JP4444037B2 - Digital pulse width modulation signal generator - Google Patents

Description

  The present invention relates to a device that applies pulse width modulation to a continuous time signal based on a digital signal, such as a full digital audio amplifier.

  Switching amplifiers called class D amplifiers are beginning to be used as audio amplifiers for driving speakers due to their high power efficiency. As a method for realizing this switching amplifier, there are a type using an analog signal as an input signal and a type called a full digital amplifier using a digital signal (discrete time discrete value signal) as an input signal. A full digital amplifier can perform all signal processing digitally when the sound source signal is a digital signal such as a CD or MD, so that it can generate high-quality sound at low cost. Yes.

  In a full digital amplifier, signal processing such as electronic volume is performed on a sound source signal, and then the signal sampling frequency is increased to a value such as 16 times or 32 times by an oversampler. Then, it is re-quantized to a low resolution such as 5 bits or 6 bits by a Δ-Σ modulator, and pulse width modulation is performed with the re-quantized signal to generate a pulse width modulated signal. The switching amplifier is driven by this pulse width modulation signal, and a speaker or the like is driven after passing through a low-pass filter for the output.

  In a series of signal processing in this full digital amplifier, noise is generated by re-quantization. Therefore, it is necessary to sufficiently reduce the spectrum in the audible range. Since the lower limit value of the total power of noise generated by requantization is determined by the requantization resolution, the component of the audible range is suppressed by shaping the spectrum of the requantized noise by the noise shaping filter. At that time, the higher the sampling frequency of the requantized signal, the better the noise component in the audible range can be suppressed. However, if the sampling frequency of the requantized signal is increased, the carrier of the pulse width modulation signal It is necessary to increase the frequency, which causes problems such as deterioration in power efficiency of the output stage and increase in electromagnetic noise radiation. On the other hand, if the sampling frequency of the requantized signal is lowered, problems such as power efficiency are improved, but it becomes difficult to suppress requantization noise in the audible range.

JP-T 6-508484 Japanese Patent Publication No. 2000-507759 Japanese Patent Application No. 2004-047113 Japanese Patent Application No. 2004-236617 Hiroyuki Kawanishi: Overview of signal processing required for full digital amplifier, Transistor technology, July 2003, pages 205-222

  The problem to be solved is to sufficiently suppress the audible range component of the requantization noise while keeping the carrier frequency in pulse width modulation low in a full digital amplifier.

  As the pulse width modulation, symmetrical pulse width modulation is used, and the sampling period is set to half of the pulse period, whereby the pulse width modulation is performed twice in the pulse signal of one period. By doing in this way, the sampling frequency of a noise shaping filter can be made into twice the frequency of a pulse width signal, and the requantization noise in an audible range can be suppressed well.

  Also, as a pulse width modulation, two pulse width signals that are 180 degrees out of phase with each other are generated, and a differential output type using the difference between the two signals as an output signal is used to modulate the two pulse width signals. By allowing the difference in degree, the resolution in pulse width modulation can be improved by a factor of two, and as a result, the re-quantization resolution can also be doubled, thereby reducing the re-quantization noise. Can do.

  FIG. 3 shows a general configuration of a full digital audio amplifier. However, the switching amplifier portion is omitted. The input signal s [i] is a digital discrete-time signal. For example, when audio data is reproduced from a CD, the signal is a sampling period Fs = 44.1 kHz and a resolution of about 16 bits. The input signal s [i] has its sampling period changed by the oversampler 6 and is converted (oversampled) into a signal having a sampling frequency Ff that is 16 times Fs, for example. The oversampled signal x [k] is converted into a requantized signal y [k] by the noise shaping filter 3 and the quantizer 1. The quantization resolution of the quantizer 1 is, for example, 32, and the signal y [k] is a signal with a high sampling frequency that is oversampled due to rough quantization. The spectrum of the quantization noise accompanying the re-quantization in the signal y [k] is shaped by the noise shaping filter 3, and the quantization noise component in the audible range is small. The signal y [k] is input to the pulse width modulator 2 and converted into a pulse width modulation signal w (t) having a pulse width determined by the signal y [k]. After that, although not shown in FIG. 3, the pulse width modulation signal w (t) is power amplified by the switching amplifier, and the signal is supplied to the speaker after passing through the low pass filter by the LC filter. As a result, sound that is a reproduction signal is reproduced from the speaker.

