WO2004036746A1 - デジタルフィルタの設計方法および装置、デジタルフィルタ設計用プログラム、デジタルフィルタ - Google Patents
デジタルフィルタの設計方法および装置、デジタルフィルタ設計用プログラム、デジタルフィルタ Download PDFInfo
- Publication number
- WO2004036746A1 WO2004036746A1 PCT/JP2003/013381 JP0313381W WO2004036746A1 WO 2004036746 A1 WO2004036746 A1 WO 2004036746A1 JP 0313381 W JP0313381 W JP 0313381W WO 2004036746 A1 WO2004036746 A1 WO 2004036746A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- filter
- numerical sequence
- unit
- digital filter
- tap
- Prior art date
Links
Classifications
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03H—IMPEDANCE NETWORKS, e.g. RESONANT CIRCUITS; RESONATORS
- H03H17/00—Networks using digital techniques
- H03H17/02—Frequency selective networks
- H03H17/0223—Computation saving measures; Accelerating measures
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03H—IMPEDANCE NETWORKS, e.g. RESONANT CIRCUITS; RESONATORS
- H03H17/00—Networks using digital techniques
- H03H17/02—Frequency selective networks
- H03H17/06—Non-recursive filters
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03H—IMPEDANCE NETWORKS, e.g. RESONANT CIRCUITS; RESONATORS
- H03H17/00—Networks using digital techniques
- H03H17/02—Frequency selective networks
- H03H17/0219—Compensation of undesirable effects, e.g. quantisation noise, overflow
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03H—IMPEDANCE NETWORKS, e.g. RESONANT CIRCUITS; RESONATORS
- H03H17/00—Networks using digital techniques
- H03H17/02—Frequency selective networks
- H03H17/0283—Filters characterised by the filter structure
- H03H17/0286—Combinations of filter structures
Definitions
- the present invention relates to a digital filter design method and apparatus, a digital filter design program, and a digital filter.
- the present invention includes a tapped delay line composed of a plurality of delay units, and multiplies the signal of each tap by several times and adds the signals.
- IIR Infinite Impulse Response
- FIR Finite Impulse Response
- the filters When the filters are classified according to the arrangement of passbands and stopbands, they are mainly divided into four types: mouth-to-pass filters, high-pass filters, band-pass filters, and band-elimination filters.
- the mouth-pass filter is the basis for IIR filters and FIR filters, and the other high-pass filters, band-pass filters, and band-stop filters perform processing such as frequency conversion from low-pass filters.
- the impulse response represented by the finite time length is directly used as the filter coefficient. Therefore, designing a FIR filter means determining the filter coefficient so that the desired frequency characteristics are obtained.
- the frequency characteristics of the filter obtained by the conventional design method depend on the window function and the Chebyshev approximation formula, and unless these are set properly, good frequency characteristics cannot be obtained.
- the present invention has been made to solve such a problem, and an object of the present invention is to enable a FIR digital filter having a desired frequency characteristic to be simply designed.
- Another object of the present invention is to provide a FIR digital filter capable of realizing a desired frequency characteristic with high accuracy on a small circuit scale. Disclosure of the invention
- the design method of the digital filter according to the present invention is such that the numerical value sequence is asymmetric, the sum of the numerical value sequence is non-zero, and the value of each of the above-mentioned numerical value sequences is equal to each other with the same sign.
- the first unit filter having a basic filter coefficient set with, the numerical sequence is asymmetric, the total value of the numerical sequence is zero, and the jump value of the numerical sequence is inverse sign Using at least one of the second unit filters having the basic filter coefficients whose values are set to be equal to each other, zero or more of the first unit filters and zero or more of the first unit filters Arbitrary vertical connection with the second unit filter It is characterized in that the filter is designed by connecting continuously.
- a non-symmetric filter coefficient whose value is set such that the total value of a numerical sequence is non-zero and the total value of every other digit of the numerical sequence is equal to each other with the same sign.
- the first unit filter is composed of multiple basic filters that are cascaded so that the whole number sequence is symmetrical, and the total value of the number sequence is zero and the value sequence is one jump
- a second basic filter is constructed by cascade-connecting a plurality of basic filters having asymmetric filter coefficients whose values are set to be equal to each other with opposite signs so that the entire numerical sequence is symmetric.
- zero or more of the first unit filters and zero or more of the second unit filters are arbitrarily connected in cascade to design a filter. The feature is that That.
- a delay of n clocks is inserted between taps corresponding to the asymmetric filter coefficients constituting the first and second unit filters, by inserting n clock delays. It is characterized in that the pass frequency band is adjusted.
- a digital filter according to the present invention includes a delay line with a tap comprising a plurality of delay devices, and a signal of each tap is transmitted to the digital filter according to any one of claims 1 to 7. It is characterized by multiplying each by the fill coefficient obtained by the above, adding, and outputting.
- delay means for inserting a delay for n clocks between each evening corresponding to the asymmetric filter coefficient is configured as follows. That is, the delay means is provided with a delay circuit that holds and outputs input data in accordance with a second clock having a cycle of lZn times the reference clock.
- a digital filter is designed by arbitrarily combining one or more unit filters having a predetermined asymmetrical numerical sequence as a filter coefficient. The coefficient of the digital filter having a desired frequency characteristic can be automatically obtained only by combining the filters, and the design of the filter can be performed extremely easily even by a non-skilled technician.
- the predetermined asymmetrical numerical sequence corresponds to one side of the predetermined symmetrical numerical sequence divided in half at its center.
