JPH0770952B2 - Octave multiple filter - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial application] The present invention relates to an octave multiplex filter configured by regularly combining digital filters having low-pass characteristics and band-pass characteristics.

[Outline of the Invention] The octave multiplex filter of the present invention includes, for example, a plurality of filter stages each including a pair of digital filters (FIR type filters) each having a bandpass characteristic and a lowpass characteristic, arranged between input and output in accordance with a predetermined rule. The octave multiplex filter that is set allows the characteristics to be specified at logarithmically uniform intervals over a wide band in the frequency domain.

[Prior Art] Since a conventional digital filter processes discrete sample values at equal time intervals, it operates at a predetermined clock frequency. Therefore, characteristics in the frequency domain are specified at equal frequency intervals. It was being appreciated.

For this reason, a conventional filter that specifies characteristics over a wide band at logarithmic intervals has been realized only by combining a plurality of analog filters.

[Problems to be Solved by the Invention] Since the former digital filter adopts the characteristic designating method at equal frequency intervals as described above, the low frequency resolution is too coarse, while the high frequency resolution is too fine. However, widening the band is very difficult.

Generally, the characteristics of audio equipment are evaluated or specified at logarithmic intervals in the frequency domain, so when using a conventional digital filter, such as a FIR type (finite impulse response type) filter, it is possible to obtain accurate characteristics in the low range for the reasons described above. However, in the high frequency range, the characteristics were specified with higher precision than necessary, which was extremely inefficient in realizing the circuit. In particular, when trying to realize a complicated frequency characteristic like a graphic equalizer with a conventional FIR filter, such a constraint cannot be escaped, and a graphic equalizer using a FIR filter has not yet been put to practical use. .

On the other hand, since the latter filter adopts the addition or multiplication method of the analog active filter, there is interference between the filters and it is difficult to specify the phase and the amplitude independently. In particular, it was very difficult to arbitrarily specify the phase characteristic.

SUMMARY OF THE INVENTION An object of the present invention is to provide an octave multiplex filter capable of designating characteristics at logarithmically uniform intervals over a wide band in the frequency domain in order to solve the problems of the prior art.

[Means for Solving the Problems] In order to achieve the above-mentioned object, the first invention of the present application has a bandpass characteristic and a lowpass characteristic which operate with a sampling time interval of an input signal as a clock at the time interval. A plurality of digital filter groups are provided, the low-pass characteristic digital filter groups are connected in series, and the band-pass characteristic digital filter groups are connected in parallel. The input signal is supplied to each of the band-pass characteristic digital filter groups. Given that each output is given to each input of the digital filter group of low-pass characteristic, if the numbers are assigned in the order of connection in each digital filter group, the unit delay element per stage of the maximum number digital filter If the delay time is longer than the sampling time interval of the input signal, In the digital filter group, the delay time per stage of the digital filter becomes A ^B (A> 1, B> 0) times as the number decreases by one, and the unit delay of the digital filter having the low-pass characteristic of a certain number. The gist is that the delay time per one stage of the element is an integral multiple of the delay time per one stage of the unit delay element of the digital filter having the band pass characteristic with the same number.

Further, in the second invention of the present application, while the first invention is the input parallel type, each output of the output parallel type, that is, the low pass characteristic digital filter is given to each digital filter of the band pass characteristic. The gist is that the respective outputs are added together and taken out.

[Operation] In the first invention of the present application described above, for example, an FIR type filter is used as the digital filter having the low-pass and band-pass characteristics, and the digital filter having the low-pass characteristic is an odd-numbered digital filter having the band-pass characteristic. It operates so as to eliminate the pass band, and each filter stage is configured by using both as a pair, and the unit delay time of the delay element constituting the digital filter of each stage is changed and used in multiple.

The second invention of the present application is basically equivalent to the first invention, and therefore its operation is also exactly the same.

BEST MODE FOR CARRYING OUT THE INVENTION The present invention will be described below with reference to the embodiments shown in the drawings.
FIG. 1 shows an implementation of an octave multiple filter (5 octaves) according to the first invention of the present application. In the figure, IN is an input terminal, OUT is an output terminal, A / D is an AD converter, D / A is a DA converter, LPF is an ordinary analog low-pass filter, and S _{0 to} S ₄ are, for example, Digital filter with band-pass characteristics composed of FIR type filters, F _{0 to} F ₄
Is a digital filter having a low-pass characteristic formed by using the same filter, and D _{0 to} D ₄ are adder circuits.