  In the above process, in order to sufficiently reduce the audible range component of the requantization noise included in the signal y [k], the sampling frequency of the signal y [k] needs to be sufficiently high. However, in the conventional technique, the sampling frequency of the signal y [k] and the carrier frequency of the pulse width modulation signal w (t) are made equal. This is shown in FIG. This is an example of symmetric uniform pulse width modulation, but the pulse period and sampling period are the same. Therefore, there is a problem that the carrier frequency of the pulse width modulation signal w (t) inevitably increases.

  As a method for reducing the requantization noise itself included in the signal y [k], it is conceivable to increase the quantization resolution of the quantizer 1. However, the pulse width modulator 2 also needs to have a resolution corresponding to the signal y [k]. Since the pulse width modulator 2 is realized by a digital circuit, there is a problem that the clock frequency must be set high in order to have high resolution.

  In order to deal with these problems, first, an approach is adopted in which the pulse width is controlled by sampling the input signal twice during one pulse of the pulse width modulation signal w (t). FIG. 1 shows the relationship between the sampling timing and the pulse width signal w (t) in the present invention. In this example, it is assumed that the value of y [k] changes from -7 to 7. The input signal sample timing is set to an intermediate time during the period when the pulse width modulation signal w (t) is H or L. Since the signal value of the pulse signal changes twice in one period, there are two control timings of the leading edge and trailing edge of the pulse in one pulse period, and this fact is used.

  In digital amplifiers, complementary pulse width modulation is often used. FIG. 4 shows a portion after pulse width modulation of a digital amplifier by complementary pulse width modulation. The pulse width signal y [k] is directly input to one pulse width modulator 21 to generate a pulse width modulation signal w1 (t), and the other pulse width modulator 22 receives a pulse width signal. Are inverted and a pulse width modulation signal w2 (t) is generated. The two pulse width modulation signals W1 (t) and w2 (t) are amplified by the switching amplifiers 41 and 42, respectively, and these signals pass through a low pass filter by LC and are then BTL connected to the speaker 5. The two pulse width modulation signals w1 (t) and w2 (t) are often set in phase.

  FIG. 5 shows the state of the signal waveform when sampling is performed and controlled every half of the pulse period when the two pulse width modulation signals w1 (t) and w2 (t) are set in phase. Is. The two pulse width modulation signals w1 (t) and w2 (t) themselves are controlled every half cycle of the pulse, but w1 is the difference between those signals corresponding to the actual output signal. The waveform of (t) -w2 (t) has a symmetric pulse waveform having a period half that of w1 (t) and w2 (t). That is, controlling the pulse width modulation at a sampling frequency that is twice the carrier frequency of the pulse width modulation corresponds to sampling and controlling every period of the signal waveform of w1 (t) -w2 (t). Therefore, even if the pulse width modulation is controlled at half the carrier width of the pulse width modulation as in the present invention, when the complementary pulse width modulation is used, the carrier frequency of the pulse width modulation is doubled. You can expect almost the same results.

  FIG. 6 is a waveform example of a pulse width modulation signal for explaining a method of doubling the resolution of pulse width modulation. In order to double the resolution of pulse width modulation while keeping the clock frequency and carrier frequency of the pulse width modulator constant, it is allowed to shift the modulation degree in the two pulse width modulators by the minimum unit. For example, when performing pulse width modulation for an input value of 1.5, perform pulse width modulation for an input value of 2 for the pulse width signal w1 (t) and -1 for the pulse width signal w2 (t) Desired pulse width modulation is realized by performing pulse width modulation on the input value. As a result, the signal waveform of w1 (t) -w2 (t) becomes a quasi-symmetric pulse width modulation waveform.

  By using the digital pulse width modulation signal generator of the present invention, the frequency of the pulse width signal in pulse width modulation can be halved while maintaining the sampling frequency of the noise shaping filter for requantization noise as it is. The power efficiency of the amplifier can be increased and electromagnetic noise emission can be suppressed.

  Further, by introducing non-equilibrium operation in complementary pulse width modulation, the resolution of pulse width modulation can be doubled, and re-quantization noise can be reduced.

  The best mode for carrying out the present invention will be described through examples.

