Note : Les descriptions sont présentées dans la langue officielle dans laquelle elles ont été soumises.
CA 02181691 2000-02-18
METHOD AND APPARATUS FOR SCRAMBLING
AND DESCRAMBLING OF AUDIO SIGNALS
BACKGROUND
This invention relates to techniques for low cost scrambling and descrambfing
of audio information signals. More particularly, this invention relates to a
lower cost
Hi Fi descrambler with an improved performance over the prior art.
The prior art in the art of audio scrambling and descrambling utilized various
frequency shifting techniques. The prior arts in audio descrambling suffer
from hiss
in the form of "white noise", and more importantly in band carrier "whistle
caused by
inter-modulation of the two carrier frequencies. The prior arts also use
expensive
circuitry such as band pass filters for mixer circuits, wide band 0 degree and
90
degree all pass networks and 0 degree and 90 degree circuits for varying the
carrier
frequencies with constant amplitude and the need for adjustments to balance
gain of
quadrature mixers for sideband elimination. In addition, since the mixers used
in the
prior art are generally not stable in time, their drift results in an audible
whistle as the
result of carrier leak through.
The prior art requires mixers that require a pure sine wave modulation,
therefore a truly analog multiplier is needed. Truly analog multipliers tend
to have
noise problems because of their circuit configuration that cause white thermal
or
shot noise components that degrade the signal to noise {SNR) of the audio
scrambling system.
Prior art systems having one or more of the identified problems include US
patents # 4,636,853 ('853), DYNAMIC AUDIO SCRAMBLING SYSTEM, by Forbes
issued on Jan. 13, 1987, #5,058,159, METHOD AND SYSTEM FOR SCRAMBLING
AND DESCRAMBLING AUDIO INFORMATION SIGNALS by Quan issued on
October 15, 1991 and # 5,159,631, AUDIO SCRAMBLING SYSTEM USING IN
BAND CARRIER, by Quan et al. issued on October 27, 1992 ('159).
1
CA 02181691 2000-02-18
A review of the prior art for a full understanding of the present invention
will be
helpful. Turning now to the drawings, Fig 1 is a block diagram of the key
elements of the
Fortes '853 prior art. The Fortes '853 descrambler 10 has a scrambted audio
input 34
which is connected to an all pass phase shifter 20 containing a 0 deg. output
38 and a
90 deg. output 39. The scrambled audio signal has an offset frequency 36 f~-
f2 as
shown in Fig. 2a. This shows the scrambled audio offset by an offset frequency
determined by the scrambling process. The phase shifted outputs are connected
to a
first input of linear modulators 21 and 27.
A frequency generator 22 generates a square wave frequency ( f,) which is fed
to
band pass filter 24 to remove any harmonics, thus producing a pure sine wave.
This fj
sine wave is connected to a 0 deg. and 90 deg. phase shifter 25. The outputs
of phase
shifter 25 are in turn connected to second inputs of linear modulators 21 and
27
respectively. The outputs of the first and second linear modulators are added
in summer
28 to produce signal 37. This output signal 37 is connected to a first input
of a second
mixer 30 via high pass filter 29 which passes only f, and the upper sideband
as shown
in Fig. 2b.
A second square wave frequency generator 23 generates a signal fz as shown
Figs. 1 and 2b. This square wave is filtered by band pass filter 26 to remove
any
harmonics to produce a pure sine wave signal. This pure sine wave signal is
connected
to a second input of third mixer 30. The output of the third mixer 30 is
connected to a
low pass filter 31 to produce a descrambled output signal 35.
The second spectral diagram in Fig. 2b shows the input to the 3rd mixer 30.
The
frequency f, here represents the residual carrier feed through from mixers 21
and 27.
Fig. 2c shows. the relationship of a carrier f2 to f~ in Fig. 2b and the
scrambled
audio signal shown in Fig. 2a. Fig 2d, shows the relationship of the spectral
characteristics of the descxambled signal 35 and the residual difference
frequency (f, f2)
component to the spectra! characteristics of the signals in Figs. 2a 2c.
