Coded data generation or conversion – Analog to or from digital conversion – Differential encoder and/or decoder
Reexamination Certificate
2003-10-09
2004-11-23
Tokar, Michael (Department: 2819)
Coded data generation or conversion
Analog to or from digital conversion
Differential encoder and/or decoder
C341S144000
Reexamination Certificate
active
06822594
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates in general to the field of signal processing, and more specifically to a system and method for providing overload protection and stability in high order, 1-bit delta-sigma modulators.
2. Description of the Related Art
Many signal processing systems, such as audio signal processing systems, utilize delta-sigma modulators to provide output data with a high, in-band signal to noise ratio.
FIG. 1
depicts a conventional digital signal processing (“DSP”) system 100 that includes a P-order digital loop filter
112
with a z-domain transfer function of L(z). DSP system
100
receives an R-bit digital input signal
104
from source
106
. Source
106
can be any digital signal source, such as an analog-to-digital converter, a compact disk player and/or recorder system or a digital versatile disk (DVD) player and/or recorder. Digital input signal
104
often undergoes one or more preprocessing operations
108
prior to modulation by delta-sigma modulator
102
. The type of pre-processing operation depends upon the purpose of DSP system
100
. Example preprocessing operations are decimation and interpolation. Often preprocessing operation(s)
108
increase or decrease the sampling frequency of digital input signal
104
by a factor of “k”. A summer
110
adds the now k-bit digital input signal
104
with the negative of output data q from delta-sigma modulator
102
. The P-order digital loop filter processes the oversampled digital input signal
104
minus feedback signal q and generates an output signal u.
Delta-sigma modulator
102
represents a one-bit delta-sigma modulator. One-bit delta-sigma modulators provide a full-scale “high” or full scale “low” quantization output data q. The full-scale “high” and full-scale “low” are referred to herein as respective logical values +1 and −1. Thus, if the input signal, u, to the 1-bit quantizer
114
resides above a predetermined threshold level, q is +1. Output data q is otherwise −1. Output data q provides feedback to loop filter
112
. Output data q is provided to one or more post-processing components
116
. The output data q can be used directly by some digital recording components to directly encode digital media. In other embodiments, post-processing components include a digital-to-analog converter (“DAC”), such as a switched capacitor filter, to process groups of M output signals q to reconstruct the digital input signal
104
into an analog waveform. Post-processing components
116
can also include an analog low pass filter to filter out out-of-band noise and provide the filtered output signal to a load, such as a speaker system. “Delta-sigma modulators” are also commonly referred to using other interchangeable terms such as “sigma-delta modulators”, “delta-sigma converters”, “sigma delta converters”, and “noise shapers”.
Many DSP systems utilize delta-sigma modulators because of their noise shaping capabilities. The delta-sigma modulator
102
shapes the noise frequency components of digital input signal
106
so that nearly all the noise energy lies out-of-band, i.e. outside a bandwidth of interest. In an audio system, the bandwidth of interest typically lies within the range of 0-25 kHz. P-order loop filter
112
is well-known to those of ordinary skill in the art and includes P integration stages. In general, higher order delta-sigma modulators provide better in-band noise shaping characteristics by shifting more noise energy to frequencies outside the bandwidth of interest. The inherently coarse quantization of 1-bit delta-sigma modulators results in a higher noise floor. Because of the better in-band noise shaping characteristics of higher order loop filters, 1-bit delta-sigma modulators often include at least a third order loop filter. Third order and higher order delta-sigma modulators are referred to herein as “high order” delta-sigma modulators.
Stability is a primary concern in the design and operation of delta-sigma modulators. “Instability” of a delta-sigma modulator means that the delta-sigma modulator exhibits large, bounded (or unbounded) states and poor signal to noise ratios, especially as compared with predictions by linear models. Stability in high order delta-sigma modulators is even more challenging than 1
st
and 2
nd
order delta-sigma modulators because the analytical complexities of high order delta-sigma modulators in actual operational environments have defied conventional attempts to develop accurate instability prediction mechanisms. Thus, although empirical evidence can be obtained to suggest stability for high order delta-sigma systems over a variety of operational regions, stability cannot be guaranteed for all possible input signals.
Many causes for instability exist with high order delta-sigma modulators. One such cause is quantizer overload. Quantizer overload generally occurs when a quantizer receives input data that is either excessively high or low. Higher order delta-sigma modulator systems exhibit an increasingly lower tolerance to input signals that approach reference levels (i.e. output limits) of the quantizer. For example, an input signal as small as one-half of a reference level can cause a sixth order delta-sigma modulator to overload. Quantizer overload conditions can cause instability via a number of mechanisms including nonlinear feedback to the high order loop filter, large quantization noise, and low signal gain. For example, if quantizer
114
has a limited range of output signals, such as signals between reference levels of +/−5. An input signal outside the range will cause the quantizer
114
to unsuccessfully attempt to track the output signal. Signals can be unintentionally outside the range due to a variety of well-known reasons including the Gibbs phenomenon overshoot that occurs when a discontinuity exists between input samples. Even when the input signal falls within the range of quantizer
102
but lies near the range boundary, the quantizer
102
could overload. The output signal of delta-sigma modulator
102
can generally be approximated by a probability density function having an approximately Gaussian distribution. Thus, although the input signal
104
to the delta-sigma modulator
102
may originally lie within a non-overload range, the input signal u to the 1-bit quantizer
114
may exceed the non-overload range. Accordingly, quantizer overload conditions can exist when a probability of quantizer overload exists. Such error can be attributed to a number of sources such as internal noise and other system perturbations. Thus, if the delta-sigma modulator
102
receives an input signal of +4.9, a probability that the DS modulator will attempt to provide a series of output signals q representative of an output >+5 exists. Such an output can cause nonlinear feedback to the loop filter. For a P-order delta-sigma modulator, the greater P is the less tolerance delta-sigma modulator
102
has for input signals close to the reference levels.
Instability of delta-sigma modulators causes many undesirable effects. In audio systems, instability can cause oscillations resulting in undesirable, audible tones. Instability can also cause abrupt signal magnitude and frequency changes, which also result in undesirable noise.
Many techniques exist for handling quantizer overload conditions. A limiter can be used to confine the input signal to a no-overload region through clipping operations in analog systems or through bit truncations in discrete systems. Other techniques rely on complicated designs, which can require a significant amount of design time, chip real estate, and implementation difficulties. However, despite an enormous amount of effort, due to the analytical complexities of high order delta-sigma modulators, accurate instability prediction mechanisms have yet to be developed.
SUMMARY OF THE INVENTION
In one embodiment of the present invention, a delta-sigma modulation system includes an M-order filter to process input data, wherein M is great
Chambers Kent B.
Melanson John L.
Chambers Kent B.
Cirrus Logic Inc.
Hamilton & Terrile LLP
Nguyen Linh Van
Tokar Michael
LandOfFree
Overload protection and stability for high order 1-bit... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Overload protection and stability for high order 1-bit..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Overload protection and stability for high order 1-bit... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3295417