JP3381109B2 - Transfer function measurement device - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a transfer function measuring device for obtaining a transfer function of an object to be measured from an input signal supplied to the object to be measured and an output signal from the object to be measured. The present invention relates to a measuring device for measuring a transfer function.

[0002]

2. Description of the Related Art Conventionally, data converted by Fourier transform are linearly arranged at equal intervals on a frequency. When this data is viewed on the logarithmic frequency axis, if the number of points of the Fourier transform is small, the data intervals are vacant at low frequencies and accuracy cannot be obtained.

Therefore, in order to increase the frequency resolution at a low frequency on the logarithmic frequency axis, the number of points of the fast Fourier transform is increased until the desired frequency resolution is satisfied, or the transfer function is measured, or , The band is divided, sampling is performed at different sampling frequencies for each band, and the transfer function for each band is measured by performing fast Fourier transform with a small number of points.
There is a method of expressing this measurement result on one logarithmic frequency axis.

As a method of dividing the band and increasing the frequency resolution at a low frequency without increasing the number of points of the fast Fourier transform, two-channel fast Fourier transform (FFT) set at different sampling frequencies is used.
The analyzers are prepared by the number of the divided bands, the transfer characteristic data is measured by each FFT analyzer, and this transfer characteristic data is used, for example, in GPIB (General Purpose Inter
There is a method in which face bus) is used as an interface and is taken into a computer, and each transfer characteristic data is displayed on one logarithmic frequency axis in this computer.

[0005]

By the way, in the above-mentioned method, if the number of points of the fast Fourier transform is increased until the desired frequency resolution is satisfied, the processing time is significantly increased and the hardware is also larger. You need a configuration.

When performing band division and fast Fourier transform for each band using the above-mentioned conventional FFT analyzer, a 2-channel FFT analyzer for each band is provided in addition to the computer for displaying the measurement result. Required,
A system equipped with these devices becomes large-scale.

In view of the above-mentioned circumstances, the present invention provides a transfer function measuring device capable of reducing the hardware configuration and processing time.

[0008]

A transfer function measuring device according to the present invention uses a transfer function of an object to be measured by using an input signal input to the object to be measured and an output signal from the object to be measured to which the input signal is input. In the transfer function measuring device for calculating, a band to be divided into a plurality of bands by passing the input signal and the output signal through a low-pass filter, and repeatedly performing downsampling processing for thinning out the signal data obtained from the low-pass filter. Dividing means, orthogonal transformation means for performing orthogonal transformation on the input signal and output signal for each band from the band dividing means, and the input using the spectrum of the input signal and output signal for each band from the orthogonal transformation means From the power spectrum calculating means for calculating the power spectrum of the signal and the output signal for each band, and the power spectrum calculating means And a transfer characteristic calculating means for calculating the transfer characteristics of the measured object for each band using the power spectra of the input signal and the output signal, and combining the transfer characteristics for each band from the transfer characteristic calculating means to measure the measured object. The above problem is solved by obtaining the transfer characteristics of an object.

Further, display means for displaying the above-mentioned transfer characteristic as one graph can be provided.
Further, the low-pass filter in the band dividing means is a finite length impulse response (FIR) filter, and the finite length impulse response (FIR) filter for each band has the same tap number and coefficient value. You can Furthermore, the low-pass filter in the band dividing means is 1 / of the operating clock frequency.
It is possible to cut off components having a frequency of 4 or more and pass at least components having a frequency of ⅛ or less of the operating clock frequency. Further, the power spectrum calculating means is characterized in that the spectrum of the input signal from the orthogonal transforming means is multiplied by the spectrum of the output signal from the orthogonal transforming means to calculate a cross spectrum for each band.

