JP5102710B2 - Digital filter - Google Patents

Description

  The present invention relates to a digital filter, and more particularly, to an IIR type (cyclic type) digital filter that enables high-speed arithmetic processing.

As a conventional digital filter, a 1D type second-order IIR digital filter as shown in FIG. 9 is known (see, for example, Non-Patent Document 1 and Patent Document 1).
In this 1D type second-order IIR digital filter, input data S (mT) (m: natural number) is input to the adder 1001, and the adder 1001 obtains a calculation result Q (mT). Delay data Q (mT−T) and Q (mT−2T) obtained by giving a delay of time T (= 1 / Fs) by the delay elements 1002 and 1003 to Q (mT) are multiplied by multipliers 1004 and 1005, respectively. It is multiplied by b1 (−b1 × Q (mT−T)) and −b2 times (−b2 × Q (mT−2T)), and is added to the input data S (mT) by the adder 1001.

Q (mT), Q (mT-T), and Q (mT-2T) are multiplied by a0, a1 and a2 by multipliers 1006, 1007 and 1008, respectively, and the results are added by adder 1009. Is output as the final calculation result P (mT). Q (mT) and P (mT) are expressed as follows.
Q (mT) = S (mT) + (− b1 × Q (mT−T)) + (− b2 × Q (mT−2T)) (1)
P (mT) = (a0 × Q (mT)) + (a1 × Q (mT−T)) + (a2 × Q (mT−2T)) (2)
Furthermore, the transfer function H (z) of the IIR digital filter shown in FIG. 9 is expressed by the following equation.

H (z) = (a2 × Z ⁻² + a1 × Z ⁻¹ + a0) / (b2 × Z ⁻² + b1 × Z ⁻¹ +1) (3)
As a conventional digital filter, a 2D type second-order IIR digital filter as shown in FIG. 10 is known (see, for example, Non-Patent Document 1 and Patent Document 2). The transfer function of this IIR digital filter is also expressed by the above equation (3), and the same filter characteristics as the 1D type second order IIR digital filter shown in FIG. 9 are obtained.

Furthermore, as a conventional digital filter, a 3D type second-order IIR digital filter as shown in FIG. 11 is known (see, for example, Non-Patent Document 1 and Patent Document 1). The transfer function of this IIR digital filter is also expressed by the above equation (3), and the same filter characteristics as the 1D type second order IIR digital filter shown in FIG. 9 are obtained.
The second-order IIR digital filters represented by the 1D type in FIG. 9, the 2D type in FIG. 10, and the 3D type in FIG. 11 are called biquad circuits, and are connected in cascade or in parallel to provide higher-order filter characteristics. It is used as a basic section for realization.

FIG. 12 shows an example of cascade connection constituted by 1D type biquad circuits. In the example shown in FIG. 12, i + 1 (i: natural number) 1D type biquad circuits from 4000 to 400i are cascade-connected, and the 2 × (i + 1) th order or ((2 × (i + 1)) − 1 ) The following filter characteristics.
FIG. 13 shows an example of parallel connection composed of 1D type biquad circuits. In the example shown in FIG. 13, i + 1 1D type biquad circuits from 5000 to 500i are connected in parallel, and the 2 × (i + 1) th order or ((2 × (i + 1)) − 1) th order filter characteristics are obtained. .
Ochihiro, "Digital signal processing learned by simulation", CQ Publisher, February 2004 JP-A-9-186554 JP-A-5-291880

However, the conventional IIR digital filter has the following problems in realizing high-speed arithmetic processing.
The delay elements 1002 and 1003 in FIG. 9, the delay elements 2004 and 2005 in FIG. 10, and the delay elements 3001, 3002, 3007, and 3008 in FIG. 11 are basic elements that give a delay of time T. Results must be confirmed. However, an actual arithmetic circuit such as an adder or a multiplier has a delay of elements (element delay). The more complicated the operation is, and the higher the operating frequency Fs and the shorter the time T are due to the higher speed, the more serious the element delay becomes. Furthermore, the element delay tends to increase as the number of bits to be calculated increases.