2
CA 02181691 2000-02-18
The Forbes encoder uses sine wave type modulators for their carriers. Switch
type modulators as disclosed in the instant invention produce a lower white
noise
component and do not require bandpass filters for filtering the carriers. See
44 and 62 in
Figure 1 of Forbes.
Forties' decoder assumes that their mixers or multipliers have no noise or
carrier
feed-through. All practical mixers or multipliers have residual white random
noise and
carrier leakage from parasitic capacitance and small mismatches of circuit
elements
within (i.e., transistor offset voltages). Forties thus does not take into
account of the
residual carrier feed through particularly from the first mixer output to
cause in-band
whistle tones caused by the intermodulation of the two carrier frequencies and
combinations of harmonics of the two carrier frequencies used in decoding the
scrambled audio signal. The invention firstly takes into account of residual
carrier feed
through as a problem to high quality (signal to noise ratio) decoding. It
identifies
primarily that the carrier feed through from the first mixer or multiplier
must follow with a
specific filter. This filter after the first mixing step must filter out
substantially the carrier
feed through and harmonics of the carrier feed through in order to have a
whistle free
decoded output. Forties does not identify this problem.
Secondly, there is a need to reduce random noise via switch type mixers versus
the sine wave mixers in Forties. Switch type mixers will out perform the
analog types by
a substantial amount (i.e. >10 db). Forties does not identify the random white
noise
characteristics of their multipliers or mixers.
Fig. 4 shows the scrambled audio input of the Quan prior art descrambler 11.
This shows the scrambled audio 40 offset by an offset frequency determined by
the
original scrambling process. The scrambled audio input signal 40 is connected
to an all pass
2a
CA 02181691 2000-02-18
shifter 41 which provides 0 deg. and 90 deg. phase shifted outputs 42 and 43
to first
inputs of first and second mixers 44 and 45.
Carrier frequency generator 46 generates a sine wave signal f~ 47 with a
frequency of 1 l4iz or 2-3 ichz. The carrier frequency 47 is filtered by a low
pass filter 48
to remove any harmonics to produce a pure sine wave 49. This pure sine wave
signal
49 is connected to an all pass phase shifter 50 to produce 0 deg. and 90 deg
signals 51
and 52 which in tum are connected to second inputs of mixers 44 and 45. The
outputs
of mixers 44 and 45, signals 53 and 54 are connected to summer 55 to produce
descxambled output 56.
Fig. 4b shows the relationship of the in band descrambling carrier f~ to the
scrambled audio signal. Figure 4c shows the descrarnbled audio spectrum with
the
residual carrier f~ that is typically -60 db below the descambled audio
program, but is still
audible during silent passages of the audio program.
It is therefore an object of this invention to provide a higher performance
descrambler andlor lower cost frequency shifted scrambled audio signals. The
method
and apparatus described 1 ) eliminates the use of 0 degree and 90 degree phase
shift
circuits, 2) eliminates the use of quadrature mixer circuits, 3) eliminates
the need for
band pass filters or low pass filters far the modulating carrier, 4) reduces
white noise
and cost using switching type mixer circuits instead of linear mixers, 5)
eliminates in-
band audible whistle via filtering out the residual first carrier whistle; fi)
eliminates the
need to adjust mixers for minimum in-band carrier whistle and 7) since the SNR
has
been improved the need for noise reduction circxrits has been eliminated.
BRIEF DESCRIPTION OF THE DRAWINGS
Fig. 1 is a block diagram of the key elements of the Fortes prior ark
Figs. 2a-2d are spectral diagrams of the system in the Fortes prior art;
Fig. 3 is a blod'c diagram of the key elements of the Quan et al. prior art;
Figs. 4a-4c are spectral diagrams of the Quan et al. prior art;
Fig. 5 is a block diagram of the prefen~ed embodiment;
Figs. 6a-6g are spectral diagrams of the preferred embodiment of the
descrambler described in Fig. 5;
3
CA 02181691 2000-02-18
Fig. 7 is block diagram of a switch type low noise modulator;
Fig. 8 is a block diagram of a first implementation of a descrambler using the
concepts of the invention;
Fig. 9 is a block diagram of a second implementation of a descrambler using
the
concepts of the invention;
Fig. 10 is a block diagram of a third implementation of a descrambier using
the
concepts of the invention;
Fig. 11 is a block diagram of a preferred embodiment of scrambler using the
concepts of the invention;
Figs. 12a-12g are spectral diagrams of the scrambler described in Fig. 11; and
Figs. 13a-13c are schematic diagrams of implementations of the 1st and 2nd low
pass filters of the invention.
