JP2011029787A - Signal processor - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To reduce the folding noise of a higher harmonic generated in nonlinear processing. <P>SOLUTION: A signal processor 100 includes: an over-sampling circuit 20 for performing the over-sampling processing of PCM data Din sampled by a sampling frequency Fs; an interpolation circuit 30 for raising a sampling frequency to 128 Fs by interpolation processing; a slice circuit 40 for performing the slice processing of an output signal of the interpolation circuit 30; a noise shaper 50; and a PWM circuit 60. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to a technique for reducing harmonic aliasing noise generated by nonlinear processing preceding a class D amplifier.

  The class D amplifier generates a binary output signal in which the pulse width and the pulse time density are modulated in accordance with the input signal. When the level of the input signal exceeds a predetermined range, the output signal of the class D amplifier is distorted. Therefore, Patent Document 1 discloses a technique for providing an amplitude limiting circuit for limiting the level of an input signal.

JP 2007-124625 A (Claims 1 and 2)

  However, when the amplitude is limited by the amplitude limiting circuit, a harmonic component is generated. For example, as shown in FIG. 6, when the input signal IN is limited by the first level L1 and the second level L2, the amplitude-limited waveform has a corner in a circled portion. Such corners contain many harmonic components. This harmonic component is folded at the sampling frequency, mixed into the audible band as folded noise, and may be reproduced as an abnormal sound.

  The present invention has been made in view of the above circumstances, and an object of the present invention is to reduce harmonic aliasing noise generated by nonlinear processing such as a slicer or clipper.

  In order to solve the above-described problems, a signal processing device according to the present invention includes a sampling unit that samples an analog signal at a first frequency and converts the sampled signal into a first digital signal, and a sampling frequency of the first digital signal. Frequency converting means for converting the first frequency to a second frequency higher than the first frequency and outputting a second digital signal; and applying a non-linear process to the second digital signal to output a third digital signal And non-linear processing means for performing pulse width modulation on the basis of the third digital signal to generate an output signal.

In the non-linear processing, a distortion component is generated in the harmonic frequency with respect to the fundamental frequency of the signal. The fundamental frequency is a frequency within the audible band, and the distortion component gradually attenuates as the frequency increases. However, the distortion component is folded at the Nyquist frequency. If non-linear processing is performed before frequency conversion, aliasing occurs in the audible band because aliasing occurs at half the sampling frequency. On the other hand, in the present invention, nonlinear processing is performed after frequency conversion from the first frequency to the second frequency by the frequency conversion means. Therefore, it is possible to suppress the aliasing noise from being mixed in the audible band.
Note that the nonlinear processing includes processing for limiting the amplitude, such as slice processing and clip processing.

Here, the frequency conversion means may use any method as long as it can convert the sampling frequency. For example, the frequency conversion unit may include an oversampling unit that oversamples the first digital signal to generate the second digital signal, or the frequency conversion unit converts the first digital signal to the first digital signal. An oversampling unit that performs oversampling and an interpolation unit that performs an interpolation process on the output signal of the oversampling unit and outputs the second digital signal may be provided.
Oversampling is performed in order to improve the SN ratio by shifting aliasing noise due to the sampling theorem to a high frequency region outside the audible band. According to the present invention, frequency conversion for oversampling and nonlinear processing are performed. The frequency conversion for reducing the aliasing noise can be also used. Therefore, the configuration can be simplified.

  In the above-described signal processing device, the pulse width modulation unit generates a pulse width modulation signal by applying a noise shaping process to the second digital signal and performing a pulse width modulation on the output signal of the noise shaping unit. It is preferable to provide a pulse width modulation unit. The noise shaping process can shift the quantization noise in the audible band to a high frequency region outside the audible band.

  In the signal processing device described above, the pulse width modulation means includes a filter unit having a frequency characteristic in which a gain at a carrier frequency of pulse width modulation is smaller than an audible band, a noise shaping unit that performs noise shaping processing, and a digital And a digital pulse width modulation unit that generates a pulse width modulation signal by performing pulse width modulation, supplies the second digital signal to the filter unit, and supplies an output signal of the filter unit to the nose shaping unit Then, the output signal of the nose shaping unit is supplied to the digital pulse width modulation unit, or the second digital signal is supplied to the nose shaping unit, and the output signal of the nose shaping unit is supplied to the filter unit. And supplying the output signal of the filter unit to the digital pulse width modulation unit. , It is preferable.

