Note: Descriptions are shown in the official language in which they were submitted.
1 Inve or: R~ M. DOLBY
2 Title: IMPROVEMENTS IN CIRCUIT ARRANGEMENTS
3 FOR MODIFYING DYNAMIC RANGE
4 The present invention is concerned in general with
circuit arrangements which alter the dynamic range of audio and
6 other signals, namely compressors which compress the dynamic
7 range and expanders which expand the dynamic range. More
8 particularly, it relates to improvements in compressors and
9 expanders that reduce their susceptibility to control by
undesired signals. Such improvements are designated "modulation
11 control" for reasons explained herein.
12 Compressors and complementary expanders are often used
13 together (a compander system) to effect noise reduction; the
14 signal is compressed before transmission or recording and
expanded aFter reception or playback from the transmission
16 channel. However compressors may be used alone ~o reduce the
17 dynamic range, e.g., to suit the capacity of a transmission
18 ch~nnel, without subsequent expansion when the compressed signal
19 is adequate for the end purpose. In addition, compressors alone
are used in certain products, especially audio products which
21 are intended only to transmit or record compressed broadcasts or
22 prerecorded signals. Expanders alone are used in certain
23 ~ products, especially audio products which are intended only to
241 receive or play back already compressed broadcasts or prerecorded
2~1 signals. In certain products, particularly audio recording and
26 playback products, a single device is often configured for
271 switchable mode operation as a compressor to record signals and
28¦ as an expander to play back compressed broadcasts or prerecorded
291 signals.
301 The amount of compression or expansion may be expressed
31¦ in dB. For example, 10 dB of compression means that an input
32 ¦ dynamic range of N dB is compressed to an output range of (N-10)
I .~
1¦ dB. In a noise reduction system 10 dB of compression followed
21 by 10 dB of complementary expansion is said to provide 10 dB of
31 noise reduction.
41 The present invention relates in particular to improve-
sl ments in circuit arrangements for modifying the dynamic range of
61 an input signall which circuit arrangements have a bi-linear
71 characteristic (where "linear" in this context denotes constant
31 gain) composed of:
91 1) a low level linear portion up to a threshold,
10¦ 2) an intermediate level non-linear tchanging gain)
11¦ portiont above the threshold and up to a finishing
12 ¦ point, providing a predetermined maximum compression
13 ¦ ratio or expansion ratio, and
14 ¦ 3) a high level linear portion having a gain
15 I different from the gain of the low level portion.
16 ~The characteristic is denoted a bi-linear characteristic
17 ~ because there are two portions of substantially constant gain.
18 ~ In practice, the threshold and finishing point are
19 ~ not always well defined "points". The two transition regions
20 ¦ where the intermediate level portion merges into the low
21 ¦ level and high level linear portions can each vary in shape
22 from a smooth curve to a sharp curve, depending on the control
23 characteristics of the compressor and expander.
24 It is also pointed out tha~ circuit arrangements with
bi-linear characteristics are distinguished from two other known
26 classes of circuit arrangement, namelyo
27 (a) a logarithmic or non-linear circuit arrangement
28 with either a fixed or changing slope and with no linear portion:
29 the gain changes over the whole dynamic range.
lb) circuit arrangements with a characteristic
31 having two or more portions of which only one portion is linear
32 ///
2-
~"uni-linear"). The invention is also applicable to uni-linear circuits~ as
is explained further below.
A circuit arrangement with a bi-linear characteristic has part-
icular advantages and is widely used. The threshold can be set above *he in-
put noise level or tra~smission channel noise level in order to exclucle the
possibility of control of the circuit by noise. ~le high level portion of
substantially constant gain avoids non-linear treatment of high level
signals which would otherwise introduce distortion.
Two well known types of bi-linear circuits are referred to as slid-
ing band circuits and fixed band ~or split band or multiband) circuits.
Sliding band circuits create the specified desirable character-
istic for the case of high frequency audio compression or expansion by apply-
ing high frequency boost ~for compression) or cut (for expansion) by way of
a high pass filter with a variable lower corner frequency. As the signal
level in the high frequency band increases, the filter corner frequency
slides upwardly so as to narrow the boosted or cut band and exclude the use-
ful signal Erom the boost or cut. Examples of such circuits are to be found
in US-PS Re 28,426, US-PS 3,757,254, US-PS 4,072,914, US-PS 3,93~,190. Such
circuits can also be configured to act at low frequencies in which case low
~0 frequency boost or cut is provided by way of a low pass Eilter with a vari-
able upper corner frequency.
In fixed ban~ circuits the frequency spectrum is split into a
plurality of bands by corresponding band-pass filters and the compression or
expansion is effected in each band by a gain control device ~whether an auto-
matically responsive, diode type of limiting device or a controlled limiting
device) in the case of a compressor, with some form of reciprocal or comple-
mentary
l circuitry for an expander. Examples of such circuits are to be
2 ~ound in US PS 3,84~,719, US~PS 3,903,485 and in Journal of the
3 Audio Engineering Society, Vol. 15, NoO 4, October~ 19Ç7, pp.
4 383-388. These fixed band circuits provide independent action
in the various frequency bands.
~ It is known to construct bi-linear compressors and
7 expanders, of both sliding band and fixed band type, by the use
3 of only a single signal path. However, it is generally preferred
9 to construct such devices by providing a main signal circuit
which is linear with respect to dynamic range, with a combining
l1 circuit in the main circuit, and a further circuit which derives
12 its input from the input or output of the further circuit and
13 has its output coupled to the combining circuitO The further
14 circuit includes a limiter (self-acting or controlled) and the
limited further circuit signal boosts the main circuit signal in
16 the combining circuit for the case of compression but bucks the
17 main circuit signal for the case of expansion. The limited
18 further path signal is smaller than the main path signal in the
19 upper part of the input dynamic range. The main and further
2n circuits are preferably and most conveniently separately
21 identifiable signal paths~ More than one further circuit i5
22 usually provided in the case of fixed band devices. A bi-linear
23 device having main and further circuits is often referred to as
24 a dual path device.
Such known dual path compressors and expanders are
26 particularly advantageous because they enable the desired kind
27 of transfer characteristic to be established in a precise way
28 without problems of high level distortion. The low level
29 portion of substantially constant gain is established by giving
the further path a threshold above the noise level; below this
3l ¦ threshold the further path is linear. The intermediate level
32 ¦ portion is created by the region over which the f~rther path
Il -4-
limiting action becomes partially efective and the high l~vel portion of
substantially constant gain arises ater the limiter has become fully effec-
tive so that the further path si.gnal ceases to increase and becomes negligi-
ble compared to the main path signal. At the highest par-t of the input
dynamic range, the output of the circuit arrangement is effectively only the
signal passed by the linear main path, i.e. linear with respec~ to dynamic
range.
Examples of these known circuits are to be found in IJS-PS
3,8~6,719, US-PS 3,903,~85 and US-PS Re 28,~26. lhere are also known analo-
gous circui.ts which achieve like results but wherein the further path hascharacteristics inverse to limiter characteristics and the further path out-
put bucks the main path signal for compression and boosts the main path
signal for expansion (US-PS 3,828,280 and US-PS 3,875,537).
The invention is applicable to any of these known bi-linear cir-
cuits in order to obtain the advantages inherent therein. The invention is
not limited to bi-linear circuits, but also may be employed to improve the
operation of the a:Eorementioned mi-linear circuits. As discussed further
below, the invention may also be applied to logarithmic circuits provided
that a departure from a logarithmic transfer function can be tolerated.
