U.S. patent application number 09/890537 was filed with the patent office on 2002-12-19 for oversampling circuit and digital/analog converter.
Invention is credited to Koyanagi, Yukio.
Application Number | 20020190883 09/890537 |
Document ID | / |
Family ID | 18467670 |
Filed Date | 2002-12-19 |
United States Patent
Application |
20020190883 |
Kind Code |
A1 |
Koyanagi, Yukio |
December 19, 2002 |
Oversampling circuit and digital/analog converter
Abstract
It is object to provide an oversampling circuit and a digital to
analog converter capable of realizing a smaller circuit and
reducing a cost of parts. The oversampling circuit comprises
multiplying section 1, four data holding sections 2-1 through 2-4,
four data selectors 3-1 through 3-4, an adding section 4, and two
integrating circuits 5-1 and 5-2. Input data is multiplied by four
multiplicators by the multiplying section 1, and four
multiplication results held, as one set, in the data holding
sections. The data selectors read out the data held in the four
data holding sections in a predetermined order and generate step
function data. The adding section adds the values of four step
functions outputted from the respective data selectors, and then
digital integrating operations corresponding to the sum are carried
out by means of two integrating circuits.
Inventors: |
Koyanagi, Yukio; (Saitama,
JP) |
Correspondence
Address: |
DELLETT AND WALTERS
310 S.W. FOURTH AVENUE
SUITE 1101
PORTLAND
OR
97204
US
|
Family ID: |
18467670 |
Appl. No.: |
09/890537 |
Filed: |
July 31, 2001 |
PCT Filed: |
December 15, 2000 |
PCT NO: |
PCT/JP00/08901 |
Current U.S.
Class: |
341/144 |
Current CPC
Class: |
H03M 3/508 20130101;
H03H 17/0657 20130101; H03H 17/028 20130101 |
Class at
Publication: |
341/144 |
International
Class: |
H03M 001/66 |
Foreign Application Data
Date |
Code |
Application Number |
Dec 17, 1999 |
JP |
11-360053 |
Claims
1. An oversampling circuit, characterized by comprising: a
multiplying unit for performing a plurality of multiplying process
using plural multiplicators on plural pieces of digital data
inputted at a predetermined intervals; a plurality of step function
generation unit for generating step functions corresponding to each
of the plural pieces of digital data using plural of multiplication
result obtained by the multiplying unit synchronized with an input
timing of each of the plural pieces of digital data; an addition
unit for performing a process of adding up values of the step
functions generated by the plurality of step function generation
unit; and an integrating unit for performing a digital integrating
process plural times on output data from the addition unit.
2. The oversampling circuit according to claim 1, characterized in
that each of the multiplicators used in the multiplying processes
by the multiplying unit corresponds to each of the values of step
functions obtained by differentiating plural times piecewise
polynomials for a predetermined sampling function configured by the
piecewise polynomials.
3. The oversampling circuit according to claim 2, wherein said step
function comprises a positive region and a negative region set to
have an equal area.
4. The oversampling circuit according to claim 3, wherein said
sampling function is differentiable only once over the whole range
and has values of local support.
5. The oversampling circuit according to claim 2, characterized in
that said step function consists of eight piecewise sections in
equal width with a weight of -1, +3, +5, -7, -7, +5, +3, and -1 in
a predetermined range corresponding to said five digital data
arranged at an equal interval, and that the eight weight
coefficients are set as the multiplicators of said multiplying
unit.
6. The oversampling circuit according to claim 5, characterized in
that a multiplying process performed by said multiplying unit is
realized by adding said digital data to an operation result of an
exponentiation of 2 by a bit shift.
7. The oversampling circuit according to claim 1, characterized in
that times of said digital integration is two, and data whose value
changes like a quadric function is output from said integrating
unit.
8. The oversampling circuit according to claim 1, characterized in
that said digital integration performed by said integrating unit is
an operating process of accumulating input data, and n times of an
oversampling process is performed by repeatedly performing the
operating process n times in one period of inputting the digital
data.
9. A digital-to-analog converter, comprising at a stage subsequent
to said oversampling circuit according to claim 1: voltage
generation unit for generating an analog voltage corresponding to a
value of data output by said integrating unit; and smoothing unit
for smoothing the analog voltage generated by said voltage
generation unit.
Description
TECHNICAL FIELD
[0001] The present invention relates to an over sampling circuit
for interpolating input data discretely and a digital-to-analog
converter to which the oversampling circuit is applied. In this
specification, it is assumed that a case where function values have
finite values except zero in a local region and become zero in
regions different from the region is called a "local support."