  In the 1D type IIR digital filter of FIG. 9, the element delay of the calculation result is most problematic. From the output Q (mT−T) of the delay element 1002 through the multiplier 1004, the adder 1001 calculates the operation result Q (mT ), And further through a multiplier 1006 and an adder 1009 to obtain P (mT), and from the output Q (mT−2T) of the delay element 1003 through the multiplier 1005, the adder 1001 calculates the operation result Q ( mT), and further passes through a multiplier 1006 and an adder 1009 to obtain P (mT).

Further, in the 2D type IIR digital filter of FIG. 10, the path from the output of the delay element 2004 to the adder 2006, the multiplier 2009, and the adder 2007, and the output of the delay element 2004 to the adder 2006, the multiplier 2010, and the addition The element delay of the calculation result becomes the problem in the two types of paths passing through the device 2008.
Further, in the 3D type IIR digital filter of FIG. 11, the path from the input data S (mT) to the delay element 3007 through the multiplier 3003 and the adder 3006, and the multiplier 3004 and adder 3006 from the output of the delay element 3001. The three types of paths from the output to the delay element 3007 and the output of the delay element 3002 to the delay element 3007 through the multiplier 3005 and the adder 3006 are the most problematic element delay.

If the computation time exceeds the time T due to element delay, the transfer function expressed by the above equation (3) cannot be realized.
In order to solve this, it is conceivable to use an adder or a multiplier capable of high-speed calculation, but this also has a limit and does not provide a fundamental solution to the element delay problem.
Accordingly, an object of the present invention is to provide an IIR digital filter that solves the problem of element delay and can perform high-speed arithmetic processing.

To achieve the above object, the digital filter according to the present invention, the first operation in which the first input data at the first timing with the input, the first output data at the first timing is output The circuit and k-th input data at the k-th timing before the (k-1) period delayed by a predetermined time from the first timing are input, and the k-th output data at the k-th timing are output (K is 2, 3, 4,..., N, and n is an integer of 2 or more), and the first arithmetic circuit and the kth arithmetic circuit include the nth arithmetic circuit . An IIR digital filter is configured in which 1-input data and k-th input data are input in parallel , the n first output data and k-th output data are output in parallel , and the biquad circuit is configured. a respective said first operational The first IIR digital filter is the IIR digital filter road has, the multiplication result of the adder quad circuit has to give n period delay before which is delayed a predetermined time than the first time at least first The kIIR digital filter which has a delay element and is the IIR digital filter of the kth arithmetic circuit is delayed by a predetermined time from the kth timing with respect to the operation result of the adder of the biquad circuit. A k-th delay element that gives a delay before a period, and the first IIR digital filter outputs the output result of the first delay element to at least one kIIR digital filter as a delay result that forms a term of its transfer function together are configured to output as at least one of said first kIIR digital The calculation result of the adder having the filter of biquad circuit is characterized in that it is configured to enter a delay result to make the section of the transfer function of itself.

This ensures, by comparing the operating frequency to a conventional it can be a low speed to 1 / n. Since each delay element gives a delay of time nT (T is one sampling time), each operation has only to determine the result within time nT. Therefore, even if the calculation is complicated or the time T is shortened due to high speed, the calculation result is not adversely affected. In this way, it is possible to solve the problem of delay (element delay) of the elements constituting the digital filter and to enable high-speed arithmetic processing.

  Further, n pieces of output data are obtained simultaneously from the arithmetic circuit every time nT. Of the n pieces of output data Y (mT) to Y (mT− (n−1) T), Y ( mT) is the newest operation result, Y (mT) to Y (mT− (n−2) T) are temporally intermediate operation results, and Y (mT− (n−1) T) is the oldest operation result. By using it, it is possible to obtain an operation result equivalent to a conventional IIR digital filter in which Y (mT) to Y (mT− (n−1) T) are sequentially output at every time T.

Furthermore, the digital filter according to the present invention, in the above, n number of the IIR digital filter is characterized in that a structure of connecting the respective plurality of the biquad circuit cascade or in parallel.
Thereby, higher-order filter characteristics can be obtained.
The digital filter according to the present invention, in the above, in front of the first arithmetic circuit and the first k calculating circuit, the input serial data serial - to-parallel conversion, the n first input data and the k A serial-parallel conversion circuit for outputting input data is provided.

Thereby, parallel input data input to the arithmetic circuit every time nT can be obtained from the input serial data.
Furthermore, the digital filter according to the present invention, in the above, the n first output data and the k output data outputted from said first arithmetic circuit and the first k arithmetic circuit parallel - to-serial conversion, A parallel-serial conversion circuit that outputs output serial data is provided.

