WO2009147379A1

WO2009147379A1 - Switching power amplifier

Info

Publication number: WO2009147379A1
Application number: PCT/GB2009/001368
Authority: WO
Inventors: Jukka Typpo
Original assignee: Ntnu Technology Transfer As; Jackson, Robert
Priority date: 2008-06-02
Filing date: 2009-05-29
Publication date: 2009-12-10
Also published as: GB0902999D0; GB0810017D0

Abstract

A switching power amplifier comprising an electronic switch in parallel with a resonant network, wherein the electronic switch is DC-connected to a first node of the resonant network, and a power supply is DC-connected to a second node of the resonant network.

Description

SWITCHING POWER AMPLIFIER

The present invention relates to a switching power amplifier, in particular a class E amplifier. A power amplifier is a device for increasing the power of a signal. The amplifier responds to the signal to be amplified by taking power from an external power supply and converting it to an output signal that reproduces the input signal's waveform, but has higher power.

Power amplifier circuits are usually divided into two groups, depending on how the active element of the circuit is used. In the classic Class A, B, and C amplifiers, the active element is used as controllable current source. Switching power amplifiers, like Class D, E, and F amplifiers, use the active element as a switch.

While the classic amplifiers have many desirable properties, their modest efficiency or other performance limitations when operating under high efficiency conditions are known problems. All commonly known switching power amplifiers provide theoretically 100% efficiency. While this high efficiency is not possible to achieve in practice, in many applications switching amplifier circuits nevertheless provide superior efficiency, compared with the classic amplifier alternatives. Switching power amplifiers use an electronic switch, such as a transistor, which operates in on/off states. Generally, the input signal to the switching amplifier circuit is a periodic series of pulses with typically 50% duty cycle, and the output is a closely sinusoidal signal with increased power at the fundamental operating frequency. A pre-processing circuit can be used to adjust the impedance level of the amplifier input and to reshape the input signal to the most appropriate waveform.

Sometimes the input signal is first divided into separate amplitude and phase signals, and only the phase signal is used for driving the switch, while the amplitude signal controls the power supply of the amplifier for reproducing the amplitude information of the signal. A post-processing circuit may be used to suppress the remaining undesired frequency components of the output signal waveform, and to match the amplifier output to the load impedance. Various classes of switching power amplifiers exist. Each class has its own characteristic circuit topology, and typical voltage and current waveforms. All switching amplifier circuits have at least one associated dual or 'inverse' amplifier circuit. The circuit topology of the inverse amplifier may look quite different, but the most important internal waveforms look exactly the same, except that the voltage signals have been replaced with currents, and currents have been replaced with voltages. Since only the product of the voltage and the current is relevant from the efficiency and the output power point of view, the inverse amplifier will operate with the same efficiency and deliver the same output power to the load as its associated dual circuit. Since the circuit topology and the component values of the inverse amplifier in most cases are quite different, the inverse amplifier circuit may be easier to implement in practice, or may provide some other advantages than the original circuit.

The operation of all switching amplifiers described in this specification is explained in terms of the steady state, that is, after a long series of periodic pulses has appeared in the amplifier input. The amplifier response may be very different when only one or a small number of pulses appear in the input, but this is irrelevant in most applications.

Class E amplifiers are switching power amplifiers of high efficiency, often designed to operate at such high frequencies that the switching time of the switching device becomes several percent of the signal period. A typical class E amplifier circuit is shown in Figure 1. The switch Ti is connected via the inductor Li and the serial resonance circuit L₀C₀ to the load R_L. In addition, T₁ is DC connected to the supply voltage source Vcc via the large choke inductance L₂, and a capacitor Cj is connected across the switch.

The operation of the class E amplifier in the steady state can be explained by describing what happens during one complete cycle of the signal period. Since the final conditions of a cycle are exactly the same as the initial conditions of the following cycle, we can freely choose the starting time point of this cycle. For simplicity, we can say that the switch is opened in the beginning of the first half of the cycle, and is closed in the beginning of the second half. In the beginning of the cycle, the voltage across the switch is zero, since the switch has been closed during the second half of the previous cycle. When the switch is opened, the currents from Lj and L₂ start to charge capacitor Ci, shown in Figure 1. Because of the presence of Ci, the voltage will rise slowly from zero, allowing the switch to turn off relatively slowly. Because of the DC current from the power supply and the damped oscillation response of Lo, Li, C₀, Ci, and R_L, the voltage across Ci will first rise, then decrease, and finally approach the zero in the end of the first half of the cycle.