BACKGROUND ART
[0002] A recent digital audio apparatus, for example, a CD (Compact
Disk) player, uses a D/A (digital-to-analog) converter to which an
over-sampling technique is applied to obtain a continuous analog
audio signal from discrete music data (digital data). Such a D/A
converter generally uses a digital filter to raise a pseudo
sampling frequency by interpolating input digital data, and outputs
smooth analog audio signals by passing each interpolation value
through a low-pass filter after generating a staircase signal
waveform with each interpolation value held by the sample holding
circuit.
[0003] A data interpolation system disclosed in WO99/38090 is well
known as a method of interpolating data into discrete digital data.
In this data interpolation system, differentiation can be performed
only once in the whole range, and a sampling function is used such
that two sampling points each before and after an interpolation
position, that is, a total of four sampling points, can be
considered. Since the sampling function has values of a local
support unlike the sinc function defined by sin (.pi.ft)/(.pi.ft)
where f indicates a sampling frequency, there is a merit that no
truncation errors occur although only four pieces of digital data
are used in the interpolating operation.
[0004] Generally, oversampling is performed by using a digital
filter in which the waveform data of the above mentioned sampling
function is set to a tap coefficient of an FIR (finite impulse
response) filter.
[0005] If the oversampling technology of performing an
interpolating operation for discrete digital data using the above
mentioned digital filter, a low pass filter having a moderate
attenuation characteristic can be used. Therefore, the phase
characteristic with a low pass filter can approach a linear phase
characteristic, and the sampling aliasing noise can be reduced.
These effects are more outstanding with a higher oversampling
frequency. However, if the sampling frequency becomes higher, the
number of taps of the digital filter is also increased. As a
result, there arises the problem of a larger circuit. In addition,
the performance of the delay circuit or multiplier comprises the
digital filter is also sped up. Therefore, it is necessary to use
expensive parts appropriate for the quick performance, thereby
increasing the cost of the required parts. Especially, when the
oversampling process is performed using a digital filter, an actual
value of a sampling function is used as a tap coefficient.
Therefore, the configuration of a multiplier is complicated, and
the cost of the parts furthermore increases.
[0006] Moreover, although a digital-to-analog converter can be
configured by connecting a low pass filter after the oversampling
circuit, the above mentioned various problems with the conventional
oversampling circuit have also occurred with the digital-to-analog
converter configured using the circuit.
BRIEF SUMMARY OF THE INVENTION
[0007] The present invention has been achieved to solve the above
mentioned problems, and aims at providing an oversampling circuit
and a digital-to-analog converter having a smaller circuit at a
lower cost of parts.
[0008] In the oversampling circuit according to the present
invention, a multiplying unit performs a plurality of multiplying
processes using plural multiplicators on a plurality of digital
data input at predetermined intervals, and using the plurality of
multiplication result, step function is generated corresponding to
each inputted digital data. By performing digital integration
plural times on the addition results obtained by addition unit
adding up values of the step function corresponding to each digital
data, digital data whose values change stepwise is output along a
smooth curve. Thus, the values of step function corresponding to
sequentially input plural pieces of digital data are added up, and
then the digital integration is performed on the addition result.
As a result, output data whose values smoothly change can be
obtained. Therefore, when an oversampling frequency is high, it is
necessary only to speed up the digital integration, thereby
avoiding the conventional complicated configuration, that is,
simplifying the configuration, and reducing the cost of parts.
[0009] Each of the multiplicators used in the multiplying processes
by the multiplying unit is desired to correspond to each of the
values of step functions obtained by differentiating plural times
piecewise polynomials for a predetermined sampling function
configured by the piecewise polynomials. That is, by integrating
plural times the above mentioned step function, a waveform
corresponding to the predetermined sampling function can be
obtained. Therefore, a convolution operation using a sampling
function can be equivalently realized by generating a step
function. As a result, the contents of the entire process can be
simplified, and the number of processes required oversampling can
be successfully reduced.
[0010] In addition, the above mentioned step function is desired to
equally set the positive and negative areas. Thus, the divergence
of integration results of the integrating unit can be
prevented.
[0011] Furthermore, it is desirable that the above mentioned
sampling function has a value of local support with the whole range
differentiable only once. It is assumed that a natural phenomenon
can be approximated if the whole range is differentiable only once.
By setting a smaller number of times of differentiation, the times
of the digital integration performed by the integrating unit can be
reduced, thereby successfully simplifying the configuration.
[0012] It is further desirable that the above mentioned step
function contains an area of eight piecewise sections in equal
width weighted by -1, +3, +5, -7, -7, +5, +3, and -1 in a
predetermined range corresponding to five pieces of digital data
arranged at equal intervals, and that the eight weight coefficients
are set as the respective multiplicators of multiplying unit. Since
simple weight coefficients represented by integers can be used as
the multiplicators in the multiplying unit, the multiplying process
can be simplified.
