JPH03169124A - Oversample a/d converter - Google Patents

Abstract

PURPOSE:To prevent deterioration in S/N by providing a means to correct a gain error after an oversample coder and a digital filter. CONSTITUTION:The output of a band pass filter 5 passes through a full wave rectifier 6 and a low pass filter 7, the output of the low pass filter 7 goes to the amplitude B of a reference signal including a gain error. The amplitude is inputted to a comparator 9 together with the amplitude A of a reference signal, and an error due to a gain error included in the value B is obtained. A correction coefficient calculator 10 stores a correction coefficient H=A/B to correct the error in the gain error included in the value B. Since a gain error caused by step imbalance is a linear gain error, the gain error generated inside of a coder 2 is corrected by multiplying a gain error correction coefficient with the signal inputted at normal operation at all times. Thus, the S/N is improved.

Description

Detailed Description of the Invention [Field of Industrial Application] The present invention relates to an oversampling A/D converter, and in particular to correction of errors caused by step imbalance of a 1-bit D/A converter inside an oversampling encoder. Regarding technology.

[Prior Art] Conventionally, this type of oversampling encoder includes a circuit as shown in FIG. In the same figure, 21 and 23 are analog adders. 22.24 is an analog integrator, 25 is a quantizer 26 is a digital delay device, 27.28 is a 1-bit D
/A converter.

The two transformations of the input signal, output signal of this circuit, and quantization noise added by the quantizer are respectively X(Z), y(z),
as well as. Expressed as Q(Z), the input/output transfer characteristic is expressed as Y(Z)-X(Z)+ (1-Z-') 2 Q(
Z)... (1) It is known that it is expressed as.

({) in the expression. The coefficient (1-Z-')' of the second term on the right side can be attenuated sufficiently by increasing the sampling frequency compared to the input analog signal, so the contribution of quantization noise (1-Z-') 2Q(
Z) can be made extremely small.

Therefore, as shown in Figure 3. By combining the oversampling encoder 32 with a prefilter 31 for removing aliasing distortion and a digital low-pass filter for removing high-frequency quantization noise, highly accurate encoding (A/D conversion) can be performed. .

This type of encoder is called an oversampled encoder 32 because the sampling frequency is much higher than the Nyquist frequency. This type of encoder is internally equipped with a local D/A converter that converts the digital signal into an analog signal.

Also, as shown in equation (1), the quantization noise Q (Z)
An encoder that multiplies (1 - Z-') by a coefficient is called a noise shaping type encoder.

In general, oversampled encoders are broadly classified into two types: predictive type and noise shaping type. (Compared to a predictive encoder, a noise shaping type oversampled encoder as expressed in equation 1 requires a smaller number of bits in the D/A converter inside the encoder to obtain the same accuracy. It has the advantage that only 1 bit is required.

However, even in a 1-bit D/A converter, noise occurs due to step imbalance.

[Problems to be Solved by the Invention] However, the above-mentioned noise shaping type oversampled A/D converter generates noise due to step imbalance of the 1-bit D/A converter inside the encoder, and the A/D This has the disadvantage that the S/N of the converter deteriorates.

Here, FIG. 4 shows the relationship between the input level and the S/N in the time domain determined by simulation.

In Fig. 4, the circle indicates S/ when the inside of the encoder is in an ideal state.
N is shown, and the mark is the S/N when -1% step imbalance is applied to the local D/A converter (28 in FIG. 2) inside the encoder.

Furthermore, FIG. 5 shows the definition of S/N calculation (time domain evaluation).

In Figure 5, 51. 3 is a digital filter, and 52 is an oversampling encoder.

Incidentally, IN indicates an input signal, and it is assumed that X (nT>). In FIG.

(2) A digital filter is connected to the encoder 52, and the output when operating normally is set to V2 (nT).

Binding. The definition of S/N in the time domain is as follows.

S/N- From FIG. 4, it can be seen that the S/N when step imbalance is applied is significantly degraded at high input levels.

This time, simulations have confirmed that the noise that occurs when step imbalance is applied is a linear gain error that has an offset component and a frequency component that is equal to the input signal.

SUMMARY OF THE INVENTION In view of the above drawbacks, it is a technical object of the present invention to provide an oversampling A/D converter that improves S/H deterioration.

