Jointly nonlinear delta sigma modulators

• US 7,190,294 B2
• Filed: 01/13/2005
• Issued: 03/13/2007
• Est. Priority Date: 01/16/2004
• Status: Active Grant

• Current Assignee: Cirrus Logic Incorporated
• Original Assignee: Cirrus Logic Incorporated
• Inventors: Melanson, John L.
• Primary Examiner(s): JEAN PIERRE, PEGUY
• Application Number:
  US11/035,288
• Publication Number:
  US 20050156770A1
• Time in Patent Office: 789 Days
• Field of Search: 341/143, 341/155, 341/144
• US Class Current: 341/143
• CPC Class Codes: H03M 7/3011 of non-linear distortion, e.g. by temporarily adapting the operation upon detection of instability conditions avoiding instability by structural design H03M7/3035
• CPC Class Codes: H03M 7/3013 Non-linear modulators
• CPC Class Codes: H03M 7/3028 the quantiser being a single bit one
• CPC Class Codes: H03M 7/304 with distributed feedback, i.e. with feedback paths from the quantiser output to more than one filter stage

Abstract

A signal processing system includes a jointly non-linear delta sigma modulator. In one embodiment, the jointly non-linear delta sigma modulator includes a non-linear quantization transfer function, and the output of the delta sigma modulator is defined, at least in part, by a non-linear interrelationship of multiple noise-shaping filter state variables. A look-ahead delta-sigma modulator can be implemented as a noise shaping filter and a function generator. State variables of the noise shaping filter provide the input data from which the function generator determines a quantizer output signal. Latter state variables are more dominant in determining the quantizer output signal. Accordingly, earlier state variables can be approximated to a greater degree than latter state variables without significant compromise in quantization accuracy.

Claims

1. A signal processing system comprising:

a jointly nonlinear delta sigma modulator comprising;

a noise shaping filter to process a signal and generate N state variables, wherein N is greater than or equal to two; and

a quantizer to quantize quantizer input data determined from at least a subset of the N state variables in accordance with a nonlinear quantization transfer function, wherein one or more boundaries between quantization regions for at least one quantization level are characterized at least in part by a nonlinear interrelationship between at least two of the state variables.

2. The signal processing system as in claim 1 wherein the boundary between at least two quantization regions is nonconvex.

3. The signal processing system as in claim 2 wherein N is greater than or equal to three, and the boundaries between at least two quantization regions are nonconvex in N-dimensions.

4. The signal processing system as in claim 2 wherein N is greater than or equal to three, and the boundary between at least two quantization regions is nonconvex in less than N-dimensions.

5. The signal processing system as in claim 1 wherein the noise shaping filter comprises N integrators and the state variables represent output values of at least a subset of the integrators during successive, discrete times.

6. The signal processing system as in claim 5 wherein the subset of the integrators consists of the Nth integrator through the N-x^th integrators, wherein 2≦

x≦

N−

1.

7. The signal processing system as in claim 1 wherein the noise shaping filter comprises an infinite impulse response filter.

8. The signal processing system as in claim 1 further comprising:

a processor; and

a memory coupled to the processor and storing processor executable code to implement the noise shaping filter and the quantizer.

9. The signal processing system as in claim 1 wherein during operation of the delta sigma modulator the quantizer input data is at least a subset of the N state variables.

10. The signal processing system as in claim 9 wherein during operation of the delta sigma modulator the quantizer input data comprises state variables from a first integrator and a second integrator of the filter and the state variables of the first and second integrators have a non-linear interrelationship with respect to boundaries in the quantization transfer function.

11. The signal processing system as in claim 1 wherein the quantizer input data is derived from audio input signal data.

12. The signal processing system as in claim 1 further comprising:

signal processing and recording equipment to process output data from the quantizer and record the processed output data on storage media.

13. The signal processing system as in claim 1 wherein the quantizer input data comprises an approximation of at least a subset of the N state variables.

14. A method of processing a signal using a jointly nonlinear delta-sigma modulator that includes a noise shaping filter having N state variables, the method comprising:

filtering an input signal to the jointly nonlinear delta-sigma modulator using the noise shaping filter to generate respective quantizer input data from the N state variables, wherein N is greater than or equal to two; and

quantizing the quantizer input data in accordance with a nonlinear quantization transfer function, wherein one or more boundaries between quantization regions for at least one quantization level are characterized at least in part by a nonlinear interrelationship between at least two of the state variables.

15. The method as in claim 14 wherein the boundary between at least two quantization regions is nonconvex.

16. The method as in claim 15 wherein N is greater than or equal to three, and the boundaries between at least two quantization regions are nonconvex in N-dimensions.

17. The method as in claim 14 wherein N is greater than or equal to three, and the boundary between at least two quantization regions is nonconvex in less than N-dimensions.

18. The method as in claim 14 wherein the noise shaping filter comprises N integrators and the state variables represent output values of at least a subset of the integrators during successive, discrete times.

19. The signal processing system as in claim 18 wherein the subset of the integrators consists of the Nth integrator through the (N-x)th integrators, wherein 2≦

x≦

N−

1.

20. The method as in claim 14 wherein filtering an input signal to the jointly nonlinear delta-sigma modulator using the noise shaping filter to generate respective quantizer input data further comprises:

generating state variables from at least a first integrator and a second integrator of the filter and the state variables of the first and second integrators have a non-linear interrelationship with respect to boundaries in the quantization transfer function.

21. The method as in claim 14 wherein the input signal data sample comprises audio input signal data.

22. The method as in claim 14 further comprising:

recording quantized quantizer input signal data on storage media.

23. A method of quantizing an input signal using non-linearly related input data and a nonlinear quantization transfer function, the method comprising:

receiving input signal samples;

filtering the input signal samples to generate digital filter state variables; and

applying a nonlinear quantization transfer function to at least a subset of the state variables to generate an output signal, wherein one or more boundaries between quantization regions for at least one quantization level are characterized at least in part by a nonlinear interrelationship between at least two of the state variables.

24. The method as in claim 23 wherein the input signal is comprised of audio digital signal samples.

25. An apparatus comprising:

means to filter input signal samples to generate digital filter state variables representing state variables of a multi-order digital filter; and

means to quantize at least a subset of the state variables using a nonlinear quantization transfer function, wherein one or more boundaries between quantization regions for at least one quantization level are characterized at least in part by a nonlinear interrelationship between at least two of the state variables.