JP3214981B2 - Delta-sigma A / D converter with AGC function - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention
In the field of multimedia and the like, an A for converting an analog signal such as audio or voice into a digital signal.
More specifically, the present invention relates to an A / D converter with an automatic gain control (GC) function, and more specifically, a delta-sigma A / A converter with an AGC function that maintains a predetermined signal amplitude level of a digital signal even when the signal amplitude level of an analog signal changes. It relates to a D converter.

[0002]

Conventionally scheme 1 Figure 1 shows an example of a conventional AGC function A / D converter which is used in the field of digital audio.

[0003] In FIG. 1, an audio input signal is amplified by a variable amplifier 10 whose amplification rate changes by external control, and an output thereof is input to a high-pass filter 11. The high-pass filter 11 is for removing a DC component included in the output signal of the variable amplifier 10, and generally has a cut-off frequency of about 10 Hz and is realized by a first-order CR filter including a capacitor and a resistor. Is done.

The maximum amplitude level (peak level) of the output of the high-pass filter 11 is detected by a peak detector 12. The detected peak level is compared with reference data by the comparator 13, and the gain of the variable amplifier 10 is controlled according to the comparison result. With this configuration, an AGC loop that works so that the peak level becomes equal to the reference data is created. By this AGC loop, even if the AC signal amplitude level of the audio input signal changes, the peak level of the AC signal from the variable amplifier 10 is maintained at a predetermined level. This predetermined level is called an AGC output level. In digital audio applications, only the AC component of the signal is important, so such an AGC method is suitable.

[0005] The output of the variable amplifier 10 is input to an anti-aliasing filter 14 together with the high-pass filter 11 in the AGC loop. This anti-aliasing filter 14
Converts the signal frequency component of the output of the variable amplifier 10 before the A / D conversion into a sampling frequency (f
This is to limit the band to less than half of s). When sampling a continuous analog signal, according to the Nyquist theorem, all frequency components of fs / 2 or more are DC and fs /
Appears between two. This is called folding. The anti-aliasing filter 14 is a low-pass filter for eliminating the influence of the aliasing.

For digital audio applications, fs = 48
Required frequency band (pass band) is 20k for kHz
Hz, so to prevent aliasing, 20k
It must have a low-pass characteristic in which the gain is sharply attenuated between Hz and 24 kHz.

The output of the anti-aliasing filter 14 is A /
The signal is converted into a digital output (A / D output) having a sampling frequency fs by the D converter 15. A 16-bit A / D converter is used here because a 16-bit dynamic range is required for audio applications.

As described above, in FIG. 1, the variable amplifier 1 which holds the peak level of the AC signal component at the AGC output level
0 is band-limited by the anti-aliasing filter 14,
The conversion into a digital signal by the 16-bit A / D converter 15 realizes an A / D converter with an AGC function.

However, in this conventional system, since the peak detector 12 and the comparator 13 are realized by analog circuits, there is a disadvantage that an error occurs in the detection result and the comparison result, and the AGC output level shifts.

A general peak detector is composed of a rectifier and a hold circuit for holding the previous peak level.

As the comparator, an analog comparator for judging the magnitude of the peak level and the reference data is used. And each circuit is an analog circuit,
It will be affected by noise and offset.

Conventional system 2 FIG. 2 shows A with AGC function which has solved the disadvantages of conventional system 1.
4 shows an example of a / D converter. The biggest difference from the conventional method 1 is that the 16-bit A / D converter 22 is located in the AGC loop. The output of the variable amplifier 21 is 16
Since the data is converted into a 16-bit digital output (A / D output) of fs = 48 kHz by the bit A / D converter 22, the high-pass filter 23, the peak detector 24, and the comparator 25 can be realized by a digital circuit. Becomes

Therefore, the peak detection and comparison with the reference data are not affected by analog noise or offset unlike the conventional method 1. As a result, an A / D converter with an AGC function with an accurate AGC output level can be realized.

Another difference from the conventional system 1 is that the anti-aliasing filter 20 is located before the variable amplifier 21. As described in the description of the conventional method 1, the low-pass characteristics necessary for preventing aliasing are steep. In general, such a low-pass filter has a large number of orders, such as ninth and higher, and has a high Q value for each order. Therefore, in the passband,
The gain variation (ripple) with respect to frequency is (+/-) 0.
It becomes as large as about 5 dB.