[0013] Especially, it is desirable that a multiplying process
performed in the multiplying unit is represented by adding digital
data to an operation result of the exponentiation of 2 by a bit
shift. Since the multiplying process can be replaced with a bit
shift process and an adding operation, the configuration can be
simplified and the process can be sped up by simplifying the
contents of the processes.
[0014] It is also desirable that the times of the digital
integration is two, and a data whose value changes like a quadric
function is output from the integrating unit. For smooth
interpolating plural pieces of discrete data, it is necessary at
least to change a value like a quadric function. Since it can be
realized only by setting the number of times of the digital
integration to 2, the configuration of the integrating unit can be
simplified.
[0015] Furthermore, the digital integration performed by the
integrating unit is a process of accumulating input data, and it is
desirable that the process is repeated n times in a period of
inputting digital data. Thus, the operation of accumulating data
can be realized only by adding the input data. Therefore, the
configuration of the integrating unit can be simplified, and the
process can be easily and more quickly repeated. As a result, the
value of the multiple n of the oversampling can be set to a large
value without complicating the configuration and largely increasing
the cost of parts.
[0016] In addition, the digital-to-analog converter can be
configured only by providing voltage generation unit and smoothing
unit at the stage after the above mentioned oversampling circuit.
Accordingly, the digital-to-analog converter according to the
present invention can be realized with a simplified configuration
and reduced cost of parts. Furthermore, the above mentioned
oversampling circuit can easily set a high oversampling frequency
without complicating the configuration or largely increasing the
cost of parts. As a result, the distortion of the output waveform
of the digital-to-analog converter to which the oversampling
circuit is applied can be minimized.
BRIEF DESCRIPTION OF THE DRAWINGS
[0017] FIG. 1 is a diagram showing a sampling function used in an
interpolating operation in the oversampling circuit according to an
embodiment;
[0018] FIG. 2 is a diagram showing a relationship between the
sampling values with an interpolation values;
[0019] FIG. 3 is a diagram showing a waveform obtained by
differentiating once the sampling function shown in FIG. 1;
[0020] FIG. 4 is a diagram showing a the waveform obtained by
further differentiating the polygonal line function shown in FIG.
3;
[0021] FIG. 5 is a diagram showing a the configuration of an
oversampling circuit of an embodiment;
[0022] FIG. 6 is a block diagram showing a detailed configuration
of an integrating circuit included in the oversampling circuit
shown in FIG. 5;
[0023] FIGS. 7A through 7L are charts showing the operation timings
of the oversampling circuit of an embodiment;
[0024] FIGS. 8A and 8B are diagrams showing detailed data output
from the integrating circuits;
[0025] FIG. 9 is a diagram showing a detailed configuration of the
multiplying section; and
[0026] FIG. 10 is a diagram showing a configuration of the D/A
converter to which the oversampling circuit shown in FIG. 5 is
applied.
BEST MODE FOR CARRYING OUT THE INVENTION
[0027] An embodiment of the oversampling circuit according to the
present invention is described below in detail by referring to the
attached drawings. FIG. 1 shows a sampling function used in an
interpolating operation in the oversampling circuit according to
the present embodiment. The sampling function H(t) is disclosed by
WO99/38090, and represented by the following expressions. 1 ( - t 2
- 4 t - 4 ) / 4 ; - 2 t < - 3 / 2 ( 3 t 2 + 8 t + 5 ) / 4 ; - 3
/ 2 t < - 1 ( 5 t 2 + 12 t + 7 ) / 4 ; - 1 t < - 1 / 2 ( - 7
t 2 + 4 ) / 4 ; - 1 / 2 t < 0 ( - 7 t 2 + 4 ) / 4 ; 0 t < 1 /
2 ( 5 t 2 - 12 t + 7 ) / 4 ; 1 / 2 t < 1 ( 3 t 2 - 8 t + 5 ) / 4
; 1 t < 3 / 2 ( - t 2 + 4 t - 4 ) / 4 ; 3 / 2 t 2 ( 1 )
[0028] where t=0, .+-.1, .+-.2 indicates the sampling position. The
sampling function H(t) shown in FIG. 1 can be differentiated only
once in the whole range, and is a function of local support
converging into 0 with the sampling position t=.+-.2. By performing
an overlapping process using the sampling function H(t) based on
each sampling value, the interpolating process can be performed
using a function differentiable only once in the sampling
values.