SUMMARY
The present invention is directed to a method and system for descambling
frequency shifted scrambled audio signals that satisfies the needs described
above. The
invention comprises a method and system for descrambling frequency shifted
scrambled
audio signals.
The descrambling system described descrambles a scrambled frequency
translated audio information signal by generating a modulation carrier signal
at a
frequency lying outside the original frequency spectral range of a scrambled
audio
signal of about 50 Hz to about 1514~z by first generating a first modulation
can~ier signal
having a frequency greater than the highest frequency in the original audio
signal. This
first modulation carrier is used for double sideband modulating the scrambled
audio
signal into a first modulation frequency, a first upper sideband signal and a
first lower
sideband signal. This set of signals is filtered by a filter to filter out the
first modulation
frequency, all its harmonics, arid the upper sideband signal and its harmonics
from the
double sideband signal and passing the first lower sideband signal.
A second modulation carrier frequency having a frequency less than the first
modulation frequency is generated. This second modulation frequency is
connected to
a second modulating means for double sideband modulating the first lower
sideband
4
CA 02181691 2000-02-18
signal with the second modulation carrier frequency to produce a second
modulating
frequency,~a second upper sideband signal and a second lower sideband signal.
A second filter passes the second lower sideband signal to produce a
descrambled audio signal.
The modulators used are low noise switch type modulators that improve the
signal to noise ratio in the descrambled signal over the previously used
linear
modulators. The use of switch type modulators provides a lower cost device
with
improved performance.
A companion scrambling device uses similar techniques to provide improved
pertonnance at a lower cost. The method of scrambling of an original audio
signal of
about 50 Hz to about 15 Khz comprises: generating a first modulation carrier
signal
having a frequency greater than the highest frequency in the original audio
signal;
quadrature modulating said original audio signal into a first lower sideband
signal;
filtering out the first modulation frequency and all its harmonics,at feast
part of the upper
sideband signal and all the hamlonics from the modulated signal and passing
said f rst
lower sideband signal; generating a second modulation carrier frequency having
a
frequency higher than the first modulation frequency; double sideband
modulating the
first lower sideband signet with the second modutation carrier frequency to
produce a
second modulating frequency, a second upper sideband signal and a second lower
sideband signal; filtering the second modulating frequency,part of the second
upper
sideband signal and the second lower sideband signal to pass the seoand lower
sideband signal to produce a scrambled audio signal.
From a method standpoint, the invention broadly comprises frequency
translating
the original speehum of audio information signals to produce scrambled audio
information signals by generating a modulation carrier signal having a
frequency lying
outside the frequency spectral range of the audio information signals, and
first single
side band modulating followed by double side band modulating the original
information
signals with the modulation carrier signal to translate the frequency of the
original audio
information signal in a given direction. Preferably, the frequency of the
modulation
carrier signals) are varied during generation in a pseudo random fashion,
particularly
5
CA 02181691 2000-02-18
by sweeping the frequency of the modulation carrier signal between
predetermined
limits. The step of varying the frequency of the modulation carrier signal
preferably
includes the steps of initiating a frequency varying operation in response to
a first
control signal at a rate determined by a second control signal.
For a fuller understanding of the nature and advantages of the invention,
reference should be made to the ensuing detailed description taken in
conjunction with
the accompanying drawings.
Fig. 5 shows a block diagram and Figs. 6a-6g show spectral diagrams of the
preferred embodiment of the instant disclosure. Fig. 6a shows the spectral
characteristic
of the scrambled audio input of the preferred embodiment. This shows the
scrambled
audio offset by an offset frequency determined by the scrambling process. Fig.
6b
shows the relationship of the first mixer's carrier and the output of the
first mixer. Both
the upper and lower sidebands and the residual carrier fA plus the harmonics
of all of
these are at the first mixer's output. Fig. 6c shows the filter
characteristics of the first
LPF following the first mixer's output. It is crucial that this first LPF
filter out the residual
carrier and its upper sideband harmonics. Fig. 6d shows the spectral
characteristic of the output of the first LPF following the first mixer's
output.
