JP2622962B2 - Zooming device for FFT analyzer - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION <Industrial Application Field> The present invention relates to an improvement of a zooming device used for an FFT (fast fourier transform) analyzer.

<Prior Art> An FFT analyzer performs an arithmetic process on time-series data to measure its spectrum. In order to obtain a more detailed spectrum, an operation of expanding a frequency axis thereof is performed by a zooming device. To perform zooming, the sampling frequency may be reduced by thinning out a time-series signal, which is time-series data. However, such processing causes an aliasing phenomenon in which a high frequency component overlaps a low frequency component. Therefore, high frequency components must be cut using a low-pass filter. A device using a digital filter is used as the low-pass filter. FIG. 3 shows the configuration of a zooming device using a digital filter. In FIG. 3, reference numeral 1 denotes a multiplier, which receives digitized time-series data and a sine wave signal as a carrier signal, and multiplies these signals.
Since the frequency component of the time-series data is shifted to a lower band by the multiplier 1, the subsequent processing is simplified. 2 is IIR (inf
inite implse response) is a digital filter to which the output of the multiplier 1 is input and filtering is performed by digital operation. Reference numeral 3 denotes a RAM to which the output of the IIR digital filter 2 is input and stored. The data stored in the RAM 3 is thinned out and returned to the IIR digital filter 2 again. In this way, a plurality of stages of filtering are executed and output. This output is not shown
The spectrum is analyzed by the FFT analyzer. That is, the high zoom ratio and the variability of the zoom ratio are satisfied by using the IIR digital filter in multiple stages.

FIG. 4 shows the principle configuration of the IIR digital filter. IIR
The digital filter outputs the n-th time series output as y
(N), and let the n-th time series input be x (n), Is represented by The configuration shown in FIG. 4 is for executing this operation, wherein 4 and 5 are adders, 6 is a register for holding time-series data, and 7 is a multiplier. The multiplier 7 outputs multiplies the data input to the multipliers and the constant _{a 1 ~a N-1, b} 0 ~b N-1 listed on it. The time-series input data x (n) is input to the adder 4, and the sum of the data previously stored in the register 6 and a value obtained by multiplying the data by constants a _{1 to} a _N−1 is obtained and stored in the register 6. Is done. The value obtained by multiplying the output of the adder 4 by the constant b _O is input to the adder 5, and the sum of the value stored in the register 6 and the value obtained by multiplying the data by the constants b _{1 to} b _N−1 is obtained. (N). The data stored in the register 6 is sequentially shifted. In this way, the calculation of equation (1) is performed.

<Problems to be solved by the invention> However, such a zooming device of the FFT analyzer uses an IIR digital filter, and thus has the following problems.

(1) The operation accuracy is low because of the recursive type, and an extremely long word length is required to improve the characteristics of the filter.

(2) The phase is not linear and it is difficult to use.

(3) Since it is a recursive type, an operation is required for all input data despite being a thinning filter.

(4) There is no special design method or transfer function structure that reduces the number of multiplications, and it is difficult to increase the speed.

<Object of the Invention> An object of the present invention is to provide a zooming device of an FFT analyzer having a simple configuration and having a high-precision zoom function.

<Means for solving the problems> In order to solve the above problems, in the present invention, a zooming device used in an FFT analyzer that performs arithmetic processing on time series data and obtains the spectrum of this time series data, A multiplexer to which time-series data is input, a half-band filter to which the output of the multiplexer is input, a storage unit to which the output of the half-band filter is input, and control the multiplexer, the half-band filter, and the storage unit, respectively. A controller, a register for receiving the output of the half-band filter when the required number of stages is completed, temporarily holding the output of the half-band filter, a RAM for storing data held in the register, and a predetermined number of When data is stored in the RAM, the spectrum of the time-series data An FFT analyzer that is obtained by an operation, under the control of the controller, the output of the storage unit is thinned out and input to the multiplexer, and the multiplexer selects the time-series data at the first stage of the filter operation in the half-band filter. In the second and subsequent stages, data is selected from the storage means.

<Embodiment> FIG. 1 shows an embodiment of a zooming apparatus for an FFT analyzer according to the present invention. In FIG. 1, reference numeral 10 denotes a multiplier to which time-series data and a sine wave signal as a carrier signal are input, and these inputs are multiplied to shift the frequency band of the time-series data to a lower frequency side. Reference numeral 11 denotes a multiplexer to which the output of the multiplier 10 is input. The data stored in the storage means 13 is also input to the multiplexer 11. Reference numeral 12 denotes a half-band filter to which the output of the multiplexer 11 is input. This half-band filter
The output data of 12 is input to the storage means 13 and stored.
Reference numeral 14 denotes a register to which the output of the half-band filter 12 is input and temporarily stored. Reference numeral 15 denotes a RAM, into which data temporarily stored in the register 14 is input and stored. Reference numeral 16 denotes an FFT analyzer which performs an FFT operation based on the data stored in the RAM 15 to obtain a spectrum. 1
Reference numeral 7 denotes a controller that controls the multiplexer 11, the half-band filter 12, and the storage unit 13.

Next, the operation of this embodiment will be described. Since a filter used in a zooming apparatus requires high performance and a high zoom ratio and variability of the zoom ratio are required, it is difficult to satisfy the required performance by only one filtering. Therefore, a multistage that supplies data to the same filter a plurality of times is used.