Further, the orthogonal transform processing in the orthogonal transform means is repeated a plurality of times, and the power spectrum calculation means uses the spectra of the plurality of input signals and output signals from the orthogonal transform means to obtain a plurality of input signals and The power spectrum of the output signal and the cross spectrum are calculated, and the average value of the power spectrum of the input signal and the output signal and the average value of the cross spectrum are obtained by using the power spectra of the plurality of input signals and the output signal and the cross spectrum. The transfer characteristic calculating means calculates the transfer characteristic of the device under test using the average value of the power spectrum of the input signal and the average value of the cross spectrum.

Further, there is a coherence calculating means for calculating a coherence value using the average value of the power spectra of the input signal and the output signal and the average value of the cross spectrum, and the discriminating means has the power of the input signal. It is characterized in that it is judged whether or not the level of the average value of the spectrum and the value of the coherence from the coherence calculating means are above a certain value.

[0012]

In the present invention, the input signal and the output signal are divided into a plurality of bands by repeatedly performing the down-sampling process of passing the signal data through the low pass filter and thinning out the signal data, and the input signal and the output signal are orthogonal to each other. Obtaining the transfer characteristics of the DUT after performing the conversion to obtain the spectrum of the input signal and the output signal, calculating the power spectrum of the input signal and the output signal for each band using the spectrum of the input signal and the output signal. By this, since the same number of points are orthogonally transformed in each band, the processing operation at the time of measuring the transfer function of the DUT is simplified.

Here, the power spectrum calculating means calculates the cross spectrum for each band by multiplying the spectrum of the input signal from the converting means by the spectrum of the output signal from the converting means, thereby calculating the transfer function. Can be measured with high accuracy.

Further, the orthogonal transform processing in the orthogonal transform means is repeated a plurality of times, and the power spectrum calculation means uses the spectra of the plurality of input signals and output signals from the orthogonal transform means to obtain a plurality of input signals and The power spectrum of the output signal and the cross spectrum are calculated, and the average value of the power spectrum of the input signal and the output signal and the average value of the cross spectrum are obtained by using the power spectra of the plurality of input signals and the output signal and the cross spectrum. The transfer characteristic calculation means calculates the transfer characteristic of the object to be measured using the average value of the power spectrum of the input signal and the average value of the cross spectrum, thereby improving the reliability of the transfer function to be obtained. it can.

Further, there is a coherence calculating means for calculating a coherence value using the average value of the power spectra of the input signal and the output signal and the average value of the cross spectrum, and the discriminating means has the power of the input signal. By determining whether or not the level of the average value of the spectrum and the value of the coherence from the coherence calculating means are above a certain value, it is possible to obtain a highly reliable measurement result as a transfer function.

[0016]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT A preferred embodiment of the present invention will be described below with reference to the drawings. FIG. 1 shows the configuration of a system using the transfer function measuring apparatus according to the present invention.

The information processing apparatus 63 such as the personal computer shown in FIG. 1 converts the input audio signal into a digital signal for each of the analog audio input terminals for two channels of ch1 and ch2 and each of the analog audio input terminals of these two channels. An A / D (analog / digital) converter for conversion is provided. Further, a display device 64 such as a CRT (cathode ray tube) monitor for displaying the measurement result of the transfer function is connected to the information processing device 63.

The input signal input from the input terminal 61 is
It is supplied to the DUT 62 and is also input to the analog audio input terminal indicated by ch1 of the information processing device 63. The output signal from the DUT 62 is the information processing device 63.
It is input to the analog audio input terminal indicated by ch2 and output from the output signal terminal 65.

In the information processing device 63, the input 2
The transfer function is calculated using the analog audio signal of the channel, and characteristics such as amplitude and phase are displayed on the connected display device 64.

FIG. 2 shows signals used in the transfer function measuring apparatus. For example, the input signal x (t) is supplied to the DUT 1 and is output to the input terminal 2 as a signal of the input IN _1. To be done. An output signal y corresponding to the input signal x (t) output from the device under test 1 is output.
(T) is output to the input terminal 3 as a signal of the input IN ₂ . The transfer function measuring device measures the transfer function of the DUT using the input signal x (t) and the output signal y (t) output to the input terminals 2 and 3.

