PH12013000153A1 - Linear signal reconstruction system and method - Google Patents

Abstract

Linear signal reconstruction systems and methods that use mathematical relationships that exist between discrete time signals, digital to analog conversion characteristics and digital signal processing to produce a highly accurate, low noise, arbitrary analog signal from a discrete digital representation. This analog signal is produced by connecting discrete digital voltages through the use of a segmented straight line curve fit. This approach significantly reduces out of band harmonics that are normally associated with the stair-step approach while improving the output signal amplitude and phase accuracy. More specifically, the present system and method provides for reconstructing original analog signals from a digitized representation thereof. Digitized signals corresponding to the original analog signals are differentiated and then D to A converted into differentiated analog signals. The differentiated analog signals are then integrated to provide for reconstructed analog signals that correspond to the original analog signals.

Description

oT To transport 11. The transport 11 includes a digital filter 13, such as a digital brick wall oy filter, for example, a digital to analog converter 15, a current to voltage converter 17 vl and a synchronizing clock that synchronizes the digital processing. The present system = includes, in addition to the above components, a digital signal processor 14 - interposed between the digital filter 13 and the digital to analog converter 15 and an : analog reconstruction circuit 18 that processes the output from the current to voltage . converter 17. B

FIG. 2 shows a more detailed block diagram of the digital signal processor 14 " employed in the system 10 of FIG. I. The digital output from the digital filter 13 is Tr applied to a serial to parallel data converter 21. Outputs from the serial to parallel data @ converter 21 are applied in parallel to first and second differentiation circuits 22, 23. 1

Many commercial digital to analog converters use a serial data input method to reduce implementation costs and pin count. However, some digital to analog converters employ parallel data inputs. Therefore, the serial to parallel data converter 21 and the parallel to serial data converter 28 are optional. The first and second differentiation circuits 22, 23 process the left and right channel of the audio signals extracted from the recorded audio signals on the compact disk. Each differentiation circuit 22, 23 is comprised of first and second data latches 24, 25, a digital differentiator 26 that is ] comprised of a subtraction circuit 29, and an output data latch 27. The output of the first data latch 24 is coupled to the second data latch 25 and the respective outputs of the first and second data latches 24, 25 are coupled to the subtraction circuit 29. The digital differentiator circuit 26 subtracts the first and second outputs of the first and second data latches to provide a differentiated output signal. The outputs of the data ] latches 27 of the respective differentiation circuits 22, 23 are coupled to a parallel to serial data converter 28 whose output is coupled to the digital to analog converter 15.

FIG. 3 shows a more detailed block diagram of the analog reconstruction circuit 18 employed in the system 10 of FIG. 1. The analog reconstruction circuit 18 is comprised of a linear reconstructor circuit 37 that comprises an integrator, and a bass compensation circuit 38 that comprises a bass boost circuit. FIG. 4a shows a detailed block diagram of the linear reconstruction circuit 37 of FIG. 3. The linear reconstruction circuit 37 is a simple integration circuit arrangement whose bandpass is = shown in FIG. 4c. FIG. 4b shows a detailed block diagram of the bass compensation o circuit 38 of FIG. 3. The bass compensation circuit 38 is adapted to improve the bass pt response by improving the low frequency response that is limited by the "real world" o integrator (shown by the dashed line in FIG. 4c). “

It should be clear from FIGS. 1-4 that the essence of the system 10 and signal fo processing method of the present invention is that the digitized analog signals extracted = from the compact disk are differentiated, converted to analog signals, and then integrated to reconstruct the analog signals. The results of this processing is to smooth Bn out the step output normally provided by compact disk digital processors, which nN provides for a more realistic and improved representation of the original recorded audio . signal.

In order to better understand the system 10 and method of the present invention, the theory of operation thereof is presented below. The linear signal reconstruction system and method uses mathematical relationships that exist between discrete time signals, ] digital to analog conversion characteristics and digital signal processing to produce a highly accurate, low noise, arbitrary analog signal from a discrete digital representation.

This analog signal is not produced form a stair-step approximation, which is performed by current systems and processing methods, but by connecting discrete digital voltages through the use of a segmented straight line curve fit. This approach significantly reduces out of band harmonics that are normally associated with the stair-step approach : while improving the output signal amplitude and phase accuracy.

A general mathematical proof and analysis of the present invention is presented below.

Referring to FIGS. Sa and Sb, which show an original digital signal and a discretely differentiated signal, respectively, linearly reconstructing an analog signal from its digital representation requires that the digital signal be discretely differentiated.

