 |
Description  |
|
|
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to Class D switching audio amplifiers. More
particularly, the present invention relates to a Class D switching audio
amplifier making use of four state modulation, input-to-output drive and
feedback signal isolation, a dual topology output filter, and a low
inductance board layout.
2. Description of the Prior Art
It is often desirable to amplify audio signals using a Class D switching
audio amplifier. Basic circuit layout of the Class D amplifier is
substantially similar to that of linear amplifiers, such as Class A, B,
and AB, with a major difference being in the signals provided to an output
stage. Rather than feeding an audio waveform directly to the output stage,
as is done in linear amplifiers, the Class D amplifier first feeds the
audio waveform into a Pulse Width Modulator (PWM) circuit which feeds
modulated pulses to the output stage. By quickly switching the output
stage completely on and completely off with varying pulse widths, the
Class D amplifier is able to recreate waveforms of almost any shape, and,
by filtering the switching output, sound is produced by a loudspeaker
connected thereto. In practice, the pulses are fed to the output stages at
a frequency between 100 and 300 kHz, or 100 to 300 thousand pulses per
second, which is required to produce a smooth waveform at the loudspeaker.
An advantage of the Class D amplifier is that the output stage transistors
are switched either completely on or completely off. Amplifier topologies
that operate in a partially on state, such as Class A and AB, act like
resistors and produce heat, thereby wasting energy. Thus, Class D
amplifiers are substantially more efficient than non-switching linear
amplifiers. Higher efficiency and less waste heat allows the Class D
amplifier to utilize a smaller power supply and to be offered in a more
compact package than a comparable linear amplifier.
Unfortunately, existing Class D amplifier designs suffer several
disadvantages, including disadvantages related to modulation, isolation,
feedback, and board layout. Existing Class D amplifier designs incorporate
a full H-bridge output stage and use a single PWM signal to derive four
FET gate drive signals providing two H-bridge switch states. Both H-bridge
switch states result in a differential voltage across the outputs leading
to current flow through the load. These two-state Class D amplifiers
typically compare a reference triangle waveform to an audio error waveform
(audio feedback) using a single comparator. The output of the comparator
is a single PWM signal with the same frequency as the reference triangle
waveform. The PWM signal is then passed through a logic circuit that
generates four drive signals used to drive the H-bridge, resulting in a
180.degree. phase difference between output.sub.NEG and output.sub.POS.
Thus, a differential voltage is always present at the output causing power
to be lost via the loudspeaker or low pass filter even in the absence of
an audio input to the amplifier.
Existing Class D amplifiers typically require large power transformers to
accommodate a relatively inefficient output stage and to meet government
regulations requiring high voltage isolation between AC mains and all
user-accessible inputs and outputs. This isolation is typically achieved
by incorporating one or more power transformers between the AC mains and
the input and output stages. Unfortunately, such power transformers are
large and expensive. Furthermore, because 99% of any incoming power is
required to drive the output stage and the loudspeakers connected thereto,
a power transformer isolating the output stage must be substantially
larger than a power transformer isolating the input stage.
Even in applications where the outputs are not user-accessible, no effort
is typically made to isolate the input stage from the output stage. Where
input-to-output isolation is attempted, small-signal audio transformers
are typically used. Unfortunately, these transformers suffer from limited
frequency response, making implementation difficult.
Typical output efficiencies for prior art linear amplifiers are
approximately 60%, with the remaining 40% of supplied power being
dissipated as heat. Consequently, expensive heat-sinking is required, and
large, expensive power transformers are needed to deliver 66% more power
than the desired output power of the amplifier. With the development of
Class D amplifiers, output efficiencies increased to 85%, thereby reducing
power supply requirements and waste heat. Unfortunately, expected
theoretical efficiencies of 90+% for the Class D amplifier have not been
achieved, due primarily to the many problems and disadvantages set forth
herein.