Next, a schematic structure of the transfer function measuring device is shown in FIG. 3 and will be described below.

Here, the number of samples of the voice signal necessary for measuring the transfer function is the fast Fourier transform (FFT) described later.
Is calculated using the number of points, the number of bands to be divided, and the number of times of averaging the power spectrum and cross spectrum described later. For example, when the number of samples required to measure the transfer function is S, the number of points of the fast Fourier transform is P, the number of bands to be divided is M, and the number of times of averaging is N, it is necessary to measure the transfer function. The number of samples S is expressed by the following equation (1).

S = P × 2 ^M × N (1) Specifically, the number of points P of the fast Fourier transform is 51.
When the number of divided bands M is 7 and the number of times of averaging N is 10, the number of samples S has a value of 512 × 2 ⁷ × 10 = 655360.

As shown in FIG. 2, the analog audio signals of the input signal and the output signal input to the input terminals 2 and 3 of the two channels are digitally converted, respectively, and then the band division processing circuits 10 and 20 of FIG. To the input IN ₁ and the input IN ₂ respectively. Here, considering the case where the signal input IN ₂ to the input IN ₁ of the signal is delayed beyond the point number in the range of the fast Fourier transform, the head of the intact signal data as a signal input IN ₂ From the beginning of the signal data for the correction as an input IN ₁ signal.

The band division processing circuits 10 and 20 perform the band division processing by repeatedly performing the downsampling processing for the sampling frequency FS of the input signal for the required number of bands. Specifically, the data to be down-sampled is passed through a low-pass filter, which is a linear phase finite impulse response (FIR) filter, and its output is extracted so as to extract one data from two data. Downsampling processing is performed by thinning out.

Here, the characteristic of the ideal low-pass filter used in the down-sampling process is as shown in FIG. 4 when the operating clock frequency of the filter is f _CL , which is realized by the linear phase FIR filter. In that case, the number of taps of the coefficient of the FIR filter becomes enormous, and the processing time cannot be shortened. Therefore, in order to realize a low-pass filter using an FIR filter with a small number of taps, the operating clock frequency of the filter is f
When _CL, as shown in FIG. 5, flat up to a frequency band f _CL / 8, the transient region of _{f CL / 8~f CL / 4,} f CL / 4~f
_Use an FIR filter with a characteristic that _CL / 2 is the stop band. Then, the data of f _CL / 8 or less is down-sampled and then used as data.

When the number of taps of the coefficient of the linear phase FIR filter is 28 and the values of the respective taps are shown in Table 1, the amplitude characteristics of the 28 tap linear phase FIR filter are as shown in FIG. Become.

[0028]

[Table 1]

FIG. 7 concretely shows the measurement section of each band when the number of bands to be divided is 7, and the measurement section of the band D ₁ is f when the original sampling frequency is f _{s. s} / _{2~f s} / 8, the measurement interval of the band D ₂ is f _s
/ 8~f _s / 16, the measurement interval of the band D ₃ is f _s /. 16 to
f _s / 32, the measurement interval of the band D ₄ is f _s / 32~f _s /
64, the measurement section of the band D ₅ is f _s / 64 to f _s / 12
8, the measurement section of the band D ₆ is f _s / 128 to f _s / 25
6, the measurement section of the band D ₇ is f _s / 256-0.

As described above, the number of bands is set to 7, and FF
FIG. 8 shows a specific downsampling processing procedure when the number of points of T is 1024.

The FFTs 50-56 in FIG.
This corresponds to the FFT analyzer 14 or the FFT analyzer 24 in FIG. First, 1024 × 64 pieces of time series data are sent to the FFT 50 and also input to the downsampling processing circuit 41. This FFT50 is
The continuous 1024 samples at arbitrary positions are taken out of the 1024 × 64 time series data and subjected to the 1024-point fast Fourier transform process. F shown below
Similar to the FFT 50, the FTs 51, 52, 53, 54 and 55 also take out 1024 consecutive samples at arbitrary positions in the input time series data and apply the 1024 point fast Fourier transform process.