Let S(n.DELTA.t) be an arbitrary digital signal, where .DELTA.S(n.DELTA.t)=S[n- 1].DELTA.t), for 0.ltoreq.n.Itoreq.N, where N is the last point in digital signal representation, and S(-n)=0. An examination of FIGS. 5a and 5b reveals that after - differentiating the signal S (n.DELTA.1), the resultant signal .DELTA.S (n.DELTA.{) 3 has the same number of points but is displaced in time. The first point, represented by ” .DELTA.S (-.DELTA.1), establishes the initial conditions at time t=0. Furthermore, it = will be shown later that the last point of the analog signal is reconstructed in the analog o domain using point .DELTA.S([N-11.DELTA.1). -

Referring to FIG. 6, it shows a digital to analog converted signal derived from the - discretely differentiated signal of FIG. 5b, and at this point, signal S(n.DELTA.t) has y been differentiated and is ready to be applied to the digital to analog converter 15. The ber output of the digital to analog converter 15 is a stair-step representation of the or differentiated signal. In order to recover the original signal, the output of the digital to o analog converter 15 must be integrated. This effectively creates a mathematical Hi multiplication of the original signal by one. ##EQU I ##

By evaluating the equation for F(t), over values oft ranging from 0 to N.DELTA.t, it can be shown that F(t) equals S(n.DELTA.t) for values of t equal to n.DELTA.t.

Mathematically this means that F(0)=S(0), F(DELTA.t)=S(.DELTA.t)...

F(N.DELTA.t)=S(N.DELTA.t). With reference to FIGS. 7a and 7b, which show the linear reconstruction of the digital waveform and a stair step approximation of the same waveform, respectively, the function F(t) is a linear reconstruction of the digital : waveform and represents a better approximation of the true analog waveform when compared against the traditional stair-step approximation.

The following is a mathematical analysis of linearly reconstructed waveforms versus stair-step approximations. Linearly reconstructed waveforms differ significantly from traditional stair-step approximations. The major differences lie in the areas of image rejection (at multiples of the sampling frequency), voltage accuracy, and phase accuracy. To analyze these issues, an understanding of how the point-to-point approximation schemes implemented by the two techniques differ from the original signal.

IER

For purposes of comparison, it will be assumed that the original signal is a sinusoidal in the form of: v(t)=sin(.omega.t). It is necessary to examine the original signal segment . by segment in order to outline the approximation errors introduced by the stair-case and o linear approximation techniques. If it is assumed that v(t) is sampled at a sampling ~ frequency f.sub.s (DELTA. t=1/f.sub.s =sampling period) and N is the number of points & sampled, the original signal may be rewritten as a summation of point to point . segments: ##EQU2## »

The term u(t) is a STEP function. The term u(t-t.sub.1)-u(t-t.sub.2) is a pulse signal bod where the pulse ranges from t.sub.1 to t.sub.2. The summation equation describing the - original signal is multiplied by the pulse function [u(t-n.DELTA.t)-u(t(n+1).DELTA.t)] = so that the time argument used is valid from point n to point n+1 of the summation, this on forces the time argument to be in the range - [n.DELTA.t].ltoreq.t.Itoreq.[(n+1).DELTA.t] from segment to segment. This operation is used repeatedly during the following description.

With reference to FIG. 8, which illustrates the stair-case approximation technique, it is a traditional digital to analog conversion technique. FIG. 9 shows the unfiltered error function derived from the stair-case approximation. The error function contains the fundamental frequency at a 90 degree phase shift from the original signal. This phase shifted signal causes the reconstructed signal to be reduced in amplitude and shifted in phase (vector summation mathematics). It uses a digital representation of the original signal sampled at frequency f.sub.s. Each digital value of the represented signal is input to the digital to analog converter and held at that value until the next digital value is selected. #EQU3## The term s(n,t) has a constant value from time n.DELTA.t to time (n+1).DELTA.t. It is also multiplied by a pulse function so that the time argument is valid from segment to segment.