Existing high-power Class-D amplifier designs incorporate a control or
feedback loop to minimize distortion. Conventional control theory requires
filtering, attenuating, and summing the output signal with the input
signal. This typically involves a feedback loop comprising a differential
RC low pass filter, followed by an attenuating differential amplifier, and
then a summing amplifier to combine the feedback signal with the input
signal. For high power applications where common-mode voltages can exceed
70 Vdc, precision matching of feedback resistors is a critical concern.
Resistor tolerances greater than 1% in the differential amplifier and the
RC low pass filter sections result in reduced common-mode rejection,
potentially damaging voltages at the differential amplifier, and degraded
product reliability. The RC low pass filter is required to attenuate the
PWM switching energy and to pass the audio signal to the differential
amplifier. This can result in decreased efficiency as power is lost in the
RC low pass filter even in the absence of an audio input signal. High
power applications require the use of high power resistors (>1 W) that
can effectively dissipate the switching energy. Unfortunately, precision
matching and increased power handling requirements for the RC low pass
filter resistors result in increased cost and size. For example, surface
mount 1 W 1% resistors are 7.5 times larger and 18 times more expensive
than standard 1/4 W 5% surface mount resistors.
Existing Class-D amplifier designs incorporate pairs of multi-pole
differential LC low pass filters to filter the ever-present differential
switching output voltage. Typical multi-pole differential LC filter
designs dissipate a majority of attenuated energy in the first LC low pass
filter pair. No advantage is gained from common-mode filtering because the
output of the H-bridge continues to be a differential voltage. As a
result, high power designs are required to incorporate expensive high
power inductors that can dissipate the switching energy even when no audio
input signal is present.
Existing Class D amplifiers typically exhibit high harmonic distortion
above 1 kHz as a result of pulse transient damping issues and poor
triangle waveform damping generation. Excessive pulse undershoot and
overshoot result from high inductance board layouts and power supplies.
Some existing designs attempt to reduce pulse overshoot and undershoot on
H-bridge outputs by incorporating large, expensive RC snubbers. Such
undershoot and overshoot can degrade reliability for many standard FET
driver ICs such as Harris' HIP4080A. Additionally, pulse transient damping
issues also lead to increased EMI emissions that increase the cost of
shielding the amplifier.
Triangle waveform generation has always been a source of distortion in
Class D amplifier designs. Triangle waves are typically generated using RC
oscillators made of operational amplifiers or logic gates. These Class D
amplifier designs suffer from high frequency noise superimposed on the
triangle waveform; in turn, the high frequency noise results in increased
harmonic distortion. Thus, existing Class D amplifiers typically exhibit
undesirable harmonic distortion much greater than 0.5%.
Due to the above-identified and other problems and disadvantages in the
art, a need exists for an improved Class-D audio switching amplifier.
SUMMARY OF THE INVENTION
The present invention overcomes the above-identified as well as other
problems and disadvantages in the art of Class D and linear audio
amplifiers by providing a Class D switching amplifier operable to provide
increased efficiency, increased reliability, and reduced distortion
through use of four state modulation, input-to-output driver and feedback
signal isolation, dual topology output filtration, and a low inductance
board layout. Though not limited thereto, the amplifier is particularly
ideal for applications without user-accessible outputs, such as powered
loudspeakers, wherein isolation of input-to-output drive and feedback
signals allows for the elimination of large expensive power transformers
required by the prior art. Furthermore, though not limited thereto, the
amplifier is particularly ideal for high-power applications involving, for
example, 50 W or more.
The preferred Class D switching amplifier broadly comprises an input stage;
a triangle stage; a gate drive stage; an output stage; a filter stage; and
a feedback stage. The input stage is operable to receive first and second
feedback signals, FDBK_P and FDBK_N, and an audio input signal, AUDIO_IN,
and to therefrom derive first and second error signals, ERROR and
ERROR_INV. The error signals represent the combined audio input and error
for both positive and negative swings.
The triangle stage is operable to derive a low noise triangle waveform,
TRIANGLE, having reduced high frequency noise that might otherwise lead to
excessive distortion.
The gate drive stage is operable to generate four optically isolated gate
drive signals, DRV_Q1, DRV_Q2, DRV_Q3, and DRV_Q4. Within the gate drive
stage, the ERROR signal and the TRIANGLE waveform are also compared to
produce an output signal, PWM_A; and the ERROR_INV signal and the TRIANGLE
waveform are compared to produce an output signal, PWM_B. The PWM_A and
PWM_B signals are then input, respectively, to first and second
optoisolators. The optoisolators preferably allow isolated pulse
transmission with minimal delay and pulse width distortion. The outputs of
the optoisolators are taken directly to produce, respectively, the DRV_Q2
and DRV_Q3 gate drive signals; and inverted to produce, respectively, the
DRV_Q1 and DRV_Q4 gate signals.
