WO2005112255A1

WO2005112255A1 - Frequency stabilization for free running class-d amplifier

Info

Publication number: WO2005112255A1
Application number: PCT/IB2005/051537
Authority: WO
Inventors: Hendrikus J. J. Domensino
Original assignee: Koninklijke Philips Electronics N.V.
Priority date: 2004-05-17
Filing date: 2005-05-11
Publication date: 2005-11-24

Abstract

A self-oscillating class-D amplifier comprising a controller, an amplifier stage, and an output filter is disclosed. The amplifier has a feedback loop arranged to input a signal representing a switching frequency of the amplifier stage and output a frequency control signal to the controller. Further, a method for frequency control of the self oscillating class-D amplifier is disclosed. The frequency control is performed by charging a capacitor during a positive voltage half-period of an output from an amplifier stage of the amplifier; discharging the capacitor during a negative voltage half-period of the output from an amplifier stage of the amplifier; forming a frequency control signal based on the charge of the capacitor; and controlling the oscillating frequency of the amplifier based on the frequency control signal.

Description

Frequency stabilization for free running class-D amplifier

FIELD OF INVENTION The present invention relates to a self-oscillating class-D amplifier and a method for controlling the oscillation frequency of it.

BACKGROUND OF INVENTION Switch mode amplifiers, e.g. class-D amplifiers, are known from the early years of the transistor. A class-D amplifier is a switch mode amplifier since when one transistor is off, the current through it is zero, and when it is on, the voltage across it is small, ideally zero. The power dissipation is very low, which implies a high efficiency, thus requiring less power from a power supply and smaller heat sinks. Although the class-D amplifier is a switch mode amplifier, the operation of the class-D amplifier is based on analog principles. The class-D amplifier comprises a comparator driving a complementary transistor pair. Each transistor operates as a switch. The comparator has two inputs: one for the utility signal to be amplified and one for an oscillating signal. The output of the comparator switches on one of the transistors of the complementary pair at a time such that a high frequency signal is provided at the common electrode of the complementary pair. The average of the high frequency signal represents the amplified utility signal. Thus, a low-pass filter is provided at the output of the complementary pair. A solution with low complexity, as an output inductor and a capacitor to ground, will do, at least if the oscillating frequency is high enough compared to the frequency of the utility signal to be amplified. Most class-D amplifier solutions are based on a fixed frequency, which is normally generated by a RC oscillator. However, recent implementations with a self- oscillating power circuit add a parameter of freedom to the system to reproduce the input signal as good as possible. WO03/090343 discloses a self-oscillating switching power amplifier of class-

D which, has switching means for generating a block wave signal by alternately switching the block wave signal between supply voltages. An output filter generates an amplified utility signal corresponding to an input utility signal. A control circuit provides feedback between the output power utility signal and the input for controlling the gain in the operational frequency and the alternate switching of the switching means. The control circuit is provided with oscillation feedback elements having a high feedback, i.e. low impedance, at high frequencies such that the amplifier will be self-oscillating at a high frequency. However, the frequency can deviate due to deviation in component values and in the load of the amplifier. If the oscillation frequency decreases to a certain level, the output filter will not be able to suppress the oscillation, and high- frequent noise will be present at the output. This is devastating, particularly in applications where some of the frequency range is slightly above the audible range is used for signaling, e.g. in public speaker systems. A problem with prior art is hence that noise related to the oscillating signal is present in the output utility signal.