The values of C₀, Ci, L₀, L], and R_L are chosen in such a way that at the desired operating frequency the voltage across the switch is zero and is varying very slowly exactly when the switch is closed in the beginning of the second half period. This condition is commonly referred to as 'Zero Voltage Switching', or 'ZVS', in the literature. In this way, a relatively long turn-on time of the switch device can be tolerated. During the second half of the cycle, current flows through the switch, but the voltage across it will be zero. The switch now shorts Ci, and the remaining LCR circuit performs a damped oscillation, completing the cycle.

Throughout the cycle, the switch current is always zero when the voltage across it is nonzero, or the voltage is zero when current is flowing through it. These arrangements ensure that the product of the voltage and the current in the switch is always zero, and no electric power is dissipated in the switch.

The amplifier output signal current will flow through the serial resonance circuit L₀C₀ to the load. Since the serial resonance circuit is tuned to resonate at the desired operating frequency and it has a very narrow pass band, only the fundamental sinusoidal component of the relatively complex waveform across the switch is coupled to the load.

The sensitivity of the conventional class E amplifier to the parasitic resistance of the switch device is discussed in United States patent application US 2005/0218977, and a new inverse class E amplifier circuit is proposed for mitigating the associated problems. Another known problem with the conventional class E circuit is that the switching device has to stand a relatively high peak voltage in the off-state. The peak voltage across the switch in the inverse class E circuit is lower - A -

for the same output power and load resistance, which may be a good reason for favouring the inverse circuit.

The circuit topology of this prior art inverse E amplifier shown in Figure 2 is quite different from the standard class E amplifier of Figure 1. Instead of a series resonant LC circuit connecting the transistor (active device) to the load, a parallel resonant circuit L₂Ci is connected across the load. The transistor is connected to this resonant circuit by an inductor Li and a capacitor C₂, which acts as a DC block. The DC block is required to prevent the DC supply voltage from entering the resonance circuit inductor L₂, which would short the supply voltage source to the ground. The supply voltage V_dd is applied to the transistor via the inductor L₁ and the RF choke L₃. The RF choke is required to ensure a constant flow of current throughout an entire RF cycle by only permitting a DC current to pass through it.

To describe the steady state operation of this prior art inverse class E amplifier, we can say that the switch is closed in the beginning of the first half cycle, and is opened exactly in the middle of the cycle. Since the switch was open during the last half of the previous cycle, no current is flowing in the switch or inductor Li before the beginning of the cycle. The voltage across Li is also zero. The voltage across the switch may be nonzero, but it is not relevant, as long as the switch is ideal. When the switch is closed, the current starts to flow. The switch connects the inductor Li in parallel with L₂, which speeds up the damped oscillation response of the L_IL₂C_IR_L circuit. The DC blocking capacitor does not affect the response; it represents a short circuit at the operating frequency since it is much larger than any other capacitor in the circuit. By suitable component value choices, the current first increases, then decreases, and finally approaches slowly zero in the middle of the cycle. This arrangement corresponds to 'Zero Current Switching', as opposed to the ZVS condition of the conventional Class E amplifier.

When the switch is opened in the middle of the cycle, the inductor Li is disconnected from the resonance circuit. During the second half of the cycle, the remaining RLC circuit continues the damped oscillation response. The time constant of this part of the response is now larger since L) was disconnected from the circuit. The inverse class E amplifier of Figure 2 has some drawbacks. A large DC bias choke inductor is needed for biasing the transistor. The large physical size of the choke, self-resonances, and resonances with the other connected components may make it difficult to design the practical circuit. Large inductance also limits the modulation bandwidth of the circuit if controlling the supply voltage is used for controlling the amplitude of the output signal. In practice, the choke may be replaced with a smaller inductor, but in this case some RF current is coupled through the inductor to the power supply connection node. This may cause unpredictable effects, since anything that is connected to this node will also affect the internal waveforms of the circuit. A large DC blocking capacitor C₂ is also required between the choke L₃ and the resonance circuit CiL₂. Since it should be much larger than the other capacitors of the circuit, a relatively small amount of parasitic capacitance from either capacitor node to the ground will detune the resonance circuit. In addition, the physical size of the DC blocking capacitor may be large compared with the other components of the circuit, which makes the layout design of the most critical parts of the circuit difficult. The DC blocking capacitor also isolates the switch from the load at low frequencies, which may cause low-frequency stability problems in the case that the switching device is capable of amplifying low- frequency signals. The present invention seeks to minimize the disadvantages of the known inverse switching amplifiers, or at least to provide an alternative design.

Viewed from a first aspect the present invention provides a switching power amplifier comprising an electronic switch in parallel with a resonant network, wherein the electronic switch is DC-connected to a first node of the resonant network, and a power supply is DC-connected to a second node of the resonant network.