The pulse width modulation unit performs pulse width modulation on the output signal of the noise shaping unit at the carrier frequency. When the distortion component due to nonlinear processing is superimposed on the output signal, the distortion component near the carrier frequency is intermodulated at the carrier frequency. Wake up. This generates noise in the audible band. According to the present invention, since the filter section has a small carrier frequency gain, it is possible to reduce distortion components associated with nonlinear processing. In addition, since the filter unit only needs to suppress distortion components associated with nonlinear processing near the carrier frequency before the pulse width modulation is performed, the filter unit may be configured in the order of the nose shaping unit → the digital pulse width modulation unit. However, it may be configured in the order of nose shaping unit → filter unit → digital pulse width modulation unit.
The filter unit may be configured with a low-pass filter, or may be configured with a trap filter whose center frequency is the carrier frequency.

  Further, in the signal processing apparatus described above, instead of the digital pulse width modulation unit, a DA conversion unit for converting a digital signal into an analog signal, and an output signal of the DA conversion unit is subjected to analog pulse width modulation to generate a pulse. And an analog pulse width modulation unit that generates a width modulation signal. That is, pulse width modulation may be performed digitally or analogly.

It is a block diagram which shows the structure of the signal processing apparatus which concerns on 1st Embodiment of this invention. It is explanatory drawing which shows the frequency component of the distortion generate | occur | produced when a slice process is performed with the frequency Fs. It is explanatory drawing which shows the frequency component of the distortion which generate | occur | produces when a slice process is performed with the frequency of 128Fs. It is a block diagram which shows the structure of the signal processing apparatus which concerns on 2nd Embodiment of this invention. It is explanatory drawing which shows the frequency component of the distortion generate | occur | produced by the slice process output from a PWM circuit. It is a wave form diagram for demonstrating the harmonic distortion in a slice process.

<1. First Embodiment>
Next, an embodiment suitable for the present application will be described with reference to the drawings. In the embodiment described below, the number of processing systems is one, but a similar processing system may be provided to be a signal processing apparatus corresponding to the L channel and the R channel.

FIG. 1 is a block diagram illustrating a configuration of a signal processing device 100 according to the first embodiment. The signal processing apparatus 100 of this embodiment includes an audio interface circuit 10, an oversampling circuit 20, an interpolation circuit 30, and a slice circuit 40.
The audio interface circuit 10 samples an analog audio signal Vin supplied from the outside at a sampling frequency Fs and converts it into audio data Din in PCM format (hereinafter simply referred to as “PCM data”). In this example, the sampling frequency Fs is 48 KHz. Further, the audio interface circuit 10 converts the data format from the serial format to the parallel format, and adjusts the timing when the PCM data Din is transferred to the oversampling circuit 20. Note that the AD conversion function may be separated from the audio interface circuit 10 and an AD conversion circuit may be provided in the preceding stage. In this case, PCM data Din sampled at the sampling frequency Fs is supplied to the audio interface circuit 10, and data format conversion and timing adjustment are performed in the audio interface circuit 10.

The oversampling circuit 20 performs oversampling processing on the PCM data Din to generate first data D1. The oversampling circuit 20 is configured by an oversampling filter of a predetermined multiple (in this example, four times), and is used to increase the Nyquist frequency and reduce the level of quantization error noise of the input PCM data Din. .
The oversampling circuit 20 can be configured by an FIR filter or the like.

  The interpolation circuit 30 interpolates the first data D1 to increase the sampling frequency to 128 Fs and generates the second data D2. The interpolation process is performed by linear interpolation, for example. In this example, the sampling frequency is increased by 32 times. In this case, a certain sample of the first data D1 and the next sample are divided into 32 equal parts in time, and 31 samples are linearly generated between the samples. For example, the interpolation circuit 30 can be configured by a bit shift circuit and an addition circuit without using a multiplier.

  Here, the oversampling circuit 20 and the interpolation circuit 30 function as frequency conversion means for converting the sampling frequency of the PCM data Din from the first frequency Fs to the second frequency 128Fs.