The output stage is operable to receive the gate drive signals and to
derive therefrom intermediate output signals, OUT_HP and OUT_HN, and
broadly comprises first and second H-bridge halves which combine to form a
full H-bridge.
The filter stage is operable to reduce EMI emissions by attenuating
switching energy, and is essential for series connecting, or "daisy
chaining", the floating output stages of multiple instances of the Class D
switching audio amplifier. The filter stage receives as input the OUT_HP
and OUT_HN signals, and broadly comprises a four-pole LC low pass filter
combining common-mode filter topology for lowering inductor current in the
absence of an audio input signal, and differential filter topology for
attenuating high frequency differential signals. The filter stage provides
final output signals, OUT_P and OUT_N, to drive the loudspeaker or other
load.
The feedback stage is operable to provide the processed feedback signals,
FDBK_P and FDBK_N, to the input stage, and broadly comprises first and
second optoisolators 84,86 and first and second RC low pass filters 88,90.
Within the feedback stage, the OUT_HP and OUT_HN signals produced by the
output stage are optoisolated and filtered through RC low pass filters to
result in the FDBK_P and FDBK_N signals.
As mentioned, the present invention introduces a unique four state
modulation scheme that advantageously increases efficiency and allows for
common-mode filtering to reduce loss during no-audio conditions. Using the
four state modulation scheme of the present invention, in the absence of
an audio input signal the H-bridge outputs are common-mode (in phase) and
no current is delivered to the load. In the presence of an audio input
signal, the H-bridge outputs differentially drive current through the load
at double the frequency of the triangle waveform.
The input stage is isolated from the output stage using optoisolators.
Alternatively, small signal transformers may be used in place of the
optoisolators; however, the optoisolators, being more cost and space
effective, are preferred. Many available optoisolators provide fast data
transmission while minimizing pulse distortion effects. By isolating the
input from the output, applications without user-accessible outputs can
advantageously eliminate expensive high power transformers commonly found
in existing amplifiers, thereby resulting in an estimated 75% weight
savings and 40% cost savings over typical prior art amplifiers.
Furthermore, isolating the input stage from the output stage advantageously
allows the output stage to float with respect to the chassis or input
ground, which, in turn, allows for series connecting or "daisy chaining"
multiple amplifiers to increase power delivered to the loudspeaker.
Another benefit of floating the output stage is reduction of typical Class
D chassis referenced DC voltage present at the amplifier output.
Additionally, the present invention improves upon prior art Class D
feedback topology by isolating feedback signals and referencing the RC low
pass filter to the input stage ground. This improvement eliminates
potentially damaging differential and common-mode voltages present in the
feedback circuit. As a result, precision resistor matching is no longer
required, and less power is lost in the RC low pass filter. Thus,
isolating the feedback signals substantially reduces costs and increases
efficiency and design reliability.
The filter stage includes an LC low pass output filter operable to
attenuate the high frequency switching, pass the amplified audio signal,
reduce radiated emissions, and smooth the output current. In prior art
Class D amplifiers, differential LC low pass filter designs are used with
3 dB cutoffs at no less than 25 kHz. Regardless of whether audio is
present at the output or not, the filter is absorbing energy at the
switching frequency. With prior art modulation schemes no advantage was
gained from common-mode filtering because the output of the H-bridge was a
differential voltage waveform. As a result, high power designs were
required to incorporate expensive high current, low resistance inductors
in the LC low pass filters that could absorb the switching energy with or
without an audio signal present.