The approximation error associated with the stair-case approximation technique is derived by simply subtracting the stair-case approximation equation from the original signal equation. Let e.sub.1 (n,t) be the stair-case approximation error function given by

HHEQUA4H# : !

i

An equation describing the phase shifted fundamental may be derived in order to - subtract it out from the error function. The equation is in the form k.sub.1 cos(.omega.t- & phi.) where k.sub.1 is the peak amplitude and .phi. is the phase shift of the phase i shifted fundamental signal. It is known that the peak value of the fundamental occurs at - the first sampled point (at time t=.DELTA.t), it is also one half the peak amplitude of - “ the original signal. The phase shift must be evenly distributed over all segments, h therefore, the phase shifted fundamental signal may be represented as f(t)=k.sub.1 - cos(.omega.t-.pi./N), where, k.sub.1 =sin(.omega.t)/2 and is evaluated at t=.DELTA.t o (.DELTA.t=1/f.sub.s) [i.e., k.sub.1 =sin(.omega.t)/2]. or

It is necessary to subtract the phase shifted fundamental signal from the error function.

It is thus necessary to describe this signal as a summation of point to point segments, h given by #4EQUSH#

The filtered stair-case approximation error function is the error function (e.sub.1 (n,t)) minus the fundamental and phase shift function (f(n,t)). The filtered stair-case approximation error function is thus e.sub.1f (n,t) where e.sub.1f (n,t)=e.sub.1 (n,t)- f(n,t), or #HEQUO6##

FIG. 10 shows the stair-case approximation filtered error function. The fundamental and phase shift are no longer present. The resultant signal is a double-sideband suppressed carrier signal.

A spectral analysis of the stair-case approximation is as follows. Fast Fourier transform : (FFT) computations were performed on the filtered error function of the stair-case ] approximation technique using various values of N. The results show that the sampling images are replicated at multiples of the sampling frequencies. The magnitude of these images vary as a function of the number of points used to generate the desired baseband signal. That is, the magnitude of the out-of-band images increase as a function of i decreasing point density. 19 1

.

An analysis of the phase error of the stair-case approximation is discussed below. £

Another disadvantage with the stair-case approximation technique is that it introduces a - phase lag into the reproduced signal. This phase lag occurs due to the sample-and-hold = scheme that is employed. FIG. 11 shows how the stair-case approximation lags the - original signal. The amount of phase lag introduced by the stair-case approximation is oo proportional to the number of points used to generate the desired signal. Larger point o densities generate smaller phase lag errors. A derivation of the phase lag error follows:

The phase shifted fundamental is given by f(t)=k.sub.1 cos(.omega.t-.pi./N), and the Fo filtered original signal is given by s(t)=sin(.omega.t)-k.sub. 1 cos(.omega.t-.pi./N), where 5

SO, -

By definition, and therefore, ] and 50,

Using the substitution -sin(a+b)=cos(b)sin(a)+sin(b)cos(a) results in

Therefore, s(t)=Asin.omega.t-Bcos.omega.t, where

The phase (.differential.) is therefore differential. =arctan(-B/A).

Plots of the phase lag associated with the stair-case approximation technique with ] respect to point density (N) are shown in FIGS. 12a and 12b.

j .

A fundamental is present in the stair-case approximation technique. Mathematically incorporating this fundamental into a base band signal (subtracting it) results in a signal = whose amplitude was less than that of the original signal. This is an undesired - amplitude error. -

The derivation of the phase error associated with the stair-case approximation method ~ showed that the system output signal could be characterized as follows: 2

Therefore, s(t)=Asin.omega.t-Bcos.omega.t, where, : or "

The magnitude of s(t) may be computed as follows: LE

The amplitude of the original signal is 1 and the amplitude of the system output signal is C. Therefore the amplitude error is (1-C). FIG. 13 shows the percentage amplitude error [(1-C).multidot.100] as a function of point density (N). It can be seen that the amplitude error decreases as the point density increases.

The total harmonic distortion may be calculated by taking the integral of the error ~~ function squared. The total harmonic distortion for the stair-case approximation technique is as follows: ##EQU7## Let a=.omega.n.DELTA.t, j b=.omega.n.DELTA.t+.omega..DELTA.t (or .omega..DELTA. .t(n+1)), and 2.pi.f.sub.c i =.omega., and 1/f.sub.s =. DELTA.t.

Substituting variables, ##EQUS8## Squaring, ##EQU9## Integrating, ##EQU 1 0##

The total harmonic distortion (THD) for the stair-case approximation technique as a function of point density is shown in FIGS. 14a and 14b. The theoretical THD for a stair-case approximation has been calculated to be approximately 100.multidot..pi./(.sqroot.3.multidot.N). The results obtained from our analysis closely J matched this prediction.

.