[0015] The AGC loop works to maintain the peak level of the A / D output at the AGC output level. Therefore,
If a low-pass filter having poor ripple characteristics is present in the AGC loop, when the main frequency component of the audio input changes, that is, when the frequency that most affects the peak level changes, the amplification factor of the variable amplifier 21 according to the ripple characteristics changes. Will change.

The frequency dependence of the amplification factor is expressed by A /
It appears as distortion at the D output. This distortion is a problem in digital audio. Therefore, in the conventional example of FIG.
The anti-aliasing filter 20 will be located before the variable amplifier 21. As a result, since the output of the variable amplifier 21 is not low-passed, all the noise generated by the variable amplifier 21 is turned back to a frequency of fs / 2 or less. For example, if the noise band of the variable amplifier 21 is 50 times fs, fs
The noise of the variable amplifier 21 existing in the band equal to or less than / 2 is
It is 100 times larger than that of the conventional method 1.

Since the noise generated by the variable amplifier 21 increases with the amplification factor, the effect of the noise appears more remarkably especially when the amplitude level of the audio input is small. As described above, one of the drawbacks of the system shown in FIG. 2 is that the influence of noise generated by the variable amplifier 21 is large.

Further, as a common drawback between the system shown in FIG. 2 and the conventional system 1, since the order of the low-pass filter for preventing aliasing is large, the circuit scale becomes large, and the noise of the low-pass filter itself is large. . In addition, a common drawback is that the group delay characteristics are poor because of the multi-order and high Q value of each order.

Another disadvantage of the method shown in FIG. 2 is that the circuit size of the high-pass filter 23 realized by a digital circuit is large. The filter required here has a simple high-pass characteristic with a cutoff frequency of 10 Hz and an order of 1st to 2nd order. As a basic circuit for realizing a filter by digital signal processing, a multiplier, a register, and an operation coefficient ROM are required. With the simple high-pass characteristics described above, the number of registers and the number of words in the operation coefficient ROM are small. However, the multiplier does not get smaller.

The circuit scale of the multiplier is determined by the bit length of the data to be multiplied. Here, the A / D of the A / D converter 22
It is determined by the output and the bit length of the coefficient ROM required for the high-pass filter. The bit length of the coefficient ROM depends on the precision of the cutoff frequency. However, since the A / D output is as large as 16 bits, the multiplier does not become small. Thus, here, a large-scale multiplier is provided only to realize the high-pass filter 23, which is inefficient when considering the circuit scale.

[0021]

As described above, the conventional system 1 has a disadvantage that the AGC output level cannot be accurately controlled. Further, the conventional system 2 has a disadvantage that the circuit scale of the high-pass filter required for amplitude level detection cannot be reduced due to the influence of the noise of the variable amplifier.

SUMMARY OF THE INVENTION Accordingly, an object of the present invention is to provide a high-pass filter circuit required for accurately maintaining an AGC output level, reducing the influence of noise of a variable amplifier, and detecting an amplitude level, in view of the above-mentioned drawbacks of the prior art. An object of the present invention is to provide a delta-sigma A / D converter with an AGC function, which can be reduced in scale.

[0023]

In order to achieve the above object, the present invention maintains the amplitude of an A / D conversion output signal at a predetermined level even when the amplitude level of an analog input signal fluctuates. A / D with AGC (automatic gain control) function
In the converter, a variable amplifier capable of variably setting the input / output amplitude ratio of the analog signal, and oversampling the analog signal amplified by the variable amplifier at a frequency higher than a finally required sampling frequency to obtain 1 bit or multiple bits. A delta-sigma modulator for converting to bit quantized data, and inputting the quantized data, performing decimation processing and frequency band limiting processing, and finally converting the data to an A / D conversion output having a required sampling frequency and resolution A digital filter section, a level detection section for detecting an amplitude level of a digital signal generated by the digital filter section, and an input / output amplitude ratio of the variable amplifier according to a detection result of the level detection section. Gain control means for controlling the gain.

Here, to prevent aliasing, it is preferable to insert a low-pass filter between the delta-sigma modulator and the variable amplifier. At this time, the cutoff frequency of the low-pass filter can be switched according to the sampling frequency of the A / D conversion output. Further, it is preferable that the digital filter section has a high-pass signal frequency characteristic.