  The first embodiment of the present invention is an audio full digital amplifier. The configuration includes a re-quantized signal y [k] generation part shown in FIG. 3, a complementary pulse width modulation part and a speaker driving part shown in FIG. The noise shaping filter 3 is described in Japanese Patent Application No. 2004-047113, and its block diagram is shown in FIG. The output from the quantizer 1 y [k] is fed back in a non-linear manner to suppress harmonics contained in the pulse width modulation signal and prevent the noise floor from rising in the audible range.

  The sampling period of s [i], which is a sound source signal, is Fs = 44.1 kHz, and has a sufficiently high resolution. The oversample ratio in the oversampler 6 is 16, and the sampling frequency of the requantized signal y [k] is Ff = 16Fs = 705.6 kHz. The number of quantization levels of the quantizer 1 is 31.

  In the first embodiment of the present invention, the carrier frequency for pulse width modulation is half the sampling frequency of the requantized signal y [k], so the carrier frequency for pulse width modulation is Fc = 8Fs = 352.8 kHz. . Further, in the first embodiment of the present invention, the number of levels of pulse width modulation is 31, which is the same as the number of quantization levels of the quantizer 1, so the clock frequency required for pulse width modulation is Fk = 32Ff = 512Fs = 22.5792MHz.

  FIG. 8 is an example of the spectrum of the pulse width modulation signal in the first embodiment of the present invention. Since complementary pulse width modulation is used, a spectrum is taken with respect to the difference signal between the two pulse width modulation signals. The sound source signal s [i] is a sine wave having a frequency of 2.7563 kHz. The modulation rate is 82%.

  Since the noise shaping filter described in Japanese Patent Application No. 2004-047113 is used, an increase in the audible noise floor is suppressed in the pulse width modulation signal. Since the complementary type is used for the pulse width modulation, even harmonics of the signal are not generated in principle. Further, the third harmonic of the signal appears. This is also reduced to some extent because the noise shaping filter described in Japanese Patent Application No. 2004-047113 is used. However, the third harmonic is described in Japanese Patent Application No. 2004-236617. Further reduction can be achieved by using this technique together. In the first embodiment of the present invention, the harmonic component of the signal is suppressed to a size that does not cause a practical problem as an audio amplifier, so the technique described in Japanese Patent Application No. 2004-236617 is not used.

  As can be seen from FIG. 8, in the first embodiment of the present invention, by operating the noise shaping filter at the sampling frequency twice the carrier frequency of the pulse width modulation and controlling the pulse width modulator, Although the carrier frequency is as low as 8 times the sampling frequency of the source signal and the clock frequency of the pulse width modulator is as low as 512 times the sampling frequency of the source signal, the requantization noise in the audible range is sufficiently small A pulse width modulated signal is obtained. A high-fidelity full digital audio amplifier can be realized by amplifying the pulse width modulation signal with a switching amplifier and driving a speaker.

  The second embodiment of the present invention is a full digital amplifier for audio, except that the resolution of the requantizer and the pulse width modulator is 62 times that of the first embodiment of the present invention. The apparatus configuration and parameters are the same as those in the first embodiment of the present invention. That is, in the second embodiment of the present invention, the resolution of the pulse width modulation is doubled by allowing the modulation degrees of the two complementary pulse width modulation signals to be shifted by a minimum unit. The clock frequency of the pulse width modulator is also the same as Fk = 32Ff = 512Fs = 22.25792 MHz as in the first embodiment of the present invention.

  FIG. 9 shows an example of the spelling of the pulse width modulation signal in the second embodiment of the present invention. Since complementary pulse width modulation is used, a spectrum is taken with respect to the difference signal between the two pulse width modulation signals. The sound source signal s [i] is a sine wave having a frequency of 2.7563 kHz. The modulation rate is 82%.

  In the second embodiment of the present invention, since the resolution of the pulse width modulation is twice as high as that of the first embodiment of the present invention, the magnitude of the requantization noise should ideally be 6 dB smaller. But the difference is not so great. This is because the difference signal between the two pulse width modulation signals is not a symmetric pulse width modulation signal but a quasi-symmetric pulse width modulation signal, and noise is increased by a noise shaping filter accordingly. Because it is. Nevertheless, noise in the audible range is clearly reduced compared to the first embodiment of the present invention. Further, it is predicted that the second harmonic of the source signal is generated by using the quasi-symmetric pulse width signal. However, in the example of FIG. 8, it is small enough to be buried in the noise floor.