The FIR type filter is composed of a delay element SR, a coefficient multiplier ML and an adder AD as shown in FIG. 2, and is shown in FIG.
The coefficient a _{-5 to} a ₆ of the coefficient multiplier ML so that the target frequency characteristic can be obtained by the DFT (Discrete Fourier Transform) method.
Has been set.

FIG. 3 (a) shows the target frequency characteristic (low-pass characteristic).
, And FIG. 6B shows a function in the time domain obtained by performing an inverse Fourier transform on the function.

Each band (S ₀ , F ₀ ) of the digital filter having the band-pass characteristic and the low-pass characteristic having such a configuration
Each filter stage is composed of (S ₄ , F ₄ ), and is arranged in parallel between the input and the output.

In each of the filter stages, each of the digital filters S _{0 to} S ₄ having the band pass characteristic has a delay time nT per stage with the sampling time interval of the input signal in the AD converter A / D as a reference time interval T. has a delay element S ₀ -1~S ₄ -5 operating in one of the time intervals of the integral multiples of the time interval group T, 2T, 3T · · · for this reference time interval. The low-pass characteristic digital filters F _{0 to} F ₄ correspond to the delay time per stage, and the delay time of each stage of the delay elements of the band-pass characteristic digital filters S _{0 to} S ₄ corresponds to an integer fractional delay time. Working delay element
It has F ₀ -1 to F ₄ -5.

Further, each filter stage is coupled as shown via the adder circuits D ₀ -D ₄ . That is, if each filter stage has a bandpass characteristic or low-pass characteristic digital filter stage having a delay element that operates with a longer delay time, it is defined as an upper stage (in the order of S _{4 to} S ₀ , F _{4 to} F ₀ ). , The output of the digital filter of the band pass characteristic of each filter stage is added to the output of the digital filter of the low pass characteristic of one stage higher than that stage, and is connected so as to be input to the digital filter of the low pass characteristic of its own stage. There is.

Now, in the octave multiplex filter having the above-mentioned configuration, the input signal given to the input terminal IN is A / D converter A /
Sampling frequency f ₀ (sampling rate 1 / f
It is AD-converted in ₀ sec) and added to the digital filters S _{0 to} S ₄ having band pass characteristics. This filter, for example, S _4, has a delay time of 1 / f ₀ 2 ^-4 s per stage of delay elements that compose it.
ec. Normally, the digital filter of this configuration is used with a clock frequency of f ₀ 2 ^-4 Hz and a sampling frequency of the input signal of f ₀ 2 ^-4 Hz. for that reason,
Since the clock frequency and the sampling frequency are different, the intended processing cannot be performed as it is. Therefore, in the present invention, first, as described above, the unit delay time is set to 1 / f ₀ 2 ^-4 sec.
Then, the coefficient of the FIR type filter is determined, and then the clock frequency is set to f ₀ Hz for the input signal with the actual sampling interval of 1 / f ₀ sec, and the coefficient remains as it is for each stage of the delay element S ₄ -1. Set the delay time to be nT = 1 / f ₀ 2 ^-4 sec.
Similarly, the digital filters of the other stages are also set.

The output characteristic of the digital filter S ₄ is as shown in FIG. 4 (a). Next, since the odd-numbered pass band is eliminated, the delay time per delay element is 1 / f ₀ 2 ^-3 Input to digital filter F ₄ with low pass characteristic of sec. The output characteristic of F ₄ is as shown in FIG. 4 (b). The output of this F ₄ is the addition of the output of the digital filter S _{3 with} the bandpass characteristic that the delay time per delay element is 1 / f ₀ 2 ^-3 sec.
Added by D ₄ . FIG. 4 (c) shows this added characteristic.

Further, in order to cancel the odd-numbered pass band of S ₃ , the delay time per stage of the delay element is input to the digital filter F ₃ having a low-pass characteristic of 1 / f ₀ 2 ^-2 sec. The output characteristic of F ₃ is as shown in FIG. 4 (d).

The same process is repeated thereafter, and the final digital filter F ₀
As a result, the output having the desired characteristics as shown in FIG. 5 is obtained.

6 and 7 show respective embodiments of the octave multiplex filter according to the second and third inventions of the present application, respectively.
The embodiment of FIG. 6 is an input parallel type, that is, digital filters BP1 to BP5 having bandpass characteristics in which input signals are connected in parallel.
And the respective outputs are applied to the amplitude and phase coefficient units COE1 to COE
The digital filters LP1 to LP4 having a low-pass characteristic connected in series via E5 are provided, but the embodiment of FIG. 7 is an output parallel type, that is, a digital filter having a low-pass characteristic connected in series. The respective outputs of LP4 to LP1 are given to the digital filters BP5 to BP1 having the band-pass characteristics connected in parallel, and the respective outputs are added by the adder D5 via the amplitude / phase coefficient units COE5 to COE1 and taken out. It is like this.