Further, the digital filter section is composed of a cascade connection of two or more filter stages, and the filter stage for outputting a digital signal to be input to the level detecting means is different from the filter stage for outputting an A / D conversion output. It is also possible to make it.

Also, the input amplitude of the delta-sigma modulator is
It is also possible to have limiter means for limiting the amplitude below a limiter level set higher than the full-scale analog input level of the delta-sigma A / D converter with AGC function.

The amplification factor control means compares the detection result of the level detection means with the reference digital data, and outputs a digital signal indicating a magnitude or a difference, and generates an analog level dependent on the output of the comparison means. And a time constant control means for generating a gain control signal by controlling a time constant at which the output of the D / A converter changes. It is also possible to adopt a configuration that changes linearly according to the rate control signal. In this case, it is possible to switch the time constant of the time constant control means depending on the output of the comparison means.

[0028]

[Action]

1) According to the delta-sigma A / D converter according to the present invention, the delta-sigma modulator oversamples the output of the variable amplifier at a frequency sufficiently higher than the finally required sampling frequency of the A / D conversion output. Therefore, it is possible to insert a simple anti-aliasing filter having less pass band ripple after the variable amplifier. Therefore, the influence of the noise of the variable amplifier due to the folding is eliminated. Also, noise of the anti-aliasing filter itself is reduced.

2) In the present invention, the digital filter section originally has a steep low-pass filter characteristic in order to remove quantization noise in a high frequency range generated by the delta-sigma modulator. And a basic circuit such as a register, a multiplier, and a calculation coefficient ROM. Therefore, an increase in the circuit size for performing additional filtering processing, for example, high-pass filtering for peak detection is small. Further, by using a multi-stage configuration of the digital filter unit, it is possible to realize an additional digital processing function with a small increase in circuit scale without affecting the AGC operation.

3) Further, according to the present invention, by limiting the input amplitude of the delta-sigma modulator, it is possible to maintain the stability of the delta-sigma modulator even if the input amplitude of the variable amplifier increases rapidly. It is possible.

4) Further, according to the present invention, the comparison result between the peak level and the reference data is D / A converted, and the time constant at which the D / A conversion output changes is controlled, so that the time constant is linear with respect to time. Since the gain of the variable amplifier is changed, the change in the gain does not cause a problem in hearing.

[0032]

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

Embodiment 1 FIG. 3 shows an embodiment in which the present invention is applied to digital audio applications.

In this embodiment, the audio input signal is amplified by the variable amplifier 30 and the anti-aliasing filter 31
Is input to The output of the anti-aliasing filter 31 is 4
The next delta-sigma modulator 32 performs sampling at an oversampling frequency (64 fs) that is 64 times the final sampling frequency (fs) = 48 kHz.
It is converted to bit quantized data.

The first digital filter 33 limits the band of the 1-bit quantized data to a pass band of 20 kHz.
Generate a 16-bit digital output (A / D output) at the fs rate. The second digital filter 34 has a cut-off frequency of 10 to remove the DC component of the A / D output.
Hz high-pass characteristics.

The peak detector 35 detects the peak level of the output of the second digital filter 34. Comparator 36
Compares the peak level with the reference digital data, and controls the amplification factor of the variable amplifier 30 according to the comparison result.

With the above configuration, the audio input is set to 20
It is converted to a 16-bit A / D output with a sampling rate of 48 kHz band-limited to a kHz pass band. Also,
Since the AGC loop works so that the output of the peak detector 35 and the reference digital data become equal, the A / D converter with the AGC function does not change even if the amplitude level of the audio input changes. Can be realized.

The fundamental difference between the present embodiment shown in FIG. 3 and the conventional method 2 is that the A / D converter is not a Nyquist sampling type but is a delta sigma type A / D converter which is one of oversampling types. It is a vessel. The delta-sigma A / D conversion method is a method for achieving a wide dynamic range at a low cost when a CMOS LSI is formed using a miniaturization process. It has become.

Here, in order to realize a dynamic range equivalent to 16-bit resolution, a delta-sigma modulator of fourth-order noise shaping and 64 times oversampling is used. The circuit is realized by a switched capacitor circuit that can be easily integrated in CMOS.

In the AGC loop, since the peak detector 35 and the comparator 36 are realized by digital circuits, unlike the conventional method 1, the AGC output level can be accurately adjusted without being affected by analog noise or offset. Can be held.