The FIR filter 41a in the down-sampling processing circuit 41 has an operating clock frequency f _CL = f _s.
The 1024 × 64 input time series data are passed through the FIR filter 41a and then thinned by the thinning processing circuit 41b to set the clock frequency to f _s / 2, and the down sampling processing is performed. Circuit 4
1 outputs 1024 × 32 time series data. This time series data is sent to the FFT 51 and
It is input to the downsampling processing circuit 42.

In this down-sampling processing circuit 42, the input 1024 × 32 time series data is passed through an FIR filter 42a which operates at an operating clock frequency f _CL = f _s / 2, and then decimated by a decimation processing circuit 42b. By performing the processing, the clock frequency is set to f _s / 4, and the down sampling processing circuit 42 outputs 1024
× 16 pieces of time series data are output. This time-series data is sent to the FFT 52 and also to the downsampling processing circuit 43.

In this down-sampling processing circuit 43, the input 1024 × 16 time-series data is passed through an FIR filter 43a which operates at an operating clock frequency f _CL = f _s / 4 and then decimated by a decimation processing circuit 43b. By performing the processing, the clock frequency is set to f _s / 8, and 1024 × 8 time series data is output. This time series data is sent to the FFT 53 and also to the downsampling processing circuit 44.

In this down-sampling processing circuit 44, the input 1024 × 8 time series data is passed through an FIR filter 44a operating at an operating clock frequency f _CL = f _s / 8 and then decimated by a decimation processing circuit 44b. By performing the processing, the clock frequency is set to f _s / 16, and 1024 × 4 time series data is output. This time-series data is sent to the FFT 54 and also to the downsampling processing circuit 45.

The thinning at a thinning processing circuit 45b after being passed through the FIR filter 45a in this down-sampling processing circuit 45, which input 1024 × 4 pieces of time-series data is to operate at an operating clock frequency f _CL = f _s / 16 The clock frequency is f _s / 32 due to the processing.
Then, 1024 × 2 time series data are output.
This time-series data is sent to the FFT 55 and also to the downsampling processing circuit 46.

In this down-sampling processing circuit 46, the input 1024 × 2 time-series data is passed through an FIR filter 46a operating at an operating clock frequency f _CL = f _s / 32, and then thinned out by a thinning-out processing circuit 46b. The clock frequency is f _s / 64 due to the processing.
Then, 1024 time series data are output. This time series data is sent to the FFT 56. This FFT56
Then, fast Fourier transform processing is applied to 1024 samples of frequency f _s / 64.

In this way, the sampling frequency of the signal from the input IN ₁ and the sampling frequency of the signal from the input IN ₂ are each divided into seven bands, and the data corresponding to each band is extracted.

The signal of the input IN ₁ divided for each band is switched and selected by the signal switcher 11 for each band and output to the multiplier 13, and the signal switched and selected by the signal switcher 11 is selected. The signal in the band corresponding to the band is selected by the signal switch 21 and output to the multiplier 23. The signal switchers 11 and 21 can switch the number of times N to be averaged.

Here, the band division processing circuits 10 and 20.
In, the voice waveform of a predetermined number of samples sampled in a predetermined cutout section is cut out, and at the time of cutout of this voice waveform, not to cause a sudden change at both ends of the cutout section, In the spectral domain, it is necessary to multiply the original waveform by the time window in order to perform the convolution of the Fourier transform of the window function on the spectrum of the signal, that is, the weighted moving average. Therefore, the window function generators 12 and 22 generate window functions corresponding to the bands of the signals sent to the multipliers 13 and 23, and the multipliers 13,
23. Thereby, the multipliers 13 and 23 are
Then, the signal for each band is multiplied by the window function, and the multiplied signals are sent to the FFT analyzers 14 and 24, respectively.

In the FFT analyzers 14 and 24, the frequency spectrum of the signal at the input IN ₁ and the input IN ₁ are obtained by performing the fast Fourier transform processing on the sent signal data.
The frequency spectrum of the _second signal is determined.