  Thereby, the parallel output data output from the arithmetic circuit every time nT can be output in order from the oldest output data every time T, and an operation result equivalent to the conventional IIR digital filter can be obtained. it can.

  As described above, the digital filter of the present invention has an effect that it can be a digital filter that can solve the element delay problem and can perform high-speed arithmetic processing.

Hereinafter, embodiments of the present invention will be described with reference to the drawings.
FIG. 1 is a diagram illustrating a digital filter according to an embodiment of the present invention.
The digital filter according to the present embodiment includes an arithmetic circuit including n (n is an integer of 2 or more) IIR (infinite impulse response) digital filters 10, 11,..., 1 (n−1) arranged in parallel.

  Input data U (mT) at a certain timing mT is input to the IIR digital filter 10, and input data U (mT-T) one cycle before mT (mT-T) is input to the IIR digital filter 11, Input data U (mT− (n−1) T) before (n−1) periods of mT is input to the IIR digital filter 1 (n−1). The output data Y (mT) from the IIR digital filter 10, the output data Y (mT-T) from the IIR digital filter 11, and the output data Y (mT- (n-) from the IIR digital filter 1 (n-1). 1) T) is output.

  In the present embodiment, the calculation result M0 (1- (n-1)) of the IIR digital filter 10, the calculation result M1 (0, 2- (n-1)) of the IIR digital filter 11, ..., the IIR digital filter 1 ( The calculation result Mn-1 (0 to (n-2)) of (n-1) is input (cross-tapped) to each other IIR digital filter, so that the operation frequency is changed to the operation frequency of the conventional IIR digital filter. The speed is reduced from Fs to Fs / n.

  The output data Y (mT) to Y (mT− (n−1) T) are simultaneously output at the operating frequency Fs / n, that is, every time nT, and Y (mT) is used as the newest calculation result Y (mT− T) to Y (mT− (n−2) T) are temporally intermediate calculation results, and Y (mT− (n−1) T) is the oldest calculation result, so that it operates at the operating frequency Fs. An operation result equivalent to that of a conventional IIR digital filter is obtained.

  As described above, the digital filter of the present embodiment receives n pieces of input data delayed every predetermined time, and outputs n pieces of output data delayed every predetermined time. With IIR digital filters. These n IIR digital filters input the output results of the delay elements of the respective IIR digital filters.

Next, the case where the digital filter in the present invention is a 1D type secondary IIR digital filter will be described in detail.
FIG. 2 is a diagram illustrating a configuration of a 1D type second-order IIR digital filter in which the parallel processing number n = 2.
As shown in FIG. 2, the parallel arithmetic unit 20 includes an adder 201, a delay element 202, multipliers 203, 204, 205, 206, and 207, and an adder 208, and the parallel arithmetic unit 21 includes an adder 211 and a delay element 212. , Multipliers 213, 214, 215, 216, 217, and adder 218.

Then, the operation result M0 (1) of the parallel operation unit 20 is input to the parallel operation unit 21, and the operation result M1 (0) of the parallel operation unit 21 is input to the parallel operation unit 20.
The input data U (mT) is input to the adder 201 of the parallel operation unit 20, and the operation result X (mT) is output by the adder 201. Delay data X (mT−2T) obtained by giving a delay of 2T (= 2 / Fs) to the X (mT) by the delay element 202 is multiplied by −b2 by the multiplier 203 and input data U ( mT). Further, the parallel calculation unit 20 is supplied with the calculation result X (mT−T) of the adder 211 as the calculation result M1 (0) from the parallel calculation unit 21, and X (mT−T) is The multiplier 204 multiplies -b1 and the adder 201 adds the input data U (mT).

X (mT), X (mT−T), and X (mT−2T) are multiplied by a0, a2, and a1 by multipliers 205, 206, and 207, respectively, and the result is added by adder 208. Is output as the final calculation result Y (mT).
The input data U (mT−T) is input to the adder 211 of the parallel calculation unit 21. Delay data X (mT−3T) obtained by giving a delay of 2T by delay element 212 to X (mT−T) is multiplied by −b2 by multiplier 213 and input data U (mT−T) by adder 211. Is added. Further, the parallel operation unit 21 receives the operation result X (mT-2T) of the delay element 202 as the operation result M0 (1) from the parallel operation unit 20, and X (mT-2T) is The multiplier 214 multiplies −b1 and the adder 211 adds the input data U (mT−T).