The modulation scheme of the present invention results in a common-mode
voltage in the absence of audio that allows for use of a combination
common-mode and differential LC low pass filter constructed with
inexpensive 5022 series surface-mounted inductors. The first two-pole LC
low pass filter combination is arranged in a common-mode topology; the
second two-pole LC low pass filter combination is arranged in a
differential topology. Through use of a four pole combined common mode and
differential LC output filter, inductor current is reduced in the absence
of an audio signal. With the first two-pole combination typically
absorbing more power, a common-mode topology results in less power
dissipation by reducing inductor current 37% over prior art filters. By
incorporating a differential second two-pole combination, the filter
maintains beneficial rejection of high frequency differential signal
components.
The present invention utilizes a unique low inductance board layout and
modularization that advantageously lowers pulse overshoot and undershoot,
leading to reduced distortion, reduced radiated emissions, and increased
efficiency. The low inductance design allows for elimination of expensive
RC snubbers common in prior art Class D amplifiers. With the improved
board layout, harmonic distortion has been reduced to less than 0.2%
typical at 200 W.sub.rms. Furthermore, the unique board layout reduces
overall size and allows for small lightweight construction.
These and other important features of the present invention are more fully
described in the section titled DETAILED DESCRIPTION OF A PREFERRED
EMBODIMENT, below.
BRIEF DESCRIPTION OF THE DRAWINGS
A preferred embodiment of the present invention is described in detail
below with reference to the attached drawing figures, wherein:
FIG. 1 is a circuit diagram of a preferred embodiment of the present
invention;
FIG. 2 is a circuit diagram of an alternate circuit topology for an input
stage portion of the present invention;
FIG. 3 is a circuit diagram of an alternate circuit topology for a gate
drive stage portion of the present invention;
FIG. 4 is a circuit diagram of a first alternate circuit topology for a
feedback stage portion of the present invention;
FIG. 5 is a circuit diagram of a second alternate circuit topology for the
feedback stage portion of the present invention;
FIG. 6 is a depiction of a switching portion of the output stage in a
configuration corresponding to a first switch state;
FIG. 7 is a depiction of the switching portion of the output stage in a
configuration corresponding to a second switch state;
FIG. 8 is a depiction of the switching portion of the output stage in a
configuration corresponding to a third switch state;
FIG. 9 is a depiction of the switching portion of the output stage in a
configuration corresponding to a fourth switch state;
FIG. 10 is an side elevation view of a preferred hardware layout scheme of
the present invention;
FIG. 11 is a front elevation view of the preferred hardware layout scheme
shown in FIG. 10;
FIG. 12 is a top plan view of a topside amplifier board hardware mounting
scheme; and
FIG. 13 is a top plan view of a bottomside amplifier board hardware
mounting scheme.
DETAILED DESCRIPTION OF A PREFERRED EMBODIMENT
Referring to FIG. 1, a Class D switching audio amplifier 20 is shown
constructed in accordance with a preferred first embodiment of the present
invention. The amplifier 20 is operable to provide increased efficiency,
increased reliability, and reduced distortion through use of four state
modulation, input-to-output drive and feedback signal isolation, dual
topology output filtration, and a low inductance board layout. Though not
limited thereto, the amplifier 20 is particularly ideal for applications
without user-accessible outputs, such as powered loudspeakers, wherein
isolation of input-to-output drive and feedback signals allows for the
elimination of large expensive power transformers required by the prior
art. Furthermore, though not limited thereto, the amplifier is
particularly ideal for high-power applications involving, for example, 50
W or more. The preferred Class D switching amplifier 20 broadly comprises
an input stage 22; a triangle stage 24; a gate drive stage 26; an output
stage 28; a filter stage 30; and a feedback stage 32.
The input stage 22 is operable to derive first and second error signals,
ERROR and ERROR_INV, and broadly comprises a differential amplifier 40; an
error amplifier 42; and an inverting amplifier 44. The differential
amplifier 40 receives from the feedback stage 32 first and second feedback
signals, FDBK_P and FDBK_N, and combines said feedback signals into one
single-ended signal. The single-ended signal is input to the error
amplifier 42 where it is negatively summed with an audio input signal,
AUDIO_IN. Thus is the ERROR signal derived, representing the combined
audio input and error for both positive and negative swings. The ERROR_INV
signal is derived by inverting the ERROR signal using the inverting
amplifier 44.
It will be appreciated by those with ordinary skill in the electrical arts
that alternative circuit topologies may be devised substantially
equivalent in function to those described herein and shown in the figures.