When the complex data of the spectrum of the input IN ₁ is X (k), this complex data X (k) is sent to the multiplier 16 and the complex conjugate converter 15. In the complex conjugate converter 15, the sent complex data X (k) is converted into complex conjugate data X ^* (k) and sent to the multiplier 16. In the multiplier 16, the complex data from the FFT analyzer 14 is sent. X (k) and the complex conjugate data X
^* (k) is multiplied to obtain the power spectrum X ^* (k) X (k) of the input IN ₁ , and this power spectrum X ^* (k) X (k) is stored in the register 31a.

Similarly, assuming that the complex data of the spectrum of the input IN ₂ is Y (k), this complex data Y
(K) is sent to the multiplier 26 and the complex conjugate converter 25. In the complex conjugate converter 25, the sent complex data Y (k) is converted into complex conjugate data Y ^* (k) and sent to the multiplier 26.
The complex data Y (k) from the FT analyzer 24 and the complex conjugate data Y ^* (k) are multiplied to obtain the power spectrum Y ^* (k) Y (k) of the input IN ₂ , and this power spectrum Y ^* (k) Y (k) is stored in the register 31c.

Further, the complex conjugate data X ^* (of the spectrum of the input IN ₁ output from the complex conjugate converter 15)
k) and the input IN ₂ output from the FFT analyzer 24
The complex data Y (k) of the spectrum is multiplied by the multiplier 17 to obtain the cross spectrum X ^* (k) Y (k). The cross spectrum X ^* (k) Y (k) is stored in the register 31b. Remembered.

Here, the registers 31a, 31b, 3
1c is composed of N pieces corresponding to the averaging times N, and uses the data output from the band processing circuits 10 and 20 by switching the signal switchers 11 and 21 N times respectively. A power spectrum X ^* (k) X (k) of N input IN ₁ , a power spectrum Y ^* (k) Y (k) of input IN ₂ , and a cross spectrum X ^* (k) Y (k), respectively. It is something to remember.

After that, the registers 31a, 31b, 3
The power spectrum X ^* (k) X (k) of the N input IN ₁ respectively stored in 1c, the power spectrum Y ^* (k) Y (k) of the input IN ₂ , and the cross spectrum X ^*.
(k) Y (k) is an averaging circuit 32a, 32b, 32
In c, the average value is obtained.

Power spectrum of the input IN ₁ X ^* (
k) The average value of X (k) is P ₁ (k), and the average value of the power spectrum Y ^* (k) Y (k) of the input IN ₂ is P
₂ (k) and the average value of the cross spectrum X ^* (k) Y (k) is C (k), the average value P ₁ (k) of the power spectrum of the input IN ₁ is expressed by the equation (2), The average value P ₂ (k) of the power spectrum of the input IN ₂ is represented by the equation (3), and the average value C (k) of the cross spectrum is represented by the equation (4).

[0048]

[Equation 1]

[0049]

[Equation 2]

[0050]

[Equation 3]

Further, the average value of the power spectrum of the input IN ₁ , the average value of the power spectrum of the input IN ₂ ,
And the average value of the cross spectrum is, as shown in Table 2,
It is calculated for each frequency section used for measurement of each band.

[0052]

[Table 2]

A in each band D represents a ratio to the original down-sampling frequency f _s , and B represents an actually measured frequency. For example, in the band D ₂ , the range 1 shown in A is shown. / 8 to 1/4 represents the ratio to the down sampling frequency f _s / 2,
The frequency section of the actual measurement data is the range f _s / 1 shown in B.
6 to f _s / 8. Also, for example, in FIG.
When the frequency characteristic of the FIR filter 41a in the down sampling processing circuit 41 of FIG. 5 is as shown in FIG. 5, the operating clock frequency of the FIR filter 41a is f _s , and therefore, as shown in B of Table 2, the band D frequency range of the data that is actually taken as ₂ f _s / _{16~f s} / 8
Corresponds to the flat part of the frequency characteristic curve in FIG.