This equation may be rewritten as a summation of a sampled signal at sampling i frequency f.sub.s. The equation for the fundamental is f(n,t) and it is ##EQU13## -

Thus, the filtered linear approximation error function is the error function (e.sub.2 (t)) o minus the fundamental function (f(t)). The filtered linear approximation error function i is thus e.sub.2f (n,t) where ##EQU 1 4## .

FIG. 17 shows the linear approximation filtered error function. The fundamental is no wi longer present, The resultant signal is a double-sideband suppressed carrier signal. Fo

Fast Fourier transform (FFT) computations were performed on the filtered error = function of the linear approximation technique using various values of N. The sampling os : images generated using this method are substantially smaller than those generated using - the stair-case approximation method.

The linear approximation phase error is as follows. The equation describing the fundamental was calculated to be f(t)=k.sub.2 sin(.omega.t). The fundamental has the same frequency as the original signal and it does not contain a phase lag with reference to the original signal. The fundamental's only contribution to the error function is in amplitude (k.sub.2). This is an improvement over the stair-case approximation technique which introduced an unwanted 90 degree phase shift component. ]

As was discussed previously, there exists a fundamental that is 180 degrees out of phase in the linear approximation method. Mathematically incorporating this fundamental into the base band signal (subtracting it) results in a signal whose amplitude was less than that of the original signal. This is an undesired amplitude error. ‘

The system output signal may be described with the following equation or,

The amplitude of the original signal is 1, the amplitude of the system output signal is

EEE

(1-k.sub.2). Therefore, the amplitude error is 1-(1-k.sub.2) or simply just k.sub.2. FIG. fo 18 shows the percent amplitude error (k.sub.2 .multidot. 100) as a function of point “ density (N). It can be seen that the amplitude error decreases as the point density @ © increases. .

The total harmonic distortion may be calculated by taking the integral of the error - function squared. The total harmonic distortion for the linear approximation technique - is as follows: #¥EQUI S## 5

Let a=.omega.n.DELTA.t, and b=.omega.n. DELTA .t+.omega..DELTA.t (or .omega..DELTA.t(n+1) and 2.pi.f.sub.c =.omega., and 1/f.sub.s =.DELTA.t oh

Substituting variables, ##EQU16## Simplifying, #EQU 1 7## Squaring, #4#EQU 18##

Integrating, #4EQU | 9##

The total harmonic distortion for the linear approximation technique as a function of point density (N) is shown in EIGS. 18a and 18b. Comparing the total harmonic distortion content of this method with the stair-case approximation method shows that this method is substantially better.

In conclusion, the above mathematical analysis has shown that the linear reconstruction 1 system 10 and method of the present invention offers substantial improvements over the traditional stair-case approximation technique in several important areas. These i areas include alias rejection, group delay, amplitude accuracy, and total harmonic distortion. For the sake of clarity, each area is summarized individually. Spectral analysis of both techniques reveals that the linear reconstruction technique of the present invention provides better alias rejection. Test data is summarized in Tables 1 { through 8. It was also shown that the stair-case approximation technique introduces an f unwanted phase lag error (group delay). This phase lag error is a function of point density and the relationship shown in FIGS. 12a and 12b. The linear reconstruction technique of the present invention does not introduce this error. 1 24 1

Both techniques have an amplitude error associated with them that results from the Ty presence of an out of phase fundamental signal resulting from the sampling process. -

This amplitude error is summarized in FIGS. 13 and 18a. As can be seen, the linear = reconstruction technique has an amplitude error which is roughly 50% less than that of ~ the stair-case approximation technique. The total harmonic distortion (THD) was also - substantially less using the linear reconstruction technique. FIG. 19 shows the ratio of . the THD of the stair-case approximation versus the THD of the linear reconstruction » technique. As can be seen, with higher point densities, the linear reconstruction wl technique of the present invention offers far lower THD. Theoretically, the linear . reconstruction technique offers substantial improvements in many key areas when ” compared against the stair-case approximation technique. It also offers superior = performance over all the present conventional approaches. : %

TABLE 1 Point Sensity Per Cycle: 1764 General Actual Piecewise Linear Frequency Frequency Approx. (Ideal) Recon. (Ideal) fc =1fs/N 11,337 Hz +15 dBm +15 dBm fs +/- fc 20 MHz +/- fc -52 dBm -118 dBm 2fs +/- fc 40 MHz +/- fc -59 dBm -130 dBm 3fs +/- fc 60 MHz +/- fc -62 dBm -135 dBm 4fs +/- fc 80 MHz +/- fc -65 dBm -138 dBm