- the number of taps required is very small, and the number of filter coefficients required for each tap output is very small.
- the digital filter to be designed has a very simple configuration consisting of repeating the same pattern called a unit filter, so that the man-hours required for integration can be reduced, and IC integration can be facilitated.
- FIG. 1 is a diagram showing a circuit configuration of a 5-tap unit filter L10, HI0 and a numerical sequence of filter coefficients.
- Figure 2 is a diagram illustrating the algorithm for generating the filter coefficients of the 5-tap unit filter L10 and HI0.
- FIG. 3 is a diagram for explaining the meaning of the fill coefficient of the five-tap mouth-to-pass unit file L 10.
- FIG. 4 is a diagram for explaining the meaning of the fill coefficient of the 5-tap high-pass unit fill H 10.
- FIG. 5 is a diagram illustrating an algorithm for generating filter coefficients of the 5-tap one-pass unit filter L11.
- FIG. 6 is a diagram illustrating frequency-gain characteristics of the 5-tap one-pass unit filters L10 and L11.
- FIG. 7 is a diagram illustrating an algorithm for generating a filter coefficient of the 5-tap high-pass unit filter H11.
- FIG. 8 is a diagram illustrating frequency-gain characteristics of the 5-tap high-pass unit filters H I0 and H I1.
- FIG. 9 is a diagram illustrating a frequency-gain characteristic of a 5-tap port one-pass unit filter (L10) ffl .
- FIG. 10 is a diagram illustrating a frequency-gain characteristic of a 5-tap high-pass unit filter (H 10) ⁇ .
- FIG. 11 is a diagram for explaining extraction of a frequency band.
- FIG. 12 is a diagram showing another example of extracting a frequency band.
- FIG. 13 is a diagram showing a circuit configuration of two types of three-tap unit filters and a numerical sequence of filter coefficients which are the most basic in the filter design method according to the first embodiment.
- Figure 14 is a diagram showing the frequency characteristics of the 3-tap high-pass unity filter ⁇ 10 ′ when the simple numerical sequence ⁇ 8, —9, 0, 1 ⁇ / 16 is used as the filter coefficient.
- FIG. 15 is a diagram illustrating a frequency characteristic of the 3-tap high-pass unit filter H I 0 ′ when only the filter coefficient ⁇ 3 is adjusted.
- FIG. 17 is a diagram showing a circuit example of band pass filtering according to the first embodiment.
- FIG. 18 is a diagram showing the frequency characteristics of the bandpass filter shown in FIG.
- FIG. 19 is a diagram showing a circuit configuration of unit filters L10 "and H10" according to the second embodiment and a numerical sequence of filter coefficients.
- FIG. 20 is a diagram illustrating a frequency characteristic of the low-pass unit filter L10 "according to the second embodiment.
- FIG. 21 is a diagram illustrating the frequency characteristics of the high pass unit filter H 10 "according to the second embodiment.
- FIG. 22 is a diagram illustrating an algorithm for generating the filter coefficients of the 5-tap unit secondary filters L 20 and H 20. BEST MODE FOR CARRYING OUT THE INVENTION
- the digital filter according to the present embodiment includes a delay line with taps formed of a plurality of delay devices, and a FIR of a type in which an output signal of each tap is multiplied by a given filter coefficient and then added and output. Filter.
- FIG. 1 is a diagram showing a 5-tap unit filter L10, H10, in which (a) shows the circuit configuration and (b) shows a numerical sequence of filter coefficients.
- the filter coefficient of the 5-tap low-pass unit filter L10 is a very simple numerical sequence ⁇ 1-1,0,9,16,9,0,1-1 ⁇ 3 2 (However, the part with a value of "0" has no tap output as shown in Fig.
- the filter coefficient of the 5-tap high-pass unit filter H 10 is a very simple numerical sequence ⁇ 1, 0, — 9, 16, — 9,0,1 ⁇ no 3 2 (However, the part with the value “0” has no tap output, and the filter coefficient and Not used).
- FIG. 2 is a diagram for explaining a method of generating a numerical sequence forming the filter coefficients of the 5-tap unit filter L10 and HI0.
- the numerical value sequence that composes the fill coefficient of the 5-tap one-pass unit filter L10 has a data value of ⁇ 11, 1, 8, 8, 8 This is obtained by performing a single moving average operation on a predetermined digital basic function that changes as ⁇ 1, 1, 1 ⁇ / 16.
- the numerical sequence forming the filter coefficient of the 5-tap high-pass unit filter HI0 is such that the data value is ⁇ 1, -1, -8, 8 , — 1, 1 ⁇ / 16 and a digital basic function that changes the data value every clock CK to ⁇ 1, 1 1, 8, 1 8, — 1, 1 ⁇ / 16 Is obtained by performing a moving average operation on
- FIG. 3 is a diagram showing a result obtained by performing a four-fold oversampling and a composition operation on a numerical sequence forming a filter coefficient of the 5-tap low-pass unit filter L10. Note that in order to make the explanation easier to understand, the integer number sequence ⁇ 11,0,9,16,9,0,1-1 ⁇ that is 32 times the original number sequence is overwritten. An example is shown in which one sampling and a compo- lation operation are performed.
- Fig. 3 (a) the series of numerical values shown in the leftmost column is four times larger than the original numerical sequence ⁇ —1, 0, 9, 16, 16, 9, 0, —1 ⁇ . Overpower This is the value that was printed. Also, the four numerical sequences from left to right are the numerical sequences shown in the leftmost column shifted down one by one. The column direction in Fig. 3 (a) shows the time axis, and shifting the numerical sequence downwards corresponds to gradually delaying the numerical sequence shown in the leftmost column. I do.