The present embodiment shown in FIG.
Another difference is that the anti-aliasing filter 31 can be located after the variable amplifier 30. Since the delta sigma modulator 32 oversamples at 64 fs, the anti-aliasing filter 31 has a frequency between the pass band frequency and (64 fs−pass band frequency), that is, 20 fs.
Any low-pass characteristic that attenuates between kHz and 3.052 MHz may be used. In addition, unnecessary frequency components of 20 kHz or more are removed by the first digital filter 33, so that the cutoff frequency can be sufficiently separated from the pass band.

Accordingly, a primary filter having a cutoff frequency of about 50 kHz is sufficient for the anti-aliasing filter 31. Here, the simplest and small-scale RC filter of the first order of resistance and capacitance is used. Primary R
Since there is no passband ripple in the C filter,
Even if it is located in the AGC loop, it does not cause the distortion described in the conventional method 2.

Further, since the cutoff frequency is sufficiently higher than the passband and there is no droop with a gain near 20 kHz, the frequency dependency of the AGC output level in the passband can be completely eliminated.

As described above, since the delta-sigma modulator 32 performs oversampling, the anti-aliasing filter 31 that does not cause a problem of droop of gain having good ripple characteristics can be replaced with the variable amplifier 30 and the delta-sigma modulator 32.
It becomes possible to insert between.

As a result, the variable amplifier 30 by folding back
Can be eliminated. The RC filter used here has a small circuit scale, and does not itself become a noise source. Further, even if the anti-aliasing filter 31 is not inserted, the influence of the noise of the variable amplifier 30 is smaller than that of the conventional system 2. The delta-sigma modulator 32 has 64
Since oversampling is performed at fs, in this case, the noise of the variable amplifier 30 returns to 32 fs = 1.536 MHz or less. This is about 1/32 compared to the prior art 2 when considering the noise in the pass band.

As described above, a major difference between the present embodiment shown in FIG. 3 and the conventional system 2 is that A / D conversion is performed using a delta-sigma modulator and a digital filter unit. . Here, the digital filter section has a two-stage configuration of a first digital filter 33 and a second digital filter 34. The first digital filter 33 removes quantization noise outside the pass band included in the 64 fs quantized data output from the delta-sigma modulator 32, and removes the 16-bit A / D signal decimated to fs.
Generate a D output. The second digital filter 34 is a high-pass filter with a cutoff frequency of 10 Hz for removing a DC component from the output of the second digital filter 33. Here, since the high-pass filter is necessary only for the peak detector 35, the digital filter unit has a two-stage configuration, and the output of the first digital filter 33 in the first stage is A
/ D output.

The first digital filter 33 is constituted by an FIR type digital filter in order to realize a steep low-pass characteristic, and has a large-scale register, an operation coefficient ROM, and a multiplier. The second digital filter 34 includes 1
Since the following IIR digital filter configuration is used, the operation coefficient R
Only a few OMs and registers are needed. Further, the second digital filter 34 includes a multiplier as the first digital filter 33.
And time sharing. Further, the operation coefficient R
As for OM, it also serves as a ROM control unit. As described above, in the configuration shown in FIG.
Smaller than in the case of

As described above, by configuring the embodiment as shown in FIG. 3, the influence of the return of the variable amplifier noise is eliminated, and the circuit size of the high-pass filter required for peak level detection is small. An A / D converter with an AGC function for audio can be realized.

In this embodiment, the fourth-order noise shaping and the 1-bit quantization delta-sigma modulator are used. However, the present invention can be applied to other noise-shaping orders and multi-bit quantization delta-sigma modulators. Can be implemented. In this embodiment, the method of detecting the peak level is used to detect the amplitude level of the digital signal. However, other detection means such as effective value detection may be used.

Embodiment 2 FIG. 4 shows an embodiment of the present invention suitable for making an A / D converter with an AGC function for audio input from a microphone used in a multimedia personal computer into an LSI. Here is an example.

In FIG. 4, a voice input (microphone input) signal from a microphone is amplified by a variable amplifier 40 whose amplification factor changes by external voltage control. The output of the variable amplifier 40 is limited in amplitude in a limiter circuit 41 and is input to an anti-aliasing filter 42.