TABLE 2 Point Density Per Cycle: 882

General Actual Piecewise Linear Frequency Frequency Approx. (Ideal) Recon. (Ideal) fo = f5/N 22,675 Hz +.dBm +15 dBm fs +/- fc 20 MHz +/- fc -47 dBm -106 dBm 2fs +/- fc 40 MHz +/- fc -53 dBm -118 dBm 3fs +/- fc 60 MHz +/- fc -55 dBm -125 dBm 4fs +/- fc 80 MHz +/- fc -58 dBm - 128 dBm

TABLE 3 Point Density Per Cycle: 196

General Actual Piecewise Linear Frequency Frequency Approx. (Ideal) Recon. (Ideal) fo = fs/N 107,040 Hz +15 dBm +15 - dBm fs +/- fc 20 MHz +/- fc -34 dBm -80 dBm 2fs +/- fc 40 MHz +/- fc -40 dBm -90 dBm 3fs +/- fc 60 MHz +/- fc -43 dBm -95 dBm 4fs +/- fc 80 MHz +/- fc -46 dBm -98 dBm o

TABLE 4 Point Density Per Cycle: 98 o

General Actual Piecewise Linear Frequency Frequency Approx (Ideal) Recon. (Ideal) ° - fc = fs/N 204,081 Hz +15 dBm +15 o dBm fs +/- fc 20 MHz +/- fc -28 dBm -68 dBm 2fs +/- fc 40 MHz +/- fc -34 dBm -79 - dBm 3fs +/- fc 60 MHz +/- fc -37 dBm -85 dBm 4fs +/- fc 80 MHz +/- fc -40 dBm -98 -

TABLE 5 Point Density Per Cycle: 49 ro

General Actual Piecewise Linear Frequency Frequency Approx. (Ideal) Recon. (Ideal) io fo = f5/N 408,163 Hz +15 dBm +15 a dBm fs #/- fc 10 MHz +/- fc -21 dBm -55 dBm 2fs +/- fc 40 MHz +/- fc -28 dBm -68 oo dBm 3fs +/- fc 60 MHz +/- fc -31 dBm -75 dBm 4fs +/- fc 80 MHz +/- fc -33 dBm -78 dBm

TABLE 6 Point Density Per Cycle: 18

General Actual Piecewise Linear Frequency Frequency Approx. (Ideal) Recon. (Ideal) fo = fs/N 1,111,111 Hz +15 dBm +15 dBm fs +/- fc 20 MHz +/- fc -13 dBm -37 dBm 2fs +/- fc 40 MHz +/- fc -19 dBm -49 dBm 3fs +/- fc 60 MHz +/- fc -22 dBm -55 dBm 4fs +/- fc 80 MHz +/- fc -25 dBm -58 dBm

TABLE.7 Point Density Per Cycle: 12

General Actual Piecewise Linear Frequency Frequency Approx. (Ideal) Recon. (Ideal) fc = fs/N 1,566,666 Hz +15 dBm +15 dBm fs +/- fc 20 MHz +/- fc -9 dBm -30 dBm 2fs +/- fc 40 MHz +/- fc -15 dBm -42 dBm 3fs +/- fc 60 MHz +/- fc -19 dBm -48 dBm 4fs +/- fc 80 MHz +/- fc -22 dBm -55 dBm

TABLE 8 Point Density Per Cycle: 9

General Actual Piecewise Linear Frequency Frequency Approx. (Ideal) Recon. (Ideal) ]

fc = fs/N 2,222,222 Hz +15 dBm +15 . dBm fs +/- fo 20 MHz +/- fc -6 dBm -25 dBm 2fs +/- fc 40 MHz +/- fc -13 dBm -37 io dBm 3fs +/- fc 60 MHz +/- fc -17 dBm -45 dBm 4fs +/- fc 80 MHz +/- fc -18 dBm -48 - dBm

Experimental results of tests performed on an embodiment of the present system 10 are co as follows. An arbitrary waveform generator was used to generate sampled sinusoidal = signals of varying point densities. [n order to establish a baseline, the spectral o characteristics of the piecewise approximation were examined using a spectrum - analyzer. An integrator circuit capable of generating point-to-point linear reconstruction - of the discretely differentiated signals in accordance with the present invention was oo built. The arbitrary waveform generator was used to generate a discretely differentiated sinusoid. The output of the arbitrary waveform generator was fed into the integrator 5 circuit where the signal was linearly reconstructed. The output of the integrator circuit was then connected to a spectrum analyzer to examine the spectral characteristics of the linearly reconstructed signal.