- the second numerical sequence from the left indicates that the numerical sequence shown in the leftmost column is shifted by 1Z4 phase of the 4 ⁇ frequency clock 4CK.
- the numerical sequence in the third column from the left is a numerical sequence in which the numerical sequence shown in the second column from the left is shifted by 1 Z 4 phase of the 4 ⁇ frequency clock 4 CK
- the numerical sequence in the fourth column from the left Indicates that the numerical sequence shown in the third column from the left is a numerical sequence further shifted by 14 phases of the clock 4CK of the quadruple frequency.
- the fifth numerical column from the left is a value obtained by adding the first to fourth numerical columns in the corresponding rows.
- quadruple oversampling involving four-phase convolution operation is performed digitally.
- the ninth column from the left is the sum of the fifth to eighth columns in the corresponding rows.
- Fig. 3 (b) is a graph of the finally obtained numerical sequence shown in the rightmost column of Fig. 3 (a).
- the function having the waveform shown in Fig. 3 (b) has a finite value other than "0" only when the sample position along the horizontal axis is between t1 and t4.
- the case where the value of the function has a finite value other than "0" in a local region and becomes “0” in other regions is called "finite base”.
- Fig. 3 (b) takes a maximum only at the center sample position t5, and the value becomes "0" at the four sample positions tl, t2, t3, and t4.
- This is a sampling function with the characteristic that all the sampling points necessary to obtain a smooth waveform are passed.
- Fig. 4 is a diagram showing the results of applying a four-fold oversampling and a compilation operation to a numerical sequence constituting the fill coefficient of the 5-tap high-pass unit file H10. is there.
- the oversampler is applied to the integer number sequence ⁇ 1,0, —9,16,19,0,1 ⁇ , which is 32 times the original number sequence.
- the figure shows an example of performing the calculation and the compilation operation.
- FIG. 4 (a) shows the same operation process as in FIG. 3 (a).
- Graphing the finally obtained numerical sequence shown in the first column of Fig. 4 (a) results in Fig. 4 (b).
- the function shown in Fig. 4 (b) is also a sampling function that takes a maximum value only at the center sampling position t7 ', and is differentiable once in all regions, and furthermore, sampling positions t1' and t6 ' Is a finite function that converges to zero.
- FIG. 6 is a diagram showing the frequency-gain characteristics of the result of FFT (Fast Fourier Transfer) of the numerical sequence of the 5-tap one-pass unit filters L1Q and L11. Here, the gain and frequency are normalized by "1".
- the gain at the center frequency of the five-tap mouth-to-pass unit filter L10, L11 is 0.5, and the overshoot in the low frequency region and the high frequency region Good mouth-to-pass filter characteristics with no ringing at the filter are obtained. If the number of "0" inserted between each filter coefficient is n, the frequency axis (period in the frequency direction) of the frequency-gain characteristic is lZn.
- FIG. 7 is a diagram showing the filter coefficients of the 5-tap high-pass unit filter H11. As shown in FIG. 7, the filter coefficient of the 5-tap high-pass unit filter H11 is between the filter coefficients of the 5-tap high-pass unit filter HI0. It is generated by inserting one "0".
- Figure 8 shows the frequency-gain characteristics of the 5-tap high-pass unit filters H10 and H11. Again, the gain and frequency are normalized by "1". As can be seen from FIG. 8, in the 5-tap high-pass unit filters H10 and H11, the gain is 0.5 at the center frequency, and overshoot in the high-frequency region and low-frequency Good high-pass filter characteristics with no ringing in the region are obtained. If the number of "0" s inserted between the filter coefficients is n, the frequency axis (period in the frequency direction) of the frequency-gain characteristic is lZn.
- Fig. 2 (b) which is the basis for realizing such high-pass filter characteristics, is also the basis for the finite-unit sampling function as shown in Fig. 4 (). Therefore, by using this numerical sequence as a filter coefficient, it is not necessary to cut off the coefficient using a window function, and a good high-pass filter characteristic can be obtained.
- FIG. 9 is a diagram illustrating the frequency-gain characteristics of the 5-tap low-pass unit filters L10, (L10), (L10) 4 , and (L10) 8 .
- the gain and frequency are normalized by "1".
- the clock at the position where the amplitude is 0.5 is 0.25.
- FIG. 10 is a diagram showing the frequency-gain characteristics of the 5-tap high-pass unit filter H10, (H10) (HI0) (HI0) 8 .
- the gain and the frequency are normalized by "1".
- the clock at the position where the amplitude is 0.5 is 0.25.
- FIG. 11 is a diagram for explaining extraction of a frequency band.
- the extraction of the frequency band is performed using a cascade connection of four or more unit filters as described above.
- Fig. 11 (a) shows the frequency-gain characteristics of the 5-tap low-pass unit filter (L10) 4 , (L11), (L13), (L17) 4 on one graph.
- FIG. In Fig. 11 (a) the gain and frequency are normalized by "1".
- Figure 11 (b) shows these four types of 5-tap low-pass unit filters (L10), (L11), (L13) 4 , and (L17) 4 cascaded. The frequency characteristics obtained at this time are shown. According to this, a one-pass filter having almost good attenuation characteristics, in which a very narrow region of low frequency is a passband, is obtained. Although a slight ringing occurs, this ringing occurs in a portion that is reduced by more than 106 dB, and is almost ignored.
