Complex digital demodulator employing Chebychev-approximation derived synthetic sinusoid generation

• US 5,361,036 A
• Filed: 08/12/1993
• Issued: 11/01/1994
• Est. Priority Date: 08/12/1993
• Status: Expired due to Term

• Current Assignee: BEI Technologies, Incorporated
• Original Assignee: Rockwell International Corporation
• Inventors: White, Stanley A.
• Primary Examiner(s): MIS, DAVID C
• Application Number:
  US08/105,326
• Time in Patent Office: 446 Days
• Field of Search: 329/361, 329/304, 329/308, 329/309, 329/357, 329/345, 329/346, 329/310, 329/305, 329/306, 329/307, 329/356, 340/669, 340/870.05, 340/870.3, 340/870.06, 324/162, 324/727, 310/318, 364/426.01, 364/566, 73/489, 73/514, 73/517 R, 73/517 AV, 73/517 A, 73/518
• US Class Current: 329/361
• CPC Class Codes: G01C 19/5607 using vibrating tuning forks double-ended tuning forks using planar vibrating masses suspended at opposite ends G01C19/5719
• CPC Class Codes: H03D 1/2254 and a phase locked loop
• CPC Class Codes: H03D 3/006 by sampling the oscillations and further processing the samples, e.g. by computing techniques H03D3/007 takes precedence
• CPC Class Codes: H03D 3/007 by converting the oscillations into two quadrature related signals H03D3/245 takes precedence
• CPC Class Codes: H03L 7/0991 the oscillator being a digital oscillator, e.g. composed of a fixed oscillator followed by a variable frequency divider H03L7/0995 takes precedence; fixed oscillators with means for selecting among various phases H03L7/0814

Abstract

To demonstrate a signal, the signal is sampled and each sample of the signal is multiplied by the sine and the cosine of a phase angle indicating when the sample was taken from the signal. The sinusoidal signal for producing an in-phase demodulated signal is Chebychev-approximation derived and computed as a selected even or odd polynomial, depending on whether the phase angle falls within one of a plurality of angular ranges. So that the error in the synthetic sinusoid is minimax and so that the even and odd polynomials have similar computational complexity, the angular ranges for the even polynomial exceed the angular ranges for the odd polynomial. Preferably, the sinusoid for producing a quadrature-phase demodulated signal is computed as a differential of the sinusoid spliced from the odd and even polynomials. Therefore the quadrature-phase demodulated signal can be provided with a minimal increase in computational complexity. The demodulation method permits a signal to be demodulated by a reference frequency signal to produce a demodulated digital signal that is synchronized to a system clock. The system clock need not be synchronized to the reference frequency signal. The demodulation method is computationally efficient, and permits a digital signal processor to be programmed for demodulating an angular rate signal from a quartz angular rate sensor vibrating at about 10 kilohertz.

Claims

1. A method of demodulating a digital signal by a reference frequency signal, said digital signal including a series of digital values, said reference frequency signal including a series of phase angle values, each of said digital values having a corresponding one of said phase angle values, said method comprising the steps of:

a) computing, with computing means, a numerical approximation of a sinusoidal function of each phase angle value;

b) multiplying, with a multiplier, each digital sample by the numerical approximation of the sinusoidal function of the phase angle corresponding to said each digital value, to produce a series of products of said sinusoidal function; and

c) digitally filtering, with a filter, the series of products.

2. The method as claimed in claim 1, wherein said numerical approximation is a Chebychev-derived approximation.

3. The method as claimed in claim 1, wherein said numerical approximation of said sinusoidal function is computed by comparing, with a comparator, said each phase angle to predefined angular limits, transforming, with transforming means, said each phase angle by a linear transformation to produce a transformed phase angle in response to the comparing of said phase angle to the predefined angular limits, and computing, with computing means, said numerical approximation for said sinusoidal function as either an even function or an odd function of said transformed phase angle in response to the comparing of said each phase angle to the predefined angular limits.

4. The method as claimed in claim 3, wherein said even function is a polynomial of said transformed phase angle, and said odd function is also a polynomial of said transformed phase angle.

5. The method as claimed in claim 3, wherein said even function is used for said numerical approximation of said sinusoidal function over a greater angular range of said each phase angle than said odd function.

