Digital filter circuit

Claim 2

In a recursive digital filter including: (a) a forward path including an input terminal for receiving an input signal X(n) and an output terminal for supplying an output signal Y(n); (b) a first feedback loop including a first delay element and a first digital multiplier circuit, said first feedback loop being arranged to form a first signal given by the product -B 1 D 1 (n), wherein -B 1 ≡ the multiplier coefficient of said first digital multiplier circuit, D 1 (n) ≡ Y(n-1), n≡ the time at which a variable is sampled, (c) a second feedback loop including a second delay element and a second digital multiplier circuit, said first and second feedback loops together being arranged to form a second signal given by the product -B 2 D 2 (n), wherein -B 2 ≡ the multiplier coefficient of said second digital multiplier circuit, and D 2 (n)≡ Y(n-2)≡ D 1 (n-1), (d) first means for summing said first and second signals and said input signal to form said output signal, said output signal being given by Y(n)= X(n)- B 1 D 1 (n)-B 2 D 2 (n), and (e) second means for randomly switching between rounding and truncating of at least one of said products, the improvement comprising: (f) third means operable to control said random switching means, said third means including (1) fourth means for determining the polarity relationship between D 1 (n) and D 2 (n) and the polarity of B 1 , (2) fifth means for determining the magnitude relationship between D 1 (n) and D 2 (n), and (3) sixth means jointly responsive to said fourth and fifth means for selectively inhibiting said second means from truncating said at least one product.

Claim 3

The invention defined in claim 1 wherein (a) said second means effects switching of said -B 2 D 2 (n) product, and (b) said sixth means inhibits truncation if B 1 > 0, sign D 1 (n)≠ sign D 2 (n), and | D 1 (n)| > | D 2 (n)| .

Claim 4

The invention defined in claim 1 wherein (a) said second means effects switching of said -B 2 D 2 (n) product, and (b) said sixth means inhibits truncation if B 1 < 0, sign D 1 (n)= sign D 2 (n), and |D 1 (n)| > | D 2 (n)| .

Claim 5

The invention defined in claim 1 wherein (a) said second means effects switching of said -B 1 D 1 (n) product, and (b) said sixth means inhibits truncation if B 1 < 0, sign D 1 (n)= sign D 2 (n) and | D 2 (n)| > | D 1 (n)| .

Claim 6

The invention defined in claim 1 wherein (a) said second means effects switching of said -B 1 D 1 (n) product, and (b) said sixth means inhibits truncation if B 1 > 0, sign D 1 (n). sign D 2 (n), and | D 2 (n)| > | D 1 (n)|.

Claim 7

The invention defined in claim 1 wherein (a) said second means effects switching of the sum of said -B 1 D 1 (n) and -B 2 D 2 (n) products, and (b) said sixth means inhibits truncation if B 1 < 0, sign D 1 (n)= sign D 2 (n), and | D 2 (n)| > | D 1 (n)| .

Claim 8

The invention defined in claim 1 wherein (a) said second means effects switching of the sum of said -B 1 D 1 (n) and -B 2 D 2 (n) products, and (b) said sixth means inhibits truncation if B 1 > 0, sign D 1 (n)≠ sign D 2 (n), and | D 2 (n)| > | D 1 (n)|.

Claim 9

The invention defined in claim 1 further including: (a) seventh means for comparing the magnitudes of D 1 (n) and D 2 (n) for two consecutive values of n, and (b) eighth means for reducing the magnitude of Y(n) by one if said consecutive magnitudes of D 1 (n) and D 2 (n) are the same.

Claim 10

The invention defined in claim 8 wherein said eighth means includes an adder circuit disposed in said forward path, and said filter further includes ninth means arranged to selectively apply the number minus one (-1) to said adder circuit if Y(n)> 0 and to apply the number plus one (+1) to said adder circuit if Y(n)< 0.