SUMMARY OF THE INVENTION It is therefore an object of the present invention to provide a self-oscillating class-D amplifier with reduced noise at the output related to the oscillating signal. The above object is obtained according to a first aspect of the present invention by a self-oscillating class-D amplifier for amplifying a utility signal, comprising a controller and an amplifier stage. The self-oscillating class-D amplifier is characterized by a feedback circuitry arranged to input a signal representing a switching frequency of said amplifier stage, and output a frequency control signal to said controller. An effect of this is provision of frequency control to a self-oscillating class-D amplifier such that the oscillation frequency is always kept high enough to be filtered out by an output filter of the amplifier. An advantage of this is reducing high frequency noise. This is based on the understanding that the frequency of the high frequency part of the switching amplifier, e.g. at the output from the amplifier stage, is the oscillation frequency of the self- oscillating amplifier. The feedback loop may comprise an energy balancing circuit and an averaging means. This will enable forming a frequency dependent signal, which is averaged to allow a certain controlled frequency deviation. The averaging will allow a certain controllable slack in the frequency control. An advantage of this is improvement of amplifier performance. The feedback loop may be arranged such that the input to the averaging means is the output from the energy balancing circuit, and the output of said averaging means is the frequency control signal. This makes the high-frequency path short, i.e. close to the output of the amplifier stage, which is advantageous for high-frequency design. The averaging means may be an integrator. Advantages of this is impedance matching between the input of the averaging means and the controller, and makes the analog design easy. The integrator may be provided with a controllable set-point means. An advantage of this is that the frequency control signal is then a direct measure on the average frequency deviation from a given set-point, reducing the need for further arrangements for comparing the actual frequency with a nominal frequency to control the oscillation frequency. The energy balancing circuit may comprise a capacitor that is arranged to be charged and discharged for each oscillation period. An advantage of this is a simple, compact, and cost efficient implementation of converting a high frequency to a manageable quantity, i.e. a charge in a capacitor, which automatically gives a voltage across the capacitor. The energy balancing circuit may comprise a pair of diodes for determining polarity of the signal from the point between the amplifier stage and the output filter. An advantage of this is a simple, compact, and cost efficient circuit for determining the oscillation periods. A further advantage is the enablement of analog design. The above object is obtained according to a second aspect of the present invention by a method for frequency control of an oscillation frequency of a self oscillating class-D amplifier, characterized in the steps of: charging a capacitor during a positive voltage half-period of an output from an amplifier stage of said amplifier; discharging said capacitor during a negative voltage half-period of the output from an amplifier stage of the amplifier; forming a frequency control signal based on the charge of the capacitor; and controlling said oscillating frequency of the amplifier based on the frequency control signal. An advantage of this is provision of frequency control to a self-oscillating class-D amplifier such that the oscillation frequency is high enough to be filtered out by an output filter of the amplifier, thus reducing high frequency noise. The method may further comprise the step of averaging the charge of the capacitor. This will enable forming a frequency dependent signal, which is averaged to allow a certain controlled frequency deviation by the frequency control signal. Allowing this frequency deviation will improve overall amplifier performance. Forming the frequency control signal may comprise comparing the averaged charge with a set-point. An advantage of this is that the frequency control signal can be a direct measure on the average frequency deviation from a given set-point. BRIEF DESCRIPTION OF THE DRAWINGS The above, as well as additional objects, features, and advantages of the present invention, will be better understood through the following illustrative and non- limiting detailed description of preferred embodiments of the present invention, with reference to the appended drawings, wherein: Fig. 1 shows a self-oscillating class-D amplifier; Fig. 2 shows a self-oscillating class-D amplifier with frequency control feedback; Fig. 3 shows a feedback loop according to the present invention; Fig. 4 shows an energy balancing circuit according to the present invention; Fig. 5 shows an integration with set-point according to the present invention; and Fig. 6 shows a flow chart illustrating a method of managing frequency deviations.

DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS Fig. 1 shows a self-oscillating class-D amplifier 100 provided with an input 102, a controller 104, a complementary pair amplifier stage 106, an output filter 108, a feedback loop 110, and an output 112. The complementary pair amplifier stage 106 comprises a first and a second transistor 114, 116, which work as switches. The controller 104 is fed by an input signal at the input 102, and the input signal is modulated by a high- frequency signal provided by the oscillating properties of the self-oscillating class-D amplifier 100. The modulated signal is used for controlling a switching of the amplifier stage 106, which will provide a signal to the output filter 108. The output filter 108 comprises an inductor 118 connected between the amplifier stage 106 and the output 112 and a capacitor 120 connected between the output 112 and ground 122. Other configurations of output filters are possible, and the main object of the filter 108 is to provide low-pass filtering to achieve an amplified replica of the input signal. The output 112 is connected to a load 124, e.g. a speaker. The feedback loop 110 is connected between the output 112 and the controller 104 to control gain in the operational frequency and the alternate switching of the complementary pair amplifier stage 106. Like the self-oscillating class-D amplifier 100, a self-oscillating class-D amplifier 200 of Fig. 2 is provided with an input 202, a controller 204, a complementary pair amplifier stage 206, an output filter 208, a first feedback loop 210, and an output 212. The complementary pair amplifier stage 206 comprises a first and a second transistor 214, 216, which work as switches. The controller 204 is fed by an input signal at the input 202, and the input signal is modulated by a high-frequency signal provided by the oscillating properties of the self-oscillating class-D amplifier 200. The modulated signal is used for controlling a switching of the amplifier stage 206, which will provide a signal to the output filter 208. The output filter 208 comprises an inductor 218 connected between the amplifier stage 206 and the output 212, and a capacitor 220 connected between the output 212 and ground 222. Other configurations of output filters are possible, e.g. symmetrical or asymmetrical second order LC filters, or higher order filters, and the main object of the filter 208 is to provide low-pass filtering to achieve an amplified replica of the input signal. The output 212 is connected to a load 224, e.g. a speaker. The first feedback loop 210 is connected between the output 212 and the controller 204 to control gain in the operational frequency and the alternate switching of the complementary pair amplifier stage 206. Other configurations of the first feedback loop 216 is possible, e.g. where the current through an output capacitor or the output of the amplifier, i.e. the half bridge point, is used for input signal to the first feedback loop 216. In order to better control the frequency deviation of the self-oscillating frequency, a second feedback loop 226 is provided for the high frequency domain, i.e. the oscillation. The second feedback loop 226 is fed by a high-frequency switched signal from the output of the amplifier stage 206, before the output filter 208, and the high-frequency switched signal is processed by a loop circuitry 228 to provide a frequency control signal 230. The frequency control signal 230 is then fed to the controller 204. The controller 204 can use the frequency control signal 230 in different ways, e.g. by providing a phase shift to the oscillating signal. The principal aim of the present invention is to provide a frequency control signal 230, and preferably by a robust, compact and low-cost implementation. This aim is reached by a solution, which will be described below. For the sake of simplicity, the amplifier stage 206 in Fig. 2 is depicted as a complementary pair amplifier stage. However, the amplifier stage 206 can as well be a full- bridge amplifier stage with four power switches and an asymmetrical power supply. In Fig. 2, the input to the second feedback loop 226 is connected to a point of the signal path between the amplifier stage 206 and the output filter 208. However, an input signal to the second feedback loop 226 can be provided from other parts of the amplifier circuitry. For example, it can be derived from a gate drive of one of the power switches of the amplifier stage. Alternatively, the input signal of the gate drive circuit can be used as input signal to the second feedback loop 226. The main request on the input signal of the second feedback loop 226 is that the signal is representing the switching frequency of the free running class-D amplifier. Fig. 3 shows an outline of the feedback processing circuitry 300, which functionally can be divided into an energy balancing circuit 302 and an averaging means 304. The energy balancing circuit 302 provides a predetermined electric charge for each period of the high-frequency oscillating signal of an input 306 of the feedback processing circuit 300, i.e. from the output of the amplifier stage 206 in Fig. 2. Thus, the amount of charge provided over a certain time by the energy balancing circuit 302 is proportional to the frequency, and the energy balancing circuit 302 will thus provide a signal 308 which depends on the frequency. However, an certain frequency deviation is allowed in the self-oscillating class-D amplifier, at least for a short moment, since a certain slack will provide improved overall amplifier performance. Therefore, the averaging means 304 will provide an output signal 310 which depends on an average of the frequency, or, preferably, on an average of the frequency deviation. The controller 204 in Fig. 2 can then perform actions to restore the oscillation frequency. Fig. 4 shows an embodiment of an energy balancing circuit 400. An input 402 of the energy balancing circuit 400, i.e. the high-frequency signal from the amplifier stage 206 in Fig.2, is fed to a first capacitor 404 and a first resistor 406 in series for providing the high-frequency component to a first and a second diode 408, 410. During positive voltage half-period, current is flowing through the first diode 408 to a resistor 412 and a second capacitor 414, parallelly connected to ground 416, which current will charge the second capacitor 414 to a certain level. The first capacitor 404 is chosen such that the first capacitor 404 will reach a fully charged state during the half-period, i.e. at the end of the half-period, the current through the first capacitor 404 is small. Therefore, the charge transferred and stored in the second capacitor 414 is well defined for each switching cycle of the self- oscillating class-D amplifier. During the negative voltage half-period, current is flowing through the second diode 410, i.e. from ground 416 to the input 402 of the energy balancing circuit 400, and the first capacitor 404 will reach a fully charged state with an opposite polarity during the negative half-period, i.e. at the end of the half-period, the current through the first capacitor 404 is small. At the same time, the second capacitor 414 partly discharges through the second resistor 412 during the negative half-period. Then, at the next half-period, the procedure starts all over again. Thus is a certain charge provided at the second capacitor 414 for each period, and thus a certain voltage over the second capacitor 414. Therefore, the output 418 of the energy balancing circuit 400 is provided across the second capacitor 414, and is fed to the output 418 of the energy balancing circuit 400, i.e. to the averaging means 304 in Fig. 3. Fig. 5 shows an embodiment of an averaging means implemented as an integrator 500. The input 502 of the integrator 500 is fed from the output 308 of the energy balancing circuit 302 in Fig. 3, and a set-point is provided by a controlled supply 504. The set-point is preferably controllable to enable changes of the switching frequency. The low- pass characteristic of the integrator 500 provides that a certain slack in the frequency control signal at the output 506 of the integrator 500 is allowed. The integrator 500 is implemented with an amplifier 508, preferably an operational amplifier, with a capacitive feedback comprising a capacitor 510 and a resistor 512. The output 506 of the integrator 500, which is the frequency control signal, is fed to the controller 204 in Fig. 2. Fig. 6 shows a flow chart illustrating a method of managing frequency deviations in the oscillation frequency of a free-running class-D amplifier. Depending on the polarity of the voltage output of a switching amplifier stage, prior output filtering, actions are performed. It is determined in a positive/negative voltage determination step 602 if the output voltage from the amplifier stage is positive or negative. If the voltage is positive, i.e. during the positive voltage half-period of the switching, i.e. the transistor connected to positive voltage is in a conducting state, a capacitor is charged in a capacitor charging step 604. The charge of the capacitor will be represented by a voltage across the capacitor, which is integrated. Thus is an average charge of the capacitor formed in a charge averaging step 606, which charge represent an average frequency, or a low-pass filtered frequency deviation signal if the integration is provided with a set-point. Thus is a frequency control signal provided in a frequency control signal provision step 608, on which frequency control signal frequency control is based in a frequency control step 610. If the voltage is negative, i.e. during the negative half-period of the switching, i.e. the transistor connected to negative voltage is in a conducting state, the capacitor is discharged in a capacitor discharging step 612, and an averaging of the charge and provision of a frequency control signal is performed in steps 608 and 610 as described above. Then, the procedure starts all over again by returning 600 to step 602. In a preferred embodiment, this is performed continuously, and the division into steps is of functional nature and should not be construed as discrete time steps. Of course, the voltage is either positive or negative. Therefore, the two branches 614, 616 of the flow are entered one at a time.