The output of the anti-aliasing filter 42 is supplied to the fourth-order delta-sigma modulator 43 at the sampling frequency (f / F) of the digital output (A / D output) finally required.
s) is oversampled at 64 times (64 fs), and is converted into 1-bit quantized data at a rate of 64 fs.

The first digital filter 44 converts the 1-bit quantized data into an 18-bit digital signal decimated to a 1/8 frequency (8 fs).
The second digital filter 45 has a high-pass characteristic and generates an 8 fs rate 18-bit digital signal from which the DC component of the output of the first digital filter 44 has been removed. The output of the second digital filter 45 is a peak detector 47
And the third digital filter 46.

The third digital filter 46 band-limits the input signal to a pass band of 4/10 (0.4 fs) of fs, further decimates by 1/8, and outputs a 16-bit A / F signal of fs rate. Convert to D output. Further, the third digital filter 46 has an attenuator function in which the amount of attenuation of the signal amplitude changes according to an externally controlled attenuation amount control signal.

The peak detector 47 detects the peak level of the output amplitude of the second digital filter 45.

The comparator 48 compares the peak level with the reference digital data, and outputs the magnitude as 1-bit polarity data.

The 1-bit D / A converter 49 includes a comparator 48
Outputs one of the two analog voltages according to the polarity data of the output. The output of the 1-bit D / A converter 49 is input to a time constant circuit 50 composed of a resistor and a capacitor, and the changing speed is controlled. The amplification factor of the variable amplifier 40 changes according to the output voltage of the time constant circuit 50.

With the above configuration, the microphone input is set to 0.4 f
An A / D converter that converts the data into a 16-bit A / D output with an fs rate band-limited to s and generates an output from the third digital filter 46 is realized. The output of the second digital filter 45 is the peak detector 47 and the comparator 4
Since an AGC loop for controlling the amplification factor of the variable amplifier 40 is configured via the 8-bit and 1-bit D / A converter 49 and the time constant circuit 50, even if the amplitude level of the microphone input fluctuates, the A / D An A / D converter with an AGC function in which the output amplitude level does not change is realized.

FIG. 4 shows a configuration utilizing the advantages of the present invention described in the first embodiment. Further, several measures different from those of the first embodiment are devised in consideration of multimedia applications. Hereinafter, each device and advantage will be described in detail.

There is a great difference in the AGC range between multimedia use and audio use. The AGC range is a ratio between a minimum and a maximum of the input amplitude level that is required to keep the amplitude level of the A / D output at the AGC output level, and is 10 dB to 20 dB for audio applications and 30 dB to 40 dB for multimedia applications.

FIG. 5 shows the A / A with AGC function shown in FIG.
4 shows the DC input / output characteristics of the D converter. This FIG.
When the loop has a microphone input amplitude of 10 mVpp to 320
This shows that the A / D output operates so as to be maintained at the AGC output level within the range of mVpp. Although the A / D output is actually a digital code, the digital code is normalized by setting the full-scale digital output to 2.8 Vpp.

Here, the AGC output level is 2 Vpp, which is set to a level about 3 dB lower than the full scale voltage.

The maximum / minimum gain of the variable amplifier 40 is 4
Since it is 6 dB / 16 dB, the AGC range is 30 dB. As described above, in multimedia applications for microphone input, the AGC range is wide, and therefore, some measures are taken in this embodiment.

As one of them, a voltage-controlled variable amplifier (VGA) whose amplification factor changes linearly according to an external control voltage is used for the variable amplifier 40. Generally, when controlling the gain by digitally detecting the peak, a method of digitally selecting the gain from several gain settings by switching resistance or the like is used.

However, when the AGC range is wide, the circuit scale becomes large in order to reduce the interval of setting the gain so as not to cause a problem in hearing when switching the gain. In the VGA, if the external control voltage (VGA control voltage) changes linearly, the amplification factor changes linearly, so that there is no problem in hearing.

The method of linearly changing the VGA control voltage will be described in detail below. Although the VGA is used here, other means for linearly controlling the amplification factor such as current control may be used.

As a second contrivance, a limiter circuit 41 is inserted after the variable amplifier 40. The fact that the AGC range is wide means that the amplification factor of the variable amplifier 40 is kept large when the amplitude level of the microphone input is minimum.