Photographs were taken to compare the actual piecewise approximation against the linear reconstruction technique, and FIGS. 20 and 21 are representative of these photographs. FIG. 20 shows several points of a piecewise approximation (stair-case approximation). It can be seen that the point to point transitions are not very smooth and look like an ascending stair case. FIG. 21 shows the same points connected using the linear reconstruction technique of the present invention. It should be clear that the present linear reconstruction technique is far superior. It generates a signal free of "stair steps” that more closely approximates the original signal.

The linear reconstruction technique of the present invention offers a substantial improvement in the spectral characteristics of the regenerated signal as compared to the stair-case piecewise approximation technique. Tables 9-16 tabulate the results of the ] tests performed on the stair-case approximation and linear approximation techniques.

The ideal results are those mathematically calculated from the filtered error function equations of the two techniques and converted to dBm. The measured results are those that were observed using the arbitrary waveform, integrator circuit, and spectrum 0 analyzer. The measured distortion of the stair-case approximation appears to be better . than the theoretical prediction. This is caused by the natural filtering (band-limiting) of the amplifiers used within the arbitrary waveform analyzer. Furthermore, the - instruments noise floor account for the additional noise measured for the linear reconstruction method. .

TABLE 9 oh

Point Density Per Cycle: 1764 Wavetek Piecewise Piecewise Linear Linear -

General 1375 Approx. Approx. Recon. Recon Frequency Frequency (Ideal) (Meas.) 0 (Ideal) (Meas.) fc =fs/N 11,337 Hz +15 dBm +15 dBm +15 dBm +15 dBm fs +/- fc 20 MHz +/- fc -52 dBm -61 dBm -118 dBm -98 dBm 2fs +/- fc 40 MHz +/- fc -59 dBm -74 dBm - 130 dBm -105 dBm 3fs +/- fc 60 MHz +/- fc -62 dBm -83 dBm -135 dBm -111 dBm 4fs +/- fc 80 MHz +/- fc -65 dBm -91 dBm -138 dBm -88 dBm :

TABLE 10

Point Density Per Cycle: 882 Wavetek Piecewise Piecewise Linear Linear

General 1375 Approx. Approx. Recon. Recon Frequency Frequency (Ideal) (Meas.) (Ideal) (Meas.) fc =fs/N 22,675 Hz +15 dBm +15 dBm +15 dBm +15 dBm fs +/- fc 20 MHz +/- fc -47 dBm -55 dBm -106 dBm -101 dBm 2fs +/- fc 40 MHz +/- fc -53 dBm -69 dBm - 3 118 dBm -101 dBm 3fs +/- fc 60 MHz +/- fc -55 dBm -78 dBm -125 dBm -118 dBm 4fs +/- fc 80 MHz +/- fc -58 dBm -85 dBm -128 dBm -85 dBm

TABLE 11 .

Point Density Per Cycle: 196 Wavetek Piecewise Piecewise Linear Linear >

General 1375 Approx. Approx. Recon. Recon Frequency Frequency (Ideal) (Meas.) wo (Ideal) (Meas.) 2 fc = fs/N 107,040 Hz +15 dBm +15 dBm +15 dBm +15 dBm fs +/- fc 20 MHz +/- = fc -34 dBm -40 dBm -80 dBm -110 dBm 2fs +/- fc 40 MHz +/- fc -40 dBm -55 dBm - fo 90 dBm -114 dBm 3fs +/- fc 60 MHz +/- fc -43 dBm -64 dBm -95 dBm -114 dBm 4fs +/- fc 80 MHz +/- fc -46 dBm -73 dBm -98 dBm -92 dBm =

TABLE 12

Point Density Per Cycle: 98 Wavetek Piecewise Piecewise Linear Linear General 1375 Approx. Approx. Recon. Recon Frequency Frequency (Ideal) (Meas.) (Ideal) (Meas.) : fc = fs/n 204,081 Hz +15 dBm +15 dBm +15 dBm +15 dBm fs +/- fc 20 MHz +/- fc -28 dBm -35 dBm -68 dBm -90 dBm 2fs +/- fc 40 MHz +/- fc -34 dBm -50 dBm -79 dBm -114 dBm 3fs +/- fc 60 MHz +/- fc -37 dBm -58 dBm -85 dBm -104 dBm 45 +/- fc 80 MHz +/- fc -40 dBm -67 dBm -98 dBm -90 dBm