Referring also to FIG. 2, for example, an alternative topology for the
input stage 122 is shown wherein two error amplifiers 141,143 and an
inverting amplifier 145 are used in a substantially similar manner to
generate the ERROR and ERROR_INV signals.
The triangle stage 24 is operable to derive a low noise triangle waveform,
TRIANGLE, and broadly comprises an ultra-low noise voltage regulator 46; a
ramp capacitor 48; and a low ESR capacitor 50; and a switched current
source IC 52. The ultra-low noise voltage regulator 46 reduces high
frequency noise that might otherwise lead to excessive distortion, and
provides the ramp capacitor 48 with a clean supply rail with minimized
high frequency transients. The low ESR capacitor 50 provides a low
impedance path to ground for high frequency transients. The switched
current source IC 52 provides a switched current capacitive charging
circuit; a suitable switched current source IC 52 is available from
Unitrode in the model UC3637. By incorporating such a switched current
source IC, a majority of the components required for modulation signal
derivation are self-contained, which results in reduced cost and minimizes
board real estate usage. The TRIANGLE waveform is taken from the positive
terminal of the ramp capacitor 48.
The gate drive stage 26 is operable to generate four optically isolated
gate drive signals, DRV_Q1, DRV_Q2, DRV_Q3, and DRV_Q4, and broadly
comprises a first comparator 54; a second comparator 56; a first
optoisolator 58; a second optoisolator 60; a first invertor 62; and a
second invertor 64. The ERROR signal and the TRIANGLE waveform are input
to the first comparator 54 to produce an output signal, PWM_A; the
ERROR_INV signal and the TRIANGLE waveform are input to the second
comparator 56 to produce an output signal, PWM_B. The PWM_A and PWM_B
signals are then input, respectively, to the first and second
optoisolators 58,60.
The optoisolators 58,60 preferably allow isolated pulse transmission with
minimal delay and pulse width distortion. A suitable optoisolator is
available from NEC Corporation as model number PS9701, which allows for
fast data transmission with a standard LED input drive and CMOS/TTL logic
output. The outputs of the optoisolators 58,60 are taken directly to
produce, respectively, the DRV_Q2 and DRV_Q3 gate drive signals; and
inverted by the invertors 62,64 to produce, respectively, the DRV_Q1 and
DRV_Q4 gate signals.
Referring also to FIG. 3, an alternative topology for the gate drive stage
126 is shown wherein a high-speed inverting amplifier 153 and a unity gain
amplifier 155 precede, respectively, the first and second comparators
154,156. In this topology, rather than compare the TRIANGLE signal to the
ERROR and ERROR_INV signals, the ERROR signal is compared to the TRIANGLE
and a TRIANGLE_INV signal. The high-speed inverting amplifier 153 receives
the TRIANGLE signal and produces the TRIANGLE_INV signal. The unity gain
amplifier 155 buffers the TRIANGLE signal.
It will be appreciated that the gate drive stage 26, regardless of which
circuit topology is used, includes a first internal branch and a second
internal branch. In the illustrated embodiment, for example, the first
internal branch includes the first comparator 54, the first optoisolator
58 and the first inverter 62, while the second internal branch includes
the second comparator 56, the second optoisolator 60, and the second
inverter 64. Also regardless of which topology is used, separate reference
or triangle signals, whether TRIANGLE or TRIANGLE_INV, and separate error
signals, whether ERROR or ERROR_INV, must be applied to each of the
branches. Thus, as a matter of convenience, one may refer to a first
triangle or error signal (applied to the first internal branch) and a
second triangle or error signal (applied to the second internal branch) to
encompass both situations where the first and second signals are identical
and situations where the first and second signals are different in some
way (e.g., one is inverted relative to the other). Use of the terms "first
signal" and "second signal" should not be interpreted to preclude these
signals from being identical signals branching from a single source
signal.