In FIG. 5, when the microphone input is 10 mVpp or less, the amplification factor is the maximum of 46 dB. When the microphone input amplitude suddenly increases in a state where the amplification factor is large, the input amplitude of the delta-sigma modulator 43 also rapidly increases. Generally, 2
Delta-sigma modulators above are conditionally stable and can only maintain stable operation over a limited input range.

Therefore, when the input increases sharply, the AGC
Before the gain becomes small due to the function of the loop, the delta-sigma modulator 43 enters an unstable operation state, causing problems such as oscillation and deterioration of S / N. To prevent unstable operation,
It is effective to insert a limiter circuit 41 for limiting the input amplitude of the delta-sigma modulator 43.

After the amplitude of the microphone input sharply increases,
During the period until the amplification rate decreases, the A / D conversion operation may be performed as long as the A / D output is a full-scale digital output. Therefore, if the limiter level of the limiter circuit 41 is larger than the full-scale voltage (+/-) 1.4 V corresponding to the full-scale digital output, the A / D conversion operation is not affected and the delta-sigma modulator is not affected. 43 can be kept stable. Here, the limiter level is (+
/-) 1.5V.

In FIG. 4, the limiter circuit 41 is located between the variable amplifier 40 and the anti-aliasing filter 42, but may be located at any other position as long as the input amplitude of the delta-sigma modulator 43 can be limited. Further, the variable amplifier 40 and the anti-aliasing filter 42 may have a limiter function. Further, the full-scale voltage can be set to a level close to the power supply voltage of the LSI, and the amplitude can be naturally limited by the variable amplifier 40 by the power supply voltage.

Another difference between the multimedia application and the audio application is that fs is not fixed. For audio applications, fs is 48 kHz or 4
It is fixed at either 4.1 kHz, but varies between 5.5 kHz and 44.1 kHz for multimedia applications.

Since the digital filter and the delta-sigma modulator are sampling circuits that perform a sampling operation, the signal frequency characteristics change in conjunction with fs. However, the cutoff frequency of the anti-aliasing filter, which is a linear circuit, is determined by linear elements such as a resistor and a capacitor, and is not linked to fs. In order to solve this problem, the anti-aliasing filter 42 is a primary RC filter including a resistor R1 and a capacitor C1, and C1 is connected to the outside of the LSI. The cutoff frequency fc is

[0074]

Fc = 1 / (2π · R1 · C1) (1) Therefore, since the cutoff frequency can be freely controlled by changing C1 externally, the influence of the aliasing of the noise of the variable amplifier 40 can be effectively eliminated even if fs changes. Here, the cutoff frequency can be varied by externally connecting a linear element that determines the cutoff frequency. Alternatively, the cutoff frequency can be switched by automatically detecting a change in fs inside the LSI. .

In this embodiment, the digital filter section has a three-stage configuration. The difference from the first embodiment described above is that the filtering for limiting the band to the pass band and the digital filter processing for performing the decimation are divided into two stages of a first digital filter 44 and a third digital filter 46. is there.

For multimedia applications, there are various input media such as CDROM and FM sound source in addition to microphone input. Therefore, in consideration of mixing signals from various media, it is necessary to independently control the amount of attenuation of signals from each medium. This means that an A / D converter with an AGC function that receives a microphone signal as input receives an attenuator function that operates independently of the AGC loop.

Here, filtering and decimation processing are performed in two stages, only the first digital filter 44 in the former stage is positioned in the AGC loop, and the third digital filter 46 in the latter stage has an attenuator function. . This configuration does not affect the operation of the AGC loop,
/ D output attenuation can be controlled. The attenuator function in the third digital filter 46 can be realized by making the coefficient ROM used for digital operation variable.

Therefore, the increase in the circuit scale is smaller than when the attenuator is provided independently of the A / D converter. For example, when controlling the attenuation amount in 6 dB steps, only the bit shift of the coefficient ROM is required. As described above, when the present invention is used, since a digital filter for performing a digital operation is originally present, the increase in the circuit scale due to the addition of the attenuator function required for multimedia use is smaller than that of the related art.

Further, the frequency characteristic of the peak detector 47 can be canceled by the multi-stage configuration of the digital filter section as shown in FIG. If the difference between the frequency characteristic and the ideal characteristic of the peak detector 47 is corrected by the first digital filter 44 and the third digital filter 46 cancels the correction, the A / D output and the AGC loop operation can be performed. , The frequency characteristics of the peak detector 47 have no effect. Since the addition of the digital processing is only the change of the coefficient ROM of the first digital filter 44 and the third digital filter 46,
The circuit scale does not increase.

