JPS5951779B2 - Digital/analog conversion method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a digital-to-analog conversion method that facilitates separation and removal of undesired folded signal components that occur along with a desired converted output analog signal during digital-to-analog conversion. It is something that

Generally, to convert a digital signal to an analog signal with a level corresponding to the digital value of the signal,
As shown in FIG. 1, an interpolation filter 2 is arranged in series with a digital-to-analog converter 1.

This interpolation filter 2 has a form in which the output signal of the digital-to-analog converter 1 includes a folded signal component shown by a dotted line in the frequency range of 172 or higher of the sampling frequency f5 in the conversion output energy frequency characteristic shown in FIG. Because of this, if it remains in that form, waveform distortion will occur in the converted output analog signal, making it impossible to reproduce and display normal color television images, etc. Therefore, the folded signal component mentioned above is separated and removed, and the solid line This is necessary in order to extract only the originally desired signal components.

The interpolation/interpolation filter used for the purpose of removing the aliased signal component mentioned above is originally a low-pass filter that removes only the aliased signal component without changing the amplitude or phase of the signal component. However, it is necessary to make the characteristics of the filter close to the ideal amplitude and phase characteristics shown in FIGS. 3a and 3, respectively.

However, it is not only difficult to design a low-pass filter having such ideal amplitude and phase characteristics, but even if characteristics close to the ideal characteristics can be obtained, it is extremely difficult to adjust them. It has to be complicated and expensive.

Particularly when this filter is configured using an IC filter, etc., it is common for both the amplitude and phase characteristics to be related to each other. It is almost impossible to match exactly.

However, since such a problem always occurs in the digital processing of information signals, we will first briefly explain the relationship between each signal component, such as the above-mentioned original signal component and aliased signal component, in the digital processing of information signals. In other words, when the input information signal is a digital television signal sampled and quantized by a sampling pulse of frequency f5, the signal component of the digital information signal that is effective as an information signal without signal distortion is: This is a signal component included in a frequency band up to frequency 172 of the sampling frequency fs, and corresponds to the above-mentioned original signal component.

On the other hand, a signal component in a frequency band exceeding the above-mentioned frequency corresponds to the above-mentioned aliasing component and cannot be treated as an effective signal component.

The above is a well-known fact taught by the sampling theorem, and when quantizing a color television signal, the sampling frequency f5 is usually set to 3f5C1 or 4ftic for the color subcarrier frequency fsc.
, so now the sampling frequency f is set to 3f.
5c, the upper limit frequency of the video signal frequency band is 4.5M ratio between the original upper limit frequency of the signal component and half of the sampling frequency, that is, 1/2 f5, and 1
/2 f, is 3 x 3.58 x 1/2 = 5, 3
Since it is 7MHz, there is some gap, but if we assume that even low-level signal components are within the video signal frequency band, the upper limit frequency of the signal component is approximately 1/2.
Therefore, the gap between the original signal component and the aliased signal component with 1/2 fs as the boundary becomes extremely narrow, making separation by a filter extremely difficult.

In order to avoid this kind of difficulty, when converting a digital signal to an analog signal, the interpolation filter placed on the output side of the digital-to-analog converter is replaced with a conventional low-frequency filter that requires strict conditions for both amplitude and phase characteristics. It is an object of the present invention to enable the use of a normal, for example, LC filter with less severe characteristics in place of a pass filter, and to do so without causing any deterioration in the quality of the obtained analog signal. To this end, when converting a digital signal into an analog signal, prior to the digital-to-analog conversion, digital interpolation is performed to multiply the unique sampling frequency f8 of the input digital signal by α (a rational number where α>1). Japanese Unexamined Patent Application Publication No. 52-117 discloses a method of widening the gap between the upper limit frequency of a desired original signal component and the lower limit frequency of an undesired aliased signal component in a digital-to-analog conversion output signal to facilitate separation of both signal components. It is described in Publication No. 43308.