Average value P ₁ (k) of the power spectrum of the input IN ₁ and average value C (k) of the cross spectrum
Is sent to the transfer function calculator 35, and the transfer function H (k) of the object to be measured is calculated in this transfer function calculator 35. Specifically, the amplitude and phase are calculated as a transfer function. The value of this transfer function is output from the output terminal 39.

The transfer function H (k) including the amplitude and the phase is expressed by the following equation (5), and the transfer function H (k) of only the amplitude is given.
Is expressed by the following equation (6).

[0056]

[Equation 4]

[0057]

[Equation 5]

Further, the average value P ₁ (k) of the power spectrum of the input IN ₁ , the average value P ₂ (k) of the power spectrum of the input IN ₂ , and the average value C of the cross spectrum.
(K) is sent to the coherence calculator 34.

The average value C (k) of the cross spectrum obtained by the averaging circuit 32b is sent to the complex conjugate converter 33 and the complex conjugate data C ^* of the average value C (k) of the cross spectrum ^. (k) is obtained, and the complex conjugate data C ^* (of the average value C (k) of this cross spectrum is calculated.
k) is also sent to the coherence calculator 34.

In the coherence calculator 34, the average value P ₁ (k) of the power spectrum of the input IN ₁ , the average value P ₂ (k) of the power spectrum of the input IN ₂ , and the average value C (k) of the cross spectrum. , And the complex conjugate data C ^* (k) of the average value C (k) of the cross spectrum,
Coherence, which is a property of waves that interfere with each other, is called coherence. When this coherence is r, the coherence r is expressed by the following equation (7).

[0061]

[Equation 6]

The values of the transfer function and the coherence are also calculated for each frequency section used in the measurement of each band, as shown in Table 2.

As described above, the processing time can be reduced by calculating each spectrum, transfer function, and coherence only in the section used for measurement in each band.

Then, on a display device such as a CRT monitor,
Data is displayed for each band shown in FIG. 7 as one frequency characteristic.

Here, at each data point, only the transfer characteristic of the frequency component when the level of the average value of the power spectrum of the signal of the input IN ₁ and the value of the coherence are above a certain value is displayed as the measurement result. Updated data. For example, the comparator 36 determines whether the coherence value is 0.8 or more, and if the coherence value is 0.8 or less, the coherence value is output to the output terminal 3
7, the display data is left as it is, and if it is 0.8 or more, the coherence value is output from the output terminal 38 to update the display data.

If the transfer function measurement is further continued, the signals of the input IN ₁ and the input IN ₂ are captured again, and the above calculation process is repeated.

Next, the transfer function measuring procedure will be described below with reference to the flow chart of the measuring procedure in the transfer function measuring apparatus shown in FIG.

First, in step S1, the analog voice signals from the input IN ₁ and the input IN ₂ are fetched and digitally converted, and the number of samples required for transfer function measurement is fetched from this digital signal. The extracted digital signal is subjected to band division processing in step S2 by repeatedly performing downsampling processing according to the number of necessary bands. For example, down-sampling processing is performed 6 times when dividing into 7 bands. The following steps S3 to S11 are performed for each of the divided bands.

[0069] First, in step S3, multiplied by the window function to the signal input IN ₁ of one band, determine the frequency spectrum of the input IN ₁ of the signal by performing FFT processing on this output, it is determined in step S4, The signal of the input IN ₂ is multiplied by the window function, and the output is subjected to FFT processing to obtain the frequency spectrum of the signal of the input IN ₂ . And step S
In step 5, the power spectrum is calculated using the spectrum of the signal of the input IN ₁ , and in step S6, the input IN ₂
A power spectrum is calculated using the spectrum of the signal of, and further, in step S7, the calculated input IN
_A cross spectrum is calculated using the power spectrum of _{1 and} the power spectrum of the input IN ₂ .