In the above description, it is assumed that V
He stated that if the GA is controlled, the change in the amplification rate will not cause a problem in hearing. In this embodiment, in order to generate such a VGA control voltage on a small circuit scale, 1-bit D / A
A combination of the converter 49 and the time constant circuit 50 is used.

The 1-bit D / A converter 49 includes a comparator 48
The two switches are switched according to the one-bit polarity data of the output of (1) to output an analog voltage of either Vda or GND. In this state, the output of the 1-bit D / A converter 49 is a square wave. Therefore, the time constant circuit 50
, The temporal change is limited to a transient response with a predetermined time constant, and the VGA control voltage is used. Time constant circuit 50
Is composed of two resistors R10 and R11 and a capacitor C10. Since the time constant is the product of the resistance and the capacitance, when changing in the Vda direction, it is determined by R10 and C10, and when changing in the GND direction, it is determined by R11 and C10.

If this time constant is sufficiently large (generally 2
0 msec or more), the change in the amplification factor due to the change in the VGA control voltage does not cause a problem in hearing. The conversion cycle of the 1-bit D / A converter 49 is 1/8 fs, which is sufficiently smaller than the above time constant. Therefore, when the microphone input amplitude is constant, V
The GA control voltage is a sawtooth wave having a minimum amplitude around a control voltage corresponding to an amplification factor for maintaining an AGC output level. When the amplitude of the microphone input changes, the VGA control voltage changes smoothly with the time constant of the time constant circuit 50.

As described above, according to the configuration of FIG. 4, even when the peak level and the reference digital data are compared by digital processing, a VGA control voltage that changes linearly with time can be generated. Here, in order to reduce the circuit scale, the comparator 48 having a 1-bit output and the 1-bit D / D
Although the A converter 49 is used, the same effect can be obtained by using a multi-bit output comparator and a multi-bit D / A converter.

As another contrivance necessary for multimedia use, R10 and R11 of the time constant circuit 50 are used.
Are different resistance values, and the time constant is changed according to the polarity data of the output of the comparator 48. This is when the amplitude of the microphone input suddenly increases and when the amplitude suddenly decreases.
The purpose is to switch the time constant at which the amplification factor of the variable amplifier 40 changes.

A state in which the AGC loop operates so that the input amplitude sharply increases to reduce the amplification factor is called an attack state. The AGC loop operates so that the input amplitude decreases rapidly and the amplification factor increases. The state is called a decay state.
In order to prevent the AGC operation from becoming unnatural when listening to a microphone voice input, the decay state generally has a time constant at which the amplification factor changes by 10 compared to the attack state.
Must be at least twice as large. Therefore, in this embodiment, R10 and R11 are set to 470 kΩ and 47 kΩ, respectively.
Kiloohms. Since C10 is 1μF in common,
The time constant Tdcy in the decay state is

[0086]

Tdcy = R10 · C10 = 470 [msec] (2) where the time constant Tatk in the attack state is

[0087]

## EQU3 ## The following equation is obtained: Tdcy = R11 · C10 = 47 [msec] (3), and Tdcy is ten times as large as Tatk.

FIG. 6 shows the transient operation in the decay state (however, only the envelope is shown and the signal portion is omitted from the drawing technique). The frequency of the microphone input is 1 kHz, and the amplitude is 320 mVpp to 1
The A / D output when the voltage suddenly decreases to 0 mVpp is shown.
In FIG. 6, as in FIG. 5, the full scale voltage is normalized to 2.8 Vpp as the digital code of the A / D output. In the decay state, since the polarity data “0” of the comparator 48 continues, the time constant at which the VGA control voltage changes is fixed to Tdcy.

In the state shown in FIG. 6, since the amplification factor changes from one end of the AGC range to the other end, the time for the amplitude of the A / D output to change from 10% to 90% of the AGC level (2 Vpp) is:

[0090]

Tdcy · (ln0.9−ln0.1) = 1.03 [sec] (4)

FIG. 7 shows the transient operation in the attack state (however, only the envelope is shown as in FIG. 6, and the detailed signal portion is omitted). Here, the microphone input frequency is also 1 kHz, and the amplitude is 10 m.
This shows a case where the voltage suddenly increases from Vpp to 320 mVpp. In this case, the time constant at which the VGA control voltage changes is T
Fixed to atk. During the period until the amplification rate decreases, the amplitude of the A / D output is limited to the full-scale voltage. As described above, if the method of generating the VGA control voltage used in the present embodiment is used, only two types of resistors are switched,
It is possible to easily change the time constant between the attack state and the decay state.

