims (1)

[Scope of Claims] A frequency analysis mechanism that includes a sampling means (1) that samples an analog signal input at a constant speed, and a frequency analysis means (2) that expands the sampling output into a frequency region that is 1/2 of the sampling speed. In, sampling speed changing means (3) for changing the sampling speed in the sampling means (1) by a minute value Δf.
and peak comparison means (4) for detecting as corresponding peaks where the frequency difference between two peaks of equal value in the output spectrum of the frequency analysis means (2) is an integral multiple of Δf, The sampling means (2) determines to which frequency band the peak belongs in the output of the frequency analysis means (2) from the frequency difference between the corresponding peaks.
1) The frequency component of the input signal is set to 1/1 of the sampling rate.
A frequency analysis method that is characterized by being able to analyze up to high frequency components exceeding 2.