6. A method of demodulating a digital signal by a reference frequency signal, said digital signal including a series of digital values, said reference frequency signal including a series of phase angle values, each of said digital values having a corresponding one of said phase angle values, said method comprising the steps of:

a) computing, with computing means, a numerical approximation of a sinusoidal function of each phase angle value; and

b) multiplying, with a multiplier, each digital value by the numerical approximation of said sinusoidal function of the phase angle corresponding to said each digital value, to produce a series of products of said sinusoidal function;

wherein said numerical approximation is computed by comparing, with a comparator, said each phase angle to predefined angular limits, transforming, with transforming means, said each phase angle by a linear transformation to produce a transformed phase angle in response to the comparing of said phase angle to the predefined angular limits, and computing, with computing means, said numerical approximation of said sinusoidal function as either an even function or an odd function of said transformed phase angle in response to the comparing of said each phase angle to the predefined angular limits.

7. The method as claimed in claim 6, wherein said even function is a polynomial of said transformed phase angle, said odd function is also a polynomial of said transformed phase angle, and said even function is used for said numerical approximation of said sinusoidal function over a greater angular range of said each phase angle than said odd function.

8. A method of demodulating an analog signal by a reference frequency signal to produce a demodulated signal, said method comprising the steps of:

a) sampling, with a sampler, said analog signal to produce a digital sample;

b) determining, with a determiner, a phase angle of said reference frequency signal existing when said analog signal is sampled to produce said digital sample;

c) computing, with computing means, a numerical approximation of a sinusoidal function of said phase angle; and

d) multiplying, with a multiplier, said digital sample by said numerical approximation of said sinusoidal function of said phase angle, to produce a product that is a sample of said demodulated signal.

9. The method as claimed in claim 8, further comprising the step of digitally filtering, with a filter said demodulated signal.

10. The method as claimed in claim 8, wherein said numerical approximation of said sinusoidal function of said phase angle is a Chebychev-derived approximation.

11. The method as claimed in claim 8, wherein said phase angle is determined by counting cycles of a clock signal in a counter, latching a count from said counter into a latch when a transition occurs in said reference frequency signal, and determining said phase angle from said count latched in said latch.

12. The method as claimed in claim 11, wherein a number of cycles of said clocking signal occur in a cycle of said reference frequency signal, and said phase angle is determined from said count latched in said latch by dividing, with a divider, said count by said number.

13. The method as claimed in claim 12, wherein said count latched in said latch is divided by said number by computing, with computing means, a polynomial approximation of the reciprocal of said number, and multiplying, with a multiplier, the polynomial approximation for the reciprocal of said number by said count latched in said latch.

14. The method as claimed in claim 8, wherein said numerical approximation of said sinusoid function is computed by comparing, with a comparator, said phase angle to predefined angular limits, transforming, with transforming means, said phase angle by a linear transformation to produce a transformed phase angle in response to the comparing of said phase angle to the predefined angular limits, and computing, with computing means, said numerical approximation as either an even or odd function of said transformed phase angle in response to the comparing of said phase angle to the predefined angular limits.

15. The method as claimed in claim 14, wherein said even function is a polynomial of said transformed phase angle, said odd function is also a polynomial of said transformed phase angle, and said even function is used for said numerical approximation over a greater angular range of said phase angle than said odd function.

16. The method as claimed in claim 8, further including the step of computing, with computing means, a differential of said sinusoidal function to produce a quadrature-phase function in phase quadrature with said sinusoidal function, and multiplying, with a multiplier, said sampled value by said differential to produce a product that is a sample of a quadrature-phase demodulated signal.

17. A method of demodulating a digital signal by a reference frequency signal, said digital signal including a series of digital values, said reference frequency signal including a series of phase angle values, each of said digital values having a corresponding one of said phase angle values, said method comprising the steps of:

a) computing, with computing means, a numerical approximation of a pair of sinusoidal functions of each phase angle value, said pair of sinusoidal functions being in phase quadrature with each other; and

b) multiplying, with a multiplier, each digital value by the numerical approximation of each sinusoidal function of the phase angle corresponding to said each digital value to produce a series of products of each of said sinusoidal functions.

18. The method as claimed in claim 17, further comprising the steps of digitally filtering, with a filter, the series of products of a first one of said sinusoidal functions, and digitally filtering, with a filter, the series of products of a second one of said sinusoidal functions.

