1. Field of the Invention
This invention relates to a method and circuit for stopping signals quantized using noise-shaping.
2. Description of the Prior Art
A common problem related to systems using noise-shaping concerns how to stop a noise-shaped signal without producing a disturbing inband transient. It would therefore, be advantageous and desirable to provide a circuit and method that can stop a noise-shaped signal with a minimum of inband noise related to the transition.
The present invention is directed to a circuit and method for stopping a noise-shaped signal with a minimum of inband noise related to the transition. When the noise-shaper is running (with a zero input signal), the quantized output signal f(n) consists of a finite number of discrete amplitude levels forming a signal with an inband signal content which is ideally zero. Stopping the noise-shaper, by forcing the output signal f(n) to zero, typically causes a significant inband transient which degrades the inband system performance.
More specifically, a circuit and method are provided for stopping a quantized signal from a noise-shaper with a significantly reduced inband transient, compared to a traditional random stop of a noise-shaped signal (i.e. forcing f(n) to zero at an arbitrary time instant). The noise-shaped signal is stopped at a favorable time controlled by a detector that indicates the occurrence of a good time to stop the noise-shaper such that the transient is minimized in a fashion that substantially reduces the inband disturbance.
According to one aspect, a noise-shaper stop detector is simple to implement in digital hardware containing the noise-shaper.
According to another aspect, a noise-shaper stop detector is provided having only negligible added production cost when implemented in digital hardware.
According to one embodiment a method of stopping transmission of a quantized signal from a noise-shaper comprises the steps of: feeding the quantized signal to a detector filter;
monitoring a Boolean function B(x(k)) of the state variables x(k) of the detector filter; and stopping transmission of the quantized output signal only at a time where the Boolean function is true such that a switching transient occurring when the quantized output signal transmission is stopped, is minimized to substantially reduce inband disturbance.
According to another embodiment, a system for stopping a quantized signal from a noise-shaper comprises: a filter fed by a quantized signal from a noise-shaper; a circuit configured to generate a Boolean output signal in response to the state variables of the filter; at least one combinatorial element configured to generate a stop signal in response to the Boolean output signal; and a switch configured to turn the noise-shaped signal on and off in response to the stop signal.
According to yet another embodiment, a system for generating a noise-shaper stop signal comprises: a filter fed by a quantized signal from a noise-shaper; a circuit configured to generate a Boolean output signal in response to the state variables of the filter; and a control logic circuit configured to generate a stop signal in response to a desired sample number associated with the Boolean output signal.
Other aspects, features and many of the attendant advantages of the present invention will be readily appreciated as the invention become better understood by reference to the following detailed description when considered in connection with the accompanying drawings in which like reference numerals designate like parts throughout the figures thereof and wherein:
a) is a block diagram illustrating a detector for synchronized stop of a noise-shaped signal f(n), where the detector filter is of first order;
b) is a block diagram illustrating a simplified topology for implementing the detector shown in
While the above-identified drawing figures set forth alternative embodiments, other embodiments of the present invention are also contemplated, as noted in the discussion. In all cases, this disclosure presents illustrated embodiments of the present invention by way of representation and not limitation. Numerous other modifications and embodiments can be devised by those skilled in the art which fall within the scope and spirit of the principles of this invention.
Methods for stopping a noise-shaped signal with a minimum of inband noise related to the transition are discussed herein below with reference to
When the output signal f(n) is forced to zero, the loop filter can no longer correct for past quantization errors by using corrections to future samples. This gives generally an in-band transient error if the Noise Shaper is stopped at an arbitrary time instant due to the lack of correction when the output is forced to zero.
Example 1 below now describes a method and results of implementing a traditional noise-shaper stop (ie. at an arbitrary time) using the topology shown in
A 2nd order noise-shaper is made using the topology shown in
Future Filter Response Energy
A future filter response energy (FFRE) can be calculated, given a zero future filter input and the present filter state. To do this the in-band filter is defined in state-space. Any linear filter can be defined in state-space representation by the equations
x(n+1)=Ax(n)+bf(n)
y(n)=cTx(n)+df(n) (2)
where x(n) is a column vector of the filter states, A is a square matrix, b and c are column vectors and d is a constant. The f(n) term is the filter input and the y(n) term is the filter output. A, b, c and d define the filter transfer function; and any filter can be represented in this state-space form. The T symbol denotes the transpose operation. Given a filter input at zero (f(n)=0), and using the foregoing definitions, it can be shown that
x(n+k)=Akx(n), k≧0 (3)
y(n+k)=cTx(n+k)
y(n+k)=cTAkx(n), k≧0 (4)
The energy EN(n) of the filter transient in the time-window from n to n+N (normalized by the sampling period) is given by
is a constant matrix given by the filter parameters. Then
Subtraction of the two equations gives
The {circle around (X)} operator is the Kronecker product and the vec function converts a matrix to a column vector by stacking the columns of the matrix. The vec−1 function is the inverse function of the vec function in the sense that vec−1(vec(X))=X, e.g. the vec−1 function forms a matrix from a vector. Stable filters have the property that,
The limit value of the sum in equation (6) forming MN can consequently be found for N→∞ if the filter is stable, which gives
Example 2 below describes one method for calculating the future inband transient energy E∞(n) for each sample time, assuming that the future filter input f(n) is zero.
