JPS5956170A - Transfer function measuring instrument - Google Patents

Abstract

PURPOSE:To measure a transfer function digitally with high precision, by finding frequencies of peak power spectra of an input and an output signals and those peak values, and performing averaging and dividing processing regarding only the peak values. CONSTITUTION:A sine wave signal from a sweep sine wave signal generator 11 is inputted to a body 12 to be measured and its input signal and output signals are written in buffer memories 22 and 24 through AD converters 21 and 23. Their sampled values are inputted to an FFT processor 29 through a multiplexer 25 to perform fast Fourier transform and then the product of the input and output spectrum is calculated to find a cross spectrum. Then, the frequencies and values of peak spectra are detected and when the peak value is greater than a specific level, the mean value of peak values so far is found. When sweep within a desired frequency range is completed, amplitude characteristics and phase characteristics are found by a CPU27 from the mean value of the input/ output power spectra of every frequency of each peak and the mean value of cross spectra.

Description

[Detailed Description of the Invention] This invention supplies a signal to an object to be measured, and the input signal and
The present invention relates to a transfer function measuring device that determines the transfer function of a measured object from an output signal from the measured object, and particularly to a measuring device that digitally measures the transfer function using fast Fourier transform processing.

<Prior Art> As shown in FIG. 1, the transfer function measuring instrument inputs a swept sine wave signal from a swept sine wave signal generator 11 to a device under test 12, and outputs a signal to the device under test 12 at the level of the input signal. The level of the output signal is divided by the division circuit 13 and supplied to the first Y-axis terminal of the CRT display 14, and the phase difference between the input signal and the output signal to the object under test 12 is detected by the phase difference detector 15. The signal is detected and supplied to the second Y-axis terminal of the CRT display 14, while the signal is swept W.

The frequency sweep signal from the wave signal generator 11 is supplied to the X-axis terminal of the CRT display 14, and the amplitude (gain) characteristics of the 111I constant object 12 are displayed as a curve 16 on the display surface of the display 14. The phase characteristics are displayed as a curve 17.

It has been proposed to measure the transfer function digitally instead of analogously determining o+, thereby facilitating subsequent digital processing. This is to obtain a transfer function by performing fast Fourier transform on input and output signals.

That is, the time series slipping from n+1jul samples of the input signal is subjected to daytime Fourier transform, and its frequency components Fl-Fn are respectively input as shown in FIG. 2, for example.

Signal spectrum/l/ a + +ja + '~al
+j an' and its power spectrum A+-An
get. Similarly, a time series consisting of n samples of the output signal of the object under test 12 is subjected to fast Fourier transform, and the output signal spectrum bs+jb+ is obtained using the frequency component F+−Fn.
'~bn+jbn' is obtained, and the cross spectrum, which is the product of the corresponding frequencies of these input and output waste vectors, is expressed as c+-1-jc for the frequency components F1 to Fn.
+"~Cn+jCn1. In this way, the input signal worth vector and cross spectrum are determined for each time series of the input signal and output signal, and the average value for each corresponding frequency component is determined, and for each negative frequency component, The amplitude (gain) characteristics are obtained by dividing the averaged input power spectrum and the averaged cross spectrum by the vector, and the phase characteristics are obtained from the real part and imaginary part of the average cross spectrum of each wave number component. There is.

In this way, in the past, all frequency components of the fast Fourier transformed spectrum were averaged, that is, even the spectra with small power were averaged. Dividing by a small power involves a large error,
This lowered the overall measurement accuracy.

<Summary of the Invention> An object of the present invention is to provide a transfer function measuring instrument that can digitally measure a transfer function with high accuracy and is free from nonlinear effects such as distortion.

According to the present invention, the peak frequency and peak value of the input coverworth vector are determined for the fast Fourier transform time series of each input signal and output signal. For that peak frequency, calculate the average of the power peak value of the corresponding frequency 62 of the Fourier transform for each time series that has been obtained, and the average of the cross spectrum of the corresponding frequency, and use these averages. Find the transfer characteristics. In this way, averaging and division processing is performed only on peak values, and low-level frequency components with many low error components are ignored? Flj constant accuracy has been improved. The input signal is a sine wave signal, and its spectrum usually only appears at a single frequency, and as described above, highly accurate measurement can be obtained by processing only the frequency component of the peak value.