19. The method as claimed in claim 18, wherein the numerical approximation of one of said sinusoidal functions is computed, with computing means, as a differential of the other of said sinusoidal functions.

20. The method as claimed in claim 19, wherein the numerical approximation of said other sinusoidal function is delayed, with a delayer, by one interval between said digital values before being multiplied, with a multiplier, by said each digital value, and wherein the numerical approximation of said other sinusoidal function is further delayed, with a delayer, by another interval between said digital values so that said differential is computed as the difference in the numerical approximation of said other sinusoidal function over two interval s between said digital values.

21. The method as claimed in claim 17, wherein said numerical approximation of one of said sinusoidal functions is a Chebychev-derived approximation.

22. The method as claimed in claim 17, wherein said numerical approximation of one of said sinusoidal functions is computed by comparing, with a comparator, said each phase angle to predefined angular limits, transforming, with transforming means, said each phase angle by a linear transformation to produce a transformed phase angle in response to the comparing of said phase angle to the predefined angular limits, and computing, with computing means, said numerical approximation of said one of said sinusoidal functions as either an even function or an odd function of said transformed phase angle in response to the comparing of said each phase angle to tile predefined angular limits.

23. The method as claimed in claim 22, wherein said even function is a polynomial of said transformed phase angle, said odd function is also a polynomial of said transformed phase angle, and said even function is used for said numerical approximation of said one of said sinusoidal functions over a greater angular range of said each phase angle than said odd function.

24. A system for demodulating an analog signal by a reference frequency signal, and producing a digitized demodulated signal synchronized to a high-speed system clock, said system comprising, in combination:

an analog-to-digital converter for producing digital samples of said analog signal at a predetermined sampling frequency synchronized to said system clock;

a counter clocked by said system clock for counting during at least one cycle of said system clock;

means including a set of latches connected to said counter and responsive to said reference frequency signal for determining a number of counts of said counter in a period of said reference frequency, and determining a number of counts into a period of said reference frequency when said analog signal is sampled by said analog-to-digital converter;

means for computing a phase angle value for each of said digital samples from said number of counts of said counter in a period of said reference frequency, and said number of counts into a period of said reference frequency when said analog signal is sampled by said analog-to-digital converter;

means for computing a numerical approximation of a pair of sinusoidal functions of each phase angle value, said pair of sinusoidal functions being in phase quadrature with each other;

means for multiplying each digital sample by the numerical approximation of each sinusoidal function of the phase angle for said each digital sample to produce a series of products of each of said sinusoidal functions;

a first digital filter for digitally filtering the series of products of a first one of said sinusoidal functions; and

a second digital filter for digitally filtering the series of products of a second one of said sinusoidal functions.

25. The system as claimed in claim 24, wherein said means for computing a numerical approximation of a pair of sinusoidal functions of each phase angle value includes means for computing the numerical approximation of one of said sinusoidal functions as a differential of the other of said sinusoidal functions.

26. The system as claimed in claim 25, wherein said means for computing the numerical approximation of one of said sinusoidal functions as a differential of the other of said sinusoidal functions includes means for delaying said other sinusoidal function by one period of said sampling frequency before being multiplied by said means for multiplying, and means for delaying the numerical approximation of said other sinusoidal function by a second period of said sampling frequency so that said differential is computed as the difference in the numerical approximation of said other sinusoidal function between two periods of said sampling frequency.

27. The system as claimed in claim 25, wherein said means for computing a numerical approximation of a pair of sinusoidal functions of each phase angle value includes means four computing the numerical approximation of said other of said sinusoidal functions by comparing said each phase angle to predefined angular limits, transforming said each phase angle by a linear transformation to produce a transformed phase angle in response to the comparing of said phase angle to the predefined angular limits, and computing said numerical approximation of said one of said sinusoidal functions as either an even function or an odd function of said transformed phase angle in response to the comparing of said each phase angle to the predefined angular limits.

28. The system as claimed in claim 27, wherein said even function is a polynomial of said transformed phase angle, said odd function is also a polynomial of said transformed phase angle, and said even function is used for said numerical approximation of said other of said sinusoidal functions over a greater angular range of said each phase angle than said odd function.