[0029] FIG. 2 shows the relationship between the sampling values
and the interpolation values. As shown in FIG. 2, assume that four
sampling positions are t1, t2, t3, and t4, and the distance between
two adjacent sampling positions is 1. The interpolation value y
corresponding to the interpolation position t0 between the sampling
positions t2 and t3 is obtained by the following equation. 2 y = Y
( t1 ) H ( 1 + a ) + Y ( t2 ) H ( a ) + Y ( t3 ) H ( 1 - a ) + Y (
t4 ) H ( 2 - a ) ( 2 )
[0030] where Y(t) indicates each sampling value at the sampling
position t. Each of 1+a, a, 1-a, and 2-a indicates the distance
between the interpolation position t0 and each of the sampling
positions t1 through t4.
[0031] As described above, by performing a convolution operation by
computing the value of the sampling function H(t) corresponding to
each sampling value, an interpolation value of sampling values can
be obtained theoretically. However, the sampling function shown in
FIG. 1 is a quadric piecewise polynomial differentiable only once
in the whole range. Using this feature, the interpolation value can
be obtained in another equivalent process procedure.
[0032] FIG. 3 shows a waveform obtained by differentiating once the
sampling function shown in FIG. 1. The sampling function H(t) shown
in FIG. 1 is a quadric piecewise polynomial differentiable once in
the entire range. Therefore, by performing the differentiation
once, a polygonal line function formed by the waveform of a
continuous polygonal line as shown in FIG. 3 can be obtained.
[0033] FIG. 4 shows the waveform obtained by further
differentiating the polygonal line function shown in FIG. 3.
However, the polygonal line waveform contains a plurality of corner
points, and the differentiation cannot be performed in the whole
range. Therefore, the differentiation is performed on the linear
portion between two adjacent corner points. By differentiating the
polygonal line waveform shown in FIG. 3, the step function formed
by the stepwise waveform as shown in FIG. 4 can be obtained.
[0034] Thus, the above mentioned sampling function H(t) is once
differentiated in the entire range to obtain a polygonal line
function. By further differentiating each of the linear portions of
the polygonal line function, a step function can be obtained.
Therefore, in the reverse order, by generating the step function
shown in FIG. 4, and integrating it twice, the sampling function
H(t) shown in FIG. 1 can be obtained.
[0035] In the step function shown in FIG. 4, the positive and
negative areas are set equal to each other, and the sum of the
areas equals 0. That is, by integrating such a step function plural
times, a sampling function of local support, as shown in FIG. 1,
whose differentiability in the whole range is guaranteed can be
obtained.
[0036] In computing the interpolation value in the convolution
operation shown by the equation (2), the value of the sampling
function H(t) is multiplied by each sampling value. If the sampling
function H(t) is obtained by integrating twice the step function
shown in FIG. 4, the value of the sampling function obtained in the
integrating process is multiplied by each sampling value, or
equivalently, when a step function before the integration
processing is generated, an interpolation value can be obtained by
generating a step function by multiplication by each sampling
value, and performing twice the integrating process on the result
obtained in the convolution operation using the step function. The
oversampling circuit according to the present embodiment obtains an
interpolation value as described above. This process is described
below in detail.
[0037] FIG. 5 shows the configuration of the oversampling circuit
according to the present embodiment. The oversampling circuit shown
in FIG. 5 comprises a multiplying section 1, four data holding
sections 2-1, 2-2, 2-3, and 2-4, four data selectors 3-1, 3-2, 3-3,
and 3-4, an adding section 4, and two integrating sections 5-1,
5-2.
[0038] The multiplying section 1 outputs a result of multiplying
discrete digital data sequentially input at predetermined time
intervals by a multiplicator corresponding to each value of the
step function shown in FIG. 4. Each value of the step functions
shown in FIG. 4 can be obtained by twice differentiating each
piecewise polynomial of the above mentioned equation (1) as
follows.
1 -1 ;-2 .ltoreq. t < {fraction (-3/2)} +3 ;{fraction (-3/2)}
.ltoreq. t < -1 +5 ;-1 .ltoreq. t < {fraction (-1/2)} -7
;{fraction (-1/2)} .ltoreq. t < 0 -7 ;0 .ltoreq. t < 1/2 +5
;1/2 .ltoreq. t < 1 +3 ;1 .ltoreq. t < {fraction (3/2)} -1
;{fraction (3/2)} .ltoreq. t .ltoreq. 2
[0039] Therefore, the multiplying section 1 multiplies the input
data D by four types of the value corresponding to the above
mentioned step functions as multiplicators (-1, +3, +5, and -7),
when the data D is input, and concurrently outputs a set of
four-piece data, that is, -D, +3D, +5D, and -7D.