X (mT-T), X (mT-2T), and X (mT-3T) are multiplied by a0, a2, and a1 by multipliers 215, 216, and 217, respectively, and the result is added by adder 218. The result is added and output as the final calculation result Y (mT-T).
As described above, the delay element 202 of the parallel arithmetic unit 20 and the delay element 212 of the parallel arithmetic unit 21 are not the time T but a delay element that gives a delay of the time 2T that is twice. That is, the operating frequency is Fs / 2.

  The output result X (mT-2T) of the delay element 202 of the parallel operation unit 20 is input to the parallel operation unit 21. For example, when the parallel processing number n is 3 or more, The output result X (mT-3T) of the delay element 212 is input to the parallel operation unit to which the input data U (mT-2T) is input and the output data Y (mT-2T) is output. Thus, each IIR digital filter is configured to input the output results of the delay elements of each IIR digital filter.

Here, the case where n = 2 and two parallel arithmetic units are provided has been described. However, in the case where n arithmetic units are provided, the delay element in the arithmetic unit gives a delay of n times time nT.
The input data U (mT) of the parallel computing unit 20 is input data that is newer than the input data U (mT-T) of the parallel computing unit 21 by time T.
Y (mT) and M0 (1) in FIG. 2 correspond to Y (mT) and M0 (1) in FIG. 1 described above, and are the newest calculation results. Similarly, if Y = 2 (mT−T) and M1 (0) in FIG. 2 are n = 2, then Y (mT− (n−1) T) and M (n−1) (0) in FIG. It corresponds to the oldest calculation result.

The calculation result X (mT) of the parallel calculation unit 20 and the calculation result X (mT−T) of the parallel calculation unit 21 are represented by the following equations.
X (mT) = U (mT) + (− b1 · M1 (0)) + (− b2 · X (mT−2T)) (4)
X (mT−T) = U (mT−T) + (− b1 · M0 (1)) + (− b2 · X (mT−3T)) (5)
Here, since M0 (1) = X (mT−2T) and M1 (0) = X (mT−T), these are substituted into the above equations (4) and (5), respectively.
X (mT) = U (mT) + (− b1 · X (mT−T)) + (− b2 · X (mT−2T)) (6)
X (mT−T) = U (mT−T) + (− b1 · X (mT−2T)) + (− b2 · X (mT−3T)) (7)
Get.

The final calculation results Y (mT) and Y (mT−T) are respectively expressed by the following equations.
Y (mT) = a0.X (mT) + a1.X (mT-T) + a2.X (mT-2T) (8)
Y (mT-T) = a0 * X (mT-T) + a1 * X (mT-2T) + a2 * X (mT-3T) (9)
Here, the calculation result Q (mT) and the final calculation result P (mT) in the conventional 1D type second-order IIR digital filter shown in FIG. 9 are respectively expressed by the following equations.

Q (mT) = S (mT) + (− b1 × Q (mT−T)) + (− b2 × Q (mT−2T)) (10)
P (mT) = (a0 × Q (mT)) + (a1 × Q (mT−T)) + (a2 × Q (mT−2T)) (11)
Therefore, the input data U (mT) is exactly the same data as the input data S (mT) in FIG. 9, and S (mT) = U (mT), S (mT−T) = U (mT−T). ,..., S (mT− (n−1) T) = U (mT− (n−1) T), the comparison between the expression (10) and the expression (6), and (11 The comparison between the equation (1) and the equation (8) shows that the 1D type second order IIR digital filter shown in FIG. 2 obtains the same result as the conventional 1D type second order IIR digital filter shown in FIG. That is, Y (mT) = P (mT).

Also, from the above equation (10), the calculation result Q (mT−T) one cycle before mT is:
Q (mT−T) = U (mT−T) + (− b1 · Q (mT−2T)) + (− b2 · Q (mT−3T)) (12)
It becomes.
Therefore, if the input data U (mT) and the input data S (mT) are the same data by comparing the equation (7) and the equation (12), X (mT−T) = Q (mT -T).

