Note: Descriptions are shown in the official language in which they were submitted.
21 690~9
LOUDSPEAKER OVERLOAD PROTECTION
DESCRIPTION
TECHNICAL FIELD
The invention relates to loudspeakers, especially loudspeakers which use negative
5 feedback derived by monitoring actual displacement of the loudspeaker's electroacoustic
tr~nsdllcer or cone. The invention is especially, but not exclusively, applicable to so-
called "sub-woofer" loudspeakers.
BACKGROUND ART
In order to achieve satisfactory reproduction of the useful audio range, it is usual
to use multi-speaker systems. In some systems, "sub-woofer" loudspeakers designed to
handle the lowest frequencies, say from about 20 Hz to about 150 Hz (-3dB points), are
housed in separate enclosures and powered by their own power amplifiers. At the lower
frequencies, say 20 Hz to 40 Hz, the electroacoustic transducer, usually a cone and voice
15 coil combination, travels a significant distance, perhaps as much as one inch peak-to-
peak. Non-linearities in the cone suspension system and non-uniform magnetic flux in
the gap within which the voice coil travels can lead to acoustical distortion. Also, the
relatively high power levels involved may introduce non-linearities due to heating of the
voice coil.
It is known to compensate for these non-linearities, and equalize frequency
response, by deriving a signal related to motion of the loudspeaker cone and using it to
provide negative feedback to a power amplifier driving the loudspeaker. Such "servo"
loudspeakers are susceptible, however, to distortion arising from overload conditions.
When an overload occurs, and the loudspeaker cone reaches the limit of its range of
25 movement, or the power amplifier "clips", unwanted high frequency components may
be introduced into the loudspeaker output. This is particularly undesirable in powered
sub-woofers. Low frequency signals are not considered to be localizable, so the sub-
woofer speaker need not be close to the mid-range speakers and "tweeters" but may be
placed anywhere in the listening area. These higher frequency units will determine the
30 a~al~nt position of the sound source. They may be located, for example, so as to
position the sound source behind a home theater screen. High-frequency harmonicsproduced by "bottoming" of the loudspeaker core and/or clipping of the power amplifier
may cause an appalent shifting of the sound source, which could be distracting for the
21 6~029
, ~
listener.
The problem is particularly serious in "servo" loudspeakers because the motion
sensor and feedback circuit will detect that the loudspeaker cone is not moving in
proportion to the audio input signal and try to increase the amplifier gain, thereby
5 increasing the clipping.
In servo loudspeaker systems, the amplified signal includes out-of-phase speakernon-linearities. Clipping by the amplifier would compound these non-linearities,resulting in rather unpleasant distortion of the reproduced signal.
In loudspeakers which do not use negative feedback, the gain of the power
10 amplifier is known, so overload protection may be provided by static limiters and
compressors set to apprupliate levels and placed at the input of the loudspeaker drive
circuitry. In the case of servo loudspe~kers, however, overload protection is more
complicated because sensor sensitivity, which affects closed loop gain, may vary, for a
particular sensor, with temperature and time; and the rate of such variations may be
15 different for different sensors. Moreover, the servo loop will automatically equalize the
loudspeaker, resl-ltin~ in different gains at different frequencies. Difficulties may also
result from differences in sensitivity, frequency response and Q-factor from oneloudspeaker to another from the same production line; voice coil heating increasing voice
coil resi~t~nce; and lack of predictability of output peak power in view of the non-
20 linearities in the amplitude of the ampliher signal. It is not commercially viable to
design a loudspeaker unit with sufficient excess capacity that overload conditions will not
occur.