[0040] The data holding sections 2-1 through 2-4 cyclically fetch a
set of four-piece data output from the multiplying section 1, and
hold the data until the next fetching timing. For example, a set of
four-piece data output from the multiplying section 1 corresponding
to the first input data is fetched and held in the data holding
section 2-1, and a set of four-piece data output from the
multiplying section 1 corresponding to the second input data is
fetched and held in the data holding section 2-2. Similarly, each
set of four-piece data output from the multiplying section 1
corresponding to the third and fourth input data is fetched and
held in the data holding section 2-3, and 2-4, respectively. When a
cycle of the data holding operation is completed in the data
holding sections 2-1 through 2-4, then the next output data from
the multiplying section 1 corresponding to the fifth input data is
fetched and held by the data holding section 2-1 which has first
held the data. Thus, sets of four-piece data sequentially output
from the multiplying section 1 corresponding to the input data are
cyclically held by the data holding sections 2-1, etc.
[0041] The data selectors 3-1 through 3-4 output data whose values
change stepwise corresponding to a step function by sequentially
reading four pieces of data held in the one-to-one corresponding to
the data holding sections 2-1 through 2-4 in a predetermined order.
Practically, for example, when four pieces of data (-D, +3D, +5D,
and -7D) obtained by multiplying the data D by the above mentioned
four types of multiplicators are held in the data holding section
2-1, the data selector 3-1 cyclically reads the held digital data
in the order of -D, +3, +5D, -7D, -7D, +5D, +3D, and -D at
predetermined time intervals, thereby outputting the data of step
functions having a value proportional to the input data D.
[0042] The adding section 4 adds up digitally the values of the
step functions output from four data selectors 3-1 through 3-4. The
two serially connected integrating circuits 5-1 and 5-2 perform two
integrating processes on the data output from adding section 4. A
linearly changing data (like a linear function) is output from the
integrating circuit 5-1 at the first stage, and a data changing
like a quadric function is output from the integrating circuit 5-2
at the subsequent stage.
[0043] FIG. 6 shows the detailed configuration of the integrating
circuits 5-1 and 5-2. The integrating circuit 5-1 at the preceding
stage comprises two D flip-flops (D-FF) 51a and 51c and an adder
(ADD) 51b. The adder 51b has two input terminals. Data output from
the adding section 4 and temporarily held in the D flip-flop 51a is
input into one input terminal, and data output from the adder 51b
itself and temporarily held in the D flip-flop 51c is input into
the other input terminal. Each of the D flip-flops 51a and 51c
holds the data synchronous with the clock signal CLK2 for an
integrating operation. The clock signal CLK2 corresponds to the
oversampling frequency, and is set to the frequency n times as high
as the frequency of the clock signal CLK synchronized with input
timing of the input data. Therefore, when the data output from the
adding section 4 is input into the integrating circuit 5-1 with the
above mentioned configuration, a digital integrating operation for
accumulating the input data is performed in synchronization with
the clock signal CLK2.
[0044] The integrating circuit 5-2 at the subsequent stage has the
basically the same configuration as the above mentioned integrating
circuit 5-1 at the preceding stage, and comprises two D flip-flops
(D-FF) 52a and 52c and an adder (ADD) 52b. When data output from
the integrating circuit 5-1 at the preceding stage is input into
the integrating circuit 5-2 with the above mentioned configuration,
a digital integrating operation for accumulating the input data is
performed in synchronization with the clock signal CLK2.
[0045] Since the value of the step function output from the above
mentioned data selector 3-1 is proportional to the value of the
digital data input to the multiplying section 1 at predetermined
timing, the data output from the subsequent integrating circuits
5-2 by performing twice the integrating process on the value of the
step function by the two integrating circuits 5-1 and 5-2 include
the data corresponding to the multiplication result of the sampling
function shown in FIG. 1 by the input data. Also, the adding
section 4 adds up the values of the step functions output from data
selectors 3-1 through 3-4. This can be equivalently performed by
the convolution process using a step function as shown in FIG. 1,
paying attention to an output data from the integrating circuit 5-2
at the subsequent stage.
[0046] Therefore, in the case of inputting digital data into the
oversampling circuit according to the present embodiment at a
predetermined time intervals, the outputting timing of data of the
step function from each data selector is shifted corresponding to
the input interval, and the step functions respectively generated
are added up, then the adding results are performed the integrating
operation twice, thereby obtaining digital data whose values change
stepwise along the curve smoothly connecting digital data input at
predetermined intervals.
[0047] The above mentioned multiplying section 1 corresponds to
multiplying unit, the combinations of the data holding section 2-1,
or the like, and the data selector 3-1, or the like correspond to
the step function generation unit, the adding section 4 corresponds
to the addition unit, and the integrating sections 5-1 and 5-2
correspond to integrating unit, respectively.