Accordingly, the calculation results X (mT) and Q (mT) are equal, and the calculation results X (mT−T) and Q (mT−T) one cycle before are also equal. That is, Q (mT) = X (mT), Q (mT−T) = X (mT−T),..., Q (mT− (n−1) T) = X (mT− (n−1) T ).
From the above equation (10), the final calculation result P (mT−T) one cycle before mT is:
P (mT−T) = (a0 · Q (mT−T)) + (a1 · Q (mT−2T)) + (a2 · Q (mT−3T)) (13)
It becomes.

Therefore, if the input data U (mT) and the input data S (mT) are the same data by comparing the equation (9) with the equation (13), Y (mT−T) = P (mT -T).
In the present embodiment, the calculation result Y (mT−T) and Y (mT) are simultaneously output as the final result, but Y (mT) is the latest calculation result and Y (mT−T) is time T before. As a result of the calculation, the same calculation result as that of the conventional 1D type secondary IIR digital filter of FIG. 9 operating at the operating frequency Fs can be obtained.

  As described above, in the first embodiment, a plurality of IIR digital filters are connected in parallel, and the output results of the delay elements included in each IIR digital filter are input (cross-tapped) to each other, so that the operating frequency Fs is set to 1. Even when the operation time is reduced to n times (n is the number of parallel processes) and the computation time is n times nT, a result equivalent to the conventional IIR digital filter can be obtained. As a result, the problem of calculation time due to element delay can be solved and high-speed calculation processing can be performed.

Next, a second embodiment of the present invention will be described.
In the second embodiment, a 2D type IIR digital filter is applied to the 1D type second order IIR digital filter in the first embodiment described above.
FIG. 3 is a diagram illustrating a configuration of a 2D type second-order IIR digital filter in which the parallel processing number n = 2.

M0 ′ (1) and M0 ″ (1) in FIG. 3 correspond to M0 (1) in FIG. 1, and M1 ′ (0) and M1 ″ (0) in FIG. 3 correspond to M1 (0) in FIG. And each means a cross tap.
As shown in FIG. 3, the parallel operation unit 30 includes multipliers 300, 301, and 302, an adder 303, a delay element 304, an adder 305, and multipliers 306, 307, 308, and 309. Multipliers 310 and 311, a delay element 312, an adder 313, a delay element 314, an adder 315, a multiplier 316, and a delay element 317 are provided.

The delay element 304 of the parallel arithmetic unit 30 and the delay elements 312, 314, and 317 of the parallel arithmetic unit 31 are not the time T, but are delay elements that give a delay of 2 times of time 2T. That is, the operating frequency is Fs / 2.
The 2D type second-order IIR digital filter of the present embodiment inputs input data U (mT) and input data U (mT-T) one cycle before mT in parallel, and Y (mT) every 2T. Y (mT-T) is output simultaneously. At this time, by treating Y (mT) as the latest calculation result and Y (mT−T) as the calculation result of the previous cycle, the conventional 2D type secondary IIR digital of FIG. 10 operating at the operating frequency Fs described above. The same calculation result as the filter is obtained.

As described above, in the second embodiment, similarly to the first embodiment described above, the problem of calculation time due to element delay can be solved, and high-speed calculation processing can be performed.
In the second embodiment, as shown in FIG. 3, the case where the delay elements 312 and 317 are provided in the parallel operation unit 31 has been described. However, the delay elements 312 and 317 may be provided in the parallel operation unit 30. Similar results are obtained.

Next, a third embodiment of the present invention will be described.
In the third embodiment, a 1D type second-order IIR digital filter is applied in the first embodiment described above, but a 3D type IIR digital filter is applied.
FIG. 4 is a diagram illustrating a configuration of a 3D type second-order IIR digital filter in which the parallel processing number n = 2.

M0 ′ (1) and M0 ″ (1) in FIG. 4 correspond to M0 (1) in FIG. 1, and M1 ′ (0) and M1 ″ (0) in FIG. 4 correspond to M1 (0) in FIG. And each means a cross tap.
As shown in FIG. 4, the parallel operation unit 40 includes a delay element 401, multipliers 402, 403, and 404, an adder 405, multipliers 406 and 407, and a delay element 408. The parallel operation unit 41 includes a delay element 411, Multipliers 412, 413, 414, an adder 415, multipliers 416, 417, and a delay element 418 are provided.