TABLE 13

Point Density Per Cycle: 49 Wavetek Piecewise Piecewise Linear Linear General 1375 Approx. Approx. Recon. Recon Frequency Frequency (Ideal) (Meas.) (Ideal) : (Meas.) ]

fc = fs/N 408,163 Hz +15 dBm +15 dBm +15 dBm +15 dBm fs +/- fc 20 MHz +/- - fc -21 dBm -30 dBm -55 dBm -68 dBm 2fs +/- fc 40 MHz +/- fc -28 dBm -43 dBm -68 = dBm -100 dBm 3fs +/- fc 60 MHz +/- fc -31 dBm -52 dBm -75 dBm -100 dBm 4fs +/- ~ fc 80 MHz +/- fc -33 dBm -60 dBm -78 dBm -92 dBm o

TABLE 14 y

Point Density Per Cycle: 18 Wavetek Piecewise Piecewise Linear Linear General 7 1375 Approx. Approx. Recon. Recon Frequency Frequency (Ideal) (Meas.) (Ideal) o (Meas.) he fc=fs/N 1,111,111 Hz +15 dBm +15 dBm +15 dBm +15 dBm fs -+/- fc 20 MHz +/- fc -13 dBm -20 dBm -37 dBm -59 dBm 2fs +/- fc 40 MHz +/- fc -19 dBm -35 dBm -49 dBm -82 dBm 3fs +/- fc 60 MHz +/- fc -22 dBm -45 dBm -55 dBm -98 dBm 4fs +/- fc 80 MHz +/- fc -25 dBm -51 dBm -58 dBm -108 dBm

TABLE 15

Point Density Per Cycle: 12 Wavetek Piecewise Piecewise Linear Linear General 1375 Approx. Approx. Recon. Recon Frequency Frequency (Ideal) (Meas.) (Ideal) (Meas.) fc = fs/N 1,566,666 Hz +15 dBm +15 dBm +15 dBm +15 dBm fs +/- fc 20 MHz +/- fc -9 dBm -16 dBm -30 dBm -66 dBm 2fs +/- fc 40 MHz +/- fc -15 dBm -30 dBm - . 42 dBm -72 dBm 3fs +/- fc 60 MHz +/- fc -19 dBm -42 dBm -48 dBm -74 dBm 4fs +/- fc 80 MHz +/- fc -22 dBm -50 dBm -55 dBm -98 dBm

Claims (17)

1. ¥ . = y \ Claims ~ = ; i fp What is claimed is: hei } La

   I. A signal reconstruction system for reconstructing original analog signals from a po recorded digitized representation thereof, said system comprising: o fo a source of digitized signals that are a recorded version of corresponding original B analog signals; ps differentiation circuitry coupled to the source of digitized signals for differentiating the digitized signals to provide differentiated digitized signals; digital to analog converter circuitry coupled to the differentiation circuitry for converting the differentiated digitized signals into corresponding differentiated analog signals; and integration circuitry coupled to the digital to analog converter circuitry for integrating the differentiated analog signals to provide for reconstructed analog signals that correspond to the original analog signals.
2. 2. The signal reconstruction system of claim 1 wherein the differentiation circuitry comprises: j serial to parallel conversion circuitry for converting serial digitized signals into two sets of parallel digitized signals corresponding to left and right channels; first and second digital differentiation circuits coupled to outputs of the serial to parallel conversion circuitry that each comprise first and second serially coupled latches whose : respective outputs are coupled to a digital subtraction circuit that is adapted to subtract :

   signals derived from the latches and whose output is coupled to an output latch; and = parallel to serial conversion circuitry coupled to respective outputs of the output latches of the first and second digital differentiation circuits. Ol a :
3. 3. The signal reconstruction system of claim 2 wherein the integration circuitry i comprises: i a linear reconstruction circuit that is adapted to integrate the differentiated analog fe! signals; and a low frequency compensation circuit coupled to the linear reconstruction circuit that is + adapted to increase the low frequency response of the integration circuitry.
4. 4. The signal reconstruction system of claim 1 wherein the integration circuitry comprises: a linear reconstruction circuit that is adapted to integrate the differentiated analog signals; and a low frequency compensation circuit coupled to the linear reconstruction circuit that is adapted to increase the low frequency response of the integration circuitry. :
5. 5. The signal reconstruction system of claim | wherein the differentiation circuitry comprises: serial to parallel conversion circuitry for converting serial digitized signals into two sets of parallel digitized signals corresponding to left and right channels; and first and second digital differentiation circuits that each comprise first and second serially coupled latches whose respective outputs are coupled to a digital subtraction ; circuit that is adapted to subtract signals derived from the latches and whose output is