- FIG. 12 is a diagram showing another example of extracting a frequency band.
- the 5-tap high-pass unit filter (H10) 8 and the 5-tap one-pass unit filter (L11) 8 , (L13) ) 8 and (L 17) 8 in combination to cascade them to obtain a high-pass filter in which a predetermined frequency band is a pass band.
- a 5-tap low-pass unit filter (L10) 4 , (Lll) 8 , and (L13) 4 are combined and connected in cascade. Then, a one-pass filter having a predetermined frequency band as a pass band can be obtained. Also, as shown in FIG. 12 (c), a 5-tap high-pass unit filter (H11) 8 , a 5-tap one-pass unit filter (L13) 4 , (L17) When these are connected in cascade in combination with (L 1, 15) 8 , it is possible to obtain a band-pass filter having a predetermined frequency band as a pass band.
- the combination of unit filters can be obtained by arbitrarily combining these using the 5-tap unit filter using a predetermined basic numerical sequence as a filter coefficient. Just the FIR with the desired frequency characteristics The filter coefficient of the filter can be automatically generated. Therefore, the filter design method is simple and easy to think, and the filter design can be performed extremely easily even without a skilled engineer.
- the number of taps required for a filter circuit designed by applying the above method is very small, and the types of filter coefficients required for each tap output are also small.
- the configuration of the calculation unit can be extremely simplified. Therefore, the number of circuit elements (particularly, multipliers) can be significantly reduced, and the size of the filter circuit can be reduced, and the power consumption can be reduced and the calculation load can be reduced.
- the filter circuit designed by applying the above method has a very simple configuration consisting of repetition of the same pattern, so that the number of circuits can be reduced for integration and IC integration can be facilitated. It also has the advantage of being able to do so. Also, in terms of characteristics, the cutoff characteristics can be greatly improved, and the phase characteristics are linear, and excellent filter characteristics can be obtained.
- the applicant has already filed a patent application for the contents described above (Japanese Patent Application No. 2001-321132).
- the present embodiment is a further improvement of the content of the already-filed application, and as described above, has a desired frequency characteristic only by a combination of two types of three-tap unit filters Lin 'and H1n *.
- FIR filters can be designed.
- the filter coefficients of the 3-tap unit filter L in 'and H 1 n' are obtained by dividing one half of the numerical sequence of the filter coefficients of the 5-tap unit filter L 1 n and H n in the middle at the center. It has been further adjusted.
- FIGS. 13A and 13B are diagrams showing three-tap unit filters L10 'and H10', where (a) shows the circuit configuration and (b) shows a numerical sequence of filter coefficients.
- Fig. 13 (a) shows the circuit configuration and (b) shows a numerical sequence of filter coefficients.
- Fig. 13 (a) shows the circuit configuration and (b) shows a numerical sequence of filter coefficients.
- Fig. 13 (a) shows the circuit configuration and (b) shows a numerical sequence of filter coefficients.
- Fig. 13 (a) three D-type flip-flops connected in cascade are connected to the three-tap unit fill L10 'and H10'.
- the input signal is sequentially delayed by 1 clock CK by 1-, ⁇ 1 l -3 .
- Their to respect Fetch the signal from the predetermined tap of the D-type unfavorable flop 1 11 1 -3, three coefficient filter coefficient H. 1 to H 3 shown in Fig. 1 3 (b)
- Multipliers are respectively multipli
- circuit configurations of the above two types of three-tap unit filters L 10 'and H 10' are both as shown in Fig. 13 (a), and the filter coefficients (coefficient units 1 2-
- the filter coefficients of the three-tap one-pass unit filter L 10 are such that the numerical sequence is asymmetric, the total value of the numerical sequence is non-zero, and the total value of each skip of the numerical sequence is the same sign. They have the property of being equal to each other.
- the filter coefficient of the 3-tap high-pass unit filter H 1 0 ′ is such that the numerical sequence is asymmetric, the total value of the numerical sequence is zero, and the total value of the numerical sequence at one jump is the opposite sign. They have the property of being equal to each other.
- the filter coefficients of these three-tap unit filters L 1 0 ′ and H 10 ′ have the same properties as the above-described 5-tap unit filters L 1 0 and HI 0 except that they are asymmetric. .
- the values are slightly more complicated than the filter coefficients of the 5-tap unit filters L10 and HI0. The reason will be described below.
- FIG. 14 shows the filter coefficient ⁇ 1,0, —9,16, —9,0,1 ⁇ 32 of the 5-tap high-passunit file H10, which is divided in half at the center and placed on one side.
- the filter coefficient When the corresponding simple numerical sequence ⁇ 8, 19, 0, 1 ⁇ / 16 is used as the filter coefficient (however, the part with the value "0" has no tap output and is not used as a filter coefficient)
- FIG. 4 is a diagram showing frequency characteristics of the multiplexed signal. here Scales gain and frequency by "1".
- a unit filter having such frequency characteristics is not suitable for a method of cascading a plurality of such filters to design a desired FIR filter.
- the above simple numerical sequence ⁇ 8, —9, 0, 1 ⁇ / 16 is adjusted.
- the absolute value of the filter coefficient H3 that determines the high-frequency characteristic is reduced.
- the maximum value can be adjusted using the filter coefficient ⁇ 3 used for adjusting the high frequency component and the filter coefficient ⁇ 2 of the opposite sign.
- the value of ⁇ for gain adjustment ( ⁇ value of filter coefficient ⁇ 2) should be the same as the value of ⁇ for high frequency adjustment (value of ⁇ of filter coefficient ⁇ 3) described above.