The delay elements 401 and 408 of the parallel arithmetic unit 40 and the delay elements 411 and 418 of the parallel arithmetic unit 41 are not the time T but a delay element that gives a delay of 2 times of 2T. That is, the operating frequency is Fs / 2.
The 3D type second-order IIR digital filter of the present embodiment inputs input data U (mT) and input data U (mT-T) one cycle before mT in parallel, and Y (mT) every 2T. Y (mT-T) is output simultaneously. At this time, by treating Y (mT) as the latest calculation result and Y (mT−T) as the calculation result of the previous cycle, the conventional 3D type secondary IIR digital of FIG. 11 operating at the operating frequency Fs described above. The same calculation result as the filter is obtained.

  As described above, in the third embodiment, similarly to the first embodiment described above, the problem of calculation time due to element delay can be solved, and high-speed calculation processing can be performed.

Next, a fourth embodiment of the present invention will be described.
In the fourth embodiment, the 1D type second-order IIR digital filter in the first embodiment described above is used as a biquad circuit.
FIG. 5 is a diagram showing an example in which 1D type biquad circuits are connected in cascade.
As shown in FIG. 5, in this embodiment, it is assumed that i + 1 (i: natural number) biquad circuits 50 to 5i are connected in cascade. Here, as the biquad circuits 50 to 5i, a 1D type biquad circuit with the number n = 2 of parallel processing, in which the same cross tap as in FIG. 2 is performed, is applied. That is, the operating frequency is Fs / 2.

The input data U (mT) is new input data by time T from the input data U (mT-T).
M00 (1) and M10 (0) represent cross taps of the biquad circuit 50, and M01 (1) and M11 (0) represent cross taps of the biquad circuit 51. M0i (1) and M1i (0) represent cross taps of the biquad circuit 5i.

The digital filter shown in FIG. 5 outputs Y (mT) and Y (mT-T) simultaneously every time 2T. At this time, by treating Y (mT) as the latest calculation result and Y (mT−T) as the calculation result of the previous cycle, the conventional 1D type cascade connection IIR of FIG. 12 operating at the operating frequency Fs described above. The same calculation results as the digital filter can be obtained.
The digital filter of the present embodiment is the 2 × (i + 1) -order or ((2 × (i + 1)) − 1) -order, similar to the configuration in which the conventional 1D type biquad circuits shown in FIG. 12 are connected in cascade. Filter characteristics.

As described above, in the fourth embodiment, high-order filter characteristics can be realized by connecting the biquad circuits in which cross taps are performed in cascade.
In the fourth embodiment, the case where the 1D type second-order IIR digital filter shown in the first embodiment is applied as the biquad circuit has been described. Instead, the second embodiment shows the first embodiment. The 2D type secondary IIR digital filter and the 3D type secondary IIR digital filter shown in the third embodiment can also be applied.

Next, a fifth embodiment of the present invention will be described.
In the fifth embodiment, the biquad circuits in which the cross tap is performed are connected in cascade in the fourth embodiment described above, whereas the biquad circuits in which the cross tap is performed are connected in parallel. It is a thing.

FIG. 6 is a diagram illustrating an example in which 1D type biquad circuits are connected in parallel.
As shown in FIG. 6, in this embodiment, it is assumed that i + 1 biquad circuits 60 to 6i are connected in parallel. The biquad circuits 60 to 6i are 1D type biquad circuits in which the same cross tap as in FIG. That is, the operating frequency is Fs / 2.
The input data U (mT) is new input data by time T from the input data U (mT-T).

M00 (1) and M10 (0) represent cross taps of the biquad circuit 60, and M0i (1) and M1i (0) represent cross taps of the biquad circuit 6i.
The digital filter shown in FIG. 6 outputs Y (mT) and Y (mT-T) simultaneously every time 2T. At this time, by treating Y (mT) as the latest calculation result and Y (mT−T) as the calculation result of the previous cycle, the conventional 1D type parallel connection IIR of FIG. 13 that operates at the operating frequency Fs described above. The same calculation results as the digital filter can be obtained.

The digital filter of the present embodiment is the 2 × (i + 1) -order or ((2 × (i + 1)) − 1) -order, similar to the configuration in which the conventional 1D type biquad circuit shown in FIG. 13 is connected in parallel. Filter characteristics.
Thus, in the said 5th Embodiment, a high-order filter characteristic is realizable by connecting the biquad circuit which implemented the cross tap in parallel.