In the present example is used the same noise-shaper and inband filter characteristic as in Example 1. Also here we have n0=0. The FFRE of the inband filter E∞(n) can be calculated for each sample, assuming that the future filter input is zero—if the noise-shaped signal is stopped.
The FFRE calculation for the inband filter is done using equation (5), where the constant matrix MN=M∞ is found from (10) using the inband filter state-space representation. At
Now running the system, stopping the noise-shaped signal at n0=0 gives the quantized output signal f(n) shown in
The distribution of the FFRE is given by the noise-shaper characteristic and the inband filter. In
Stop of Noise-Shaped Signal Controlled by Detector
In view of the foregoing background information,
Most noise-shaper designs cause the quantized output to have some stochastic properties (e.g. due to the added random dither signal or the use of chaotic noise shapers having loop filters with poles outside the unit circle). In this case, the actual stop time of the noise-shaped signal will have an influence on the inband disturbance caused from the stop transient. The FFRE of an inband filter on the noise-shaped signal indicates if it is a good time to stop the noise-shaped signal with respect to the inband disturbance. Exemplary variations in the FFRE of an inband filter were shown in
With continued reference now to
Detector 14 guarantees that the transient energy (seen through the detector filter) from the stop of the noise-shaped signal is lower than the limit EL; but the detector also implies a delay in the stop. A reasonable trade-off must therefore be made between low inband transient energy and low stop delay. The probability density function for the stop delay is useful in making this trade-off. Principally it cannot be guaranteed that the detector 14 will stop the noise-shaped signal 20 in finite time; and therefore a timeout may be needed—depending on the application. Example 3 below describes one method for implementing a stop of a noise-shaped signal synchronized by a detector.
An alternative detector 100 topology is shown in
The system at
In Example 3 was used a 10th order detector filter which was a close approximation to an ideal inband filter. However, a simple detector filter can be made using only a first order filter, which causes the state-space filter parameters A, b, c and d to be single-dimensioned scalars. In
In order to evaluate the performance of the simple detector 300, a number of stops of the noise-shaped signal are simulated on the same noise-shaper as used herein before. The detector filter has a first order low-pass characteristic with a −3 dB frequency at 1/20 of the sampling frequency as shown in
In
In summary explanation, a method is described for stopping a quantized signal from a noise-shaper with a significantly reduced inband transient, compared to a traditional random stop of a noise-shaped signal. The method stops the noise-shaped signal at a favorable time controlled by a detector, that monitors the noise-shaped signal—or the states of the internal loop-filter in the noise-shaper. The detector indicates the presence of a good time to stop the noise-shaped signal; and when indicated, the noise-shaped signal can selectively be stopped—resulting in a small inband transient and a correspondingly small inband disturbance.
When evaluating the efficiency of an inband transient reduction system, the statistics needs to be compared, with- and without the inband transient reduction system enabled, e.g. the reference is the inband transient energy, when the traditional way to stop a noise-shaped signal is used. The detector in Example 3 was shown to provide a mean reduction of 13.5 dB in the inband transient energy level caused by the stop of the noise-shaped signal.
This invention has been described in considerable detail in order to provide those skilled in the art with the information need to apply the novel principles and to construct and use such specialized components as are required. In view of the foregoing descriptions, it should be apparent that the present invention represents a significant departure from the prior art in construction and operation. However, while particular embodiments of the present invention have been described herein in detail, it is to be understood that various alterations, modifications and substitutions can be made therein without departing in any way from the spirit and scope of the present invention, as defined in the claims, which follow.
Number | Name | Date | Kind |
---|---|---|---|
2829323 | Steele | Apr 1958 | A |
5019819 | Kimura | May 1991 | A |
5648778 | Linz et al. | Jul 1997 | A |
5697374 | Odagiri et al. | Dec 1997 | A |
5760722 | Harris et al. | Jun 1998 | A |
6289508 | Erickson et al. | Sep 2001 | B1 |
6304200 | Masuda | Oct 2001 | B1 |
6362765 | Masuda | Mar 2002 | B2 |
6404369 | Sheen | Jun 2002 | B1 |
6532256 | Miller | Mar 2003 | B2 |
6556158 | Steensgaard-Madsen | Apr 2003 | B2 |
6950049 | Brooks et al. | Sep 2005 | B2 |
7058136 | Hellberg et al. | Jun 2006 | B1 |
20020018012 | Nakao et al. | Feb 2002 | A1 |
Number | Date | Country |
---|---|---|
04115722 | Apr 1992 | JP |
Number | Date | Country | |
---|---|---|---|
20040017854 A1 | Jan 2004 | US |