Next, the slicing circuit 40 performs slicing processing to make the signal level equal to or higher than a predetermined amplitude constant, and generates output data. As a result, it is possible to prevent a large-amplitude signal from being supplied to the speaker 90 described later.
Next, the noise shaper 50 executes a noise shaping process for reducing quantization error noise that is uniformly distributed in the entire Nyquist frequency band by limiting the word length of the pulse width modulation in the audible frequency band. To generate output data.

Next, the PWM circuit 60 may be configured with a digital circuit or an analog circuit. However, in the case of an analog circuit, the output data of the noise shaper 50 is converted into an analog signal via a DA converter (not shown) and then supplied to the PWM circuit 60.
There are various types of PWM circuit 60. For example, an integration circuit that integrates a signal obtained by combining an input signal and a feedback signal and outputs an integration signal, a triangular wave generation circuit that generates a triangular wave signal, an integration signal, A comparator that compares the triangular wave signal and generates a pulse width modulation signal that is pulse width modulated based on the comparison result, and a low-pass filter that attenuates the high-frequency component of the pulse width modulation signal and supplies it to the integration circuit as a feedback signal There is something with. In this case, the fundamental frequency of the triangular wave signal is the carrier frequency.

  The buffer amplifier 70 amplifies the output signal of the PWM circuit 60 and supplies it to the low-pass filter 80. The low-pass filter 80 removes frequency components higher than the audible band and supplies them to the speaker 95. The speaker 90 converts an electric signal into sound and emits the sound.

Next, features of the signal processing apparatus 100 will be described. In the signal processing apparatus 100, the slicing circuit 40 is arranged at the subsequent stage of the interpolation circuit 30 with a sampling frequency of 128 Fs.
If the slice circuit 40 is disposed between the audio interface circuit 10 and the oversampling circuit 20, the slice circuit 40 operates at the frequency Fs. When the slice processing is performed on the PCM data Din, a harmonic component is generated, and the harmonic component is turned back at the Nyquist frequency Fs / 2. FIG. 2 shows frequency components of distortion generated when slice processing is executed at the frequency Fs. In the figure, the spectrum with an arrow in the region of frequency Fs / 2 or less is the aliasing component. As described above, noise is generated in the audible band.

  On the other hand, in this embodiment, the slicer circuit 40 is operated at a frequency of 128 Fs. For this reason, the frequency of folding is 64 Fs. At such a high frequency, the harmonic component of the distortion accompanying the slice processing is sufficiently attenuated. Therefore, aliasing noise generated in the audible band can be sufficiently suppressed.

FIG. 3 shows frequency components of distortion generated when slice processing is executed at a frequency of 128 Fs. As apparent from comparison with FIG. 2, there is no aliasing noise at the frequency Fs / 2. However, some noise components indicated by dotted lines are generated in the low frequency region. This noise component is caused by cross modulation distortion with the carrier frequency when the pulse width modulation is performed in the PWM circuit 60. That is, the sum and difference components of the distortion frequency and the carrier frequency in the vicinity of the carrier frequency are generated as intermodulation distortion, and the difference component is generated in the low frequency region. Actually, the frequency of the noise component is complicated, for example, the difference component turns back at the frequency “0”.
If the carrier frequency is 8 Fs, the noise component indicated by the dotted line appears in the low frequency region due to the noise component near the frequency 8 Fs surrounded by the dotted line in FIG.

<2. Second Embodiment>
Next, a second embodiment will be described. The signal processing device 200 according to the second embodiment is configured in the same manner as the signal processing device 100 of the first embodiment, except that a low-pass filter 45 is provided between the slice circuit 40 and the noise shaper 50. .

  FIG. 4 shows a block diagram of a signal processing device 200 according to the second embodiment. The low-pass filter 45 is used to reduce cross modulation at the carrier frequency of the PWM circuit 60. FIG. 5 shows frequency components of distortion generated by the slice processing output from the PWM circuit 60. The noise due to the cross modulation distortion occurs because the noise component due to the slice processing is superimposed on the input signal of the PWM circuit 50 near the carrier frequency.

FIG. 5 shows frequency components of distortion generated by the slice processing output from the PWM circuit 60. As shown in FIG. 5, the low-pass filter 45 has an attenuation characteristic in which the gain starts to drop from the frequency Fs / 2 (illustrated by a one-dot chain line). For this reason, when the carrier frequency is 8 Fs, the distortion generated by the slice processing is sufficiently attenuated in the vicinity of the carrier frequency. For this reason, noise caused by cross modulation distortion does not occur in the low frequency region.
Therefore, according to the second embodiment, the noise component in the audible band can be further suppressed and the quality can be improved.