DISCLOSURE OF THE INVENTION:
2 5 The present invention seeks to elimin~te, or at least mitig~te, the above-mentioned
disadvantages of known servo-loudspeakers. To this end, according to the presentinvention, a loudspeaker unit comprises an electroacoustic tr~n~dllcer, input means for
an audio signal to be reproduced by the electroacoustic tr~n~ducer, a power amplifier
connected between the input means and the electroacoustic tr~n~ducer, sensor means for
3 o providing a motion signal in dependence upon motion of the electroacoustic tr~n~ducer,
feedback means for providing a feedback signal in dependence upon the motion signal,
and subtracting means connected between the input means and the power amplifier for
subtracting the feedback signal from the audio signal and supplying the reslllting
~1 6~029
difference signal to the power amplifier, the loudspeaker unit further comprising control
means for providing a first gain control signal and a second gain control signal in
response to the audio signal amplitude e~cee(ling a first predetermined level and a second
predetermined level, respectively, first variable gain amplifier means connected between
5 the input means and the subtracting means and responsive to the first gain control signal
to decrease amplification of the audio signal and second variable gain amplifier means
connected within the feedback loop comprising the subtracting means, power amplifier
and feedback path, the second variable gain amplifier being responsive to the second gain
control signal to decrease gain within the feedback loop, the arrangement being such that
10 the rate of change of gain of the first variable gain amplifier means in response to the
audio signal excee(ling the first predetermined level is significantly slower than the rate
of change of gain of the second variable gain amplifier means in response to the audio
signal excee~ling the second predetermined level.
In a preferred embo.liment, the means for providing the first and second gain
15 control signals comprises means for deriving a signal proportional to audio signal
amplitude, first deriving means for providing said first gain control signal in proportion
to the difference between the proportional signal and the first threshold, and second
deriving means for providing said second gain control signal in proportion to the
difference between the proportional signal and the second threshold.
2 o Each of the first and second deriving means may comprise a threshold comparison
circuit and a charging circuit connected to the output of the threshold comparison circuit.
The charging circuit of the second deriving means may have a charging time constant
that is shorter than the charging time constant of the charging circuit of the first deriving
means, and also signifis~ntly less than a period of the audio signal. The second charging
25 circuit may have a discharging time constant significantly shorter than that of the first
charging circuit.
Various objects, features and advantages of the invention will become apparent
from the following description, taken in conjunction with the attached drawings, of an
embodiment of the invention which is described by way of example only.
BRIEF DESCRIPTION OF DRAWINGS
Figure 1 is block schematic cli~gr~m of a servo-loudspeaker unit; and
Figure 2 illustrates a sample output waveform produced by the unit.
21 690~
BEST MODE FOR CARRYING OUT THE INVENTION:
The powered loudspeaker unit shown in Figure 1 comprises an electroacoustic
tr~n~ducer unit 10 in the form of a cone 12 driven by a voice coil (not shown) within a
5 magnet assembly 14. The voice coil is supplied by a power amplifier 16 connected to
a suitable mains power source which, for simplicity, is not shown in the drawings. An
audio input signal for reproduction by the electroacoustic tr~n~ducer 10 is supplied to the
power amplifier 16 by way of, in succes~iQn, a preamplifier 20, a first variable gain
device in the form of voltage controlled amplifier 22, a second variable gain device in
10 the form of voltage controlled amplifier 24 and a compensation filter 26. Thecompensation filter 26 is of known kind with characteristics chosen so as to avoid
positive feedback at some frequencies.
A motion sensor 28, for example an accelerometer, is mounted upon the voice
coil 12 and derives a signal in dependence upon motion of the cone, specifically15 proportional to its acceleration. The cone acceleration signal is amplified by a first
amplifier 30 and applied to one input of a summin~ device 32. A first integrator 34
integrates the cone acceleration signal to produce a cone velocity signal which is
amplified by amplifier 36 and applied to a second input of the sllmming device 32. A
second inl~l~tor 38 connected to the output of first integrator 34 integrates the cone
20 velocity signal again to provide a cone position signal, which is amplified by a third
amplifier 40 and applied to a third input of the sllmmin~ device 32. Hence, the output
of the summing device 32, i.e. the feedback signal, comprises three components
pl~,pollional to acceleration, velocity and position, respectively.
This feedback signal is applied to a second s~lmming device 42 connected between25 voltage controlled amplifiers 22 and 24, respectively. The second sllmming device 42
subtracts the feedback signal from the audio signal applied to the input of second voltage
controlled amplifier 24 and supplies the resulting difference signal to the compensation
filter 26 and thence to power amplifier 16. Hence, second voltage controlled amplifier
24 is included in the feedback loop, and controls loop gain, whereas first voltage
30 controlled amplifier 22 precedes the feedback loop and controls the amplitude of the
audio signal applied to the input of the feedback loop. The loop gain affects the amount
of available gain equalization and distortion reduction.
