JPH0779257B2 - Quantizer - Google Patents

Description

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a quantizer for converting a digital signal having a long word length into a digital signal having a short word length sampled at high speed.

2. Description of the Related Art In recent years, due to improvements in digital signal processing technology, signals that have been analog processed conventionally have been digitalized. Along with this, higher performance and lower cost of digital-to-analog converters have become more important. Noise-shaping quantizers are often used for these purposes. As a quantizer using noise shaping, for example, Japanese Patent Laid-Open No. 63-209334 shows a quantizer of a multi-stage noise shaping type. By using this quantizer, stable high-order noise shaping can be performed without causing oscillation. However, on the other hand, there is also a problem that the gradation of the quantizer output increases. Therefore, a method has been proposed in which this quantizer is improved to suppress the increase in gradation of the quantizer output (for example, Journal of solid state circuit, Aug.1989, Vol.
24, No. 4).

FIG. 3 shows a block diagram of the proposed method.
In FIG. 3, the integrator 10 is composed of a delay device 127 and an adder 126.
2, the integrator 102, the adder 128, the local quantizer 103, the delay unit 4 and the subtractor 101 constitute a main loop 100 which is a single integral type noise shaping quantizer having a primary shaping order. Has been done. Also, the adder 120
And the delay device 121 constitute the integrator 108, and the adder 12
2 and the delay device 123 constitute an integrator 110. These integrators 108 and 110, subtractors 107 and 109, local quantizer 6, delay device 11
2 constitutes a sub-loop 106 which is a double integral type noise shaping quantizer having a quadratic shaping order. The difference between the input and output of the local quantizer 103 is given to the sub-loop 106 by the subtracter 2. The output of the integrator 108 is added to the adder 128 after being multiplied by the coefficient a by the multiplier 130, and added to the input of the local quantizer 103. Here, the input X is a 16-bit digital signal, and the local quantizers 103 and 6 perform quantization as shown in Tables 1 and 2. The output is standardized at 16384.

Here, the quantization error generated by the local quantizer 103 is Vq1, and the quantization error generated by the local quantizer 6 is Vq1.
If q2, the relationship between the input X and the output Q1 of the main loop 100 and the relationship between the input X'and the output Q2 of the sub-loop 106 are expressed by the equations (1) and (2).

Q1 = X + (1-z -1) · Vq1 ... (1) Q2 = X '+ (1-z -1) 2 · Vq2 ... (2) On the other hand, input of the adder output of the 2 local quantizer 103 Since it is the output difference, X ′ = − Vq1 (3) Therefore, after the output Q2 of the sub-loop 106 is differentiated by the differentiator 10 composed of the subtracter 13 and the delayer 14, the adder 12 of the main loop 100 is used. When added to the output Q1, the term of Vq1 shown in the equation (1) is canceled out, and the relation between the input and output X, Y as a whole is as shown in the equation (4).

Y = X + (1−z ⁻¹ ) ³ · Vq2 (4) Here, the reason why this circuit operates stably despite the gradation of the sub-loop 106 being ± 0.5 is as follows. That is, the value of the integrator 108 is fed back by the multiplier 130 to the local quantizer 103 via the adder 128.
Therefore, when the value of the integrator 108 is large, the input of the local quantizer 103 is also large, and the value of the subtractor 2 is a large negative value. This value is transferred to the integrator 108 via the subtractor 107.
The other input of the subtractor 107 is
Since it is at most 0.5, a large negative value is input to the integrator 108, and the output of the integrator 108 gradually decreases.

In this way, by feeding back the main loop 100 in the direction in which the value of the integrator 108 becomes smaller, the input of the local quantizer 6 can be suppressed to a smaller value, and the output level of the local quantizer 6 can be reduced. The tone can be lowered.

Here, the possible values of Q1, that is, the gradation is -2, -1,
There are 5 values (5 values) of +2, and the possible values of Q2 are -0.5 and +0.5, so the possible values of Y are-.
7 values of 3, -2, ..., +3. That is, the input signal is compressed into 7 values (a little less than 3 bits). Also,
Equation (4) shows that the quantization error in the low frequency band is driven to the high frequency band. Therefore, by constructing as shown in FIG. 3, the digital output that does not impair the dynamic range of the input digital signal is output. The bit number of the signal can be compressed, and a dynamic range of about 118 dB can be obtained by operating this circuit with 64 times oversampling.

However, in the above configuration, after the value of the integrator 108 in the subloop 106 is determined, the output of the integrator 108 → the multiplier
130 → adder 128 → local quantizer 103 → subtractor 2 → subtractor 1
The calculation must be performed again through the path of 07 → integrator 108 → subtractor 109 → integrator 110 → local quantizer 6, which requires a very long calculation time. Moreover, since the feedback is applied from the output of the integrator 108, which is the first-stage integrator, and the integrator 110, which is the subsequent integrator, has no feedback,
With respect to the integrator 110, there is no measure for preventing oscillation or overflow. Therefore, for example, there is a problem that it is difficult to use a sub-loop having a shaping order of 3 or higher.