  In the fifth embodiment, the case where the 1D type second-order IIR digital filter shown in the first embodiment is applied as the biquad circuit has been described. Instead, the second embodiment shows the first embodiment. The 2D type secondary IIR digital filter and the 3D type secondary IIR digital filter shown in the third embodiment can also be applied.

Next, a sixth embodiment of the present invention will be described.
In the sixth embodiment, a circuit for transferring data is added to the preceding stage and the subsequent stage of the arithmetic circuit in the first to fifth embodiments.
FIG. 7 is a diagram illustrating a configuration of a digital filter according to the sixth embodiment.
As shown in FIG. 7, the digital filter of this embodiment includes a serial-parallel converter 70, an IIR digital filter (arithmetic circuit) 71 including n parallel arithmetic units, and a parallel-serial converter 72. Yes.

Here, the configuration of the IIR digital filter 71 is the same as that of the IIR digital filter having the cross tap shown in FIG. 1 described above, and the specific configuration is the same as that of FIGS. 2 to 6 described above. And
The serial-parallel converter 70 and the parallel-serial converter 72 are input with a clock signal SCLK having a frequency Fs, which is an operation clock of each.

The serial-parallel converter 70 and the IIR digital filter 71 are supplied with a clock signal PCLK having a frequency Fs / n, which serves as an operation clock for each.
The serial-parallel converter 70 performs serial-parallel conversion on the input serial data S (mT), and n parallel input data U (mT) to U (mT− (n−1) in synchronization with the clock signal PCLK. ) T) is output to the IIR digital filter 71.

At this time, U (mT) is the newest input data of the IIR digital filter 71, and U (mT-T) to U (mT- (n-2) T) are temporally intermediate input data of the IIR digital filter 71, U (mT− (n−1) T) is the oldest input data of the IIR digital filter 71.
From the IIR digital filter 71, operation results Y (mT) to Y (mT− (n−1) T) are simultaneously output in synchronization with the clock signal PCLK. Here, Y (mT) is the newest calculation result, Y (mT−T) to Y (mT− (n−2) T) are intermediate calculation results, and Y (mT− (n−1) T. ) Is the oldest calculation result.

  The calculation results Y (mT) to Y (mT− (n−1) T) are input to the parallel-serial converter 72, and the parallel-serial converter 72 converts them into the oldest calculation result Y (mT− (n -1) Rearrange in order from T) to the latest calculation result Y (mT), and output as P (mT) in synchronization with the clock signal SCLK in order from the oldest calculation result.

Next, the operation timing of this embodiment will be described.
FIG. 8 is a diagram showing the operation timing when the number of parallel processes n = 2.
Here, the frequency of the clock signal SCLK is Fs, and since the parallel processing number n = 2, the frequency of the clock signal PCLK is Fs / 2.
As shown by S (mT) in FIG. 8, the serial-parallel converter 70 inputs Sj (j: natural number) as input data in synchronization with the clock signal SCLK. The inputted S0 and S1 are serial-parallel converted by the serial-parallel converter 70, and are simultaneously output as U (mT-T) and U (mT) in synchronization with the clock signal PCLK.

The IIR digital filter 71 performs an operation based on the input data U (mT) and U (mT-T) corresponding to S1 and S0, and Y1 and Y0 are calculated as operation results Y (mT) and Y (mT−T). ) In synchronization with the clock signal PCLK.
The parallel-serial converter 72 performs parallel-serial conversion on the calculation results Y (mT) and Y (mT-T), and outputs the calculation result Y (mT-T) corresponding to Y0 in synchronization with the clock signal SCLK. , Y (mT) corresponding to Y1 is output later. That is, the output P (mT) can be obtained at the same transfer rate as the input S (mT).

By repeating the above operation, the same calculation result P (mT) as that of the conventional IIR digital filter is obtained.
As described above, in the sixth embodiment, the serial-parallel converter that operates at the operating frequency Fs on the input side of the IIR digital filter that operates at the operating frequency Fs / n, and the parallel that operates at the operating frequency Fs on the output side. Since each serial converter is provided, the same calculation result as that of the conventional IIR digital filter operating at the operating frequency Fs can be obtained.