The output stage 28 is operable to provide intermediate output signals,
OUT_HP and OUT_HN, and broadly comprises a first half-bridge 67, including
a first half-bridge FET driver IC (HVIC_1) 68 and first and second MOSFETs
70,72, and a second half-bridge 77, including a second half-bridge FET
driver IC (HVIC_2) 78 and third and fourth MOSFETs 74,76. The HVIC_168 is
operable to level shift the DRV_Q2 and DRV_Q1 signals prior to inputting
said drive signals to the first and second MOSFETs 70,72, with said
MOSFETs 70,72 combining to produce an output signal, OUT_HP. The HVIC_278
is operable to level shift the DRV_Q3 and DRV_Q4 signals prior to
inputting said drive signals to the third and fourth MOSFETs 76,78, with
said MOSFETs 76,78 combining to produce an output signal, OUT_HN. The
HVIC_1 and HVIC_268,78 are required to properly drive the gates of each
MOSFET 70,72,74,76 at potentially high voltages.
The filter stage 30 is operable to reduce EMI emissions by attenuating
switching energy, and is essential for series connecting, or "daisy
chaining", the floating output stages 28 of multiple instances of the
Class D switching audio amplifier 20. The filter stage 30 receives as
input the OUT_HP and OUT_HN signals, and broadly comprises a four-pole LC
low pass filter combining common-mode filter topology 80 for lowering
inductor current in the absence of an audio input signal, and differential
filter topology 82 for attenuating high frequency differential signals.
The filter stage 30 provides final output signals, OUT_P and OUT_N, to
drive the loudspeaker or other load.
The feedback stage 32 is operable to provide processed feedback signals,
FDBK_P and FDBK_N, to the input stage 22, and broadly comprises first and
second optoisolators 84,86 and first and second differential RC low pass
filters 88,90. The OUT_HP and OUT_HN PWM output signals produced by the
output stage 28 directly drive LEDs of, respectively, the first and second
optoisolators 84,86. Passing the PWM output signals through optoisolators
84,86 effectively attenuates the output signals. The differential RC low
pass filters 88,90 are referenced to the input stage ground, and operate
to further attenuate switching energy and pass the desired audio feedback
signals. Driving the RC low pass filters 88,90 with the low voltage
outputs of the optoisolators 86,88 lowers current in said RC low pass
filters 88,90 and eliminates high common mode voltages at the differential
amplifier 40. The result is increased efficiency, low resistor power
handling requirements and reduced precision matching requirements.
Isolating the feedback signals prior to the differential amplifier 40 and
RC low pass filters 88,90 is necessary for complete input-to-output
isolation and improves product reliability and lowers cost.
Referring also to FIG. 4, a first alternative topology for the feedback
stage 132 is shown wherein isolation is accomplished using a
high-frequency small signal transformer 185 rather than optoisolators. A
disadvantage of this topology is pulse distortion and poor performance as
the pulse width approaches 100% duty cycle. Nevertheless, this first
alternative topology may be desirable for some applications.
Referring also to FIG. 5, a second alternative topology for the feedback
stage 232 is shown wherein isolation is accomplished using an audio small
signal transformer 285 following the first and second RC low pass filters
288,290. A disadvantage of this topology is that it cannot servo out DC
error, and that significant power is lost in said RC low pass filters
288,290. Nevertheless, this second alternative topology may be desirable
for some applications.
As mentioned, the Class D switching audio amplifier 10 is operable to
provide increased efficiency, increased reliability, and reduced
distortion through use of four state modulation, input-to-output optical
isolation, feedback isolation, dual topology output filtration, and a low
inductance board layout.
Referring also to FIGS. 6, 7, 8, and 9, the four state modulation scheme
increases efficiency and allows for common-mode filtering to reduce loss
during no-audio conditions.
Existing Class-D amplifier designs typically use one comparator to compare
a reference triangle waveform to an error signal. The output of the
comparator is a single PWM signal with the same frequency but different
duty cycle as the reference triangle waveform. The PWM signal is used to
drive the four MOSFET output switches in a full H-bridge configuration,
wherein the four drive signals are derived from the single PWM signal by
inversion.