- (A) shows the gain by a linear scale
- (b) shows the gain. Is represented on a logarithmic scale.
- the gain and frequency are normalized by "1" ⁇
- each filter coefficient of the 3-tap high-pass unit filter ⁇ 10 ' By inserting ⁇ pieces of “0” between them, it is possible to generate a 3-tap high-pass unit filter ⁇ 1 ⁇ ′.
- the filter coefficient of the three-pass one-pass unit filter L10 ′ can be appropriately adjusted.
- the filter coefficient ⁇ 3 that determines the characteristics from “—1” to “— (1_ ⁇ / 8)”.
- the maximum value of the gain is adjusted by the filter coefficient ⁇ 3 and the filter coefficient ⁇ 2 of the opposite sign. That is, the value of the filter coefficient ⁇ 2 is reduced to “9”, such as “9— —8”.
- the filter coefficient of the three-tap low-pass unit filter L10 is adjusted, there is no waving in the passband, and only the central part has the maximum value of the gain. It is possible to obtain a mouth-pass filter characteristic whose maximum value is just "1".
- the three-tap path unit filter L10 'having such frequency characteristics is also suitable for the method of cascading them to design a desired FIR filter.
- each file of the three-tap path unit file L10' is also provided.
- ⁇ pieces of “0” between the evening coefficients it is possible to generate a 3-tap low-pass unit filter L 1 ⁇ ′.
- the five-tap unit filter is formed by cascading the two types of three-tap filter L10 ',' 10 ', which are configured as shown in Fig. 1.3, as appropriate.
- L10 ',' 10 ' which are configured as shown in Fig. 1.3
- the filters L10 and HI0 are combined, it is possible to easily design an FIR filter having the desired frequency characteristics.
- the filter coefficients of the 3-tap unit filters L10 'and H10' were obtained by dividing the numerical sequence of the filter coefficients of the 5-tap unit filter L10 and H10 by half at the center.
- the other numerical sequence ⁇ 1-1,0,9,8 ⁇ / 16, ⁇ 1,0, —9,8 ⁇ 16 may be adjusted and generated.
- FIG. 17 is a diagram showing a circuit example of a bandpass filter that achieves this target standard.
- the circuit shown in Fig. 17 has six 3-tap high-pass unit filters H1 1 ', four 3-tap low-pass unit filters L1 3', three 3-tap high-pass unit filters H1 5 ', It consists of three 3-tap low-pass unit filters L 17 'and two 5-tap high-pass unit filters HI 9 cascaded in this order.
- FIG. 18 is a diagram showing the frequency characteristics of the bandpass filter shown in FIG. 17, where (a) represents the gain on a linear scale, and (b) represents the gain on a logarithmic scale. As can be seen from Fig. 18, the bandpass filter shown in Fig. 17 satisfies the target standards mentioned above.
- the 3-tap unit filters L1n 'and H1n' have asymmetrical numerical sequences of filter coefficients, the linearity of the phase is assured by itself as shown in Fig. 16. Absent. However, when many of these are collected and cascaded as shown in Fig. 17, phase distortions cancel each other out, and the phase characteristics are somewhat improved as shown in Fig. 18.
- the three-tap unit filters LI n ′ and H 1 n ′ having a smaller number of taps are basically used, and these are cascaded.
- a bandpass filter with desired characteristics can be obtained with a very simple configuration connected. At this time, the total number of necessary taps is only 51 taps per bit.
- the filter coefficients of the band-pass filter are generated based on the above-mentioned finite-order function, and the coefficient values become small enough to be ignored near both sides of the cascade connection. Due to the nature of the finite base, no truncation error occurs even if this part is ignored, so this part is excluded. The remaining part excluded is the final value for a bandpass filter that satisfies the target specification. It is a typical filter coefficient.
- this filter coefficient is sufficient to configure as hardware, and only 51 taps are required for it.
- the number of necessary taps as a whole can be reduced as compared with the case of using only the 5-tap unit filter Lin, HIn, and the configuration of the filter circuit can be reduced. It can be extremely simple.
- the filter coefficient of the three-tap one-pass unit filter L10 ' the numerical value forming the filter coefficient of the five-tap probe unit filter L10 is used.
- the numerical sequence ⁇ 8,9,0, —1 ⁇ no 16 on one side of the half of the sequence ⁇ —1,0,9,16,9,0, —1 ⁇ 32 was further adjusted was used.
- a filter coefficient of the 3-tap high-pass unit filter HI 0 ′ a numerical sequence ⁇ 1, 0, which constitutes a 5-tap hyno, ° Sunet filter H10 filter coefficient, is used. , — 9, 16, — 9, 0, 1 ⁇ / 3 2
- the half-valued sequence ⁇ 8, — 9, 0, 1 ⁇ Z 16 was further adjusted by dividing it into two halves. . Since the filter coefficients of these three-tap unit filters L10 'and H10' are both asymmetric, the linearity of the phase is not guaranteed as described above.
- the second embodiment can reduce the number of taps by three taps. This is to make it possible to realize linear phase characteristics while using a cut filter.
- FIG. 19 is a diagram showing unit filters L10 "and H10" according to the second embodiment, where (a) shows the circuit configuration, and (b) shows a numerical sequence of filter coefficients. Is shown.
- the unit filters L10 "and HI0" of the present embodiment have the same configuration. That is, the low pass unit filter L 10 "is formed by cascading two 3-tap low pass unit filters 1 L 10 and 2 L 10.