Further, the operations of the above steps S3 to S7 are performed for the number of times of averaging, and the input IN ₁ and the input IN are input.
The power spectrum of ₂ and the value of the cross spectrum are calculated only for the number of times of averaging.

[0071] Thereafter, in step S9, by using the the obtained power spectrum of the input IN _1, the power spectrum of the input IN _2, and the value of the number of times of averaging obtained respectively for the cross spectrum, input IN ₁
Power spectrum of input, power spectrum of input IN ₂ ,
And the average value of the cross spectrum is calculated.

Then, in step S10, the coherence value is obtained using the average value of the power spectrum of the input IN ₁ , the power spectrum of the input IN ₂ , and the cross spectrum. Further, the transfer function is calculated using the average value of the power spectrum and the cross spectrum of the input IN ₁ .

In this way, the processes of steps S3 to S11 are performed for each band, and the coherence value and the transfer function for each band are obtained.

Further, in step S12, the coherence value and the transfer function obtained above are displayed on the display device as, for example, one graph.

In step S13, it is determined whether or not the coherence value is a certain value or more, specifically, 0.8 or more at each data point,
If it is 0.8 or less, the value on the displayed graph is not updated. If it is 0.8 or more, the value on the displayed graph is updated and displayed for each point. .

[0076]

As is apparent from the above description, the transfer function measuring apparatus according to the present invention uses the input signal input to the DUT and the output signal from the DUT to which the input signal is input. In the transfer function measuring device for calculating the transfer function of the device under test, the input signal and the output signal are passed through a low-pass filter, and the down-sampling process for thinning out the signal data obtained from the low-pass filter is repeatedly performed to obtain a plurality of signals. A band dividing means for dividing the input signal and the output signal for each band from the band dividing means, and an orthogonal transform means for performing an orthogonal transform on the input signal and the output signal for each band from the band dividing means, and the input signal and the output signal for each band from the orthogonal transform means. A power spectrum calculating means for calculating the power spectrum of the input signal and the output signal for each band using the spectrum; And a transfer characteristic calculating means for calculating the transfer characteristic of the DUT for each band using the power spectra of the input signal and the output signal from the torque calculating means, and the transfer characteristic for each band from the transfer characteristic calculating means By synthesizing to obtain the transfer characteristic of the DUT, the orthogonal transformation with the same number of points is performed in each band, which simplifies the processing operation when measuring the transfer function of the DUT, resulting in high accuracy. The transfer function can be obtained. Further, since the band division processing and the calculation processing of each spectrum are efficiently performed, the processing time can be shortened and the hardware circuit configuration can be reduced.

Here, in the power spectrum calculation means, the spectrum of the input signal from the orthogonal transformation means is
The transfer function can be measured with high accuracy by multiplying the spectrum of the output signal from the orthogonal transformation means to calculate the cross spectrum for each band.

Further, the orthogonal transform processing in the orthogonal transform means is repeated a plurality of times, and the power spectrum calculating means uses the spectra of the plurality of input signals and output signals from the orthogonal transform means to obtain a plurality of input signals and The power spectrum of the output signal and the cross spectrum are calculated, and the average value of the power spectrum of the input signal and the output signal and the average value of the cross spectrum are obtained by using the power spectra of the plurality of input signals and the output signal and the cross spectrum. The transfer characteristic calculation means calculates the transfer characteristic of the object to be measured using the average value of the power spectrum of the input signal and the average value of the cross spectrum, thereby improving the reliability of the transfer function to be obtained. it can.

Further, there is a coherence calculating means for calculating a coherence value using the average value of the power spectra of the input signal and the output signal and the average value of the cross spectrum, and the discriminating means has the power of the input signal. By determining whether or not the level of the average value of the spectrum and the coherence value from the coherence calculating means are above a certain value, it is possible to obtain a highly reliable measurement result as a transfer function.

[Brief description of drawings]

FIG. 1 is a diagram showing a schematic configuration of a system using a transfer function measuring device according to the present invention.

FIG. 2 is a diagram for explaining a transfer function measurement signal.