Ilotvever, the preferred embodiments rela-te to bi-linear circuits and except
wIIere specifically noted, reference is made to bi-linear circuits throughout
this specification.
As mentioned previ.ously, it is not essential to create the desired
form of bi-linear characteristic by such "dual path" techniques. Alterna-
tlves exist, operating with single paths, as described in US-PS 3,757725~,
US-PS 3,967,219, US-PS 4,072,91~, US-PS 3,909,7337 for example. Although
these alternative circuits usually are not capable of producing such good re-
sults as dual path circuits, or
. . .
may be less convenient and thereby less economical, -they can
produce generally equivalen-t results. Accordingly, the inven-
tion is also applicable to these known circuits.
The invention also pertains -to known compressors
and expanders in which series connected (e.g. multistage)
bi-linear circuits are employed. Such arrangements are des-
cribed in selgian-PS ~89,4280
In compressors and expanders~ especially frequency
selective or multi-band devices it is clearly desirable that
strong signals in one frequency range should not unduly aEfect
the behavior of signals in another frequency range. Filtering
and equalization employed in the various circuits has been
the standard method of dealing with this problem, both in
logarithmic devices and in specialized devices such as the
unilinear and bi-linear circuits which have been describedO
In these prior art circuits the DC control signal which controls
the variable gain/loss device [e.g., a variable gain device
such as~avoltage controlled amplifier (VCA) or a variolosser
such as an FET attenuator] or variable Eilter is ~ormed from
~0 the linear additive combination of the pass-band signals and
the stop-band signals reaching the control circuit. The pres-
ent invention effectively alters this simple combination, in
a level dependent way, so as to optimize the compressor or
expander performance with respec-t to pass-band versus stop-
band signals. Non-linear operations are performed, including
rectification of the signals in various portions of the spec-
trum and analyses are made of the relative and/orabsolute
amplitudes. The final control can be formed by selecting
one of the singals, by combining two or more, or by performing
non-linear operations such as limiting a-t leas-t one of the
signals.
According to one aspect, the invention contemplates
a circuit arrangement for modifying the dynamic range of an
input signal, comprising: frequency selective circuit means
for dividing the frequency spectrum in which the input signal
lies into pass-band and stop band regions, and dynamic modifi-
cation means for modifying the dynamic range of signal com-
ponents in the pass-band region in response to signal compon-
ents lying in the pass-band and stop-band regions, the dynamic
modification means being less responsive to stop-band signal
components as the level of the input signal rises.
According to another aspec-t, the invention contem-
plates a circuit arrangement for modifying the dynamic range
of an input slgnal, comprising: frequency~selec-tive circui-t
means for dividing the frequency spectrum in which the input
signal lies into pass-band and s-top-band regions, the pass-
band frequency region sliding in response to signal components
lying ln the pass-band and stop-band regions, the frequency
selec-tive circuit means becoming less responsive -to s-top-band
signal componen-ts as the level of the input signal rises,
~0 and means for modifying the dynamic range of siynal components
in the pass band region.
In other words, a-t low input signal levels the
clrcuit arrangement performs substantially as a conventional
compressor or expander. However, at high input signal levels
the compressor or expander action is modifled by the modula-
tion control circuitry of the invention.
A side effect is to modify the input-output level
transfer characteristic of the device at any particular fre-
quency or combination of frequencies. The overall effect
is unimportant and may even be unnoticeable at the dominant
frequency in bi-linear systems. However, in logarithmic sys--
--7--
tems the effect of modulation control, which is operativeprimarily in the high level portion of the dynamic range,
is to cause a departure from a purely logarithmic characteris-
tic. This may or may not be important in any particular appli-
cation.
The invention derives from the observation that,
ideally, in compressors and expanders the compression or expan-
sion is responsive only -to the levels of signals within desired
frequency pass-bands and not to the levels of signals at o-ther
frequencies, which frequencies can be said to be in the stop-
bands. For example, in an ideal circuit, compression or expan-
sion should not be affected by the levels of signals outside
the pass
-7a-
~8B9~
1 ¦ band of the fixed band or the pass band of the sliding band
2 ¦ (whether or not in its quiescent position). In the case of a
31 sliding band circuit in accordance with the invention, the amount
41 of frequency sliding of the variable band becomes no more than
51 is necessary to assure that a dominant controlling signal is not
61 boosted (in the case of compression) above a reference level.
71 As applied to bi-linear circuits, particularly those
~¦ in the dual path configuration, the invention takes further
91 advantage of an inherent characteristic of such circuits:
10¦ at high input signal levels the main path signal is substantially
11¦ larger than the signal(s) in the further (or side) path(s).
12¦ Consequently, high-level signal manipulations in the further
13¦ path are essentially inaudible, and, except for phase shifts,
14¦ are essentially not measurable (negligible level changes)O This
15¦ property of bi-linear circuits is most easily understood in the
16¦ context of a dual path circuit. However, the principle also
17¦ applies in single-path bi-linear circuits in which there are
18¦ two or more signal components in the same path instead of in
19¦ separately identifiable paths.
20¦ The invention takes advantage of the above observations
21¦ concerninq bi-linear circuit characteristics. As compared with
22 ¦ prior art bi-linear compressors and expanders, the invention
23 ¦ provides for further manipulations of the signal (modulation
24 ¦ control) in the high input signal level region where the overall
25 ¦ compressor or expander response is linear. The comparatively
26 ¦ low level noise reduction component of the signal is manipulated
27 ¦ in this extra way only at high signal levelsj thereby assuring
28 1 that any effects important to the signal channel will be over-
29 shadowed by the large main signal component~
30 1 In dual-path bi-linear circuits, an efect of the
31 invention is to modify the transfer characteristic of the side
32 (or further) path such that the side path characteristic itself
_~
1¦ becomes bi-linear instead of flattening or downturning at high
21 input signal levels. This is a consequence of the proportionality
31 aspect of modulation control. That is, at high input levels
41 the side path level does not drop below a selected proportion
51 of the main path level (e.g. one-quarter or one-tenth). This
61 is acceptable because the side path signal still remains
71 substantially smaller than the main path signal at high input
81 signal levels and because the stop-band is usually substantially
91 shifted in phase with respect to the main path signal channel.
For these same reasons, the invention may be embodied
11 in uni-linear circuits which have a linear response at high
12 si~nal levels.
13 From another point of view, the action of the invention
14 is to increase the levels of stop band signal components in the
output of the device at high signal levels, but not ~o such an
16 extent as to cause problems with the recordin~ or transmission
17 channel since, relatively speaking they are still small.
18 Increasing the levels of such stop band signal components is not
1~ in itself a particularly advantageous thing but is necessary in
order to obtain improved dynamic action and noise reduction
21 within the pass band~ Increasing the levels of s-top band
22 ~ignals in the output oE the device at high signal levels is
23 achieved by reducing the levels of stop band signal components
24 in the control signal channel at high signal levels, or by so
arranging things that the control signal is generated as if
26 there was a reduced level of stop band signal components
2~ in the signal used to produce the control signal at high signal
28 levels (e.g., by filtering and limiting in the control circuits
29 or by fre~uency dependent control signal bucking arrangements).