As will be discussed further below, the invention allows an amplifier, such as an inverse class E amplifier, to be formed without the conventional AC blocking inductor, thereby avoiding its associated drawbacks. The first and second nodes may be DC-connected to each other through one of the resonant network components. The load may then be connected to the first node, possibly through a DC blocking component or an impedance matching network. The bias current from the DC supply to the switch therefore flows through at least one of the resonator component and as the switch is DC connected to the resonant circuit, there is no DC block between the resonator and the switch.

In such a circuit, in the frequency of operation, the resonance network presents high impedance in the first node. Since the load is also connected to this node, the impedance of this node at the operating frequency is essentially the load impedance. At higher and lower frequencies, the impedance is generally much lower and determined by the resonance network and the internal impedance of the power supply. The second node of the resonance network is generally kept in low impedance at all frequencies of interest by the internal impedance of the power supply source, preferably assisted by a bypass capacitor or some other suitable network, like a set of series resonance circuits, some other passive or active circuit, or another amplifier circuit that makes this node a virtual ground node.

To avoid unnecessary power consumption in the circuit, the power supply connection node preferably presents high impedance in DC when the electronic switch is open, the power supply is disconnected, and the output load is removed or DC blocked. The source impedance of the power supply will keep this node in low impedance at very low frequencies. hi a typical application of the invention, the input of the amplifier is to the electronic switch. Optionally, when amplitude modulation of the output signal is desired, there supply voltage may be modulated and so a further input will be provided. The output is across the two output nodes of the resonant circuit, where one of them may be the ground node.

Generally, it is not necessary to separate the second node from the supply voltage source with an intervening AC blocking inductor, and the resonance circuit need not be separated from the switching device with a DC blocking element. Further, no DC blocking element is needed if it is acceptable to connect DC voltage to the load.

Thus, by means of the invention, an inverse class E amplifier is formed where the AC blocking inductor (RP choke) is not required, and/or a DC blocking component may be avoided in certain embodiments. The absence of these components is advantageous as they can be difficult to realise in practice and their physical size and associated parasitic elements may make the realisation of the amplifier circuit difficult. The absence of these components saves also space and costs in production. In addition, where the amplitude of the output signal is controlled by varying the supply voltage, the large inductance of the RF choke L₃ would limit the modulation bandwidth of the circuit. Since in the preferred embodiments of the present invention the choke is removed and since the resonance circuit inductor L₂ is very small and the loaded Q- value of the resonance circuit is relatively low, the envelope of the output signal will follow the variations of the supply voltage very rapidly. Hence, by removing the RF choke the present invention produces an amplifier that is easier to implement and can use a varying supply voltage for amplitude modulation without limiting the modulation bandwidth of the circuit.

As noted above, a significant feature of the present invention is its bias arrangement. Thus, viewed from another aspect, the invention provides an inverse class E amplifier where the transistor bias current is provided via a circuit element that forms a part of the output resonant circuit, and therefore avoids the use of a dedicated RF choke and a DC blocking element between this RP choke and the resonance circuit. The absence of these components reduces the number of parasitic circuit elements that appear in the practical circuit. Even though some new circuit elements are needed for the bypass network, they are connected to nodes where serial resonances at some frequencies are actually desired, and resonances at unpredictable frequencies are not harmful. Since the optional DC-block is connected to the output of the circuit, only sinusoidal signals at the operating frequency are coupled to it. The invention also enables the provision of a switching amplifier circuit topology where the impedance connected to the switching device is low at frequencies far away from the operating frequency. This is advantageous since many practical switching amplifiers suffer from instability: due to the small-signal gain of the switching transistor, the circuit tends to oscillate. The small-signal gain depends on the impedance of the output network. Since in the present invention the DC block of the prior art can be removed from between the output resonance network and the switch, the impedance seen by the switching transistor will be lower or equal to the load impedance at all frequencies where the transistor has gain. Thus, viewed from a further aspect the invention provides an inverse class E amplifier in which there is no DC blocking component between the switch and the resonance network. The electronic switch can be any suitable active device capable of an on or off state (closed/open state). In preferred embodiments, the electronic switch takes the form of a transistor, for example a bipolar junction transistor (BJT), heterojunction bipolar transistor (HBT), junction field-effect transistor (JFET), metal-oxide-silicon field-effect transistor (MOSFET), metal semiconductor field effect transistor (MESFET), or High Electron Mobility Transistor (HEMT).