[0092]

As described above, according to the present invention, it is possible to accurately control the AGC output level and reduce the influence of noise of the variable amplifier, which cannot be realized by the prior art.

Further, an increase in the circuit scale for adding digital arithmetic processing such as a high-pass filter for peak detection is
It is smaller than the prior art.

Further, even when the AGC range is wide, the gain of the variable amplifier can be controlled so as not to cause a problem in audibility.

[Brief description of the drawings]

FIG. 1 A for digital audio conventionally known
It is a block diagram which shows the A / D converter with a GC function (conventional method 1).

FIG. 2 shows a conventionally known digital audio A.
It is a block diagram which shows the A / D converter with a GC function (conventional system 2).

FIG. 3 is a block diagram showing a delta-sigma A / D converter with an AGC function for digital audio according to an embodiment of the present invention.

FIG. 4 is a block diagram showing a delta-sigma A / D converter with an AGC function for multimedia according to another embodiment of the present invention.

FIG. 5 is a diagram showing DC input / output characteristics in FIG. 4;

FIG. 6 is a diagram showing a transient operation in a decay state in FIG. 4;

FIG. 7 is a diagram showing a transient operation in an attack state in FIG. 4;

[Explanation of symbols]

 Reference Signs List 30 variable amplifier 31 folding filter 32 Δ modulator 33, 34 digital filter 35 peak detector 36 comparator

──────────────────────────────────────────────────続 き Continued on the front page (58) Field surveyed (Int.Cl. ⁷ , DB name) H03M 3/04

Claims (8)

(57) [Claims] An A / D converter having an AGC (automatic gain control) function for holding the amplitude of an A / D conversion output signal at a predetermined level even when the amplitude level of an analog input signal fluctuates. A variable amplifier capable of variably setting a signal input / output amplitude ratio, and over-sampling an analog signal amplified by the variable amplifier at a frequency higher than a finally required sampling frequency to obtain 1-bit or multi-bit quantized data. A delta-sigma modulator that converts the input data into a digital signal; a digital filter unit that receives the quantized data, performs decimation processing and frequency band limiting processing, and finally converts the input data into an A / D conversion output having a required sampling frequency and resolution. Level detecting means for detecting an amplitude level of a digital signal generated by the digital filter section; And a gain control means for controlling a gain for setting an input / output amplitude ratio of the variable amplifier in accordance with a detection result of the variable detector. converter. 2. The delta-sigma A / D converter with an AGC function according to claim 1, wherein said digital filter section has a high-pass signal frequency characteristic. 3. The delta-sigma A / D with AGC function according to claim 1, wherein a low-pass filter is inserted between the delta-sigma modulator and the variable amplifier to prevent aliasing. converter. 4. The delta-sigma A / D with AGC function according to claim 3, further comprising means for switching a cutoff frequency of said low-pass filter in accordance with a sampling frequency of said A / D conversion output. converter. 5. A digital filter unit comprising a cascade connection of two or more filter stages, a filter stage for outputting a digital signal to be input to the level detecting means, and a filter for outputting the A / D conversion output. 2. The delta-sigma A / D converter with AGC function according to claim 1, wherein the stages are different. 6. An input amplitude of the delta-sigma modulator,
2. The delta with AGC function according to claim 1, further comprising limiter means for limiting the amplitude below a limiter level set higher than a full-scale analog input level of the delta-sigma A / D converter with AGC function. Sigma type A / D converter. 7. The amplification factor control unit compares the detection result of the level detection unit with reference digital data, and outputs a digital signal indicating a magnitude or a difference, and an analog depending on an output of the comparison unit. A D / A converter for generating a level; and a time constant controller for generating a gain control signal by controlling a constant when the output of the D / A converter changes. 2. A according to claim 1, wherein the rate changes linearly in response to the gain control signal.
Delta-sigma A / D converter with GC function. 8. The delta-sigma A / D converter with an AGC function according to claim 7, wherein a time constant of said time constant control means is switched depending on an output of said comparison means.