A further advantage o the invention is that in
31 listening tests, "pumpiny effects" of single-ended compressors
3~ ///
_g_
1¦ and expanders are substantially reduced if not eliminated. Thus,
21 in addition to its use in complementary noise reduction systems,
I s~
~; 31 the invention is particularly useful for use in ~ com-
41 pressors and expanders (iOe., compressors for use in compressing
51 signals that are not subsequently expanded and expanders for use
6¦ in expanding signals that were not previously compressed)O
71 By way of background of the invention~ although various
81 practical embodiments of noise reduction circuits have proven
91 successful, in operation such circuits depart in some degree
10¦ from the ideal because of the problem of stop-band signals
11¦ unduly controlling compression and expansion. The efEect of
1~¦ such shortcomings is manifested in several interrelated ways:
13 ¦ 1) a reduction in noise reduction effect in a
14 ¦ portion of the noise reduction system pass band;
15 ¦ 2) noise modulation effects (eug., the level
16 ¦ of a signal at one frequency, modulating the noise
17 ¦ level in a different part of the frequency spectrum)
18 3) signal modulatiorl effects (e.g., the level of a
19 ¦ signal at one fre~uency modulating the level of a
signal at another frequency);
21 4) cross modulation effects (e.g. spurious modulation
22 products resulting from one of more of the last two
23 enumerated modulation effects)~
24 The degree to which these shortcomings are observable
2~ depends on the type of circuits employed in the noise reduction
2~ system, the recording and playback equipment, the record/playback
27 channel or medium and the nature of the signal material. In
28 many cases, the shortcomings are essentially unobservable except
29 ¦ by test instruments. Nevertheless, it is desirable to deal with
30 ¦ these shortcomings. Because the aforementioned shortcomings of
31¦ known compressors~ expanders and noise reduction systems rela-te
32 I to modulation effects, ei-ther of signals or of noise~ -the
_~o_. ,
~ 6
1¦ invention described herein for reducing such shortcomings is
21 referred to as modulation control.
31 The severity of these modulation effects depends to a
41 great extent on the uniformity of the transmission channel
51 between the compressor and expander. For example, in magnetic
61 tape recording and playback systems, a frequency response
7 phenomenon known as ~Ihead bumps" exists. Even in professional
8 systems, particularly those operating at 30 ips, the playback
9 response below 100 Hz is nonuniform due to the relationship
between the signal wavelength on tape and the playback head
11 dimension, which are of the same orderO If the compressor/
12 expander system is susceptible to signals in the head bump
13 region, such signals when played back may control the expander
14 in a non-complementary way such that signals or noise at higher
frequencies, e.g., up to 3 kHz, may be modulated by the signals
16 in the re~ion of or below 100 Hz~
17 In prior art fixed band (single band and multi-band)
18 circuits, various filterin~ techniques have been used to minimize
19 the control of compression and expansion by undesired signals.
~0 According to these techniques, sharp filters (e.g.) with steep
21 skirts) are placed in the signal path or in the control circuit
~ ~of the limiting device).
23 However, the use of signal path filters sharper than
~ 6 dB/octave (e.g., single pole filters) in multiband compressors
and expanders causes amplitude and phase efects such that when
26 the overall signal spectrum is recombined there are amplitude
V and phase errors. This problem is greatly exacerbated if
28 filters sharper than 12 dB/octave are employed. However, a
~91 filter slope of only 6 or 12 dB/octave may not adequately
30 ~ discriminate against all unwanted signals. In the multi-band
311~ (fixed band) bi~linear circuit examples of US~PS 3,846,719 and
321 in'Journal of the Audio _ i eerinq Societ~, Vol. 15, No~ 4,
l¦ October, 1967, pp. 383-388, filters having a 12 ds/octave slope
21 are used in the signal path of three of the four fixed bands. A
31 flat overall frequency response is obtained only by the use of a
41 complex filter characteristic in the frequency band adjacent the
51 sharp filters~ Such a solution obviously is not universally
61 applicable.
71 In the logarithmic multi-band (fixed band) compressor/
81 expander circuit described in Rundfunktechnl Mitteitun~
91 Jahr. 22 (1978) H~ 2, pp. 63-74, the input signal is divided
l0¦ into four bands by single pole filters. ~oweverl th~ control
ll¦ circuits for each band employ sharp 18 dB/octave fil~ers. A
12¦ sharp eontrol circuit filter (12 dB/octave~ is also employed
13¦ in a single fixed band compressor/expander circuit sold under
14¦ the trademark "dbx II." However, the use of sharp control
l5¦ eircuit filters results in excessive amplification of high
16¦ level signals outside ~he control circuit filter pass band
17¦ when high amplitude signals are not present within the control
l~¦ circuit filter pass band, resulting in the posslble overdriving
l9¦ of the transmission channel unless sharp eutof~ filters are used
20¦ in the signal ehannel as well.
21¦ A prior art technique referred to as spectral skewing
22¦ is described in Belgian-PS 889,427; Audio, May 198l, pp~ 20-26;
~31 and paper J-6 and preprint presented at November 1981 Conven-
24¦ tion~ Audio Engineering Society, New York, New York. Spectral
251 skewing is also concerned ~ith the suppression of modulation
2S¦ effects resulting from compressor/expander non-complementarity
27 ¦ due to transmission channel errors. According to the teachings
28 ¦ of spectral skewing, sharp filtering is provided at least in the
29 ¦ compressor at a frequency well within the normal bandpass of
30 ¦ the system and within the flat response region of the transmis-
31 sion channel. While spectral skewing is successful in reducing
32 ! ///
~ -12-
l ll
1 spurious sign~l modulation effects caused by channel irregulari-
2 ties, it does not address the problem of excessive frequency
3 sliding in sliding band systems or of excessive attenuation in
4 fixed band systems~
S Thus, the present invention seeks to minimize the
6 control of expansion and compression by undesired signals without
7 the attendant side effects and/or complexity of the prior art.
8 Although measurable modulation effects are not totally
9 suppressed by the inven~ion, the effects of the invention in
audio applications are supplemented by psychoacoustical masking
11 effects such that perceived effects are~ for most listeners and
12 musical material, inaudible. That is, only the modulation of a
13 signal (or signals) sufficiently spaced in frequency from the
14 modulating signal is perceived by the human ear. Such modulation
is minimized by the present invention. While the modulation
16 of a signal (or signals) by another signal closely spaced in
17 frequency is less likely to be affected or improved by the
18 invention, such phenomena will likely not be perceived by the
19 ear because of two related effects:
a) a weak signal close in fre~uency to a strong
21 signal is masked by the strong signal such that the weak signal
22 is inaudible, or
23 b) if the closely spaced signal is audible before
24 compression or i9 increased in level by the compressor such that
it becomes audible, then there is a psychoacoustic tolerance of
26 modulation effects because of the close frequency spacing.
27 Consequently, the human ear is not able to discern
28 modulation effects of signals at closely spaced frequencies and
29 thus the invention need not be fully effective for such signals.
The operating environment o the invention is a fixed
31 band or sliding band compressor or expander circuit in which
32 there is a variable circuit means, usually controlled by a DC
-13-
l l
1 control signal~ which is operative primarily in the lower part
2 of the overall dynamic range~ In accordance with the invention,
3 mod~lation control means are employed in the upper part of the
4 dynamic range to prevent the action of the variable circuit
means from becoming any greater than is necessary to provide the
6 nominal required attenuation of dominant signals, whether such
7 signals have freauencies in the pass-band or in the stop-band.
81 In pract1ce, controlling the action of the variable circuit
91 means usually comprises operating upon the control signal
10¦ controlling the circuit means.