The present invention solves the above problem, and an object of the present invention is to provide a quantizer that can obtain the same effect in a shorter calculation time.

In order to solve the above problems, the quantizer according to the present invention,
It has a first local quantizer for quantizing an input signal, a first noise shaping quantizer for performing noise shaping of a given input, and a second local quantizer for quantizing an input signal. And a feedback circuit for feeding back an output of the detection means by a predetermined transfer function, the first noise shaping type quantization The quantization error generated by the feedback circuit is added to the output of the feedback circuit and input to the second local quantizer, and a second noise shaping quantizer for noise shaping and an output level of the feedback circuit are provided. On the basis of the amplitude detection means for adding / subtracting a predetermined value to / from the input signal of the first local quantizer, or the quantization level of the first local quantizer is changed based on the level of the output of the feedback circuit. Means for, with respect to the first local quantizer output of the first noise shaping quantization circuit,
Means for differentiating and adding the second local quantizer output of the second noise shaping quantizer according to the shaping order of the first noise shaping quantizer, and outputting the addition result It is configured to be taken out as.

Action With the above configuration, depending on the level of the feedback circuit output, by adding or subtracting a predetermined value to the input of the local quantizer in the main loop, the quantization error generated by the main loop must be the feedback circuit output. And the opposite polarity,
Alternatively, the quantization level of the local quantizer in the main loop changes according to the level of the output of the feedback circuit, and the value output by the local quantizer is made larger or smaller, resulting in the quantization generated by the main loop. The error always has the opposite polarity to the feedback circuit output. Therefore, the absolute value of the input level of the local quantizer in the subloop to which the addition result of this value and the feedback circuit output is input is always smaller than that of the feedback circuit output, and the gradation of the local quantizer in the subloop is reduced. can do. Further, since the output of the feedback circuit is fixed in advance, the time required for the calculation can be shortened.

Embodiment An embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram of a quantizer showing an embodiment of the present invention. In FIG. 1, the adders 3 and 5, the first local quantizer 1, the subtractor 2, and the delay unit 4 constitute a main loop which is a single integral type noise shaping quantizer. The output is the input / output difference of the local quantizer 1, and this output is added to the adder 3 via the delay unit 4 and fed back to the input. On the other hand, the output of the subtractor 2 is also input to the adder 7, and the adder 7, the second local quantizer 6, the subtractor 8 and the feedback circuit 9 serve as a sub-loop which becomes a double integral type noise shaping quantizer. The input / output difference of the local quantizer 6 is obtained from the output of the subtractor 8, and this output is added to the adder 7 via the feedback circuit 9 and fed back to the input. Further, the output of the feedback circuit 9 is also applied to the adder 5 of the main loop via the amplitude detector 11. The output of the local quantizer 6 is differentiated by the differentiator 10 according to the shaping order of the first noise shaping quantizer, and then added to the adder 12,
It is added to the output of the local quantizer 1 of the main loop.

The local quantizer 1 of the main loop quantizes the input signal, and the relationship between the input and the output is as shown in Table 3. The output is standardized by 11264.

The relationship between the input and the output of the local quantizer 6 of the sub loop is the same as in Table 2 above. The transfer function H (z) of the feedback circuit 9 of the sub-loop is as shown in the equation (5),
Specifically, it has the configuration shown in FIG.

H (z) = − 2z ⁻¹ + z ⁻² (5) That is, the input is given to the delay circuit 41, and this delay circuit
41 signal whose output is doubled by multiplier 44 and delay circuit 41
The difference between the output and the output of the delay circuit 42 is obtained by the subtractor 43. The amplitude detector 11 receives the output β of the feedback circuit 9 and outputs a predetermined value C according to the level thereof.
This is considered a kind of quantizer, and here the input β
The values shown in Table 4 are output accordingly.

Next, the operation of the circuit shown in FIG. 1 will be described. The quantization error generated by the adder 5 and the local quantizer 1 of the main loop of the single integration type noise shaping quantizer is
Assuming Vq1, the output of the subtractor 2 becomes -Vq1. This value is fed back to the input via the delay unit 4, and the output Q1 of the local quantizer 1 of the main loop is as in the equation (6) as in the case of the conventional example.