FIG. 4 shows the basic structure of a digital-to-analog conversion system that is almost the same as that described in Japanese Patent Application Laid-Open No. 52-43308.

As mentioned above, this method performs digital interpolation on the input digital signal by multiplying the sampling frequency f8 of the signal itself by α (a rational number where α>1) before converting the input digital signal from digital to analog. Fourth
Block 3 in the figure shows a digital interpolator for this purpose.

The other blocks indicated by 1 and 2 in FIG. 4 are the same as those shown in the first figure, and their explanation will be omitted.

By placing this digital interpolator 3 immediately before the digital-to-analog converter 1, the input digital signal shown by the frequency spectrum in FIG. 5a becomes as shown in FIG. It is converted into a signal with a frequency spectrum.

This can be easily understood from the fact that the sampling frequency f5 is doubled while the frequency band of the original information signal remains the same.

Therefore, for this obtained signal, the characteristics of the interpolation filter 2 after digital-to-analog conversion by the digital-to-analog converter 1 are those of the ideal sharp cutoff characteristics shown in Figure 3 a and l). Instead, a slow roll-off characteristic as shown by the amplitude characteristic in FIG. 5C is sufficient.

Therefore, it can be seen that it is possible to manufacture a device with sufficiently good phase characteristics, and the cost is much lower than that of a device with a sharp cutoff characteristic.

Here, the digital interpolator used in the above digital-to-analog conversion method will be explained.

As is well known, the operation of interpolation, as shown in FIG. 6, is the addition of new sample points to an already sampled signal in the time domain.

Now, suppose that the points indicated by circles in Figure 6 are signals that have already been sampled, and the points indicated by It is necessary to raise the signal to the x point using a low-pass filter.

An example of the basic circuit of this interpolator 30 is
The one shown in FIG. 9 of Publication No. 308 may be used, but
Here, one example is the circuit configuration shown in FIG. 7a.

This interpolator is a linear interpolator that generates the signal at the point indicated by an x in Figure 7 using two sample signals indicated by two temporally adjacent circles.In Figure 7a, 4 is the input categorical signal. It is a delay device consisting of a register etc. that gives a delay time corresponding to the sample interval, and the input signals are added by a digital adder 5 and further weighted by ×0.5, and the obtained signal and the input signal are The switch S1 is alternately switched to the contact A and B sides every 172 sample periods to extract the interpolated output signal.

Furthermore, in order to improve the accuracy of interpolation, the signal to be interpolated should not be created from two sample signals before and after, but by high-order interpolation that is created from a larger number of sample signals.

As described above, by using the above method when converting a digital signal to an analog signal, the upper limit frequency of the desired original signal component and the undesired folded signal component in the digital-to-analog conversion output signal can be adjusted. Is it possible to widen the gap between the lower limit frequency and separate these two signal components using an interpolation filter, that is, a low-pass filter, whose amplitude and phase characteristics are not so strict? However, even with this, there are still no requirements for the interpolation filter 114 that handles analog signals, such as the characteristic within the band of the desired original signal component must be completely flat.
I can't say it's loose either.

. In particular, as seen in Japanese Unexamined Patent Application Publication No. 52-43308, in the digital-to-analog conversion method in which digital interpolation is performed on the transmitting side of information transmission, the high sampling frequency is limited by the transmission bandwidth of the transmission path. In other words, it may not be possible to obtain an interpolated digital signal with a large α in αf, and in such a case, there may not be a sufficient gap between the desired original signal component and the undesired folded signal component, and therefore, It is extremely difficult to impose all desired in-band characteristics of original signal components on an analog interpolation filter.

The present invention has been made to solve the above problems, and even in the above-mentioned digital interpolation, the signal processing called interpolation is nothing but low-pass filtering, as is well known.

From this point of view, even if the characteristics within the passband of the low-pass filter 2 in Fig. 4 are somewhat uneven, the characteristics within the normal band of the interpolator 3 used will have the opposite characteristics. The present invention was developed by focusing on the fact that by providing the following characteristics, the system as a whole can have ideal characteristics.