Claim 11

In a second order recursive digital filter including (a) an input terminal for receiving an input signal X(n), (b) an output terminal for providing an output signal Y(n), (c) and at least first and second digital multipliers having multiplier coefficients -B 1 and -B 2 , respectively, said output signal being given by Y(n)= X(n)- B.sub.1 D.sub.1 (n)- B.sub.2 D.sub.2 (n) wherein n is the time at which a variable is sampled, D 1 (n) is a version of Y(n) removed in time by one sample interval, and D 2 (n) is a version of Y(n) removed in time by two sample intervals, apparatus for eliminating limit cycles in said output signal, comprising: (1) first means for randomly truncating the output of at least one of said multipliers, and (2) second means for inhibiting said random truncating means as a function of the relative values of D 1 (n) and D 2 (n).

Claim 12

In a second order recursive digital filter including (a) an input terminal for receiving an input signal X(n), (b) an output terminal for providing an output signal Y(n), (c) and at least first and second digital multipliers having multiplier coefficients -B 1 and -B 2 , respectively, said output signal being given by Y(n)= X(n)- B.sub.1 D.sub.1 (n)- B.sub.2 D.sub.2 (n) wherein n is the time at which a variable is sampled, D 1 (n) is a version of Y(n) removed in time by one sample interval, and D 2 (n) is a version of Y(n) removed in time by two sample intervals, apparatus for eliminating limit cycles in said output signal, comprising: (1) first means for randomly truncating the output of at least one of said multipliers, and (2) second means for inhibiting said random truncating means as a function of the relative values of D 1 (n) and D 2 (n).

Claim 13

The invention defined in claim 10 wherein said first and second digital multipliers are arranged to generate products in sign-magnitude format.

Claim 14

The invention defined in claim 11 wherein truncation of said output of said second digital multiplier is controlled by said first means.

Claim 15

The invention defined in claim 12 wherein said second means is arranged to inhibit truncation if | D 1 (n)| > | D 2 (n)| .

Claim 16

The invention defined in claim 12 further including third means for inhibiting said random truncation as a function of the polarities of D 1 (n), D 2 (n) and B 1 .

Claim 17

The invention defined in claim 14 wherein said third means is arranged to inhibit truncation if sign (B 1 D 1 (n))≠ sign (B 2 D 2 (n)).

Claim 18

The invention defined in claim 11 wherein truncation of said output of said first digital multiplier is controlled by said first means.

Claim 19

The invention defined in claim 14 wherein said second means is arranged to inhibit truncation if | D 2 (n)| > | D 1 (n)|.

Claim 20

The invention defined in claim 16 further including third means for inhibiting said random truncation as a function of the polarities of D 1 (n), D 2 (n) and B 1 .

Claim 21

The invention defined in claim 18 wherein said third means is arranged to inhibit truncation if sign (B 1 D 1 (n))≠ sign (B 2 D 2 (n)).

Claim 22

The invention defined in claim 10 wherein said first and second digital multipliers are arranged to generate products in two's-complement format.

Claim 23

The invention defined in claim 20 wherein said second means is arranged to inhibit truncation if B 1 < 0, and D 1 (n)< D 2 (n).

Claim 24

The invention defined in claim 20 wherein said second means is arranged to inhibit truncation if B 1 > 0, and -D 1 (n)< D 2 (n).

Claim 25

The invention defined in claim 10 further including: third means for generating a control signal upon the occurrence of three consecutive equal values of D 1 (n); and fourth means responsive to said control signal for reducing the magnitude of Y(n) by one.

Claim 26

In a recursive digital filter including (a) first means arranged to generate a first digital product - B 1 D 1 (n) by multiplying a delayed version D 1 (n) of the output Y(n) of said filter by a multiplier coefficient - B 1 , said first product being represented by a variable number of digital bits, (b) second means arranged to generate a second digital product - B 2 D 2 (n) by multiplying a delayed version D 2 (n) of said D 1 (n) version by a multiplier coefficient - B 2 , said second product being represented by a variable number of digital bits, (c) third means for summing said first and second digital products and an input signal X(n) to said filter to form said filter output Y(n), (d) fourth means for quantizing said first and second products to a desired bit length by retaining only the most significant portions of said products and discarding the least significant portions of said products, (e) fifth means for generating a control signal, (f) sixth means for rounding the least significant bits of said most significant portions in accordance with the most significant bits of said least significant portions, in the presence of said control signal and truncating said least significant bits of said most significant portions in the absence of said control signal, and (g) seventh means for randomly inhibiting said control signal, the improvement comprising circuitry for insuring that Y(n) rapidly approaches zero when X(n)= 0, D 1 (n)≠ 0 and D 2 (n)≠ 0, said circuitry comprising (h) eighth means for generating a second control signal as a function of the values of D 1 (n) and D 2 (n), and (i) ninth means for disabling said seventh means in response to said second control signal.