[0048] FIGS. 7A to 7L are charts showing the operation timings of
the oversampling circuit in this embodiment. As shown in FIG. 7A,
if the digital data D.sub.1, D.sub.2, D.sub.3, . . . are input at a
constant time interval, each of the data holding sections 2-1
through 2-4 holds four data corresponding to these digital data
D.sub.1, D.sub.2, D.sub.3, . . . cyclically. More specifically, the
data holding section 2-1 fetches four data -D.sub.1, +3D.sub.1,
+5D.sub.1, -7D.sub.1 output from the multiplying section 1
corresponding to the first input data D.sub.1, and holds the data
till the input digital data is circulated (or till four data
corresponding to a fifth input data D.sub.5 (-D.sub.5, +3D.sub.5,
+5D.sub.5, -7D.sub.5) is input) (FIG. 7B). The data selector 3-1
reads out four data corresponding to the first input data D.sub.1
in predetermined order, and generates a step function having a
value proportional to the input data D.sub.1 (FIG. 7C).
[0049] Similarly, the data holding section 2-2 fetches four data
-D.sub.2, +3D.sub.2, +5D.sub.2, -7D.sub.2 output from multiplying
section 1 corresponding to the second input data D.sub.2, and holds
the data till the input digital data is circulated (or till four
data corresponding to a sixth input data D.sub.6 is input) (FIG.
7D). The data selector 3-2 reads out four data corresponding to the
second input data D.sub.2 in predetermined order, and generates a
step function having a value proportional to the input data D.sub.2
(FIG. 7E).
[0050] The data holding section 2-3 fetches four data -D.sub.3,
+3D.sub.3, +5D.sub.3, -7D.sub.3 output from multiplying section 1
corresponding to the third input data D.sub.3, and holds the data
till the input digital data is circulated (or till four data
corresponding to a seventh input data D.sub.7 is input) (FIG. 7F).
The data selector 3-3 reads out four data corresponding to the
third input data D.sub.3 in predetermined order, and generates a
step function having a value proportional to the input data D.sub.3
(FIG. 7G).
[0051] The data holding section 2-4 fetches four data -D.sub.4,
+3D.sub.4, +5D.sub.4, -7D.sub.4 output from multiplying section 1
corresponding to the fourth input data D.sub.4, and holds the data
till the input digital data is circulated (or till four data
corresponding to a eighth input data D.sub.8 is input) (FIG. 7H).
The data selector 3-4 reads out four data corresponding to the
fourth input data D.sub.4 in predetermined order, and generates a
step function having a value proportional to the input data D.sub.4
(FIG. 7I).
[0052] The adding section 4 adds values of step functions output
from each of four data selectors 3-1 through 3-4 in this way. By
the way, the step function generated by each of the data selectors
3-1 through 3-4 as shown in FIG. 4 is a function of a local support
having eight piecewise sections divided at every 0.5 from a region
of the sample position t=-2 to +2 in which the sampling function of
FIG. 1 has finite values. For example, a first piecewise section, a
second piecewise section, . . . , and an eighth piecewise section
are defined in a direction from the sample position t=-2 to +2.
[0053] More specifically, the adding section 4 at first adds a
value (+3D.sub.1) corresponding to the seventh piecewise section
that is output from the data selector 3-1, a value (-7D.sub.2)
corresponding to the fifth piecewise section that is output from
the data selector 3-2, a value (+5D.sub.3) corresponding to the
third piecewise section that is output from the data selector 3-3,
and a value (-D.sub.4) corresponding to the first piecewise section
that is output from the data selector 3-4 to output a result of
addition (+3D.sub.1 -7D.sub.2 +5D.sub.3 -D.sub.4).
[0054] Then, the adding section 4 adds a value (-D.sub.1)
corresponding to the eighth piecewise section that is output from
the data selector 3-1, a value (+5D.sub.2) corresponding to the
sixth piecewise section that is output from the data selector 3-2,
a value (-7D.sub.3) corresponding to the fourth piecewise section
that is output from the data selector 3-3, and a value (+3D.sub.4)
corresponding to the second piecewise section that is output from
the data selector 3-4 to output a result of addition
(-D.sub.1+5D.sub.2-7D.sub.3+3D.sub.4).
[0055] Thus, when addition results are sequentially output in the
form of steps from the adding section 4 (FIG. 7J), the integrating
circuit 5-1 at the preceding stage outputs plural pieces of data
whose values change in the form of the polygonal line by
integrating the data (FIG. 7K). The integrating circuit 5-2 at the
subsequent stage further integrates the data whose values changes
in the form of the polygonal line, and outputs plural pieces of
data whose values change along a smooth curve differentiable only
once between the input data D.sub.2 and D.sub.3 (FIG. 7L).