Fig. Ge shows the relationship of the second carrier to the output of the
first LPF
to form the last descxarnbling step. Fig. 6f shows the relationship of the
descambled
audio that has passed through a 2nd LPF with a 12 khz cut-off to filter out fB
and its
upper sideband above fB with the absence of whistle frequency component (fs
fb). The
(fs~,,) whistle frequency component is typically equal to or less than -85 db
in the
descrambled audio.
In this preferred embodiment fA is about 19 Khz and fB is about 16.4 i~z.
These
choices are for economy, since with these frequencies the first LPF can be
designed
inexpensively. If increased pertormance at a greater cost is desired, the
carrier
frequencies can be higher in order to minimize leakage of components from the
scrambled audio input so as to not interfere with the lower sideband output of
the first
mixer. Note that in Figs. 6a and 6b there is an overlap between the spectra of
the lower
6
CA 02181691 2000-02-18
sideband frequencies and the scrambled audio frequencies. If the first mixer
feeds
through enough of the scrambled audio, distortion products will occur at the
descrambled output. By setting the carrier frequencies to for example fA = 39
I~z and fB
=36.4 Khz scrambled input leak through will not cause distortion products at
the
descrambled output since it will not overlap with the lower sideband of the
first mixer i.e.
36.4 I~z. to 24 IQiz. versus 2.6 Khz to 14.6 Khz of the scrambled input.
However raising
f~, and fB two fold causes the steepness of the first t_PF to increase to
about two fold.
This would require higher order filters such as a 10 pole elliptical low pass
filter.
Minimal carrier leakage and scrambled audio leakage with lower shot noise is
achieved by using a double throw single pole analog switch such as the 74 HCT
4053
or its equivalent i.e. MC1496 switch type mixer with a carrier input equal to
or more than
350 my p-p.
It was found, for instance, with a CD 4053 analog switch the "on" resistance
resulted in a measured noise of 2.5 nvl 1Hz which translates into a noise
resistance
(/4kTBr = VN = 2.5 nvl IHz, B = 1 Hz, T = 298° Kelvin, k = Bolzman's
constant and R =
noise resistance) of 400 ohms. The "on" resistance of the CD 4053 was measured
to be
440 ohms. Thus it was found experimentally that the "on" resistance of the
analog
switch (i.e. 4063) produces the same amount of noise as a resistor component
of the
same resistance. Thus an "on" resistance of 440 ohms in a CD4053 has
essentially the
same noise as a 440 ohrn resistor.
Linear modulators such as the AD 534 produces 0.6 my RMS over a 10 iQiz
bandwidth or a noise density of 0.6 my I 110 l4oz = 60 nv I IHz Therefore the
AD 534
linear modulator produces approximately 6012.5 more noise than the CD 4053
switch.
This is equivalent to a 27 db improvement when using a CD4053 over a linear
modulator.
Gilbert modulators such as the 1496 or 1495 wiU produce kwv noise, i.e. < 5
nv/
/Hz when the carrier input of these devices switch the differential pairs on
and off This
is achieved by either overriding the carrier input with a square wave carrier
input with a
square wave of > _/ 200 my or a large sine wave of > 1 v pp. When sinusoidal
modulators such as the 1495 does not have the carrier inputs over driven to
produce
7
CA 02181691 2000-02-18
linear modulation, the noise is substantially higher versus a switch mode 1496
modulator. This is because the 2 differential pair transistors start
amplifying their own
noise. The internal base resistance of each transistor is usually about 50 -
200 ohms. If
one assumes a 100 ohm series internal base resistors on the 2 pairs of
differential pair
transistors in series in a 1495 and 1 kohrn load for one output and further
assumes that
each of these transistors has a quiescent bias of 1 ma collector current, the
output noise
is then equal to 1I2 * 1000 (gm) V nr = Vo noise. gm = 38 maN for an h = 1 ma.