The four state modulation scheme of the present invention, when used to
drive the full H-bridge output stage 28, increases efficiency by reducing
loss when no AUDIO_IN signal is present at the input stage 22, and,
furthermore, allows for a more efficient filter stage 30 design. The
modulating signals, DRV_Q1, DRV_Q2, DRV_Q3, and DRV_Q4, used to drive the
output MOSFETs 70,72,74,76 are derived in the gate drive stage 26 by
comparing the ERROR signal to the TRIANGLE waveform to produce the PWM_A
signal; and comparing the ERROR_INV signal to the TRIANGLE waveform to
produce the PWM_B signal. By comparing both the non-inverted and inverted
error signals, the audio is effectively sampled at two points. This
results in an output switching frequency two times higher than that of the
TRIANGLE and zero current drive when no AUDIO signal is present.
The DRV_Q2 and DRV_Q3 drive signals are taken directly from the PWM_A and
PWM_B signals, while the DRV_Q1 and DRV_Q4 drive signals are derived by
inversion. In the presence of an AUDIO_IN signal, the output signals,
OUT_HP and OUT_HN, differentially drive current through the loudspeaker at
double the frequency of the TRIANGLE waveform, as shown in FIGS. 6 and 7.
In the absence of an AUDIO_IN signal, however, the OUT_HP and OUT_HN
signals are common mode (in phase) and no current is delivered to the
loudspeaker, as shown in FIGS. 8 and 9.
In the gate drive stage 26, the optoisolators 58,60 are incorporated to
isolate the input from the output. Alternatively, small signal
transformers may be used in place of the optoisolators; however, the
optoisolators, being more cost and space effective, are preferred. The
optoisolation of the present invention provides a minimum 2500V isolation
barrier. Many available optoisolators provide fast data transmission and
isolated pulse transmission while minimizing data throughput delays and
pulse width distortion effects. By isolating the input from the output,
applications without user-accessible outputs can advantageously eliminate
expensive high power transformers commonly found in existing amplifiers,
thereby resulting in an estimated 75% weight savings and 40% cost savings
over a typical prior art amplifier.
Furthermore, isolating the input stage from the output stage advantageously
allows the output stage to float with respect to the chassis or input
ground, which, in turn, allows for series connecting or "daisy chaining"
multiple amplifiers to increase power delivered to the loudspeaker.
Another benefit of floating the output stage is reduction of typical Class
D chassis referenced DC voltage present at the amplifier output.
Additionally, the present invention improves upon prior art Class D
feedback topology by isolating feedback signals and referencing the RC low
pass filter to the input stage ground. This improvement eliminates
potentially damaging differential and common-mode voltages present in the
feedback circuit. As a result, precision resistor matching is no longer
required, and less power is lost in the RC low pass filter. Thus,
isolating the feedback signals substantially reduces costs and increases
efficiency and design reliability.
The filter stage 30 includes an LC low pass output filter operable to
attenuate the high frequency switching and pass the amplified audio
signal, and are required to reduce radiated emissions and smooth the
output current. In prior art Class D amplifiers, differential LC low pass
filter designs are used with 3 dB cutoffs at no less than 25 kHz.
Regardless of whether audio is present at the output or not, the filter is
absorbing energy at the switching frequency. With prior art modulation
schemes no advantage was gained from common-mode filtering because the
output of the H-bridge was a differential voltage waveform. As a result,
high power designs were required to incorporate expensive high current,
low resistance inductors in the LC low pass filters that could absorb the
switching energy with or without an audio signal present.
The modulation scheme of the present invention results in a common-mode
voltage in the absence of audio that allows for use of a combination
common-mode and differential LC low pass filter constructed with
inexpensive 5022 series surface-mounted inductors. The LC low pass filter
reduces EMI emissions by attenuating switching energy, and is essential
for series connecting, or "daisy chaining", the floating output stages 28
of multiple Class D switching amplifiers 20. The first two-pole LC low
pass filter combination is arranged in a common-mode topology; the second
two-pole LC low pass filter combination is arranged in a differential
topology. Through use of a four pole combined common mode and differential
LC output filter, inductor current is reduced in the absence of an audio
signal. With the first two-pole combination typically absorbing more
power, a common-mode topology results in less power dissipation by
reducing inductor current 37% over prior art filters. By incorporating a
differential second two-pole combination, the filter maintains beneficial
rejection of high frequency differential signal components.