In other words, the digital-to-analog conversion method of the present invention has the following features:
When performing digital interpolation by multiplying the unique sampling frequency f5 of the input digital signal by a rational number α where α>1 using the technique described in Japanese Patent Application Laid-Open No. 52-43308, the desired original signal component and undesired aliasing are A two-stage filter consisting of a filter means for forming a blank band corresponding to the gap between the signal components and a filter means for compensating for the amplitude temporal variation within the original signal component band as the whole system of digital-to-analog conversion force. Digital interpolation is performed using a digital interpolator consisting of means.

. The present invention will be described in detail below with reference to the drawings.

First, the example shown in FIG. 8, which will be explained as an embodiment of the present invention, performs high-order interpolation, but since it is impossible to perform infinite-order interpolation, finite-order interpolation is performed. .

It is a good idea to prevent harmonics from being generated in the output interpolated signal even if the accuracy of the combination and interpolation is sacrificed to some extent. It may even change the level of the signal itself.

However, after the change, it is of course necessary to compensate for the change in the signal passband.

FIG. 8 shows a high-order interpolator suitable for performing digital signal interpolation before the digital-to-analog converter when applying the digital-to-analog conversion according to the present invention to a digital television signal. An example of a configuration is shown.

In the figure, 4 is a delay device similar to that in FIG. 7a, and 5 is also a digital adder.

Further, the digital interpolator shown in Figure 8 is divided into two parts, before and after the digital adder 6, but the actual interpolation is performed in the latter stage, and the former stage is used to compensate for the characteristics of the passband that occur when performing interpolation. This is the circuit part.

Note that the numerical value appended to each signal line represents the weighting coefficient of that signal line. For example, X (-0,0680) means that the signal arriving on that signal line is sign-inverted and the original signal is This means that the value is multiplied by 0.0680 and output.

In this example, when the sampling frequency f5 is doubled, the switch S is switched to contacts A and B in the first and second half of the sampling period, respectively.

Furthermore, the interpolator of this example uses a shift register for the delay unit 4, and a read-only memory (R
This can be easily realized by using OM).

Also, to add some explanation about the front stage part of the digital interpolator in FIG. 8, that is, the passband compensation part, this is shown in the rear stage of the same figure, as well as the reference numeral 1 in FIGS. 1 and 4. The aperture characteristic when the digital-to-analog converter is a zero-order hold type, and the interpolation filter shown by code 2 to which the analog signal is supplied after digital-to-analog transformation, that is, within the overall passband of the low-pass filter. It compensates for the characteristics of

FIG. 9 shows the overall amplitude-frequency characteristics when a digital-to-analog conversion system is designed using the digital interpolator having the circuit configuration shown in FIG. 8.

In Fig. 9, the horizontal axis is the color subcarrier frequency 3, 58MH
The normalized frequency normalized by z is shown, and the vertical axis shows the response.

This characteristic is extremely good, far exceeding the above-mentioned required characteristics.

In addition, once the digital interpolator part that determines the characteristics is designed, it can be manufactured in the same way without any adjustment.

Regarding the phase characteristics of this system, the digital filter part is a symmetrical transversal type, and the interpolation filter to which the analog signal is supplied is a 5in2 filter, so the phase characteristics are extremely good. The phase characteristics in the example in-band region are almost flat.

Therefore, the deterioration of the phase characteristics is so small that it can be ignored.

As explained above, according to the method of the present invention, the characteristics required for an interpolation filter to which an analog signal after digital-to-analog conversion is supplied, which was extremely difficult in the past, can be achieved by The regulations can be made much less strict than those described in the official gazette.

For example, sample frequency 3f5c, i.e. 10.7MH
When converting a digital television signal with z to an analog signal, the characteristics of the interpolation filter required in the conventional conversion method are flat at about ±0.2 to Q, 3 dB in the frequency range of 0 to 4.1 M ratio. After that, it was necessary to drop sharply and become less than -50 dB at least at the frequency of 2f5° where the aliased signal component of the carrier color signal exists.