- the high pass unit filter H 10" is 2. It consists of two 3-tap high-pass unit filters 1 H10 and 2 H10 connected in cascade.
- one of the three-tap low-pass unit filters 2 L10 constituting the low-pass unit filter L10 is the above-mentioned five-tap low-pass unit filter L10.
- Numerical sequence ⁇ 1 1, 0, 9, 16, 9, 0, — 1 ⁇ 3 2 of one half of ⁇ 2, 9, 0, — 1 ⁇ / 1 The values obtained by further adjusting 6 are the filter coefficients H 4 to H 6, which are the same as the three-tap one-pass unit filter L 10 ′ described in the first embodiment.
- the other 3-tap low-pass unit filter 1 L10 is obtained by further adjusting the numerical sequence ⁇ 1-1,0,9,8 ⁇ Z16 on the other side of the half, as described above.
- the filter coefficients are set to H1 to H3.
- one of the three-tap high-pass unit filters 2 H 10 constituting the high-pass unit filter H 10 is a numerical sequence ⁇ 1, 0, —9, of the filter coefficients of the 5-tap high-pass unit filter H 10 described above. 1, 6, 9, 0, 1 ⁇ Z 32 is further divided into halves, and the numerical sequence ⁇ 8, -9, 0, 1 ⁇ / 16 on one side is further adjusted to filter coefficients H 4 to H 6 And You. This is the same as the 3-tap high-pass unit filter H10 'described in the first embodiment.
- the other three-tap high-pass unit fill 1H10 is further adjusted from the numerical sequence ⁇ 1,0,-9,8 ⁇ / 16 on the other side of the half, as described above. The coefficient is set as the fill coefficient H 1 to: H 3.
- connection relationship between the two 3-tap port one-pass unit filters 1L10 and 2L10 and the connection relationship between the two three-tap high-pass unit filters 1H10 and 2H10 are shown in the figure.
- the right and left may be reversed from those shown in 19 (a).
- FIG. 20 is a diagram showing the frequency characteristics of the low pass unit filter L 10 "
- FIG. 21 is a diagram showing the frequency characteristics of the high pass unit filter H I 0".
- the gain is expressed in a logarithmic scale. In FIGS. 20 and 21 as well, the gain and frequency are
- a good attenuation of 55 dB is also obtained.
- very clear linear-phase characteristics are obtained for the frequency-phase characteristics. Therefore, by cascade-connecting the unit filters Lin ", H1n" having such frequency characteristics as described in the first embodiment, the coefficient is symmetric with a very small number of taps.
- a linear phase filter can be configured.
- the unit filters L10 "and H10" shown in Fig. 19 have 6 taps as a whole, and have more taps than the 5-tap units L10 and HI0. I have.
- the unit filter L 1 0 "and ⁇ 1 0" It consists of two 3-tuple filters connected in cascade. Therefore, the hand width is already narrower than the 5-tap unit L10, H10.
- the second embodiment is particularly effective when designing a FIR filter having a narrow band width.
- the number of cascades of unit filters required to achieve the desired narrow band width is significantly smaller than that in the case of using 5-tuple filter L10, HI0. can do.
- the circuit size can be reduced as a whole.
- An apparatus for realizing the digital filter design method according to the first and second embodiments described above can be realized by any of a hardware configuration, a DSP, and software.
- the filter design device when implemented by software, is composed of a computer CPU or MPU, RAM, ROM, etc., and is stored in RAM, ROM, hard disk, etc. It can be realized by running the program.
- the present invention can be realized by recording a program that causes a computer to perform the function of the present embodiment on a recording medium such as a CD-ROM, and reading the program into the computer.
- a recording medium for recording the above program in addition to CD-R ⁇ M, a flexible disk, a hard disk, a magnetic tape, an optical disk, a magneto-optical disk, a DVD, a nonvolatile memory card, etc. Can be used. It can also be realized by downloading the above program to a computer via a network such as the Internet.
- the filter coefficients relating to various unit files are stored as information in a memory such as RAM or ROM, and the user can use the unit filter.
- the CP calculates the FIR filter by calculating the filter coefficient corresponding to the specified combination using the information of the filter coefficient stored in the above memory. It is possible to
- various unit filters are iconized (the filter coefficients are stored as information corresponding to each icon), and the user arranges these icons arbitrarily in combination on the display screen.
- the CPU may automatically calculate and obtain the filter coefficient corresponding to the array. Also, if the calculated filter coefficient is automatically FFT-converted and the result is displayed as a frequency-gain characteristic diagram, the characteristics of the designed filter can be confirmed, making filter design easier. Can be done.
- the functions of the above-described embodiments are realized not only by the computer executing the supplied program, but also by the operating system (OS) or the operating system in which the program is running on the computer.
- OS operating system
- the functions of the above-described embodiment are realized in cooperation with other application software or the like, or when all or part of the processing of the supplied program is performed by a computer function expansion port ⁇ function expansion unit
- Such a program is also included in the embodiment of the present invention when the functions of the above-described embodiment are implemented by a computer.
- the symmetrical numerical sequence shown in FIGS. 2 (a) and (b) is divided into halves to obtain asymmetrical 3-tap unit filter filter coefficients.
- Numeric sequences of type are not limited to this.
- a 5-tap rover unit second-order filter L20 generated as shown in Fig. 22 (a) or a 5-tap rover unit second-order filter L20 generated as shown in Fig. 22 (a)
- a symmetrical numerical sequence of 5-tap high-pass unit secondary filter H20 may be used.