Q1 = X + (1−z ⁻¹ ) · Vq1 (6) On the other hand, the local quantizer 6 used in the sub-loop of the double integral type noise shaping quantizer is ± 0.5 as described above.
Is output. Further, the feedback amount β by the feedback circuit 9 is up to three times the quantization error by the local quantizer 6, as is clear from the transfer function. Since the quantization error generated by the local quantizer 6 is 0.5 or less during normal operation, the maximum value of the feedback amount β by the feedback circuit 9 is 1.5.

Here, considering the input to the local quantizer 6, this value is the difference between the feedback amount β by the feedback circuit 9 and the quantization error Vq1 generated by the main loop. Considering the case where the feedback amount β input to the amplitude detector 11 is 0 to 11264, 5632 is added to the input of the local quantizer 1 from Table 4, so that the quantization output by the local quantizer 1 is The error Vq1 is 0
.About.11264, and therefore the input P to the local quantizer 6 is in the range -11264 to 11264. That is, when standardized by 11264, it is in the range of -1.0 to 10. However, β, Vq1
Since both are considered to be random values, β and Vq at this time
The average value of 1 is 5632, and the average value P _{AVE of} P, which is the difference, is almost 0.

Similarly, considering other cases, the relationship between the feedback amount β, the output C of the amplitude detector 11, the quantization error Vq1, and the average value P _AVE of the input P of the local quantizer 6 is as shown in Table 5. .
However, in this Table 5, each numerical value is standardized by 11264.

In this way, the average value of the input of the local quantizer 6 is always ±
Within 0.25, it is possible to suppress the occurrence of distortion due to the sub loop. Therefore, as in the prior also in this circuit example, as Vq2 the quantization error local quantizer 6 sub-loop occurs, the output Q2 is Q2 = -Vq1 + (1-z -1) 2 · Vq2 ... ( 7) Further, the output Y of the adder 12 is Y = Q1 + (1-z -1) · Q2 = X + (1-z -1) 3 · Vq2 ... (8) , and the the third-order noise shaping is obtained become. In this case, the output of the local quantizer 1 is 7 from -3 to +3.
Since this is a value, and the output of the local quantizer 6 is a binary value of ± 0.5, the final output Y is a nine value from -4 to +4, and the same shaping effect as the conventional example can be obtained.

The amplitude detector 11 operates as shown in Table 6 based on the input value β, and instead of the adder 5 adding a predetermined value to the input of the local quantizer 1, the local quantizer The same effect can be obtained even if 1 changes the quantization level according to the value of the output C of the amplitude detector 11 as shown in Table 7. All the numerical values shown in Tables 6 and 7 are values standardized by 11264.

In this case, four local quantizers having different quantizing levels are prepared and one of the outputs is taken out by the output C of the amplitude detector 11, or the comparison level is set to the amplitude detector 11's output. The local quantizer 1 is configured to change the output C for a while.

Here, considering the signal transmission path, as can be seen from FIG. 2, since the output of the feedback circuit 9 is fixed,
The signal transmission path is the amplitude detector 11 → adder 5 → local quantizer 1 → subtractor 2 → adder 7 → subtractor 8 or the local quantizer 6, which is significantly shortened. As for the feedback to the main loop, the feedback is applied so that the value to be input to the local quantizer 6 becomes small, so that feedback is applied to the entire sub-loop, and the degree of freedom in designing the feedback circuit 9 is increased. For example, even if the characteristics of the feedback circuit 9 are third order or higher, that is, the shaping order is third order or higher, the circuit can be operated safely.

As an example having a third-order shaping order, the local quantizer 6 shown in Table 8 is used, and the transfer function of the feedback circuit 9 is a high-order one as shown in equation (9). Is shown below.

Here, the local quantizer 6 performs the quantization as shown in Table 8 and the amplitude detector 11 uses the one as shown in Table 9.

Here, considering the input of the local quantizer 6, since this value is the difference between the feedback amount β by the feedback circuit 9 and the quantization error Vq1 generated by the main loop, first, the amplitude detector 11
Considering the case where the feedback amount β input to is from -11264 to +11264, nothing is added to the input of the local quantizer 1 as shown in Table 9, so that the local quantizer 1 outputs The quantization error Vq1 is in the range of −5632 to +5632. Therefore, the value of the input P of the local quantizer 6 is -16896 to +16896.
Range, that is, if standardized with 11264, -1.5 to +1.5
Is in the range. Therefore, the quantization error Vq2 generated by the local quantizer 6 is within ± 0.5, and it can be seen that the operation is stable. Similarly, considering other cases,
10 As shown in the table.

Here, considering the maximum value of the feedback amount β, the maximum value of the gain by the feedback circuit 9 is 2.64 from its transfer function.
During normal operation, the quantization error generated by the local quantizer 6 is within ± 5632 (0.5 when standardized), so it can be said that the value of β rarely exceeds 1.5. Therefore, as the initial value of this quantizer, the feedback amount β
It can be seen that this quantizer operates stably when is set to 0.