This level of amplitude characteristic and phase characteristic are ±2.
Since a relative delay of 0 nanoseconds or less is required, it is quite difficult to realize this even if a complex filter such as a Bode filter is used.

However, according to the present invention, such difficulties do not occur, and since the main part of the digital interpolator is a digital circuit, the digital interpolator can be operated as designed, including a filter stage that compensates for the amplitude characteristics within the signal component band. If manufactured in
A completely similar circuit device can be arbitrarily duplicated, no adjustment process is required, and it is stable over time.

In addition, the method shown in the practical design example of the present invention, that is,
A method in which a digital filter that determines the stopband and a digital filter that determines the characteristics of the passband are connected in cascade can be applied to the case of performing sampling frequency conversion from 3f8o to 4f8c, for example, to obtain the same effect.

Note that the digital interpolator used in the present invention does not necessarily need to be one that performs high-order interpolation as shown in FIG. 8, but may be one that performs low-order interpolation. Any device may be used as long as it simultaneously includes a filter stage that compensates for the amplitude characteristics within the signal component band.

For example, it goes without saying that a filter stage for compensating the amplitude characteristic within the original signal component band may be added to the linear interpolation filter shown in FIG. 7 and used.

[Brief explanation of the drawing]

Fig. 1 is a block diagram showing the basic configuration of the digital-to-analog converter, Fig. 2 is a characteristic curve diagram showing the conversion output spectrum, and Fig. 3 a and 3 are interpolation diagrams, which are also the constituent elements of the device. A characteristic curve diagram showing the ideal characteristics of the amplitude and phase of the filter, respectively, and FIG. 4 is a block diagram showing the basic configuration of a digital-to-analog converter using almost the same system as that described in JP-A-52-43308. , Fig. 5 a, l), and c are spectral diagrams sequentially showing the principle of operation, and Fig. 6 is the spectral diagram shown in Fig. 4.
A signal waveform diagram illustrating the operating principle of the digital interpolator in the illustrated configuration; FIGS. 7a and 7b are block diagrams and signal waveform diagrams respectively illustrating the basic configuration and operation of the digital interpolator; FIG. FIG. 9 is a block diagram showing an example of a specific configuration of a digital interpolator used in the conversion method of the present invention, and a characteristic curve diagram showing its response frequency characteristics. 1...Digital-to-analog converter, 2...
...Interpolation filter, 3...Digital interpolator, 4...Delay device, 5, 6...Digital adder, S...Switch.

Claims (1)

[Claims] 1. When converting a digital signal into an analog signal, prior to digital/analog conversion, the unique sampling frequency/wave number f5 of the input digital signal is set to α (α>
By performing digital interpolation that multiplies (a rational number), the gap between the upper limit frequency of the desired original signal component and the lower limit frequency of the undesired aliased signal component in the digital-to-analog conversion output signal is widened. - In a digital-to-analog conversion method that facilitates separation of both signal components, a filter means for forming a blank band corresponding to a gap between the upper limit frequency of the original signal component and the lower limit frequency of the folded signal component; Digital interpolation is performed using a digital interpolator comprising a two-stage filter means and a filter means for compensating the amplitude characteristics within the original signal component band as a whole system of the digital value Lunar-log conversion method. Digital-to-analog conversion method 52 The digital signal is a digital color television signal, the digital interpolator is configured as a finite method digital interpolator, and the frequency at which the digital interpolation is performed with perfect precision is the color of the color television signal. 2. The digital-to-analog conversion method according to claim 1, wherein the subcarrier frequency is selected as the subcarrier frequency. 3. When converting a digital signal to an analog signal, the digital
2. The digital-to-analog conversion system according to claim 1, wherein a 5-inch 2-type low-pass filter is placed after the analog converter and used as an interpolation filter. 4. The digital-to-analog conversion method according to claim 3, wherein the zero point of the 5in2 type low-pass filter is selected to be twice the sampling frequency f5.