FIG. 3 is a diagram showing a schematic configuration of a transfer function measuring device.

FIG. 4 is a diagram showing ideal low-pass filter characteristics.

FIG. 5 is a diagram showing a low-pass filter characteristic.

FIG. 6 is a diagram showing amplitude characteristics of a 28-tap linear phase FIR filter.

FIG. 7 is a diagram showing a measurement section of each band.

FIG. 8 is a diagram for explaining downsampling processing.

FIG. 9 is a flowchart of a procedure for measuring a transfer function.

[Explanation of symbols]

10, 20 Band division processing circuit 11, 21 Signal switch 12, 22 Window function generator 14, 24 FFT analyzer 15, 25, 33 Complex conjugate converter 31a, 31b, 31c registers 32a, 32b, 32c averaging circuit 34 Coherence calculator 35 Transfer function calculator 36 Comparator

─────────────────────────────────────────────────── ─── Continuation of front page (58) Fields surveyed (Int.Cl. ⁷ , DB name) G01R 27/28 G01R 23/16

Claims (8)

(57) [Claims] 1. A transfer function measuring device for calculating a transfer function of an object to be measured using an input signal input to the object to be measured and an output signal from the object to be measured which has received the input signal, wherein the input signal and the output are The signal is passed through a low-pass filter, and the down-sampling process for thinning out the signal data obtained from this low-pass filter is repeatedly performed to divide the signal into a plurality of bands, and an input for each band from the band dividing means. A power for calculating the power spectrum of the input signal and the output signal for each band using the orthogonal transforming means for performing the orthogonal transform on the signal and the output signal, and the spectrum of the input signal and the output signal for each band from the orthogonal transforming means. The spectrum calculation means and the power spectrum of the input signal and the output signal from the power spectrum calculation means are used. The transfer characteristic of the object to be measured and a transfer characteristic calculation means for calculating for each band, by combining the transfer characteristic of each band from the transfer characteristic calculation means Te
A transfer function measuring device characterized in that the transfer characteristic of an object to be measured is obtained . 2. The transfer characteristic is further made into one graph.
2. The transfer function measuring device according to claim 1, further comprising display means for displaying the transfer function. 3. A low pass filter in the band dividing means.
Is a finite-length impulse response (FIR) filter
Finite length impulse response (FIR) filter for each band.
Ruta is characterized by the same tap number and coefficient value
The transfer function measuring device according to claim 1 . 4. A low pass filter in the band dividing means.
Is equal to or higher than 1/4 of the operating clock frequency.
Of at least the above operating clock
To pass components below 1/8 of the frequency
The transfer function measuring device according to claim 1, wherein
Place 5. The power spectrum calculation means further calculates a cross spectrum for each band by using each spectrum of the input signal and the output signal from the orthogonal transformation means. Transfer function measuring device. 6. The orthogonal transform processing in the orthogonal transform means is repeated a plurality of times, and the power spectrum calculation means uses a plurality of input signal and output signal spectra from the orthogonal transform means to generate a plurality of input signals and The power spectrum of the output signal and the cross spectrum are calculated, and the average value of the power spectrum of the input signal and the output signal and the average value of the cross spectrum are obtained by using the power spectra of the plurality of input signals and the output signal and the cross spectrum. 6. The transfer function measurement according to claim 5 , wherein the transfer characteristic calculation means calculates the transfer characteristic of the device under test using the average value of the power spectrum of the input signal and the average value of the cross spectrum. apparatus. 7. A coherence calculating means for calculating a coherence value using an average value of power spectra of the input signal and the output signal and an average value of the cross spectrum, and at least a coherence value from the coherence calculating means. 7. The transfer function measuring device according to claim 6 , further comprising a determining unit that determines whether or not the value is equal to or more than a certain value. 8. The transfer characteristic data for each band displayed by the display means is updated when the determining means determines that the value of the coherence from the coherence calculating means is at least a certain value or more. The transfer function measuring device according to claim 7 .