The low inductance board layout and modularization scheme advantageously
lowers pulse overshoot and undershoot, leading to reduced distortion,
reduced radiated emissions, and increased efficiency. The low inductance
design also allows for elimination of expensive RC snubbers commonly found
in prior art Class D amplifiers, and for simpler EMI shielding. With the
improved board layout, harmonic distortion has been reduced to 0.2%
typical at 200 W.sub.rms. Furthermore, the unique board layout reduces
overall size and allows for small lightweight construction.
Referring also to FIGS. 10 and 11, the board layout calls for a power
supply board 92 and an amplifier board 94 interconnected as a module or
"power brick" to minimize inductance. The power supply board 92 includes
two high value filter capacitors 95,96 and provides AC voltage
rectification. The amplifier board 94 includes the amplifier circuit 10,
described above, with the layout of said amplifier circuit 10 being such
that the H-bridge of the output stage 28 resides directly above the high
value filter capacitors 95,96. Interconnecting the power supply and
amplifier boards 92,94 is a 12 gauge copper buss bar 97 which minimizes
lumped inductance between the power supply circuit board 92 and the output
stage 28. Standoffs 100 separating the power supply board 92 and amplifier
boards 94 also provide low inductance paths to ground. The full H-bridge
65 of the output stage 28 is arranged with each half bridge 67,77 in
parallel, and opposing MOSFET pairs 70,72 and 74,76 spaced 0.25 inches
apart as shown in.
FIGS. 12 and 13 illustrate low inductance board routing of a topside layer
98 and a bottomside layer 99 of the amplifier board 94. This arrangement
minimizes lumped inductances between the H-bridge 65 and the filter stage
30. Spacing the half bridges 67,77 0.25 inches apart also allows for use
of standard 0.25 inch aluminum bar stock heat sinking for minimized cost
and weight.
Because HVIC_168 and HVIC_278 are off-the-shelf components, their pinouts
are identical, meaning that if one is positioned for shortest lead length,
the other must be positioned to have an undesirably long lead length.
Those with ordinary skill in the art will appreciate that longer lead
lengths result in undesirable disadvantages, including increased ringing
and higher inductance. Thus, the present invention makes use of both the
topside 98 and the bottomside 99 of the amplifier board 94 to shorten lead
lengths.
By using both the topside 98 and bottomside 99 of the amplifier board 94,
the positioning of each HVIC 68,78 immediately adjacent a respective
half-bridge 67,77 minimizes inductances. By selecting an HVIC that has an
optimal pinout, routing distances from the MOSFET gates 70,72,74,76 and
sources can be minimized. One such HVIC is available from International
Rectifier as the IR2113S half bridge driver. Thus, for example, by facing
opposing MOSFETs 70,72,74,76 and placing HVIC_278 on the topside 98 of the
amplifier board 94, and HVIC_168 on the bottomside 99 of the amplifier
board 94, the pinouts of the HVICs 68,78 are made to appear identical,
thus minimizing routing distances between the MOSFET gates 70,72,74,76.
This HVIC placement advantageously reduces pulse overshoot and undershoot
on the floating high side supply and gate drive pins of the HVICs reduces
inductance between output nodes and the HVICs 68,78 high side floating
supplies, increases reliability, and allows for higher power.
Thus, from the preceding description, it can be seen that the Class D
switching audio amplifier of the present invention, in its various
described embodiments, provides a Class D switching amplifier operable to
provide increased efficiency, increased reliability, and reduced
distortion through use of four state modulation, input-to-output optical
isolation, feedback isolation, dual topology output filtration, and a low
inductance board layout.
Although the invention has been described with reference to the preferred
embodiments illustrated in the attached drawings, it is noted that
equivalents may be employed and substitutions made herein without
departing from the scope of the invention as recited in the claims. For
example, as mentioned, circuit topologies alternative to those shown are
easily devised but which do not depart from the inventive concepts
disclosed herein and which, therefore, are considered within the scope of
the present invention. Thus, by way of illustration, is the alternative
input stage circuit topology shown in FIG. 2. Therefore, the concepts of
the present invention should not be viewed as narrowly limited to the
implementing circuit topologies described herein for purposes of
illustrative disclosure.
* * * * *
|
|
 |
|
 |
Description |
|