[Means for Solving the Problems] According to the present invention, there is provided an oversampling encoder that samples an input analog signal with a clock signal having a predetermined multiple of the Nyquist frequency or more and converts it into a digital signal; an oversampled A/D converter having a digital filter that removes a signal, a switch that controls the input to the oversampled encoder, a bandpass filter that receives the output of the digital filter as input, and an output of the bandpass filter. , a low-pass filter that receives the output of the full-wave rectifier as input, a comparator that receives the output of the low-pass filter and a reference signal as input, and a comparator that receives the output of the comparator as input. An oversampling A/D converter is obtained, which includes a correction coefficient calculator and a multiplier that multiplies the digital filter output by the correction coefficient output from the correction coefficient calculator.

That is, the oversampled A/D converter of the present invention includes an oversampled encoder and a subsequent digital filter to correct linear gain errors caused by step imbalance of the local D/A converter inside the encoder. After that, means are provided for correcting the gain error.

In other words, the present invention detects a gain error correction coefficient by providing a training period, and improves the S/N based on the gain error correction coefficient during normal operation.

[Embodiments] Next, embodiments of the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram showing one embodiment of the present invention.

In Figure 1, 1 is a switch, 2 is an oversampling encoder, 3 is a digital filter, 4 is a multiplier, 5 is a band pass filter, 6 is an aftereffect rectifier, 7 is a low pass filter, and 8 is a reference voltage. , 9 is a comparator, and 10 is a correction coefficient calculator.

First, when the power is turned on, a training period is set and switch 1 is set.
is in the state φ. During this period, the input of the oversampled encoder 2 is the reference signal 8.

The output of the oversample encoder 2 is input to a band pass filter 5 after quantization noise is suppressed by a digital filter 3 .

The band-pass filter 5 is set to have the frequency fraction of the reference signal 8 as its center frequency, and therefore the output of the band-pass filter 5 becomes the digital value of the reference signal.

If there is a step imbalance in the local D/A converter inside the oversampling encoder 2, the digital value that is the output of the bandpass filter 5 mentioned above will contain a gain error component, and therefore the oversampling Analog value input to encoder 2 and band pass filter 5
This means that the digital value that is the output of , has an error equal to the gain error.

The output of the band pass filter 59 passes through the full wave rectifier 6 and the low pass filter 7, so that the output of the low pass filter 7 becomes the amplitude value B of the reference signal including the gain error.

This amplitude value is input to the comparator 9 together with the amplitude value A of the reference signal, and the error due to the gain error included in B is determined.

The correction coefficient calculator 10 calculates the gain value A included in B. The correction coefficient H-1 for correcting the error of You will end up saving it.

The gain error caused by step imbalance is
Since it is a linear gain error, the gain error generated inside the encoder 2 is corrected by always multiplying the signal input during normal operation by the gain error correction coefficient.

Figure 4 shows the simulation results when the input signal was 37.5 kHz.

In the figure, the mark Δ is the S/N after gain error correction. S clearly deteriorated to the mark when step imbalance was applied.
It can be seen that the gain error is corrected by the above procedure and the S/N becomes close to the ideal state indicated by O.

[Effects of the Invention] As explained above, the present invention provides 1 bit D/D inside the encoder.
The gain error correction coefficient is detected by providing a training period for the noise caused by the step imbalance of the A converter, and the S/N is adjusted as shown in Figure 4 by constantly performing gain error correction multiplication during normal operation.
deterioration can be improved.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing an example of an oversampled A/D converter according to an embodiment of the present invention, FIG. 2 is a block diagram showing an example of a conventional oversampled encoder, and FIG. 3 is a block diagram showing an example of a conventional oversampled encoder. FIG. 4 is an overall block diagram of the case where an A/D converter is constructed using an oversampled encoder, and shows the deterioration of S/N due to noise and the effects of the present invention. FIG. 5 is a diagram for explaining the definition of S/N calculation. FIG. 2 also provides an example of the detailed structure of the oversampled encoder portion in FIGS. 1, 3, and 5. Figure 2 Figure 5

Claims (1)

[Claims] 1) An oversampling A/D that includes an oversampling encoder that samples an input analog signal using a clock signal that is a predetermined multiple of the Nyquist frequency or more and converts it into a digital signal, and a digital filter that removes high frequency components of the digital signal. In the converter. a switch that controls input to the oversampling encoder; a bandpass filter that receives the digital filter output; a full-wave rectifier that receives the band-pass filter output; and a full-wave rectifier that receives the output of the full-wave rectifier. a low-pass filter; a comparator that receives the output of the low-pass filter and the reference signal; a correction coefficient calculator that receives the output of the comparator; and a correction coefficient that is the output of the correction coefficient calculator; An oversampling A/D converter comprising: a multiplier that multiplies the output of a digital filter;