<Example> FIG. 3 shows an example of a transfer function measuring device according to the present invention.

A sine wave signal from the swept sine wave signal generator 11 is input to the object to be measured 12 . The input signal is sampled at regular intervals, and each sampled value is converted into a digital signal by the AD converter 21 and written into the input buffer memory 22. Furthermore, sample values of the output signal of the object to be measured 12 are converted into digital signals at regular intervals by an AD converter 23, and the digital signals are taken into an output buffer memory 24. Readout output side N''ll of buffer memory 22.24: connected to bus 26 via multiplexer 25. Central processing unit so-called CPU2
7. A read-only memory 28 in which a program for overall control is stored, an FFT processor 29 for performing fast Fourier transform operations, a transform memory 31 in which fast Fourier transform outputs are stored, and an average value memory 32 in which average values are stored.
etc. are connected to the bus 26.

The CPU 27 reads out the program in the memory 28, decodes and executes it, and performs control processing, and this processing is as follows. That is, the CPU 27 controls the multiplexer 25 and first the time series of, for example, n sample values up to that point is taken in from the input buffer memory 22 to the CPU 27.
The imported time series is now processed by the FFT processor 29
As described in FIG. 2, the input loss vector a is
l-1-jat' ~ an+jan' are calculated, and furthermore, the Shibaworth vectors A1 ~ An are calculated,
The input waste vector area 31a and the input waste vector storage area 3 in the conversion memory 31 are processed via the CPU 27.
1b respectively. Next, CPU U27's ill
ill, a time series consisting of n sample values corresponding to the input time series is transferred from the output buffer memory 24 to C.
The signal is taken into the PU 27 and further subjected to fast Fourier transform by the FFT processor 29. As a result, output waste vectors t)+jbl' to b2+jb2' and their power spectra B1 to Hn are obtained, respectively, and these are also stored in the output waste vector area 31C and the output cover worth vector 31d of the conversion memory 31, respectively.

In this way, one time series of the input signal and one time series of the output signal obtained at the same time are fast Fourier-transformed and the transformation memory 31
After storing each converted frequency component F1 to l'i'n
The cross spectrum c+-4-jct is calculated by calculating the product of the input waste vector and the output waste vector for the corresponding one.
'~cn+jcn' is calculated and stored in the cross spectral region 31e in the conversion memory 31. Next, the input cover-worth vector AI-An in the conversion memory 31 is sequentially read by the CPU 27, and the frequency of the peak value and its peak value are detected. It is determined whether the peak value is above a predetermined level, and if it is above the predetermined level, the peak value at that peak frequency is determined.
The average with the peak values obtained for the time series up to this point is performed, and the average result is stored in the corresponding frequency portion of the average value memory 32. The average calculation is performed using the following formula, where the previous average value is 5, the newly obtained peak value is Xe, and the number of averages so far is subtracted by 1.

Eyebrow←Xo 1 k This value is stored as a new average value in the corresponding frequency of the average value memory 32. The initial value of tasoshi l( is 2
, and while the peak values of the same frequency are obtained continuously, the l( is sequentially +1).In this way, for example, as shown in FIG.
The average value A1 of the input Coverworth vector for n, the average value TE of the output Coverworth vector, and the average value "+r1°" of the cross spectrum are calculated respectively. In performing this averaging, k is a preset value. When it comes to the average frequency for the Middle East.

In this way, the input signal frequency is swept, the input signal and the output signal are sampled one after another, and fast Fourier transform is performed to find the frequency of the Pinnick value for each time series.
The average value of that frequency up to that point is calculated, and when the sweep within the target frequency range is completed, the average value of the input coverworth vector, the average value of the output coverworth vector, and the cross From the average value of the spectrum, calculate the amplitude (gain) characteristic by dividing the average value of the cross spectrum by the average value of the input coverworth vector for each frequency component, and then calculate the real number of the average value of the cross spectrum. The phase characteristics are determined from the part and the imaginary part. Note that, if necessary, the correlation is calculated for each frequency component from the input coverworth vector and the output coverworth vector.