As described above, when the local quantizer 6 and the amplitude detector 11 are designed as shown in Tables 8 and 9 by using the transfer function shown in the equation (9), the fourth-order noise shaping effect is obtained in the low frequency range. Obtained and operating this circuit with 32 times oversampling, a dynamic range of about 118 dB can be obtained.

In the above embodiments, the local quantizer 1 which outputs 7 values from -3 to +3 is used, but of course, the present invention is not limited to this, and the value of 8 values or more or 6 values or less is used. It can be one. Further, the case where the transfer function of the feedback circuit 9 shown in the equation (5) or the equation (9) is used and the local quantizer 6 having a binary or ternary output is shown. Of course, it goes without saying that it is not limited to this. Furthermore, the main loop does not have to be a single-integration type noise shaping circuit. It is only necessary that the value from the main loop has the opposite polarity.

EFFECTS OF THE INVENTION As described above, according to the present invention, there is provided a first local quantizer for quantizing an input signal, and a first noise shaping quantizer for performing noise shaping on a given input. A second local quantizer for quantizing the input signal; a detecting means for detecting a quantization error generated by the second local quantizer; and a feedback circuit for feeding back the output of the detecting means by a predetermined transfer function. Second noise for performing noise shaping by adding the quantization error generated by the first noise shaping quantizer to the feedback circuit output and inputting the sum to the second local quantizer. Based on the output level of the shaping type quantizer and the feedback circuit, an amplitude detecting means for adding / subtracting a predetermined value to / from the input signal of the first local quantizer, or based on the level of the output of the feedback circuit Previous A means for changing the quantization level of the first local quantizer, and a first noise shaping quantizer for outputting the output of the second local quantizer with respect to the output of the first local quantizer. Means for differentiating and adding according to the shaping order of, and taking out the addition result as an output, the gradation of the local quantizer in the sub-loop is small, and the gradation of the quantizer as a whole is reduced. Can be reduced. Further, even if a sub-loop feedback circuit having a quadratic degree or more is used, the oscillation of the feedback circuit can be suppressed without increasing the gray scale, and the time required for the calculation can be shortened. Is to have.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of a quantizer of the present invention, FIG. 2 is a block diagram showing details of a feedback circuit of the quantizer, and FIG. 3 is a block diagram of a conventional quantizer. is there. 1,6 ... First and second local quantizers, 2 ... Subtractors, 3,5
…… Adder, 4 …… Delayer, 7 …… Adder, 8 …… Subtractor, 9 …… Feedback circuit, 10 …… Differentiator, 11 …… Amplitude detector.

 ─────────────────────────────────────────────────── ─── Continuation of the front page (56) References JP 62-265810 (JP, A) JP 61-84914 (JP, A) JP 3-289810 (JP, A) JP 63- 161713 (JP, A) JP-A-4-30619 (JP, A) JP-A-4-30620 (JP, A) JP-A-3-289809 (JP, A) JP-A-3-289808 (JP, A) JP-A-4-56407 (JP, A) JP-A-3-289709 (JP, A) JP-B 6-83150 (JP, B2)

Claims (2)

[Claims] 1. A first noise-shaping quantizer for noise-shaping a given input, and a first local quantizer for quantizing an input signal, and quantizing an input signal. A first local quantizer, a detection means for detecting a quantization error generated by the second local quantizer, and a feedback circuit for feeding back the output of the detection means by a predetermined transfer function, The quantization error generated by the noise shaping quantizer is added to the output of the feedback circuit to add the second error.
A second noise shaping type quantizer which performs noise shaping by inputting to the local quantizer and a predetermined value for the input signal of the first local quantizer based on the output level of the feedback circuit. To the first local quantizer output of the first noise shaping quantizer, and the second local quantizer output of the second noise shaping quantizer to the second local quantizer output. And a means for differentiating and adding according to the shaping order of the noise shaping type quantizer of No. 1 and configured to take out the addition result as an output. 2. A first noise-shaping quantizer for noise-shaping a given input, and a first local quantizer for quantizing an input signal, and for quantizing an input signal. A second local quantizer, a detection means for detecting a quantization error generated by the second local quantizer, and a feedback circuit for feeding back the output of the detection means by a predetermined transfer function. The quantization error generated by the noise shaping quantizer is added to the output of the feedback circuit to add the second error.
A second noise shaping quantizer for inputting noise shaping to the local quantizer, and means for changing the quantization level of the first local quantizer based on the level of the feedback circuit output; For the first local quantizer output of the first noise shaping quantizer, the second
Means for differentiating and adding the second local quantizer output of the noise shaping quantizer according to the shaping order of the first noise shaping quantizer, and taking out the addition result as an output. The quantizer configured in.