In the sixth embodiment, the case where the serial-parallel converter 70, the IIR digital filter 71, and the parallel-serial converter 72 are operated at the timing when SCLK or PCLK rises at the operation timing shown in FIG. However, it can also be operated at the timing when SCLK or PCLK falls.
In the sixth embodiment, the case where both the serial-parallel converter 70 and the parallel-serial converter 72 are provided has been described. However, only one of them may be provided. If the serial-parallel converter 70 is installed on the input side of the IIR digital filter 71, n parallel input data U (mT) to U (mT) delayed every time T from the input serial data S (mT). -(N-1) T) can be generated every time nT, and if a parallel-serial converter 72 is installed on the output side of the IIR digital filter 71, it is simultaneously output from the IIR digital filter 71 every time nT. The output data Y (mT) to Y (mT− (n−1) T) can be output in order from the oldest data every time T.

  In each of the above embodiments, a case where a 1D type secondary IIR digital filter, a 2D type secondary IIR digital filter, and a 3D type secondary IIR digital filter are applied as the IIR digital filter has been described. However, the present invention is not limited to this. However, the present invention can be applied to any type of IIR digital filter.

It is a digital filter in one embodiment of the present invention. It is a figure which shows the structure of the 1D type secondary IIR digital filter in 1st Embodiment. It is a figure which shows the structure of the 2D type 2nd-order IIR digital filter in 2nd Embodiment. It is a figure which shows the structure of the 3D type secondary IIR digital filter in 3rd Embodiment. It is a figure which shows the 1D type biquad circuit (cascade connection) in 4th Embodiment. It is a figure which shows the 1D type biquad circuit (parallel connection) in 5th Embodiment. It is a figure which shows the digital filter in 6th Embodiment. It is a figure for demonstrating the operation timing in 6th Embodiment. It is a figure which shows the structure of the conventional 1D type 2nd-order IIR digital filter. It is a figure which shows the structure of the conventional 2D type 2nd-order IIR digital filter. It is a figure which shows the structure of the conventional 3D type secondary IIR digital filter. It is a figure which shows the conventional 1D type biquad circuit (cascade connection). It is a figure which shows the conventional 1D type biquad circuit (parallel connection).

Explanation of symbols

10-1 (n-1) IIR digital filter 20,21 1D type secondary IIR digital filter 30,31 2D type secondary IIR digital filter 40,41 3D type secondary IIR digital filter 50-5i, 60-6i Bike Quad Circuit 70 Serial-parallel converter 71 IIR digital filter (arithmetic circuit)
73 Parallel-serial converter

Claims (4)

Together with the first input data at the first timing is entered, a first arithmetic circuit which first output data at the first timing is output, is delayed for a predetermined time than the first time (k -1) k-th arithmetic circuit (k is 2, 3, 4,...) In which k-th input data at the k-th timing before the cycle is inputted and k-th output data at the k-th timing is outputted. N and n is an integer of 2 or more)
Said first arithmetic circuit and the first k arithmetic circuit, together with the n first input data and the k input data are input in parallel, said n first output data and the k output data in parallel It is to output, and having respectively an IIR digital filter composed of a biquad circuit,
The first IIR digital filter which is the IIR digital filter included in the first arithmetic circuit has a delay of n cycles before the first timing with respect to the arithmetic result of the adder included in the biquad circuit. Providing at least a first delay element;
The kIIR digital filter, which is the IIR digital filter included in the kth arithmetic circuit, gives a delay of n cycles before the kth timing to the arithmetic result of the adder included in the biquad circuit. A kth delay element;
The first IIR digital filter is configured to output the output result of the first delay element to at least one kIIR digital filter as a delay result that creates a term of the transfer function , and at least one A digital filter configured to input an operation result of the adder included in a biquad circuit of the kIIR digital filter as a delay result for creating a term of its own transfer function . Said n IIR digital filter, the digital filter according to claim 1, characterized in that the configuration of connecting the respective plurality of the biquad circuit cascade or in parallel. In front of the first arithmetic circuit and the first k calculating circuit, the input serial data serial - be provided with a parallel conversion circuit - to parallel conversion, serial outputting said n first input data and the k input data The digital filter according to claim 1 or 2 . Providing the serial converter circuit - the first arithmetic circuit and the n first output data and the k output data outputted from the k-th calculation circuits parallel - parallel to and serial conversion, and outputs the output serial data The digital filter according to any one of claims 1 to 3 .

Images