[0056] FIGS. 8A and 8B show the details of the data output from the
two integrating circuits 5-1 and 5-2. For example, the frequency of
the clock signal CLK2 for an integrating operation input into each
of the integrating circuits 5-1 and 5-2 is set to 20 times as high
as the sampling frequency (frequency of the clock signal CLK) of
the input data. As shown in FIG. 8A, the plural pieces of data
output from the integrating circuit 5-1 at the preceding stage have
values changing like a linear function. As shown in FIG. 8B, the
plural pieces of data output from the integrating circuit 5-2 at
the subsequent stage have values changing like a quadric
function.
[0057] In each of the integrating circuits 5-1 and 5-2 whose
configurations are shown in FIG. 6, a digital integrating process
is performed by simply accumulating input data. Therefore, since
the value of the data output therefrom becomes larger depending on
the multiple of the oversampling, it is necessary to provide a
division circuit at the output stage of each of the integrating
circuits 5-1 and 5-2 in order to make the values of input output
data coincident. For example, in the example shown in FIG. 8, since
the value of the output data is 20 times as large as the input
data, a division circuit having a divisor of 20 is provided at the
end of each of the integrating circuits 5-1 and 5-2. However, when
a multiple of the oversampling is set to a value of the power of 2
(for example, 2, 4, 8, 16, . . .), a dividing process can be
performed on output data by bit-shifting the output data of each of
the integrating circuits 5-1 and 5-2 toward lower bits, thereby
omitting the above mentioned division circuit. For example, when
the multiple of the oversampling is set to 16, the output data from
each of the integrating circuits 5-1 and 5-2 can be shifted by 5
bits toward lower bits. Therefore, the wiring at the output
terminal of each circuit can be shifted by 5 bits in advance.
[0058] Thus, the oversampling circuit according to the present
embodiment holds the four multiplication results as a unit
corresponding to each input digital data in the four data holding
sections 2-1 through 2-4 cyclically. The data selectors 3-1 through
3-4 read out the four held data in predetermined order, thereby
generating the step functions. Then, adding section 4 adds the
values of the step function while corresponds to the four input
data. And then, by performing a digital integrating process twice
by the two integrating circuits 5-1 and 5-2 on the data output from
the adding section 4, an oversampling process can be performed for
increasing in a pseudo manner a sampling frequency n times as high
as the frequency of each piece of the input digital data.
[0059] Especially, the oversampling circuit according to the
present embodiment sets how many times the sampling frequency of
the input data the oversampling frequency is to be set depends only
on the frequency of the clock signal CLK2 input into the two
integrating circuits 5-1 and 5-2. That is, the multiple of the
oversampling can be set large only by configuring the two
integrating circuits 5-1 and 5-2 using high-speed parts. Therefore,
unlike the conventional method of performing the oversampling
process using a digital filter, the entire circuit is not large
although the frequency of the oversampling is set higher, thereby
minimizing the increase of the cost of parts. Furthermore, the
contents of the operations can be simplified by using the four
multiplicators represented by integers in the multiplying process
by the multiplying section 1, thereby simplifying the configuration
of the multiplying section, and reducing the cost of parts.
[0060] Furthermore, for example, when an oversampling process is
performed to obtain a pseudo frequency n times as high as the
sampling frequency (for examples, 1024 times), it has been
necessary in the conventional method to have the operation speed of
the parts as high as the pseudo frequency. However, according to
the oversampling circuit of the present embodiment, except the two
integrating circuits, it is necessary to operate the each data
holding sections and each data selector, etc. at the sampling
frequency or the frequency twice as high as the sampling frequency,
thereby considerably reducing the operation speed of each part.
[0061] FIG. 9 shows the detailed configuration of the multiplying
section 1 shown in FIG. 5. The multiplying section 1 shown in FIG.
9 comprises two inverters 10 and 11 for inverting the logic of each
bit of the input data and outputting the result, a multiplier 12
for multiplying by the multiplicator of 2, a multiplier 13 for
multiplying by the multiplicator of 4, a multiplier 14 for
multiplying by the multiplicator of 8, and four adders 15, 16, 17,
and 18.