- these five-tap unit secondary filters L20 and H20 are preferably used for waveform adjustment in the last stage of the cascade-connected filter blocks.
- the 5-tuple secondary filter with such characteristics is divided into halves of the numerical sequence that constitutes the filter coefficients of L 2 ⁇ and H 2 n, and one side is further adjusted to obtain a 3-tap unit secondary filter. It is possible to use the filter coefficients of L2n 'and H2n'. By connecting these three-tap unit secondary filters L 2 n ′ and H 2 n ′ in a mirror-symmetrical manner, a unit secondary filter L 2 n that realizes linear phase characteristics is obtained. 2 n "and H 2 n" can also be constructed.
- a value sequence different from the above value sequence using absolute values of "1" and "8" is used as the filter coefficient of the 5-tap unit filter.
- the numerical sequence may be divided in half and used as the filter coefficient of the 3-tap unit filter.
- a bandpass filter can be designed directly only by cascade connection of the filter.
- a bandpass filter for FM can be formed by cascading only five tap high-pass unit filters H11, and the band width (passband width) is adjusted by the number m of the connections. This can be applied to a three-tap high-pass unit filter H11 'or a high-pass unit filter H11 ".
- an AM bandpass filter is configured by cascading five tap low-pass unit filters (Lin) B after the FM bandpass filter configured as described above. Is also possible. Similarly, this can be applied to a three-tap one-pass unit filter L11 'or a low-pass unit filter L11 ".
- a relatively simple low-pass filter is designed using the unit filter described above, and the pass band is shifted in the direction of the frequency axis, so that a high-pass filter having a desired frequency band as a pass band is obtained.
- Band pass filters, band elimination filters, etc. may be designed.
- the present invention is useful for a FIR digital filter that includes a delay line with taps composed of a plurality of delay devices, multiplies the signal of each tap by several times, and adds and outputs the result.
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computer Hardware Design (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Theoretical Computer Science (AREA)
- Filters That Use Time-Delay Elements (AREA)
- Picture Signal Circuits (AREA)
Abstract
Description
Claims
Priority Applications (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2004544992A JPWO2004036746A1 (ja) | 2002-10-21 | 2003-10-20 | デジタルフィルタの設計方法および装置、デジタルフィルタ設計用プログラム、デジタルフィルタ |
EP03754198A EP1557946A1 (en) | 2002-10-21 | 2003-10-20 | Digital filter design method and device, digital filter design program, and digital filter |
US10/907,943 US7529788B2 (en) | 2002-10-21 | 2005-04-21 | Digital filter design method and device, digital filter design program, and digital filter |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2002306315 | 2002-10-21 | ||
JP2002-306315 | 2002-10-21 |
Related Child Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US10/907,943 Continuation US7529788B2 (en) | 2002-10-21 | 2005-04-21 | Digital filter design method and device, digital filter design program, and digital filter |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2004036746A1 true WO2004036746A1 (ja) | 2004-04-29 |
Family
ID=32105204
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2003/013381 WO2004036746A1 (ja) | 2002-10-21 | 2003-10-20 | デジタルフィルタの設計方法および装置、デジタルフィルタ設計用プログラム、デジタルフィルタ |
Country Status (6)
Country | Link |
---|---|
EP (1) | EP1557946A1 (ja) |
JP (1) | JPWO2004036746A1 (ja) |
KR (1) | KR20050071604A (ja) |
CN (1) | CN1706100A (ja) |
TW (1) | TW200408813A (ja) |
WO (1) | WO2004036746A1 (ja) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2008096486A1 (ja) * | 2007-02-07 | 2008-08-14 | Mitsubishi Electric Corporation | 乗員保護装置 |
US20230336384A1 (en) * | 2020-10-23 | 2023-10-19 | Marvell Asia Pte Ltd | Equalization in high-speed data channel having sparse impulse response |
Families Citing this family (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CA2621175C (en) | 2005-09-13 | 2015-12-22 | Srs Labs, Inc. | Systems and methods for audio processing |
EP2005787B1 (en) | 2006-04-03 | 2012-01-25 | Srs Labs, Inc. | Audio signal processing |
TWI559679B (zh) | 2009-02-18 | 2016-11-21 | 杜比國際公司 | 低延遲調變濾波器組及用以設計該低延遲調變濾波器組之方法 |
CN101968964B (zh) * | 2010-08-20 | 2015-09-02 | 北京中星微电子有限公司 | 一种去除语音信号中直流分量的方法及装置 |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH05235701A (ja) * | 1992-02-25 | 1993-09-10 | Nippon Steel Corp | 環状畳み込みによるディジタルフィルタバンク処理方法及び装置 |
JPH05243908A (ja) * | 1991-12-06 | 1993-09-21 | Nec Corp | ろ波器 |
JPH06503450A (ja) * | 1990-12-19 | 1994-04-14 | ルンナー,トマス | 積の数を減らすことによりディジタルフィルタバンクの電力消費を減らす方法 |
JPH06326555A (ja) * | 1992-12-21 | 1994-11-25 | Sony United Kingdom Ltd | サブバンド・フィルタ装置及びその応用 |
JPH1079686A (ja) * | 1996-09-02 | 1998-03-24 | Ricoh Co Ltd | ディジタル相関器 |