11¦ The modulation control may take the form of active or
12¦ passive control signal limiting means which become operative at
13¦ high signal levels or of means employing circuits which detect
14¦ the presence of high level signals and generate signals which
15¦ oppose the increase of the control signal level. Such control
16¦ signal limiting may take place in one or more frequency selective
17¦ control signal channels; if more than one, means are provided for
18¦ selecting or combining the control signals so as to provide the
19¦ variable circuit element with an optimal control signal When a
201 high level signal detection circuit~ or modulation control
21¦ generator, is usedr it may operate in various ways which will
22¦ give a measure of signal levels in at least the upper part of ~he
231 dynamic range. For example, the modulation control signal may be
24 ¦ derived from the input or output signal of ~he compressor or
25 ¦ expander. The modulation control signal in effect provides a
26 reference for the DC control signal applied to the variable cir-
27 ¦ cuit element (VCA or voltage controlled filter~. The reference
28 ~ signal is combined in phase opposition with (eOgO, opposite
29 ¦ polarity or so as to buck) the DC control signal generated pri-
30 1 marily in response to stop-band signal components to provide a
31 limit as to how large the control signal to the variable circuit
32 j
IL181~ g6
1¦ element can become in response to signals in the stop-band, io~
21 outside the pass band of the fixed band or sliding band. In
31 practice this limit can be made relatively "hard" or relatively
41 I'soft". That is, continued increases in the control signal can be
51 rather abruptly stopped or allowed to continue at a reduced rate
61 The modulation control signal may also be derived from
7 the variable circuit (VCA or variable filter~ by measuring voltage
8 or current components of the variable circuit and, if necessary~
9 equalization in order to generate a signal usable in providing a
limit as to how large the control signal to the variable circuit
11 can become in response to signals in the stop-band.
12 In terms of resul~sr the invention as applied to
13 either fixed band or sliding band devices provides a substantial
14 immunity to signals outside the pass band of the fixed band or
the sliding band~ In ~he case of sllding band devices, the
16 invention provides a Eurther related advantage, i.eO~ the
17 sliding band slides only so far in response to a dominant signal
18 as is necessary to bring the gain at the signal fre~uency to
19 substantially unity, at least for levels at or above a reference
level. The reference level is at or near the upper area of the
21 dynamic operating range of the device, such as within about
22 6~20dB of the maximum allowable level. Prior art sliding band
23 circuits are susceptible to excessive sliding such that the
24 variable filter corner frequency is pushed farther than is
needed with hiyh level signals, causing not only potential
26 modulation effects but also resulting in a loss of noise reduc-
27 tion effect in part of the spectrumO
~8 As applied to sliding band dual-path circuits, the
29 invention provides forr in the simplest embodiment, the rectifi-
cation and smoothing of the input or output signal and the
31 combining of the resultant DC reference signal with the control
32 signal applied to the variable filter~ The level of the
-15-
l ¦ reference signal can be set for a desired proportion limit on a
2 ¦ dominant further path signal relative to the corresponding
3 ¦ component in the main path signal~ ~or example, the modulation
41 control circuit may be made to operate such that in, say, the
S ¦ upper 20dB of the dynamic range the limiter provides only that
6 ¦ attenuation required to keep the dominant signal component in
71 the further path held at a relatively constant proportion of
81 that component in the main path signal (e.g. 15 dB below).
9 ¦ As applied to fixed band dual-path circuits, the
lO ¦ invention provides for, in the simplest embodimentl as in the
ll¦ sliding band embodiment, the rectification and smoothing of the
12¦ input or output signal to generate a modulation control signal
13 ¦ that responds primarily to the high level signal components of
14 ¦ the input signal. However, in the case o fixed band circuits a
15 ¦ sharp filter is used in the pass-band control circuit to provide
16 a pass-band control signal. In addition a stop~band control
17 circuit is employed to provide a stop-band control signal. The
l3 modulation control signal provides a reference for the stop band
19 control signal (i.e. opposes it at high signal levels).
The referenced stop-band control signal is compared with the
21 pass-band control signal and the two are combined generally to
22 favor the larger, via a maximum signal selection circu t, which
23 then controls the VCA The overall effect of the circuit is to
24 provide the required attenuation (overall compression law~ in the
pass-band, while avoiding control of the pass-band attenuation
26 by larye signal components in the stop-band, and while avoiding
27 the possibility of excessive amplification of high level signals
28 in the stop-band, as seen at the output of the overall compressor
29 In these and other embodiments in which a reference
bucking signal is generated, the signal may be derived from the
3l input or output because at high signal levels, where the inven-
32 tion operates, the input and output levels are nearly the
I
I ~ ~ame. In som mbodiments the modu1ation control signa1 may be
2 subjected to filtering or equali2ation before rectification.
3 Such equaliæation works in conjunction with the fixed or variable
4 filtering or equalization employed in the signal circuits and
control circuits, to yield an overall modulation control which
6 is most effective in suppressing control by signal components in
7 the stop-band, while at the same time in~erfering as little as
8 possible with control by signal components in the pass-band.
9 Other embodiments of the invention are described
hereinafter. For example, the amplified AC output of the
11 sliding band variable filter may be divided into two or more
12 band pass channels, each channel subjected to selected limiting
13 thresholds, rectified and combined to produce a control signal.
14 By selecting appropriate thresholds, the DC control circuit of
~ the sliding band compressor or expander then has a frequency
16 dependent maximum output characteristic that functions to
17 minimize the control of the compressor or expander by signals
18 outside the slidina pass band.
19 In a variation using only one control circuit channel,
a low ~requency boost circuit is placed in the control amplifier~
21 This is followed by an amplitude limiter and then a high fre
22 ~uency boost circuit. The resulting AC signal is then rectified
23 and smoothed to form the control signal~
24 The invention will be described in more detail, by way
of example, with reference to the accompanying drawings, in which:
26 Figure 1 is an exemplary set of curves showing com~ ¦
27 plementary bi-linear compression and expansion characteristicsO
28 Figure 2 is a schematic circuit diagram of a prior art
29 sliding band compressor.
Figure 3 is a schematic circuit diagram of a prior
31 art sliding band expander.
32 ///
~ -17-
l Figure 4 is a schematic circuit diagram of a modifica-
21 tion to Figures 2 and 3.
3¦ Figure 5 is a block diagram of a prior art sliding
41 band compressor.
51 Figure 6 is a set of probe tone curves illustrating
~¦ the sliding band action of the circuit of Figures 2 and 4.
7 Figures 7 - 10 are a series of probe tone curves
81 illustrating the effects of modulation control according to the
91 invention embodied in a sliding band compressor.
lO¦ Figure 11 is a block diagram of a preferred embodiment
ll¦ f the invention embodied in a sliding band compressor
l~¦ Figures 12 - 15 are block diagrams of further embodi-
13¦ ments o~ the invention embodied in slidin~ band compressors.
14¦ Figures 16 and 17 are block diagrams o a prior
15¦ art fixed band compressor and expander~
16¦ Figures 18 - 20 are response curves illustrating
17¦ the effects oE modulation control according to the invention
18¦ embodied in a sliding band compressor.
~91 Figure 21 is a block diagram of a preferred embodiment
20¦ of the invention embodied in a ixed band co~pressor.
21¦ Figure ~2 is a block diaaram of an alternative
22¦ embodiment of the invention embodied in a Eixed band compressor.
23 ¦ Exemplary bi-linear complementary compression and expan-
24 ¦ sion transfer characteristics (at a particular frequency) are
25 ¦ shown in Figure 1, indicating (for the compression characteristic)
2~ ¦ the low level portion of substantially constant galn, the
27 ¦ threshold, the portion where dynamic actiorl occurs, the inishing
28 ¦ point, and the high level portion of s~bstantially constant gain~
~9 I Details of one dual path sliding band bi-linea
circuit are set forth in Figures 2, 3 and 4~ The sliding band
31 I embodiments of the present invention are described with reference
32 I to this circuit, although the invention is not limited to use in
j -18-
l¦ such circuits. Figures 2, 3 and 4 are the same as Figure 4, 5
21 and 10 respe~tively of US-PS Re 28~426 and further details of
31 said circuits, their operation and theory are set orth therein.