   . coupled to an output latch. -
6. 6. The signal reconstruction system of claim 5 wherein the integration circuitry a comprises: o a linear reconstruction circuit that is adapted to integrate the differentiated analog WH signals; and a low frequency compensation circuit coupled to the linear reconstruction circuit that is y adapted to increase the low frequency response of the integration circuitry. or a ho
7. 7. The signal reconstruction system of claim 1 wherein the source of digitized signals I comprises: means for converting recorded optical signals into corresponding digital electrical signals; and a digital filter for filtering the digital electrical signals.
8. : 8. The signal reconstruction system of claim 7 wherein the means for converting recorded optical signals into corresponding digital electrical signals comprises: an optical disk transport mechanism having prerecorded optical patterns that correspond to the original analog signal; and a decoder for converting the prerecorded optical patterns into the corresponding digital : electrical signals.
9. 9. The signal reconstruction system of claim 8 wherein the differentiation circuitry comprises: © serial to parallel conversion circuitry for converting serial digitized signals into two sets

   . of parallel digitized signals corresponding to left and right channels; po a! first and second digital differentiation circuits coupled to outputs of the serial to parallel i conversion circuitry that each comprise first and second serially coupled latches whose ~ respective outputs are coupled to a digital subtraction circuit that is adapted to subtract - signals derived from the latches and whose output is coupled to an output latch; and . parallel to serial conversion circuitry coupled to respective outputs of the output latches i of the first and second digital differentiation circuits. .
10. 10. The signal reconstruction system of claim 9 wherein the integration circuitry Ci Fut comprises: da a linear reconstruction circuit that is adapted to integrate the differentiated analog signals; and a low frequency compensation circuit coupled to the linear reconstruction circuit that is adapted to increase the low frequency response of the integration circuitry.
11. 11. The signal reconstruction system of claim 8 wherein the differentiation circuitry comprises: serial to parallel conversion circuitry for converting serial digitized signals into two sets of parallel digitized signals corresponding to left and right channels; and first and second digital differentiation circuits that each comprise first and second serially coupled latches whose respective outputs are coupled to a digital subtraction circuit that is adapted to subtract signals derived from the latches and whose output is coupled to an output latch. :
12. 12. The signal reconstruction system of claim 11 wherein the integration circuitry comprises: ;

    i BRR . a linear reconstruction circuit that is adapted to integrate the differentiated analog ci signals; and - a low frequency compensation circuit coupled to the linear reconstruction circuit that is - adapted to increase the low frequency response of the integration circuitry.
13. 13. A signal reconstruction method for reconstructing original analog signals from a . recorded digitized representation thereof, said method comprising the steps of: or . wh providing a source of digitized signals that are a recorded version of corresponding I analog signals; differentiating the digitized signals to provide differentiated digitized signals; converting the differentiated digitized signals into corresponding differentiated analog signals; and integrating the differentiated analog signals to provide for reconstructed analog signals that correspond to the original analog signals.
14. 14. The signal reconstruction method of claim 13 wherein the differentiating step: comprises the steps of? : converting serial digitized signals into two sets of parallel digitized signals; for each set of parallel digitized signals, sequentially latching successive ones of the parallel digitized signals; subtracting the respective sequentially latched signals to provide for an output signal comprising a differentiated signal,

    latching the differentiated signal to provide for latched differentiated signals; and “ combining the latched differentiated signals from each set of parallel digitized signals into a serial set of differentiated signals. 0
15. 15. The signal reconstruction method of claim 13 wherein the integrating step Ee comprises the steps of? pore wy wn integrating the differentiated analog signals to provide for integrated analog signals; - and boosting the low frequency response of the integrated analog signals to increase the low frequency response thereof.
16. 16. The signal reconstruction method of claim 13 wherein the source of digitized signals provides two sets of parallel digitized signals, and wherein the differentiating step comprises the steps of: for each set of parallel digitized signals, sequentially latching successive ones of the parallel digitized signals; subtracting the respective sequentially latched signals to provide for an output signal comprising a differentiated signal; latching the differentiated signal to provide for latched differentiated signals; and combining the latched differentiated signals from each set of parallel digitized signals into a serial set of differentiated signals.
17. 17. The signal reconstruction method of claim 16 wherein the integrating step comprises the steps of: 7 3