-
2003
- 2003-10-15 TW TW092128609A patent/TW200408813A/zh unknown
- 2003-10-20 JP JP2004544992A patent/JPWO2004036746A1/ja active Pending
- 2003-10-20 KR KR1020057006906A patent/KR20050071604A/ko not_active Application Discontinuation
- 2003-10-20 EP EP03754198A patent/EP1557946A1/en not_active Withdrawn
- 2003-10-20 CN CNA2003801018286A patent/CN1706100A/zh active Pending
- 2003-10-20 WO PCT/JP2003/013381 patent/WO2004036746A1/ja active Application Filing
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH06503450A (ja) * | 1990-12-19 | 1994-04-14 | ルンナー,トマス | 積の数を減らすことによりディジタルフィルタバンクの電力消費を減らす方法 |
JPH05243908A (ja) * | 1991-12-06 | 1993-09-21 | Nec Corp | ろ波器 |
JPH05235701A (ja) * | 1992-02-25 | 1993-09-10 | Nippon Steel Corp | 環状畳み込みによるディジタルフィルタバンク処理方法及び装置 |
JPH06326555A (ja) * | 1992-12-21 | 1994-11-25 | Sony United Kingdom Ltd | サブバンド・フィルタ装置及びその応用 |
JPH1079686A (ja) * | 1996-09-02 | 1998-03-24 | Ricoh Co Ltd | ディジタル相関器 |
Non-Patent Citations (1)
Title |
---|
THE INSTITUTE OF ELECTRONICS, INFORMATION AND COMMUNICATION ENGINEERS: "digital shingo shori handbook", 31 January 1993, article OHMSHA LTD., pages: 112 - 113, XP002976234 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2008096486A1 (ja) * | 2007-02-07 | 2008-08-14 | Mitsubishi Electric Corporation | 乗員保護装置 |
US8386130B2 (en) | 2007-02-07 | 2013-02-26 | Mitsubishi Electric Corporation | Occupant protection device |
US20230336384A1 (en) * | 2020-10-23 | 2023-10-19 | Marvell Asia Pte Ltd | Equalization in high-speed data channel having sparse impulse response |
US12021670B2 (en) * | 2020-10-23 | 2024-06-25 | Marvell Asia Pte Ltd | Equalization in high-speed data channel having sparse impulse response |
Also Published As
Publication number | Publication date |
---|---|
JPWO2004036746A1 (ja) | 2006-02-16 |
TWI292044B (ja) | 2008-01-01 |
CN1706100A (zh) | 2005-12-07 |
EP1557946A1 (en) | 2005-07-27 |
KR20050071604A (ko) | 2005-07-07 |
TW200408813A (en) | 2004-06-01 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US7529788B2 (en) | Digital filter design method and device, digital filter design program, and digital filter | |
JPWO2006048958A1 (ja) | デジタルフィルタおよびその設計方法、設計装置、デジタルフィルタ設計用プログラム | |
US7619546B2 (en) | Asynchronous sample rate conversion using a digital simulation of an analog filter | |
JP4300272B2 (ja) | デジタルフィルタおよびその設計方法 | |
JPH0828649B2 (ja) | ディジタルフィルタ | |
WO2004036747A1 (ja) | デジタルフィルタの設計方法および装置、デジタルフィルタ設計用プログラム、デジタルフィルタ | |
WO2004036746A1 (ja) | デジタルフィルタの設計方法および装置、デジタルフィルタ設計用プログラム、デジタルフィルタ | |
JPWO2005078925A1 (ja) | デジタルフィルタの設計方法および装置、デジタルフィルタ設計用プログラム、デジタルフィルタ | |
US7590675B2 (en) | Digital filter, design method thereof, design device, and digital filter design program | |
EP1533898A1 (en) | Digital filter designing method, digital filter designing program, digital filter | |
WO2003047097A1 (fr) | Procede de conception d'un filtre numerique, appareil de projet, programme de projet pour la realisation d'un tel filtre, et filtre numerique y relatif | |
EP1605590A1 (en) | Digital filter design method and device, digital filter design program, digital filter | |
US20050171988A1 (en) | Digital filter design method and device, digital filter design program, and digital filter | |
US20050120067A1 (en) | Digital filter designing method, digital filter designing program, digital filter | |
EP1427104A1 (en) | Sound quality adjusting device and filter device used therefor, sound quality adjusting method, and filter designing method | |
JP4989575B2 (ja) | フィルタ及びフィルタの構成方法 | |
Syed et al. | Investigation of Area-Timing Trade off for Merged Delay Transformed Interpolator | |
JP2012085177A (ja) | デシメータ回路及びデシメータ回路の演算方法 | |
JP2001177414A (ja) | オーバーサンプリング処理回路およびデジタル−アナログ変換器 | |
JPWO2003096534A1 (ja) | 音質調整装置の設計方法および設計装置、音質調整装置設計用プログラム、音質調整装置 | |
JP2010021860A (ja) | 帯域分離フィルタ及び帯域分離方法 |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
AK | Designated states |
Kind code of ref document: A1 Designated state(s): CN JP KR US |
|
AL | Designated countries for regional patents |
Kind code of ref document: A1 Designated state(s): AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HU IE IT LU MC NL PT RO SE SI SK TR |
|
121 | Ep: the epo has been informed by wipo that ep was designated in this application | ||
WWE | Wipo information: entry into national phase |
Ref document number: 2004544992 Country of ref document: JP |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2003754198 Country of ref document: EP Ref document number: 10907943 Country of ref document: US Ref document number: 1020057006906 Country of ref document: KR Ref document number: 20038A18286 Country of ref document: CN |
|
WWP | Wipo information: published in national office |
Ref document number: 1020057006906 Country of ref document: KR |
|
WWP | Wipo information: published in national office |
Ref document number: 2003754198 Country of ref document: EP |