41 Figure 5 is a block diagram of Figure 2 (with or without the
~¦ Figure 4 modification). The following description of Figures 2,
61 3 and 4 is taken in large part from US-PS Re 28,426.
7¦ The circuit of Figure 2 is specifically designed for
8 incorporation in the recording channel of a consumer tape
91 recorder, two such circuits being re~uired for a stereo recorder.
lO¦ The input signal is applied at ~erminal 10 to an emitter follower
ll¦ stage 12 which provides a low impedance signal. This signal is
12¦ applied firstly through a main, straight-through path constituted
13¦ by a resistor 14 to an output terminal 16 and secondly through a
14¦ further path the last element of which is a resistor 18 also
lS¦ connected to the terminal 16u The resistors 14 and 18 add the
16¦ outputs of the main and further paths to provide the required
17 ¦ compression law~
l8 ¦ The further path consists of a fixed filter 20, a
19 ¦ variable cut-off filter 22 including a FET 24 (these constituting
the filter/limiter), and an amplifier 26 the output of which is
2~1 coupled to a double diode limiter or clipper 28 and to the
~¦ resistor 18. The non~linear limiter suppresses overshoots of
23 ¦ the output signal with abruptly increasing input signals. The
24 ¦ amplifier 26 increases the signal in the further path to a level
25 ¦ such that the knee in the characteristic of the limiter or
26 ¦ overshoot suppressor 28, comprising silicon diodes, is effective
27 ¦ at the appropriate signal level under transient conditionsO
28 ¦ The effective threshold of the overshoot suppressor is somewhat
29 above that of the syllabic filter/limiter. The resistors 14 and
18 are so proportioned that the required compensating degree of
31 attenuation is then provided for the signal in the further path~
321 ///
-19 ll
l The output of the amplifier 26 is also coupled to an
2 amplifier 30 the output of which is ~ectified by a germanium
3 diode 31 and integrated by a smoothing filter 32 to provide
4 the control voltage for the FET 24.
Two simple RC filters are used, though e~uivalent LC
6 or LCR filters could be used. The fixed filter 20 provides a
7 cut-off frequency of 1700 Hz (now 1500 Hz), below which diminish-
8 ing compression take place. The filter 22 comprises a serles
9 capacitor 34 and shunt resistor 36 followed by a series resistor
38 and the FET 24, with i~s source-drain path connected as a
ll shunt resistor. Under quiescent conditions with zero signal on
12 the gate of the FET 24, the FET is pinched off and presents
13 substantially infinite impedance; the presence of the resistor
14 38 can then be ignored. The cut-off frequency of the filter 22
is thus 800 Hz tnow 750 Hz), which it will be noted is substan~
16 tially below the cut-of frequency of the ixed filter 20.
17 When the signal on the gate increases sufficiently for
l8 the resistance of the FET to fall to less than say 1 K, the
19 resistor 38 effectively shunts the resistor 36 and the cut-ofE
re~uency rises, markedly narrowing the pass band of the filter.
21 The rise in cut-off frequency is of course a progressive action.
22 The use of a FET is convenient because, within a suit-
23 able restricted range of signal amplitudes, such a device acts
24 substantially as a linear resistor (for either polarity signal~,
the value of which is determined by the control voltage on the
26 gate.
27 The resistor 36 and FET are returned to 2n adjustable
28 tap 46 in a potential divider which includes a temperature
29 ¦ compensating germanium diode 4~. The tap 46 enables the compres-
30 ¦ sion threshold of the filter 22 to be adjusted~
311¦ The amplifier 26 comprises complementary transistors
3~11 giving high input impedance and low output impedance~ Since
Il ,
-20-
~ 3~
1 the amplifier drives the diode limiter 28, a finite output
2 impedance is required and is provided by a coupling resistor
3 500 The diodes 28 are, as already noted, silicon diodes
4 and have a sharp knee around 1/2 volt. I
The signal on the limiter and hence on the resistor 18
6 can be shorted to ground by a switch 53 when it is required to
7 switch the compressor out of action.
8 The amplifier 30 is an NPN ~is$~ with an emitter time
9 constant network 52 givinq increased gain at high freauenciesO
Strong high frequencies ~eag. a cymbal crash) will therefore
11 lead to rapid narrowing of the band in which compression takes
12 place~ so as to avoid signal distortionO
13 The amplifier is coupled to the smoothing filter 32
14 through the rectifying diode 310 The filter comprises a series
resistor 54 and shunt capacitor 56. The resistor 54 is shunted
16 by a silicon diode 58 which allows rapid charging of the
17 capacitor 56 for fast attack, coupled wi~h good smoothing under
13 steady-state conditions. The voltage on capacitor 56 is applied
19 d.irectly t~ the gate of the FET 24.
A complete circuit diagram of the complementary expander
21 is provided in Fiyure 3, but a full description is not required as
22 substantially as the circuit is identical to Figure 2, component
23 values, are therefore not for the most part shown in Figure 3.
24 The differences between Figures 2 and 3 are as follows:
In Fi~ure 3, the further path derives its input from
26 the output terminal 16a, the amplifier 26a is inverting,
27 and the signals combined by the resistors 14 and 18 are applied
28 to the input (base) of the emitter follower 12, -the oukput
291 (emitter) of which is coupled to the terminal 16aO To ensure
low driving impedance, the input terminal 1Oa i5 coupled to the
31¦ resistor 14 through an emitter follower 60. Suitable measures
32~ must be taken to prevent bias getting in the expander~
¦ -21-
,
118!399~i
1¦ The ampliEier 26a is rendered inverting by taking the
21 output from the emitter, instead of the collector, of the second
31 ~PNP) transistor. This alteration involves shifting the 10 K
41 resistor 62 ~Figure 2) from the collector to the emitter ~Figure
51 2~, which automatically gives a suitable output impedance for
I driving the limiter. The resistor 50 is therefore omitted in
71 Figure 3.
81 It should be noted that it is important in aligning
91 a complete noise reduction system to have equal signal levels
10¦ on the emitters of the transistors 12 in both compressor
11¦ and expander. Metering terminals M are shown connected to
12¦ these emitters.
13 Figure 4 shows a preferred circuit, for replacing the
14 circuit between points A, B and C in Figures 2 and 3. When the
FET 24 is pinched off~ the second RC network 22 is inoperative,
16 and the first RC network 20 then determines the response of the
17 further path. The improved circuit combines the phase advantages
18 of having only a single RC section under quiescent conditions
19 with the 12 dB per octave attenuation characteristics of a
two-section RC filter under signal conditions.
21 In the practical circuit, using MPF 104 FE~'s, the 39 K
22 resistor 36a is necessary in order to provide a finite source
23 impedance to work into the FET. In this way the compression
24 ratio at all frequencies and levels is held to a maximum of
about 2u The 39 K resistor 36a serves the same compression
26 ratio limiting function in the improved circuit as the resistor
27 36 in the circuit of Figure 2 or Figure 3. In addition, this
28 resistor provides a low frequency path for the signal. I
29 Certain details of the circuit oE Figures 2, 3 and 4
have evolved over the years and more modern forms of the circuit
31 have been published and are well known in the ar-t. Reference to
321 ///
~ -22--
l the specific circuit in US-PS Re 28,426 is made for convenience
2 in presentation.