[0062] For example, when data D.sub.1 is input into the multiplying
section 1 which has the configuration as mentioned above, the
inverter 10 outputs the data obtained by inverting the logic of
each bit of the input data D.sub.1, the adder 15 adds 1 to the
lowest bit of each piece of the output data, thereby obtaining the
complement of the input data D.sub.1. This equivalently shows the
value (-D.sub.1) obtained by multiplying the input data D.sub.1 by
-1. Furthermore, the multiplier 12 outputs a value (+2D.sub.1) two
times as large as the value of the input data D.sub.1, and the
adder 16 adds the original input data D.sub.1, to the data, thereby
obtaining the value (+3D.sub.1) three times as large as the input
data D.sub.1. Similarly, the multiplier 13 outputs a value
(+4D.sub.1) four times as large as the input data D.sub.1, and the
adder 17 adds the value to the original input data D.sub.1, thereby
obtaining a value (+5D.sub.1) five times as large as the input data
D.sub.1. Additionally, the multiplier 14 outputs a value
(+8D.sub.1) eight times as large as the input data D.sub.1, the
inverter 11 inverts the logic of each bit of the output data, and
the adder 18 adds the original input data D.sub.1 to the inverted
data. The adder 18 has a valid carry terminal C, and adds 1 to the
lowest bit of the output data of the inverter 11, thereby obtaining
the complement of the output data of the inverter 11. Therefore, a
value (-7D.sub.1) -7 times as large as the input data D.sub.1 can
be obtained by adding the original input data D.sub.1 to the value
(-8D.sub.1) -8 times as large as the input data D.sub.1, by means
of the adder 18.
[0063] Since the multiplicators are power of 2, the above mentioned
three multipliers 12, 13, and 14 can perform the multiplying
process only by performing bit shifting operation. Thus, by
combining the multiplying process of the power of 2 by the bit
shift with the adding process, the multiplying process is performed
by four multiplicators, thereby simplifying the configuration.
[0064] A D/A converter can be configured with smaller number of
parts by adding a low pass filter, etc. at the subsequent stage of
the above mentioned oversampling circuit. FIG. 10 shows the
configuration of the D/A converter. The D/A converter has the
configuration obtained by adding a D/A converter 6 and a low pass
filter (LPF) 7 at the subsequent stage of the oversampling circuit
shown in FIG. 5. The D/A converter 6 corresponds to the voltage
generation unit, and the low pass filter 7 corresponds to the
smoothing unit.
[0065] The D/A converter 6 generates an analog voltage
corresponding to the stepwise digital data output by the
integration circuit 5-2 at the subsequent stage. The D/A converter
6 generates a constant analog voltage proportional to the value of
the input digital data, and the voltage value at the output
terminal of the D/A converter 6 also changes stepwise. The low pass
filter 7 smoothes the output voltage of the D/A converter 6, and
outputs a smoothly changing analog signal.
[0066] Since the D/A converter shown in FIG. 10 uses the
oversampling circuit shown in FIG. 5, the configuration can be
simplified and the cost of parts can be reduced. Although an output
waveform is obtained with less distortion and the oversampling
frequency set high, the configuration is not complicated with
reduced cost.
[0067] The present invention is not limited to the above mentioned
embodiment, and various types of embodiments can be set within the
scope of the gist of the present invention. For example, according
to the above mentioned embodiment, a sampling function is defined
as a function of local support differentiable only once in the
whole range, but the times of differentiation can be set to a value
equal to or larger than 2. In this case, the number of integrating
circuits is to match the number of times of differentiation.
[0068] The sampling function of this embodiment converges to zero
at t=.+-.2, as shown in FIG. 1, but may converge to zero at t=.+-.3
or beyond. For example, in a case of the sampling function
converging to zero at t=.+-.3, six data holding sections and six
data selectors may be contained in the oversampling circuit shown
in FIG. 5, to interpolate for the six digital data.
[0069] Furthermore, it is not limited to the interpolating process
using a sampling function of local support, but using a sampling
function differentiable finite times having a predetermined value
in the range from -.infin. to +.infin., an interpolation process
may be performed only for plural digital data corresponding to
finite sample position. For example, assuming that the sampling
function is defined by a quadric piecewise polynomial, a
predetermined step function can be obtained by twice
differentiating each piecewise polynomial. Therefore, a convolution
operation is performed using this step function, and an operation
result is integrated twice, thereby performing an oversampling
process.
INDUSTRIAL APPLICABILITY
[0070] As described above, according to the present invention, a
plurality of multiplying processes are performed using plural
multiplicators on a plurality of digital data input at
predetermined intervals. Using the plurality of multiplication
result, the step functions are generated corresponding to each
input digital data. By performing digital integration plural times
on the addition results obtained by adding up values of the step
function corresponding to each input digital data, digital data
whose values change stepwise is output along a smooth curve.
Therefore, when an oversampling frequency is high, it is necessary
only to speed up the digital integration, thereby avoiding the
conventional complicated configuration, that is, simplifying the
configuration, and reducing the cost of parts.
* * * * *