In the embodiment shown in FIGS. 6 and 7, the delay time per stage of the digital filters LP4 and BP5 having the highest number in the connection order is equal to or longer than the sampling time interval of the input signal, and the respective digital filter groups LP1 to LP1 ... LP4, BP1 ~ BP5
The delay time of the delay unit per stage of the digital filter each time the number is decreased one at the ^{A B (A> 1, B} > 0)
The delay time per unit delay element of the digital filter having a certain number and low pass characteristic is one n (n is an integer) of the delay time per unit delay element of the digital filter having the same number. ) It is supposed to work at double.

The embodiment shown in FIGS. 6 and 7 is basically equivalent to the embodiment shown in FIG. 1 and operates in exactly the same manner. The embodiment shown in FIG. 6 is a digital filter group BP1 to BP5 having bandpass characteristics. Unnecessary band components of the bandpass output signal are sequentially deleted by the low pass characteristic digital filter groups LP1 to LP4, while the embodiment of FIG. 7 sequentially filters each of the filters LP1 to LP4 whose band is narrowed. Filter BP1 for low-pass signal output
~ The method of multiplying each band pass characteristic output of BP5 is adopted.

FIG. 8 shows the output characteristic of each digital filter in the case of A ^B = 2 ′, n = 2 as an example of FIG. 7.

In the embodiment shown in FIG. 6, the random access memory (RAM) can be shared as the delay element of the digital filter group having the band pass characteristic of the input stage, so that the memory is saved and the design is correspondingly simplified.

Further, in the embodiment of FIG. 7, the components of each band can be separated and taken out, which is suitable for the purpose of adding other functions.

[Effects of the Invention] As is apparent from the above description, according to the present invention, the frequency domain (on the frequency axis) is set by the above-described configuration.
The characteristics can be specified logarithmically at even intervals, and the resolution can be improved because the band can be widened in the low frequency direction. Moreover, the amplitude and phase can be specified independently, and the interference between filters can be arbitrarily reduced. it can.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention, FIG. 2 is a diagram showing an example of the structure of an FIR filter used in the embodiment, FIG. 3 is a characteristic diagram of the filter, and FIG. FIG. 5 is an output characteristic diagram of each digital filter in the above embodiment, FIG. 5 is a diagram showing the output of the target characteristic, FIGS. 6 and 7 are block diagrams showing another embodiment of the present invention, and FIG. It is an output characteristic view of each digital filter in the example. S _{0 to} S ₄ ...... Digital filter having band pass characteristic F _{0 to} F ₄ ...... Digital filter having low pass characteristic, D _{0 to} D ₅ ...... Adding circuit, A / D ...... A-D converter, D / A ... DA converter.

Claims (2)

[Claims] 1. A sampling time interval of an input signal,
A plurality of digital filter groups having a bandpass characteristic and a lowpass characteristic, which operate using the time interval as a clock, are provided, the digital filter groups having the lowpass characteristic are connected in series, and the digital filter groups having the bandpass characteristic are connected in parallel. The input signal is given to each digital filter group of bandpass characteristics, and each output is given to each input of the digital filter group of lowpass characteristics. , The delay time per unit delay element of the digital filter with the highest number is equal to or longer than the sampling time interval of the input signal, and the delay per stage of the digital filter decreases by one in each digital filter group. Time is ^AB (A> 1,
B> 0) times, and the delay time per unit delay element of the digital filter with a certain number of low-pass characteristics must be an integral multiple per unit delay element of the digital filter with the same number as the band-pass characteristic. An octave multiplex filter characterized by. 2. With respect to the sampling time interval of the input signal,
A plurality of digital filter groups having a bandpass characteristic and a lowpass characteristic, which operate using the time interval as a clock, are provided, the digital filter groups having the lowpass characteristic are connected in series, and the digital filter groups having the bandpass characteristic are connected in parallel. Each output of the low-pass characteristic digital filter group is given to each of the band-pass characteristic digital filter group, and each output of the band-pass characteristic digital filter group is added and taken out. , If the numbers are assigned in the order of connection in each digital filter group, the delay time per unit delay element of the highest-numbered digital filter is equal to or less than the sampling time interval of the input signal, and the number decreases by 1 in each digital filter group. To the digital file Delay time per one stage of the data is A ^B (A>
1, B> 0) times, and the delay time per unit delay element of the digital filter with a certain number of low-pass characteristics is equal to the delay time per unit delay element of the digital filter with the same number of band-pass characteristics. Octave multiple filter characterized by being an integral multiple.