3 Figure 5 is a block diagram showing the major elements
4 of the compressor of Figures 2 and 4. Combining circuit 15
S represents the combining resistors 14 and 18 of Figures 2 and 3.
b The variable band action of the sliding band device
7 can be seen in Figure 6, showing an actual chart recorder probe
8 tone response obtained from the circuit of Figure 2 incorporating
9 Figure 4. The variable band action is shown by plotting the
compressor frequency response by means of a low-level probe tone
ll (the level of which is below the compressor threshold) in the
12 presence of a high-level signal; the probe is detected at the
13 compressor output by means of a tracking filter. The high-level
14 signal causes the compressor circuitry to operate, the graph
showing the effect on the turnover frequency of the filter.
16 In a sliding band device in accordance with the
17 invention, the amplitude of the high level or dominant signal
18 that causes the sliding band action should not cause excessive
19 sliding, nor should the presence of other high level signals
outside of the sliding band pass band cause excessive sliding.
21 Excessive slidinq means movement of the variable filter turnover
22 fre~uency farther than necessary to produce a sliding band
2~ compressor characteristic which avoids boosting the dominant
24 signals above a reference level. The absolute value of the
reference level is chosen by the system designer, but is usually
26 some 1OdB below the highest levels normally used.
27 Figure 7 shows a further se~ of actual chart recorder
281 probe tone curves for the case of a sliding band compressor
29 circuit similar in design to that of Figure 2 (with the Figure 4
30 I modification), but with a low level gain of 8 dB and a filter
31 quiescent fre~uency of 800 Hz. The probe tone level is at -40
32 dB, below the compressor threshold. Curves are taken for a
-23-
1 100 Hz signal at -20~ -10, 0, ~10 and +20 dB9 where 0 dB is the
21 reference level. Also, a curve for no 100 Hz signal is shown.
3¦ The ~10r 0, ~10 and +20 dB chart recording curves are all started
41 at about 200 Hz. This is also the case for Figure 8~ In
51 Figures 9 and 10 there are also curves for the no signal condition
~¦ Referring again to Figure 7, ideally~ there should be
71 no sliding in response to a 100 ~z signal because it is well
81 outside the pass band of the circuit at its lowest (quiescent)
91 frequency. Nevertheless, as the 100 Hz signal increases in
10¦ level, the band slides upward. The -10, 0, ~10 and ~20 dB
11 curves need not slide any farther than the -20 dB curve in order
12¦ to avoid any substantial boosting of the 100 Hz signal. The
13¦ unnecessary sliding has two effects: a) substantial noise
14¦ reduction action is lost (during playback) because no boosting
15¦ takes place at frequencies where it otherwise can take place and
16¦ b) as the amplitude of the 100 Hz signal varies it can modulate
17¦ signals at higher frequencies as the sliding band varies under
18¦ its control~ resulting in possible incorrect restoration of the
19¦ signal by the expander if the recording or transmission channel
20¦ has an irregular frequency response in the vicinity of l00 Hz.
21¦ Figure 8 shows a set of actual chart recorder probe
22¦ tone curves for the same circuit, but with the addition of
231 modulation control circuitry as described hereinafterO Essen-
?4¦ tially no sliding occurs for the same levels o~ the 100 ~æ
25¦ signal as in the Figure 7 arrangement. The sliding band
261 compressor is made essentially immune to strong signals outside
27 its pass band. The sliding band response is essentially the
28 ¦ same as its response below threshold in the presence of no
29¦ dominant signals.
30¦ The effect of modulation control for sliding band
31 ¦ compressors is further illustrated by Figures 9 and 10, which
32 are also actual chart recorder probe tone curves taken with the
1 ~2~
l l
l l
l ¦ same circuit and probe tone level as with Figures 7 and 8~ ln
~ ¦ this case, the effect of a dominant signal at 800 Hz, a fre-
3 ¦ quency within the desired active area of the circuit, is shown
4 ¦ Ideally, slidinq is required to go only so far as not to boost
5 ¦ the 800 Hz signal above the 0 dB reference levelO Thus, in the
6 ¦ Fi~ure 9 response, without modulation control, the sliding71 produced by ~he 800 Hz signal at levels of -10, 0, ~10 and +20
81 dB are excessive. Figure 10 illustrates the response of the
91 circuit with modulation control: sliding at and above 0 dB is
l~¦ greatly reduced. The effect is progressively reduced for low
lll signal levels but is o~servable to some extent at the -10 dB
12¦ signal level~
13¦ Figure 11 shows generally a preferred embodiment of
14¦ the modulation control of the present invention embodied in a
15¦ dual path bi-linear sliding band device. Reference numerals
16¦ are, so far as possible, kept the same as in Figure 5 for the
17¦ same and functionally similar elements. The probe tone response
l~¦ curves of Figures 7 10 are taken from a sliding band device
l9¦ generally as shown in Figure 1l, with the modulation control
20¦ sub-circuitry elements within dashed block 100 taken out of
21¦ the circuit for the response curves without modulation control.
22¦ For purposes of explanation, the detailed circuitry of Figure 11
231 is essentially the same as that of Figures 2 and 4. The circuit
24 may be modified as described hereinbefore witho~t affecting the
251 basic operation of the modulation control sub-circuit.
26¦ As shown in Figure 11, the modulation control sub-
~71 circuit derives a DC control signal from the circuit input (or~
28¦ optionally from the output of combining circuit 15) by means of
291 an amplifier 30'~ rectifier 31' and smoothing circuitry 32a'0
301 Potentiometer 102 is shown to indicate that the signal from
31¦ smoothing circuitry 32a' has a controlled gain. In practice
32~ ///
l -25-
1181~
1¦ the gain is usually pre-set in the design. A combining circuit
21 33 subtracts the signal provided by the sub-circuit 100 from the
31 main control signal provided by way of the amplifier 30, rectifier
4¦ 31 and s~oothing circuit 32a'a
51 The smoothing circuitry of Figure 11 is broken into
61 two stages in order to minimize the cost of circuit components~
7 Thus, blocks 32a and 32a' may be identical and each may comprise
~¦ only a single RC filter section and block 32b which further
91 smooths the combined control signal comprises a further RC
10¦ filter sectionO
11¦ The signals are rectified to DC (by rectifiers 31 and
12¦ 31') before they are combined by the circuit 33 in order to
,31 avoid the polar-ity ambiguity that would result if AC signals
14¦ were combined and then rectified (i.e., with AC signals there
15¦ would be two possible stable states).
16¦ The arrangement of the embodiment of Figure 11 thus
17¦ provides a reference level for stabilization of the DC control
18¦ signal, a reference level that is dynamically changing with
19¦ input signal level, thereby shifting or transposing part of the
20¦ dynamic action of the variahle filter to a level region deter-
21¦ mined by the reference level~ The arrangement functions to keep
~¦ the maximum amplitude of c]ominant signals in the noise reduction
231 side path at a constant proportion of the input signal at high
24 ¦ signal levels~ The relative level from the modulation control
25 ¦ sub circuit 100 is selected to minimize sliding action in
26 ¦ response to signals outside the sliding band pass band~
27 ¦ Although the embodiment of Figure 11 functions
28 ¦ effectively when the input to the modulation control sub-circuit
2~ ~ 100 is taken from the wide band inpu-t (or output), other arrange-
30 I ments giving a measure of signal levels at the top end of the
31 dynamic range are possible~ For example, some modulation
32 ///
~ -26-
l ~,
.$~
control 0fects are obtained even if the sub-circuit lOO input is taken from
the output of band-pass filter 20. Ideally, equalization is employed in
ampli~iers 30 and 30' to optimi~e the overall modulation control effects
(control by pass-band components versus by stop-band components), taking
into account the combined frequency response effects of filters 20, 22, and
the equalization employed in control amplifier 26.