<3. Modification>
The present invention is not limited to the above-described embodiments, and for example, the following modifications are possible.
(1) In the above-described embodiment, the slice circuit 40 is assumed as an element for generating a distortion component at a high frequency by signal processing. However, the present invention is not limited to this, and the distortion component is generated at a high frequency. Any processing can be applied as long as the processing is performed. That is, the present invention can be applied to a case where there is a means for performing nonlinear processing. Such nonlinear processing includes clip processing.

(2) In the above-described second embodiment, the low-pass filter 45 is employed. However, the present invention is not limited to this, and the point is that the attenuation rate is higher than the audible band in the vicinity of the carrier frequency of pulse width modulation. Any filter having a large frequency characteristic may be used. Here, the vicinity of the carrier frequency can be considered as a range of 20 KHz with the carrier frequency as the center, considering the mixing of noise components into the audible band due to cross modulation. Such a filter includes a so-called trap filter having a reverse characteristic of a bandpass filter.

(3) In the second embodiment described above, the low-pass filter 45 is provided between the slice circuit 40 and the noise shaper 50. However, the present invention is not limited to this, and a slice that is a distortion generation source. It may be provided between the circuit 50 and the PWM circuit 60. For this reason, a low-pass filter 45 may be provided between the noise shaper 50 and the PWM circuit 60.

(4) In the embodiment and the modification described above, the low-pass filter 45 is provided between the slice circuit 40 and the noise shaper 50. However, the present invention is not limited to this, and a distortion source is provided. What is necessary is just to provide between the slice circuit 50 and the PWM circuit 60 which are. For this reason, a low-pass filter 45 may be provided between the noise shaper 50 and the PWM circuit 60.

DESCRIPTION OF SYMBOLS 10 ... Audio interface circuit, 20 ... Oversampling circuit, 30 ... Interpolation circuit, 40 ... Slice circuit, 50 ... Noise shaper, 45 ... Low pass filter, 60 ... PWM circuit, 100, 200 ... Signal Processing equipment.

Claims (6)

Frequency conversion means for converting a sampling frequency of the first digital signal from the first frequency to a second frequency higher than the first frequency and outputting a second digital signal;
Non-linear processing means for performing non-linear processing on the second digital signal and outputting a third digital signal;
Pulse width modulation means for generating an output signal by performing pulse width modulation based on the third digital signal;
A signal processing apparatus comprising:   The signal processing apparatus according to claim 1, wherein the frequency conversion unit includes an oversampling unit that oversamples the first digital signal to generate the second digital signal.   The frequency conversion unit includes an oversampling unit that oversamples the first digital signal, and an interpolation unit that performs an interpolation process on an output signal of the oversampling unit and outputs the second digital signal. The signal processing apparatus according to claim 1. The pulse width modulation means includes
A noise shaping unit that performs a noise shaping process on the second digital signal;
A digital pulse width modulation unit that performs pulse width modulation on the output signal of the noise shaping unit to generate a pulse width modulation signal;
The signal processing apparatus according to claim 2, wherein the signal processing apparatus is a signal processing apparatus. The pulse width modulation means includes
A filter unit having a frequency characteristic in which a gain at a carrier frequency of pulse width modulation is smaller than an audible band;
A noise shaping unit that performs noise shaping processing and a digital pulse width modulation unit that digitally performs pulse width modulation to generate a pulse width modulation signal,
Supplying the second digital signal to the filter unit, supplying an output signal of the filter unit to the nose shaping unit, and supplying an output signal of the nose shaping unit to the digital pulse width modulation unit;
Alternatively, the second digital signal is supplied to the nose shaping unit, the output signal of the nose shaping unit is supplied to the filter unit, and the output signal of the filter unit is supplied to the digital pulse width modulation unit.
The signal processing apparatus according to claim 2, wherein the signal processing apparatus is a signal processing apparatus. Instead of the digital pulse width modulator,
A DA converter for converting a digital signal into an analog signal;
An analog pulse width modulation unit that performs pulse width modulation on the output signal of the DA conversion unit in an analog manner to generate a pulse width modulation signal;
The signal processing apparatus according to claim 4, wherein the signal processing apparatus is a signal processing apparatus.

Images