  When the method for improving the resolution of the pulse width modulation is used, the maximum allowable modulation rate is also increased. However, when the noise shaping filter described in Japanese Patent Application No. 2004-047113 is used, the output signal is a quasi-symmetric pulse width signal. As a result, the maximum allowable modulation rate is slightly smaller than the calculated value.

  In the third embodiment of the present invention, the harmonics with respect to the input signal are suppressed using the technique described in Japanese Patent Application No. 2004-236617 as compared to the second embodiment of the present invention. A block diagram for calculating the quantized signal y [k] is shown in FIG. In the figure, the feedforward compensation element 7 for suppressing harmonics outputs a value proportional to the cube of the input signal. Others are the same as in the second embodiment of the present invention, and the pulse width modulation also uses quasi-symmetric complementary pulse width modulation as in the second embodiment of the present invention.

  FIG. 11 is a spectrum of a pulse width modulation signal when a sine wave having a frequency of 2.7563 kHz is input as the sound source signal s [i] to the pulse width modulation signal generator according to the second embodiment of the present invention. The modulation rate is 82%. Since the harmonic suppression means is used, it can be seen that the harmonics can be sufficiently suppressed as compared with the case of the second embodiment of the present invention (FIG. 9).

  FIG. 12 shows a spectrum of a pulse width modulation signal when the input signal is a superposition of a 2.7563 kHz sine wave and a 4.1344 kHz sine wave (two tone signal). The maximum modulation rate is 82%. It can be seen that harmonics and mutual interference with the input signal are not recognized and a good spectrum is obtained.

  By using the digital pulse width modulation signal generator of the present invention, it is possible to realize a full digital audio amplifier that generates less audible noise, generates less electromagnetic noise, and does not require a high clock frequency.

The figure which shows the relationship between the pulse width modulation signal and sampling timing in this invention. The figure which shows the relationship between the pulse width modulation signal and sampling timing in the conventional method. Block diagram of a full digital amplifier. Explanatory drawing of the digital amplifier by complementary pulse width modulation. FIG. 6 is an operation explanatory diagram of complementary pulse width modulation in the present invention. FIG. 4 is an operation explanatory diagram of high-resolution complementary pulse width modulation in the present invention. Explanatory drawing of the pulse width modulation signal generation part used in the 1st and 2nd Example of this invention. The example of the spectrum of the pulse width modulation signal in the 1st example of the present invention. The example of the spectrum of the pulse width modulation signal in the 2nd example of the present invention. Explanatory drawing of the pulse width modulation signal generation part used in the 3rd Example of this invention. The example of the spectrum of the pulse width modulation signal in the 3rd example of the present invention. The example of the spectrum of the pulse width modulation signal with respect to the 2-tone signal in the 3rd Example of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... Quantizer 2, 21, 22 ... Pulse width modulator 3 ... Noise shaping filter 31 ... Delay device 32 ... System matrix 33 ... Output vector 34 ... Input vector 35 ... Nonlinear function vector 36 ... Zero order hold 37 ... Sampler 38 ... Integrator 41, 42 ... Switching amplifier 5 ... Speaker 6 ... Oversampler 7 ... Feed forward Compensation factor

Claims (1)

  In a digital pulse width modulation signal generator, wherein a pulse width modulation signal is used as an output signal, and a low frequency component of the output signal is proportional to a low frequency component of an input signal which is a discrete time signal, a pulse width modulator, a quantizer, and noise The noise shaping filter is a discrete-time filter that receives the input signal and the output of the quantizer as an input and outputs a signal that is input to the quantizer; and the pulse width modulator is the quantizer A discrete-time discrete value signal, which is the output of the detector, is input and the output signal is output. The sampling period of the previous noise shaping filter and the previous quantizer is half the period of the pulse output by the previous pulse width modulator. Yes, the previous pulse width modulator performs pulse width modulation every half of the pulse output by the previous pulse width modulator, Complementary pulse width modulation signal is output, and pulse width modulation is performed by shifting the pulse width modulation degree of only one of the two pulse width modulation signals constituting the complementary pulse width modulation signal by a minimum unit. A digital pulse width modulation signal generator characterized by doubling the resolution.

Images