Claims

CLAIMS:

1. A self-oscillating class-D amplifier (200) for amplifying an utility signal, comprising a controller (204) and an amplifier stage (206), characterized in a feedback circuitry (226, 300) arranged to input a signal representing a switching frequency of said amplifier stage (206) and output a frequency control signal (230, 310, 506) to said controller (204).

2. Amplifier according to claim 1, wherein said feedback loop (226, 300) comprises an energy balancing circuit (302, 400) and an averaging means (304).

3. Amplifier according to claim 2, wherein the input (308, 502) to said averaging means (304) is the output (308, 418) from said energy balancing circuit (302, 400), and the output (310, 506) of said averaging means (304) is the frequency control signal.

4. Amplifier according to claim 2 or 3, wherein said averaging means is an integrator (304, 500).

5. Amplifier according to claim 4, wherein said integrator is provided with a controllable set-point means.

6. Amplifier according to any of claims 2-5, wherein said energy balancing circuit (302, 400) comprises a capacitor (414) that is arranged to be charged and discharged for each oscillation period.

7. Amplifier according to any of claims 2-6, wherein said energy balancing circuit comprises a pair of diodes (408, 410) for determining polarity of said signal from said point between said amplifier stage (206) and said output filter (208).

8. A method for frequency control of an oscillation frequency of a self oscillating class-D amplifier characterized in the steps of: charging a capacitor (604) during a positive voltage half-period of an output from an amplifier stage of said amplifier; discharging said capacitor (612) during a negative voltage half-period of said output from an amplifier stage of said amplifier; forming a frequency control signal (608) based on the charge of said capacitor; and controlling said oscillating frequency (610) of said amplifier based on said frequency control signal.

9. Method according to claim 8, further comprising the step of averaging the charge (606) of said capacitor.

10. Method according to claim 8 or 9, wherein forming said frequency control signal comprises comparing said averaged charge with a set-point.