[
2. The average of the input coverworth vector and cross spectrum obtained for each time series up to the peak frequency is calculated. In other words, the average is not calculated for all frequency components of the power spectrum. Correct measurement results are obtained by ignoring components that may contain large errors at small levels. Even if average processing is performed only on the peak frequency in this way, it is only necessary to obtain the characteristics for each frequency component, and measurement is also performed by using a sine wave signal as an input signal and sweeping the frequency, but it is not possible to perform instantaneous measurement. Only one frequency component is input, and the characteristics need only be obtained by averaging only that component, which enables highly accurate measurement. Moreover, since the influence of non-linear coupling such as distortion is HIKO<, averaging is performed only for the frequency components.

[Brief explanation of drawings]

Fig. 1 is a block diagram showing a transfer function measuring device using analog signals, Fig. 2 is a diagram showing data used to measure a transfer function by digital processing, and Fig. 3 is an example of a transfer function measuring device according to the present invention. 41.1 is a block diagram showing an example of storage in the average value memory 32. 11: Sweeping sine wave signal generator, 12: '#Measurement object, 21
．． 23: AD converter, 22, 24: Buffer memory, 25: Multiplexer, 27: CPU 28 Niprogram storage memory, 29: FFT processor, 31: Conversion memo 'J, 32: Average value memory. Patent applicant Takeda Riken Kogyo Co., Ltd. Agent Kusano
Zhuo 7t=1 Decoy 2 4 Sha 2 Drawing procedure amendment (voluntary) Commissioner of the Japan Patent Office November 8, 1980 Indication of Tono 1 case Patent application 167371/1971, Title of invention 1 Correction of Yundatsu function measuring instrument 3 Relationship with Person @S Matter Patent applicant Takeda Riken Kogyo Co., Ltd. 5, subject of amendment Detailed explanation of the invention in the specification Column 6 Contents of the sleeve correction (]) Specification page 3, line 13, page 7, 16~ 17-line 1-input dregs vector'' is corrected to ``complex conjugate of human dregs vector'' as described above'' to ``The spectrum of its linear response comes out only at this input frequency C2.Therefore, as described above''. (3) In the same book, page 8, line 10, "average number of times k" is corrected to "average number of times -1". (4) The same book, page 8, line 15, ``The initial value is 2.'' [The initial value is 1," corrected. (5) The same book, page 8, line 19, “average value λ,” is replaced with “average value center,”
I am corrected. (6) Lines 1 and 2 of page 9 of the same book, "Average value B, ... are calculated respectively." 1 average value B1, and cross space - 〒 ctor average value C; + J C+, respectively. It will be played W. ” he corrected. (7) In lines 9-10 of the same book, ``obtained'' in ``l'' is corrected to ``obtained up to that point in the swept frequency range.'' (8) Ibid., p. 9, lines 15-19 [amplitude (gain) of 4
1 is calculated and... is calculated. '' to ``obtain the amplitude (gain) characteristics and phase characteristics.It is necessary to calculate the opening speed function for each frequency component from the average value of the human Coverworth vector and the output Coverworth vector cross spectrum. It will be corrected.” (9) Correct "the frequency component" in line 16 of page io of the same book to "the sweep frequency component". that's all

Claims (1)

[Claims] (1) Means for fast Fourier transforming an input signal supplied to an object to be measured, means for fast Fourier transforming an output signal of the object to be measured, and each of the input signal spectrum and output signal spectrum subjected to the fast Fourier transform. means for multiplying corresponding frequency components to obtain a cross spectrum;
Means for detecting the peak power frequency and its peak value in the converted spectrum of the input signal, and an average of the power spectrum of the input signal at the peak frequencies detected so far for each sample time series of the input signal. and means for calculating the average of the power spectrum of the input peak frequency and the average of the cross spectrum, and means for calculating the transfer characteristic of the object to be measured from the average of the power spectrum of the input peak frequency and the average of the cross spectrum. vessel.