GB2386011A - Class D amplifier with reduced supply ripple currents - Google Patents

Abstract

Class D, or switching, power amplifiers are becoming more commonplace as enabling technology becomes available. Much of this is directed towards the switching components and modulation methods. One of the remaining issues is the output reconstruction filter and its impact on both output and, specifically, supply ripple currents. This invention deals with the issue of supply ripple currents by relocation and reconfiguration of the inductive element, 42). As a result supply ripple currents are greatly reduced. In addition the opportunity for ground referenced switching and control becomes possible allowing further performance enhancements.

Description

<Desc/Clms Page number 1>

Description Class D Amplifier With Reduced Supply Ripple Currents A typical Class D amplifier configuration is shown in Figure 1) Input signal, 1), is applied to an error amplifier, 2), which includes feedback compensation from the voltage output node 14) Output of the error amplifier is applied to the PWM modulator, 3), comprising a comparator, ramp generator and associated logic to generate PWM signals 4) and 5) as well as a deadtime signal, 8). The deadtime signal is required to prevent cross conduction in the output switches.

These signals are applied to level shifters, 6) and 7), which drive output switches, 9) and 10), supplied from voltage sources, 15) and 16).

The switched power output, 17) is low pass filtered by filter inductor, 11), and filter capacitor, 12) before application to the load, 13).

System ground is referenced to node 18) At the moment output switches, 9) and 10) are shown as N-channel Mosfets.

In order for the control electronics to be referenced to system ground, 18), level shifters, 6) and 7), are required The circuit is reconfigured as shown in Figure 2) to allow both control and drive electronics to be referenced to system ground, 18).

Switch 10) is replaced by a complementary P-channel Mosfet, 22) with the sources of 9) and 22) referenced to the relocated system ground, 18). The original PWM modulator, 3), now produces a single PWM signal, 19), applied to driver, 20), which produces a common bi-directional drive signal 21) applied to the gates of both switching devices.

Output inductor, 11), is relocated as shown as is the output node, 14), and the switched power output, 17) is now at the common point for supply voltage sources, 15) and 16).

Now both control and drive electronics are referenced to system ground. Another resulting advantage is that the deadtime signal, 8), is no longer required. Commutation of switches, 9) and 22), occurs without the possibility of cross conduction. Furthermore the removal of relative delays in level shifters 6) and 7) along with a reduction in deadtime permits higher frequency operation.

<Desc/Clms Page number 2>

We are initially concerned with ripple currents in voltage sources 15) and 16). Typically these will be derived from some form of supply which incorporates output filtering capacitors.

In the present arrangements output current is switched between 15) and 16) at the modulation frequency and is discontinuous in 15) and 16). As a result the filter capacitors must be rated to cope with both the low frequency audio current component and the high frequency discontinuous switching component.

The configuration shown in Figure 2) also introduces another problem. In both circuit configurations the switched output node, 17), suffers a high dV/dT. In Figure 2) this is imprinted on the power supply and will cause EMI and control problems, especially if an off-line SMPS is used to generate 15) and 16).

Figure 3) demonstrates the currents in 15) (23), 16) (24) and output current (25). As suggested (23) and (24) demonstrate the low frequency audio current component and the high frequency discontinuous switching component.

Figure 4) demonstrates the output voltage at node 14), (26), and the switched power output voltage at node 17), (27).

<Desc/Clms Page number 3>

In order to overcome these problems the circuit in Figure 2) is reconfigured as shown in Figure 5). A similar reconfiguration could be applied to the circuit in Figure 1).

Here the original output filter inductor, 11) is split into two inductors which are relocated as shown, 28) and 29), in series with the voltage sources, 15) and 16). Capacitor 30) is added across the opposite nodes, 31) and 32) of the inductors to allow a path for reset of inductors 28) and 29). This results in a series resonant loop which is critically damped using capacitor 33) and resistor 34). Zener diode, 35), is included to clamp possible overvoltage.

Output node, 14), now appears at the junction of voltage sources 15) and 16).