First and second gain control signals for controlling the respective gains of
2 1 6~029
voltage controlled amplifiers 22 and 24, respectively, are provided by means of a
precision bridge rectifier 44 connected to the output of the second voltage controlled
amplifier 24, threshold amplifiers 46 and 48 and charging circuits 50 and 52. The
rectifier 44 rectifies the bipolar audio signal to produce a unipolar signal which is applied
5 to respective inputs of first threshold amplifier 46 and second threshold amplifier 48.
The outputs of threshold amplifiers 46 and 48 are applied via charging circuits 50 and
52, respectively, to first voltage controlled amplifier 22 and second voltage controlled
amplifier 24, respectively. The outputs of charging circuits 50 and 52 are the first and
second gain control signals Cl and C2, respectively.
The threshold levels and charging circuit time constants are set so that reduction
of the gain of the first voltage controlled amplifier 22 takes place at a slower rate and
begins at a lower signal amplitude than the reduction of the gain of second amplifier 24.
Thus, threshold level THl of first threshold amplifier 46 is selected such that the output
of the threshold amplifier 46 will be non-zero when the output of power amplifier 16 is
15 more than a corresponding amplitude VTHI which is, say, 0.90 of clipping level, i.e.
approximately - 1 dB with respect to the amplitude at which the output of power amplifier
will start clipping.
Threshold level TH2 of second threshold amplifier 48 is set such that the outputof threshold amplifier 48 will be non-zero when the output of power amplifier 16 is more
20 than a corresponding amplitude VTH2; say 0.95 of clipping i.e. -0.5 dB with respect to
the clipping level. These threshold values could differ so long as threshold TH2 is
greater than threshold THl.
The charging time constant or attack time T2a of charging circuit 52 is much less
than the charging time constant or attack time Tla of charging circuit 50; and also much
25 less than the period of the audio input signal. For example, for a 100 Hz. audio input
signal having a period of 10 milli~econds, T2a is 100 microseconds and Tla is 75milli~econds. As a result, the first control signal Cl will increase more slowly than
second control signal C2.
The discharging time constant or release time T2r of charging circuit 52 is less30 than the discharging time constant Tlr of charging circuit 50 and the discharging time
constant or release time Tlr of charging circuit 50 is greater than its own charging time
constant Tla. For example, Tlr may be 5 seconds and T2r may be 50 milli~econds.
Operation of the circuit will now be described with reference also to Figure 2.
21 6~02q
When the amplitude of the audio input signal is within its normal range, such that the
output V~,p of power amplifier 16 is less than VTHI~ the output of the rectifier 44 is less
than both thresholds THl and TH2. Consequently, both control signals Cl and C2 are
zero and the gain of each of the voltage controlled amplifiers 22 and 24 is at a5 maximum. At point A in Figure 2, the amplitude of the audio input signal increases
suddenly to such a level that, with the gain unchanged, the corresponding output V"",p of
power amplifier 16 would exceed the clipping level. However, as the output V""p of
power amplifier 16 ~xcee(ls the lower predetermined level VTHI, at point B in Figure 2,
the output of rectifier 44 exceeds the threshold THl, the output of threshold amplifier
lo 46 increases and control signal Cl begins to increase, causing the gain of voltage
controlled amplifier 22 to decrease. Rec~lse the control signal Cl changes relatively
slowly~ the output voltage V""p will continue to increase until it exceeds the level VTH2
and the output of rectifier 44 exçee~l~ upper threshold level TH2. At this point,
de~ n~tPd C in Figure 2, the second control signal C2 will increase rapidly c~ in~ the
15 gain of second voltage controlled amplifier 24 to decrease rapidly. As a result of this
rapid decrease in gain, the output Val"p of power amplifier 16 will exceed the level VTH2
by only a small amount, and will not reach the clipping level.