Therefore V nr =1400 ohm * 4kT = 2.5 nvlHz. Vo noise =19 * 2.5 nv I !Hz = 47.5
nv I IHz
from a 1495 modulator. This is 19 times or 25 db more noise than the CD 4053
with an
"on" resistance of 440 ohms. It should be noted that the output noise
decreases in the
1495 or 1496 modulator as the canier input is increased.
The key to the preferred embodiment is the use of a Low Pass Filter (LPF)
after
the first mixer which is to reject out a residual carrier from the first mixer
and remove all
sidebands related to harmonics of the carrier. If this is
not done, harmonics of the whistle frequency (3f,-3fb~), (5fa-5fb~) and etc.
will appear at
the descrambling output in an audible manner.This first LPF is generally a 7
pole or
more elliptical filter with at least one zero tune to notch out the first
mixers carrier
frequency, f~. In practice a 9 pole active filter with general impedance
convertors is the
best choice for a stable and accurate filter. tn the preferred embodiment the
3 db cx~t off
of the first low pass filter is about 17 IQiz with at least 40 db attenuation
at 19 l4~tz.
A detailed description of the preferred embodiment is described below with
reference to Fig.S. The descrambling apparatus 12 has a scrambled audio signal
input
60 and ooMains the descrambling process of the preferred embodiment. The
scrambled
audio 60 is inputed into a first input of a first mixer 63. The second input
of this first mixer
is a first carrier signal fA generated by frequency generator A, 61 which is
approximately
19 14~z. The output first mixer 63 contains carrier feed through of fA, all
its sideband
components and the harmonics. The output of mixer 63 is fed to a tow pass
filter 65 that
fitters out the first carrier, the upper sideband and all of the harmonics
from signal 60.
The output of low pass filter 65, signal 66 is fed into a first input of a
second mixer 67.
The second input of this second mixer is a second carrier signal fe generated
by
8
CA 02181691 2000-02-18
frequency generator B, 62 can be 16.4 .Khz t 100 Hz shifted pseudo
randomly for security reasons. See U.S. Patent 5,095,279 for a further
explanation of
this security process. The output 70 of second mixer 67 contains the baseband
descrambled audio, residual second carrier and upper sideband components above
fBs
frequency. The second low pass filter 71 with a cut-off frequency of
approximately 12
I~z removes everything above 12 i4~z, but passes the descrambled audio to the
output
of the apparatus 12.
In the above preferred embodiment the mixers utilize a switch type low shot or
thermal noise modulator as described in Fig. 7. The operation of this mixer
will be
described relative to the first mixer. The second mixer operate on the same
principle.
Scrambled audio 60 is fed into the + input of unity gain amplifier 73. The
output of
amplifier 73 is fed on line V~, 74 to one input of a double pole single throw
analog
switcher 32. The output of 73 is also fed to the input of unity gain inversion
amplifier
consisting of RZa, R2b, and ampler 65.. The output of amplifier 65 is V~, 75
which is
fed to a second input of the switcher 32. First carrier frequency fA is fed
into the
switching control input of the double pole, single throw switcher 32. The
double pole,
single throw switcher used is 113 of an 74HCT4053 or its equivalent and is fed
to
amplifier A220. A220 is the mixer output. For minimal can-ier leakage of the
output of
mixer 65, the DC zero signal voltage of the two inputs of switch 32 V;" and
V~, must be
exactly the same, i.e. Ov. In addition the inversion amplifier 73 must be a -1
unity gain to
have minimum scrambled audio in ( V",) feed through. Thus R2a = R2b within 1
°~ or
better is required for a wide band op amp 65 (i.e. NE5532.
Fig. 13a shows a conventional RLC low pass filter with zeros for the '
descrambler's first low pass filter. Inductors L, through L3 are rather large
2
milli-henries through 20 milli-henries to a~ieve a low cost. These lower cost
inductors
suffer from a just adequate Q at audio frequencies. Much more expensive
inductors with
higher Q's will provide better low pass filtering, but will be beyond the
budget of a low
cost descrambling system.