Claim 27

In a recursive digital filter including (a) first means arranged to generate a first digital product - B 1 D 1 (n) by multiplying a delayed version D 1 (n) of the output Y(n) of said filter by a multiplier coefficient - B 1 , said first product being represented by a variable number of digital bits, (b) second means arranged to generate a second digital product - B 2 D 2 (n) by multiplying a delayed version D 2 (n) of said D 1 (n) version by a multiplier coefficient - B 2 , said second product being represented by a variable number of digital bits, (c) third means for summing said first and second digital products and an input signal X(n) to said filter to form said filter output Y(n), (d) fourth means for quantizing said first and second products to a desired bit length by retaining only the most significant portions of said products and discarding the least significant portions of said products, (e) fifth means for generating a control signal, (f) sixth means for rounding the least significant bits of said most significant portions in accordance with the most significant bits of said least significant portions, in the presence of said control signal and truncating said least significant bits of said most significant portions in the absence of said control signal, and (g) seventh means for randomly inhibiting said control signal, the improvement comprising circuitry for insuring that Y(n) rapidly approaches zero when X(n)= 0, D 1 (n)≠ 0 and D 2 (n)≠ 0, said circuitry comprising (h) eighth means for generating a second control signal as a function of the values of D 1 (n) and D 2 (n), and (i) ninth means for disabling said seventh means in response to said second control signal.

Claim 28

The invention defined in claim 24 wherein said eighth means is arranged to generate said second control signal if sign (B 1 D 1 (n))= sign (B 2 D 2 (n)) or | D 1 (n)| > | D 2 (n)|.

Claim 29

The invention defined in claim 24 further including tenth means for reducing the value of Y(n) by one if D 1 (n)= D 2 (n)= Y(n).

Claim 30

In a second order recursive digital filter including (a) an input terminal for receiving an input signal X(n), (b) an output terminal for providing an output signal Y(n), (c) and at least first and second digital multipliers having multiplier coefficients -B.sub. 1 and -B 2 , respectively, said output signal being given by Y(n)= X(n)- B 1 D 1 (n)- B 2 D 2 (n) wherein n is the time at which a variable is sampled, D 1 (n) is a version of Y(n) removed in time by one sample interval, and D 2 (n) is a version of Y(n) removed in time by two sample intervals, apparatus for eliminating limit cycles in said output signal, comprising: (1) first means in said first and second multipliers for rounding the outputs thereof, (2) second means for changing the magnitude of Y(n) by one, as a function of the relative magnitude of D 1 (n) and D 2 (n), and (3) third means for randomly operating said second means.

Claim 31

In a second order recursive digital filter including (a) an input terminal for receiving an input signal X(n), (b) an output terminal for providing an output signal Y(n), (c) and at least first and second digital multipliers having multiplier coefficients -B.sub. 1 and -B 2 , respectively, said output signal being given by Y(n)= X(n)- B 1 D 1 (n)- B 2 D 2 (n) wherein n is the time at which a variable is sampled, D 1 (n) is a version of Y(n) removed in time by one sample interval, and D 2 (n) is a version of Y(n) removed in time by two sample intervals, apparatus for eliminating limit cycles in said output signal, comprising: (1) first means in said first and second multipliers for rounding the outputs thereof, (2) second means for changing the magnitude of Y(n) by one, as a function of the relative magnitude of D 1 (n) and D 2 (n), and (3) third means for randomly operating said second means.

Claim 32

The invention defined in claim 27 wherein said second means is arranged to reduce the magnitude of Y(n) by one if | D 1 (n)| > | D 2 (n)| .

Claim 33

The invention defined in claim 27 wherein said second means is arranged to increase the magnitude of Y(n) by one if | D 2 (n)| > | D 1 (n)|.