When the invention is embodied in series connected devices such as
set forth in selgian-PS 889~428J a single modulation control sub-circuit may
be used to provide a reference signal to each stage. Such a circuit advailt-
ageously derives its input from the output of the last compressor stage whenthe series stages are arranged in the preferred order such that the first
stage has the highest level threshold. By deriving the reference signal
from the output, the low level stage(s) receive the modulation control
effect at lower signal levels, thus enhancing the modulation control action.
As mentioned previously, it is also possible to achieve modulation
control o:E sliding band circuits by other means than by deriving a control
signal reference from the input (or output) signal. One or more control
signals can be deri.ved from the variable filter output and limited so as to
achieve results similar to those achieved by the bucking embod:iment o:E
~igure 11; the essential result is the same, namely to de-sensitize the
dynamic modification action of the circuit to high level signals with;n the
stop-b~md. Figures 12, 13, 14 cmd 15 are directed to SUC]l elllbOd:illlents
employing limitlng.
In the embodiment of Figure 12, the control signal generating
means (blocks 30, 31 and 33 in Figure 5) is split into three paths by ampli-
iers 30, 116, and 124 cmd filters 110~ 118, and 126, namely a high fre-
quency path, a mid-frequency
, ~. . ;,.
l l
l path and a low frequency path Each path incl~des a limiter
21 (112, 120, 128) that has a pre-set thresholdO The limiters can
3¦ be back to back diodes such as diodes 28 in Figure 2. For a
41 high frequency audio compressor having a perEormance generally
5¦ as shown in Fi~ures 7 to 10, the filter frequencies may be as
6 follows9 for example: filter 126, 200 Hz low pass, filter 118,
7l 200 to 800 Hz band pass; and filter 110, 800 Hz high pass. The
81 output of each limiter is rectified by rectifiers 114, 122, and
~¦ 130, combined (or maximum value selected) and applied to smooth
l0¦ ing network 32. Alternatively, the ]imiting Eunctions can be
ll¦ provided after rectificationO In operation, the low frequency
12¦ and mid-frequency band limiters are set to minimize the effect
13¦ on sliding by signals outside the pass band. Little or no
14¦ limiting may be re~ulred in the high frequency path, and the
15¦ control effected by this path may be enhanced by providing the
~6l amplifier 30 with high frequency boost, as represented by block
171 52.
l~ Figure 13 shows a further split path control circuit
19¦ embodiment. In this example~ two paths are employed, a high
20¦ frequency path and a low frequency path. The high frequency
21¦ path is essentially the ~ame as in the embodiment of Figure 12,
22¦ except that the limiter 112 is omitted. The low frequency path
231 has an amplifier 132 that has a high frequency attenuation
24¦ networ~ 134. The amplifier output is applied to a low pass
25 ¦ filter 136 and to a limiter 138. The limiter threshold is
26 I set along with the various filter and amplifier filter character- !
27 ¦ istics to achieve the best immunity from sliding band control by
28 ¦ stop-band signals~ The signals in the two paths are rectiEied
29 ¦ by rectifiers 114 and 140 and combined at the input to the
30 ¦ smoothing circuit 3~
31 ¦ A further simplified embodiment of the Figure 13
32 I embodiment is shown in Figure 14. The high pass filter 110,
l -28-
I
~-
l¦ the low pass Eilter 136 and the amplifier hiqh frequency
21 attenuation network 134 are omitted. The high frequency pre-
31 emphasis network 52' of amplifier 30 is modified from tha~ of
4¦ network 52 such that the hiqh frequency boost becomes effective
51 at a hi~her fre~uency. Consequently only the wide band path
61 containing amplifier 132 carries low fre~uency signals (along
7~ with high fre~uency signals). The threshold of limiter 138 is
81 adjusted along with the high frequency boost characteristics of
91 network 52' to minimize the effect on sliding by stop-band
signals~
ll Figure 15 shows an embodiment having a single path
12 control circuit which includes a frequency dependent amplifier
13 141 having a low frequency boost network 142, followed by a
14 limiter 144 and an amplifier 146 with a hi~h frequency boost
network 148. In operation the low frequency portion of the
16 spectrum which tends to cause undesirable sliding is first
17 boosted and then limited. Limiter 144 is preferably syllabic
18 with its own closed loop amplifier, rectifier, smoothing circuit
19 and controlled gain element (such as blocks 276, 280, 282 and
270 in Figure 17). Amplifier 146 having a high frequency boost
21 network 148 restores any high frequency pre-emphasis which may
22 be required. The amplifier 146 output is then rectified and
23 smoothed by blocks 114 and 32y respectively. In this single
24 path control circuit the high level stop-band signal components
are significantly reduced at the rectification point 114n
26 For convenience and simplicity the sliding band
27 embodiments have been described in connection with a particular
28 configuration of sliding band compressor. The invention is
29 equally applicable to expanders, with no change in the noise
reduction further path control circuits shown in the embodiments
3l of Figures 11 - *~. In noise reduction systems employing
32 compressors and expanders, it is preferred that the modulation
-29~
~ ',
1 control invention be applied to both devices to assure com-
2 plementarity. The invention is also equally applicable to low3 frequency sliding band circuits, in which the compression and
4 expansion action is designed to occur in the low frequency
region~
6 Figure 16 shows a block diagram of a fixed band dual
7 path bi-linear compressor and expander conEiguration. The
~ Eundamental aspects of this system are disclosed in US-PS
9 ~r~467S~-, US-PS 3,903,485 and in Journal of the Audio Engineeriny
Society, Vol. 15, No. 4, October, 1967, pp. 383-388.
llIn the known embodiment of Figure 16, the further path
12 networks 250 provide four bands. Bands 1, 3 and 4 have conven-
13 tional 12 dB/octave input filters: an 80 Hz low pass filter 252
14 at the input of band 1, a 3 kHz high pass Eilter 254 at the
15input of band 3 and a 9 kHz high pass filter 256 at the input of
16 band 4. Each is followed by an emitter follower isolation stage
17 258. Band 2 has a frequency response which is complementary to
18 that of bands 1 and 3. Such a response is derived by adding tin
19 adder 260) the outputs of the emitter followers 258 in bands 1
and 3 and subtracting that sum from the overall input signal (in
21 subtractor 26~)~ The output of emitter follower 258 in each
22 band and the output of subtractor 262 are applied to respective
23 limiters 264 and 2641 ~ Limiters 264 and 264' are identical
24 except that limiters 2641 in bands 1 and 2 have time constants
twice those in bands 3 and ~. The outputs of bands 1-4 are
26 combined with the main path signal in combiner 266r The com-
27 pressor output is applied to a noisy channel for transmission to
28 the complementary expander in which the output of the identical
29 further path networks are subtracted Erom the input signal to
provide the complementary expansion characteristicO
31 Figure 17 shows further details of the limiters 264
32 and 264'. Each includes an FET attenuator 270 that operates
~ ~30-
I
l in response to a control signal. The attenuator output is
2 amplified by si~nal amplifier 272, the gain of which is set
3 to provide the desired low level signal gain~ The outputs
4 of all the bands are combined with the main signal in such a
way as to produce a low level output from the compressor which
6 is uniformly 10 dB higher than the input signal up ~o about
7 5 kHz, above which the increase in level rises smoothly to
8 15 dB at 15 kHz.