Under the control of the controller 17, a series of time-series input data is input to the half-band filter 12 via the multiplexer 11, where a predetermined filter operation is performed. The calculated result is stored in the storage unit 13 based on the control of the controller 17.
Is stored in In order to thin out the stored data, every other data is taken under the control of the controller 17 and input to the multiplexer 11 again. That is, the capacity of the storage means 13 is large enough to hold twice as many data as the number of stages. The multiplexer 11 selects the time series data from the multiplier 10 in the first stage of the filter operation, and the storage unit 13 in the second and subsequent stages.
Is selected. When the required number of stages is completed, the output of the half-band filter 12 is input to the register 14, and the multiplexer 11 selects the time-series data from the multiplier 10 and proceeds to the processing of the next series of time-series data.
The data held in the register 14 is stored in the RAM 15, and when a predetermined number of data is accumulated, the spectrum is calculated by the FFT analyzer 16.

Next, a half-band filter will be described. The half-band filter is a kind of FIR (finite impulse filter) type filter, and this filter is, for example, Ronald E. Cro
Multirate Digi by chiere & Lawrence R. Rabiner
It is described in tal Signal Processing and the like, and is already well known. The relationship between the output y (n) and the input x (n) of this filter is as follows: y (n) = Σh (k) × (nk) (2) h (k): impulse response coefficient There is. That is, it is a non-recursive filter whose output does not return to its input. Further, in the half-band filter, there is a special relation between the impulse response coefficients h (k). If the order of the filter is 2N + 1 and k is 0 to 2N, when N is an odd number, h (1) = h ( 3) =... = H (2N-1) = 0 (where h
(Excluding (N)) h (2i) = h (2 (N−i)) {i: 0 to (N−1) / 2} When N is an even number, h (0) = h (2) =. … = H (2N) = 0 (however, h
(Excluding (N)) h (2i + 1)) = h (2 (N−i) −1) {i: 0 to N / 2} (3) FIG. 2 shows an example of the configuration of this half-band filter. This embodiment shows a case where N is an odd number. In FIG. 2, 181 to 1891 are registers,
Connected vertically. The time-series data x (n) is input to the register 181 and is sequentially shifted one by one. Reference numerals 191 to 193 denote adders, which add the outputs of registers having the same impulse response coefficient in the equation (3). For example, the outputs of the registers 181 and 1891 are input to the adder 191 and added. Since the impulse response coefficients corresponding to the outputs of every other register 182, 184,... Are 0, they are not added. Reference numerals 201 to 203 denote multipliers which multiply outputs of the adders 191 to 193 by impulse response coefficients h (0) to h (N-1). The output of the register 186 is output from the multiplier 204 by an impulse response coefficient h.
(N). 21 is an adder, multipliers 201 to 204
Add the outputs of In this way, the filter operation shown in the above equation (2) is performed.

Note that the filter characteristics of the half-band filter 12 are gradual in the first stage and stricter in the second stage.
Therefore, the controller 17 switches the impulse response coefficient of the half-band filter 12.

The controller 17 also takes out output data of the half-band filter 12 stored in the storage means 13 every other one and performs so-called thinning-out of 2 as an input to the next stage.
Therefore, the number of data becomes smaller as the stage becomes later, so that the next stage is performed once for two stages of the former stage. This scheduling is performed by controller 17
Perform at

Furthermore, in this embodiment, the number of thinnings is two, but another number may be used.

<Effects of the Invention> As specifically described above based on the embodiments, the present invention uses a half-band filter as a filter,
Under the control of the controller 17, time-series input data is used in multiple stages. That is, the controller
Under the control of 17, the signal is input to the half-band filter via the multiplexer, the output of the half-band filter is stored in the storage means, and the stored data is thinned out and returned to the read-out multiplexer, so that the data is used in the multistage. I made it. Therefore, an anti-aliasing filter for zooming can be realized with a simple configuration, so that the configuration of the device is simplified.

In addition, since the FIR filter is a non-recursive FIR filter, the filter operation only needs to be performed once for two data, the speed can be increased, and the operation error can be reduced as compared with the number of operations.

Also, the number of multipliers can be reduced to half of the order by using the symmetry of the impulse response coefficient of the half-band filter, and every other coefficient becomes 0 except for the coefficient at the center of the filter. Can be 2. Also, the number of registers or the size of other storage means instead of registers can be reduced to 3/4 of the order.

Further, if the operation attenuation in the stop band is increased from the characteristics of the half-band filter, the ripple in the pass band is automatically reduced, and the phase can be made linear.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an embodiment of a zooming device of an FFT analyzer according to the present invention, FIG. 2 is a block diagram showing an example of a configuration of a half-band filter, and FIG.
FIG. 4 is a block diagram showing a configuration of a zooming device of the FFT analyzer, and FIG. 4 is a block diagram showing a configuration of an IIR digital filter. 10 Multiplier, 11 Multiplexer, 12 Half-band filter, 13 Storage means 14,181 to 189,1890,189
1… Register, 15… RAM, 16… FFT analyzer, 17
…… Controller, 21,191-193 …… Adder, 201-20
4 ... Multiplier.

Claims (1)

(57) [Claims] 1. A zooming apparatus for use in an FFT analyzer which performs arithmetic processing on time-series data and obtains a spectrum of the time-series data, wherein a multiplexer to which the time-series data is input and an output of the multiplexer are input. A half-band filter, storage means for receiving an output of the half-band filter, a controller for controlling the multiplexer, the half-band filter and the storage means, and an output of the half-band filter when the required number of stages is completed. A register for temporarily holding the output of the half-band filter, a RAM for storing the data held in the register, and, when a predetermined number of data is accumulated in the RAM, the spectrum of the time-series data is calculated. The FFT analyzer that you want, Control, the output of the storage means is thinned out and input to the multiplexer, and the multiplexer selects the time-series data at the first stage of the filter operation in the half-band filter, and at the second and subsequent stages, outputs the data from the storage means. FFT analyzer zooming device configured to select data.