After an interval corresponding approximately to the charging time constant T2a
of charging circuit 52, the amplitude of V~,n,p falls to, about, the level VTH2, as indicated
20 at point D in Figure 2, and the output of rectifier 44 falls to, about, the threshold value
TH2. During this time, the gain of VCA 22 continues to decrease. A point in time is
reached when the gain of VCA 22 and the gain of VCA 24 are such that V,~",p falls to just
below VTH2. Charging circuit 52 then begins to discharge, causing control signal C2 to
fall rapidly and restore the gain of voltage controlled amplifier 24 to its original value,
25 as at point E of Figure 2. During the interval between point C and point E, which is
proportional to the discharging time constant T2r of charging circuit 52 plus the charging
time Tla, the level of feedb~k is reduced. Full feedback is restored at point E.Rec~llse the peak amplitude of power amplifier output signal V""p remains above
lower level VTHI for a few more periods of the audio signal, the gain of first voltage
3 o controlled amplifier 22 continues to decrease. The gain of voltage controlled amplifier
22 will continue to fall until V""p falls to about VTHI (point F in Figure 2). At this point,
the amplitude of the output signal V"",p of power amplifier 16 is stable at about VTHI
while overload conditions persist. Hence, during overload conditions, the steady-state
21 6902~
amplitude of output signal V~",p from power amplifier 16 is controlled by voltage
controlled amplifier 22 and tr~n~içnt amplitudes in excess of the steady state are
controlled by voltage controlled amplifier 24.
If the input signal amplitude falls to a lower level, as indicated at G in Figure 2,
5 the output of rectiher 44 will fall below the threshold THl and charging circuit 50 will
begin to discharge causing control signal Cl to decrease and the gain of voltagecontrolled amplifier 22 to increase. In view of the relatively long discharge time
constant of charging circuit 50, the power amplifier output Van,p is only restored to
normal after several periods of the audio signal, as at point H of Figure 2.
o It will be appreciated that, during the time interval between point C and point E,
the output signal is not a reproduction of the input signal. Nevertheless, the overload
condition is controlled without the power amplifier clipping, so distortion is reduced.
The present invention encomp~sçs various modifications to the above-described
embodiment. For example, voltage controlled amplifiers 22 and 24 could be replaced
15 by operational transconductance amplifiers or other variable gain devices whose
output/input gain, whether voltage/voltage, current/current, voltage/current or
current/voltage, can be controlled by a control signal (voltage, current or other.)
It should also be appreciated that the circuit as described has been simplified to
facilitate the description. As shown in broken lines in Figure 1, other circuitry 54, such
20 as low pass and high pass filters and buffers, might be interposed between the first
variable gain element 22 and summing device 42. Also, lead/lag compensation 56, if
desired, might be interposed between the summing device 42 and amplifier 24.
Moreover, the two summing devices 32 and 42 could be combined into a single summin~
device. It would also be possible to replace rectifier 44 by two rectifiers, one between
25 threshold amplifier 46 and charging circuit 50, the other between threshold amplifier 48
and charging circuit 52.
The rectifier 44, threshold ampli~lers 46 and 48 and charging circuits 50 and 52are shown as separate analog components, but they could be implemented digitally, using
a digital signal processor (DSP), for example, in which case they would be preceded by
3 o an A-to-D converter and followed by a D-to-A converter. The components in the audio
signal path could remain analog.
It is also envisaged that the components of the audio path could be implemented
digitally using a DSP. This would entail inserting A-to-D converters after the
21 6~0~
preamplifier 20 and the sensor 28, respectively, and a D-to-A converter before the power
amplifier 16. Of course, if the audio signal itself were digital, the A-to-D converter
after the preamplifier could be omitted.
Although the above-described embodiment uses an accelerometer mounted upon
5 the voice coil unit, the invention also encompasses loudspeaker units using other kinds
of sensor, such as optical sensors.
INDUSTRIAL APPLICABILITY
Advantageously, motion-feedback loudspeakers embodying the present invention
o are less susceptible to distortion due to signal overload because the rapid-response
variable gain ampli~ler within the feedb~ck loop inhibits the feedback signal from
exacell,ating the "clipping" effect.
Although an embodiment of the invention has been described and illustrated in
detail, it is to be clearly understood that the same is by way of illustration and example
15 only and is not to be taken by way of limitation, the spirit and scope of the present
invention being limited only by the appended claims.