Fig. 13 b shows an active 9 pole elliptic low pass filter that is not as
sensitive to
parts tolerance as many other active filters. This is important since f,, the
first carrier
9
CA 02181691 2000-02-18
frequency must be filtered out by at feast ~0 db attenuation. Fig. 13 b is a
General
Impedance Converter (GIC) active low pass filter that was found to provide
very high
performance in filtering at low cost. The capacitors can be inexpensive
5°~ mylar film
capacitors. The resistors are inexpensive 1 °~ resistors and the
operational amplifiers
can be common type such as TL082, NE5532 etc.
Fig. 13 c shows an example of the 2nd filter as an active 7 pole low pass
filter.
Ampl~ers A1000, A2000 and A3000 can be simple voltage followers of common
operational amplifiers or single transistor emitter followers. The 2nd filter
in the
descrambler can any low pass filter, passive or active with sufficient stop
band
attenuation to provide a descrambled audio signal without measurable artifacts
such
as 2nd carrier tone its upper sidebands andlor audible artifacts.
Figs. 8 -10 show various implementations using the concepts of the invention.
In addition to a descrambling system as described above many of the same
elements can be used in a scrambler to achieve many of the same advantages
achieved in the descrambler described above, i.e. lower shot noise output and
less filter
requirements than the prior art such as Forties ('853). Fig. 11 is a block
diagram and
Fig. 12 is a series of spectral diagrams of a preferred embodiment of a
scrambler.
An audio signal 91 with a spectral response of about 30 Hz. to 15 Khz. is fed
into
a low pass frlter 20 to eliminate any unwanted signals beyond 15 Khz. The
output 93 of
low pass filter 20, is connected to 0 deg. and 90 deg. all pass phase shifters
94 and 95.
The outputs of phase shifters 94 and 95 are in tum connected to first inputs
of switch
type low noise modulators 96 and 97.
Signal generator 98 generates a square wave signal at approximately 1fi.4 Khz.
with 0° and 90° outputs which are connected to second inputs of
modulators 96 and 97.
The outputs of modulators 96 and 97 are summed to produce signal 103, a
quadrarture
modulated signal resulting in a residual 16.4 Khz. carrier with a lower
sideband. Fig. 12
shows the relationship of the quadrature modulated audio components to the
origins!
audio signal 91.
This quadrature modulated signal is fed through low pass filter 104 as signal
105
and is essentially the same filter as the first filter of the descrambler
described above.
CA 02181691 2000-02-18
This signal is connected to a first input of a third modulator 106. Modulator
106 is a
switch type tow thermal or shot noise modulator as described above and shown
in
Fig. 7. A second carrier frequency is generated by a square wave oscillator 99
generating a frequency of approximately 19 i4iz. as shown in Fig. 12 e. The
output of
modulator 106 contains a 19 khz carrier and upper and lower sidebands. This
signal is
filtered by low pass filter 107 to produce a scrambled audio signal with an
offset of
approximately 2.6 l4iz.
Theoretically, to decrease the dynamic artifacts caused by fast step frequency
changes of the 16.4 I~z can-ie~ in both the scrambler and descxambler, the low
pass
filters following the first quadrature mixer and the first mixer of both the
scrambler and
descrambler respectively should be very nearly identical in group delay
responses
(transient responses). If the transient response characteristics of the low
pass filters in
the scrambler are different from the transient characteristics of the
descrambier, the step
changes of the 16.4 Khz carrier has to be slowed down to achieve minimal
descrambling artifacts. It is preferred to have faster step changes in the
secured carrier
(16 l4iz +/- 100 Hz) and have the first low pass filter in the descrambler
have the same
characteristics as filter 104 in the scrambler of Fig. 11. In addition, the
second low pass
filter in the descrambler should have the same characteristics of filter 107
of the
scrambler of Fig. 11. This permits the step shifting spectrum of the scramble
to be
tracked quickly in the descrambter without artifacts caused by time delay
skews between
scrambler and descxambler tracking the 16 IChz stepped deviations. It should
be noted
that all carriers for all mixers in this invention for descxamblers and
scramblers should
be preferably square wave signals for minimum artifacts.
While the above provides a full and complete description of the preferred
embodiment of the invention, various modfications, alternate constructions and
equivalents will~occur to those skilled in the art. Therefore , the above
descriptions and
illustrations should not be construed as limiting the scope of the invention,
which is
defined by the appended claims.
11