9 The FET attenuator is controlled by a control signal
sub-circuit that provides a compression threshold of 40 dB below
ll peak operating level. The control sub circuit includes control
12 signal amplifier 276 followed by a phase split~er 278 which
13 drives a full wave rectifier 280. The resulting DC is applied
14 to a smoothing network 282, the output of which is the control
signal. Network 282 includes an RC pre-integrator, an emitter
16 follower and a final RC integrator that operate in con~unction
17 with diodes such that both the pre- and final integrators have
l8 non-linear characteristics produced by the diodes. Fast,
19 large changes in signal amplitude are passed quickly, whereas
small changes are transferred slowly. This dynamic smoothing
21 action produces optimum results with respect to modulation
22 effects, low frequency distortion, and distortion components
23 generated by the control signal. The circuit achieves both fast
24 recovery and low signal distortion.
Figure 18 shows an actual chart recording plot of
26 response below the compression threshold of a fixed band
27 compressor having a low level gain of 8 dB and a pass-band
28 filter frequency of 800 Hz high pass~ Boost is provided within
29 the active frequency area of the device (determined by the 800
Hz corner frequency) up to levels of about -10 dB (with respect
31 ¦ to a 0 dB reference level).
32 ~ ///
-31-
I ,
9~6
l Figure 19 shows the effect on compression when a high
2 ¦ level signal (+10 dB) is presen~ at 100 Hz, which is well below
3 ¦ the 800 Hz filter corner frequency. The strong 100 Hz signal
4 ¦ in the stop-band effectively blocks the compressor and prevents
5 ¦ any compression within the pass-band. Consequently, desired
6 ¦ noise reduction in the pass~band is lost. In addi-tion, if the
7 ¦ 100 Hz signal is intermittent, compression in the pass-band will
8 ¦ come and go with the controlling 100 Hz signal causing noise
9 ¦ modulation and/or signal modulation.
l0¦ Figure 20 shows the effect of the addition of a
ll¦ modulation control sub-circuit, described hereinafter, to a
12¦ fixed band circuit. Compression is restored to the pass-band
13¦ area even in the presence of the strong (~10 dB) signal at 100
14¦ Hz~ The modulation control sub-circuit effectively makes the
15¦ fixed band circuit immune to the strong stop-band signal.
16¦ Figure 21 shows generally the preferred embodiment
17¦ of the invention as applied ~o one band of a fixed band dual
18~ path bi~linear compressor of the type described in connection
19 with Figure 16~ Two additions are made to the circuit in order
2Q¦ to provide modulation control. A modulation control sub-circuit
21¦ 198, similar to that in the sliding band embodiment of Figure l1
22¦ is provided, which includes a rectifier 208' and a first sta~e
23 of smoothing 21Oa'. The modulation control optionally may be
241 fed from the output of the compressor. Elements 208, 208' and
25~ 210a~ ~10a' may be identical (but separate). The level of the
26¦ modulation control signal from smoothing circuit 210al is set by
271 attenuator 212 or some other suitahle means and is combined by
2~1 circuit 214 in opposite polarity with the stop~band DC control
291 si~nal from smoothing circuit 210a. In addition the output of
30l¦ VCA 204 and amplifier 206 is applied to a filter 216 which
3111 preferably has the same corner frequency as filter 202, although
321 ///
l -32-
l ll
1 this is not essential; the comparative graphs Figures 19 and 20
21 were made ~ith a simple 6dB/octave 3kHz low-pass filter 2160
31 Nevertheless, filter 216 should ideally have a relatively steep
4 cutoff characteristic, such as 12dB or 18dB per octave (e.g. f a
51 2 or 3 pole filter) with about the same cutoff fre~uency as
61 filter 202. The filter 216 output is rectified and smoothed by
blocks 218 and 220-, to form the pass band control signal. The
I smoothing provided by blocks 210a, 210a' and 210a'~ may be a
9 preliminary filtering stage followed by further smoothing in
circuit 210b~ The output of the pass-band filter channel is
11 applied to maximum selector 222 that receives at its other input
12 the output of combiner 214 the modulation controlled stop-band
13 control signal. In its simplest form the maximum selector
14 comprises two diodes which pass the larger of the two input
signals in more sophisticated circuits, operational amplifiers
16 are employed to eliminate the diode voltage drops and to increase
17 accuracy.
18 In operation, signals in the s-top-band are subject to
19 the action of the sub circuit 198 if there are no dominant
signals inside the pass band where compression action is desired.
21 Thus, although a strong signal such as that of +10 dB at 100 Hz
2~ causes a large control signal to be generated by blocks 208 and
23 210a (and 210b), that control signal is bucked by the modulation
24 control sub-circuit signal so that the VCA 204 gain is not
driven down to cause a loss of compression in the pass-band.
26 If a signal of 1 OOH2 occurs in the level region of -20dB~ on the
271 other hand, the bucking action is greatly reduced, and the
28~ stop-band control signal then appropriately controls the action
291 of the compressor whenever signal conditions are such that the
30 ¦ pass-band control signal is not controlling the compressor. If
31 ¦ strong signals are present within the active area pass band, the
32 ~ ///
-33-
1 output of the sharp filter channel, the pass-band control
2 circuit will control the maximum selector and allow the VCA to
3 react accordingly~
41 The level of the modulation control sub ~ircuit
~¦ relative to the input or output is set to provide a dynamic
61 reference signal (relative to the input) of sufficient level to
71 result in substantial immunity of the compressor action to
81 strong out of pass band signals.
9 Comments made regarding equalized control and modula~
10¦ tion control amplifiers in reference to slidinq band circuits
11 are also applicable to fixed band embodiments. Thus, optionally,
~ filter/equalizers 224 and 226 may be inserted in the respective
13 paths to rectifiers 208' and 208. However, the opportunities
14 for advantageously working one frequency dependent characteristic
against another in the fixed band case are less than with
16 sliding bands; indeed, this is why an extra control circuit is
17 required in the fixed band case (3 circuits versus 2).
18 It is also possible to achieve modulation control of
19 fixed band circuits hy other means than deriving a control
signal reference from the compressor or expander input (or
21 output) signal. One or more control signals can be derived from
22 the controllable element (attenuator or VCA) output and limited
23 so as to achieve results similar to those achieved by the
24 bucXing embodiment of Figure 210 Figure 22 is directed to such
limiting embodimentsO
26 In the embodiment of Fi~ure 22, the control signal
7 generating means (blocks 276, 278, 280 and 282 in Figure 17) is
c~
28 split into two paths, one having ~ amplifier 228, a sharp
29¦ cutoff filter (as in the Figure 21 embodiment) and a rectifier
30 1 218 and the other having an amplifier 230, a limiter 232 and a
31 ¦ rectifier 218~n The threshold of limiter 232 (which can be back
32 ///
~ -34-
I
I 1,
l to back diodes, for example~ is selected such that limiting
2 action begins at relatively high levels, at about the same
3 ¦ levels at which the output from combiner 214 begins to become
4 predominant in the embodiment of Figure 210 The outputs of
51 rectifiers 218 and 218' can be combined and applied to smoothing
61 circuit 210, the output of which is applied as the control
71 signal to VCA 204 or the rectifier outputs can be applied to (or
8 serve as) a maximum selector circuit (such as block 222 in
9 Figure 21) and its output applied to smoothing network 210.
In operation, the embodiment of Figure 22 functions in
ll a similar manner to the Figure 21 embodiment
12 ///
13 ///
14 ///
///
16 ///
17 ///
18 ///
19 ///
///
21 ///
22 ///
23 ///
24 ///
///
26 ///
27 ///
28 ///
29
30 ~ ///
31 I ///
3~ ///
-3S-
I 1~