Figure 6) demonstrates the currents in 15) (36), 16) (37), and output current (38). Output voltage at node 14) is also shown, (39) As required voltage sources 15) and 16) are no longer subjected to the previous discontinuous ripple currents. Furthermore the high dV/dT applied to the power supply is markedly reduced, effectively it becomes the output voltage ripple.

One consequence of splitting inductor 11) as shown is that the two new inductors, 28) and 29), appear in parallel to the load and output ripple current as demonstrated in (38) is doubled unless 28) and 29) are scaled appropriately.

<Desc/Clms Page number 4>

In order to achieve the required result Inductor 11) has been split and relocated as inductors 28) and 29) with capacitor 30), and associated components, added to provide for reset of inductors 28) and 29).

As a result of the way the circuit operates there is the opportunity to incorporate inductors 28) and 29) on the same core resulting in a single magnetic component.

Figure 7) demonstrates the voltages across inductors 28) (40) and 29) (41). Over each switching cycle the same voltage is applied to each inductor. There is a residual volt second difference due to the use of capacitor, 30), for reset.

As a result, inductors 28) and 29) may be wound on the same core as a coupled inductor resulting in a single magnetic component. Figure 8) demonstrates the new circuit arrangement with inductors 28) and 29) replaced by the coupled inductor, 42).

Figure 9) demonstrates the result where the windings on coupled inductor, 42) have equal number of turns and the coupling coefficient, k, is set to unity.

Current in 15) (43) and 16) (44) are shown along with output current (45) The circuit has reverted to it's original single inductor form with discontinuous supply currents in 15) and 16). However output ripple current has reverted to the same value as shown in figure 3) (24).

Figure 10) demonstrates the result where the windings on coupled inductor, 42) have equal number of turns and the coupling coefficient, k, is set to less than unity. Current in 15) (46) and 16) (47) are shown along with output current (48) and output voltage at node 14) (49). The circuit shows the desired continuous currents in 15) and 16) and maintains the same output npple current. As before the output voltage at node 14) is no longer subject to a high dV/dT.

As before a similar arrangement could be applied to Figure 1) In both cases, either using uncoupled inductors, 28) and 29), or coupled inductor, 42), there is the opportunity to vary relative turns or inductance ratios along with the coupling coefficient. Also coupled inductor, 42) may be used with additional series inductance in one or both legs. However the optimal configuration is as demonstrated when using the criteria given for Figure 10).

Claims (8)

1. CLAIMS 1. A power amplifier for delivering power to a load from a bi-polar supply comprising a positive voltage source and a negative voltage source connected in series with their common node connected to one side of said load the other side of said load being connected to the common node of a pair of switches said switches being driven on and off by complementary drive signals such that the load is alternatively connected via a first inductor to the positive voltage source by the first switch and then via a second inductor to the negative voltage source by the second switch where said inductors are in series with said voltage sources and switches and represent separate components the common node of the first inductor and switch being connected via a capacitor to the common node of the second inductor and switch.
2. 2. A power amplifier as claimed in claim 1 where said inductors are wound on the same core to form a single component as a coupled inductor.
3. 3. A power amplifier as claimed in claim 2 whereby self inductances of said inductors are equal and designed such that the coupling coefficient of said inductors is less than unity.
4. 4. A power amplifier as claimed in claims 1,2 or 3 where said switches comprise the same polarity bi-polar transistors and or mosfets.
5. 5. A power amplifier as claimed in claims 1,2 or 3 where said switches comprise complementary bi-polar transistors and or mosfets with a common emitter or source connection such that said drive signals may be referenced to the common node or replaced with a single bi-polar drive signal similarly referenced such that conduction overlap of said switches is automatically prevented.
6. 6. A power amplifier as claimed in claims 1,2, 3 4 or 5 including a means for adjusting the duty cycle of said drive signals or signal responsive to some parameter of a control signal whereby output power to said load is proportional to said parameter of said control signal to achieve an open loop power amplifier.
7. 7. A power amplifier as claimed in claim 6 including a means whereby said control signal is generated as a function of the difference between some parameter of the load and an input signal to achieve a closed loop amplifier.
8. 8. A power amplifier substantially as herein described and illustrated in the accompanying drawings.