In a preferred embodiment the supply current flows through the inductive component in the resonant circuit. Typically, the inductive component will be a simple inductor, but it may alternatively be a transformer primary winding or a short segment of a transmission line that provides inductive reactance and a DC path. The function of the inductive component is to provide inductive impedance at the operating frequency. The power supply is connected to one end of this inductive component. This end of the inductive component is kept in low impedance by the internal source impedance of the power supply, possibly assisted by a bypass capacitor and/or a set of series resonators. This bypassing network is designed to present low impedance to the currents that flow through the resonator inductor at the operating frequency, and possibly to some of its harmonic frequencies. Additional resonators may be used at potential interfering frequencies in the power supply line, for suppressing interfering signals from the power supply. The load is connected to the other end of the resonator inductive component, which may be also DC- connected to the switch through a second inductive component. The common node of the two inductive components is used as the output node of the circuit.

In this way, the impedance of the power supply connection node is kept low, but the other end of the resonator inductive component is in high impedance at the operating frequency since the inductive component is resonating. The closely sinusoidal output voltage in the output node is coupled to the load, and the current from the power supply can flow through the inductive component to the electronic switch without an additional RF choke. An inductive component is preferably connected in series with the electronic switch. This component forms a part of the DC connection of the switch through the resonant network to the power supply. In some cases, the parasitic inductance of the switch may be sufficient, and it is not necessary to use an additional inductive component.

In a preferred embodiment of the invention, the resonant network comprises a first capacitor, a second capacitor and an inductor, wherein the second capacitor and the inductor form a series LC circuit having the second capacitor connected to ground, and the first capacitor is connected in parallel with this series LC circuit. With this arrangement, the power supply may be connected at the node between the second capacitor and the inductor, and the switching element may be connected at the node common to the inductor and first capacitor. Preferably, the second capacitor has a much larger capacitance than the first capacitor in order to suppress signal coupling to (and from) the power supply and to avoid disturbing the frequency response of the resonance circuit.

Using a large capacitor for bypassing the power supply in this way has some drawbacks. However, since only some distinctive sinusoidal AC current components will flow through the inductor, it is sufficient to bypass the supply source only at these frequencies. Thus, in an alternative embodiment, a plurality of tuned series resonance circuits, i.e. one for each frequency of interest, may be used.

As a further alternative, the low impedance in the power supply connection node for some of the most important frequency components may be created by employing a virtual ground node by using two identical amplifiers that share the same power supply connection node. Further details of such circuits are given below where certain embodiments of the present invention will now be described, by way of example only and with reference to the accompanying drawings, in which: Figure 1 shows a prior art class E amplifier; Figure 2 shows a prior art inverse class E amplifier; Figure 3 shows an embodiment of the present invention

Figure 4 shows an alternative embodiment of the invention; Figure 5 shows another alternative embodiment of the invention, employing a virtual ground as the low impedance node for the power supply connection;

Figure 6 shows a further embodiment, where the load is connected to a node of a resonant circuit formed by splitting a capacitor into two series capacitors; Figure 7 shows an alternative further embodiment, where the new node is formed by splitting an inductor into two series inductors; and

Figure 8 shows a final embodiment, where the inductive component of the resonant circuit is provided by a transformer.

An embodiment of the present circuit invention is illustrated in Figure 3. It is an inverse Class E amplifier whose principle of operation and the most important internal current and voltage waveforms are similar to the prior art circuit of Figure 2, described above. However, the bias arrangements and the resonance network are significantly different and this enables some unique advantages to be achieved.

The circuit of Figure 3 can best be described as a series of modifications to the circuit of Figure 2. First, the RF choke L₃ is removed, and the DC blocking capacitor C₂ is replaced with a short circuit. The grounded end of the parallel resonance circuit inductor L₂ is disconnected and the power supply is connected between L₂ and the ground. These arrangements ensure that DC current from the power supply can flow to the switch device through the inductors L₂ and L₁. The power supply output impedance is in practice always nonzero. In order to suppress signal coupling to (and from) the power supply and to avoid disturbing the frequency response of the resonance circuit C₁L₂, the power supply is bypassed with a large capacitor C₂, as is shown in Figure 3. In this embodiment, the bypass capacitor should be much larger than the resonating capacitor. To avoid a shift in the resonance frequency, the value Of C₁ must be increased slightly even if C₂ » C₁.

Using a large capacitor for bypassing the power supply has some drawbacks. However, since only some distinctive sinusoidal AC current components will flow through L₂, it is sufficient to bypass the supply source at these frequencies. Thus, in an alternative embodiment, a tuned series resonance circuit for each frequency of interest, as is shown in Figure 4, is used (note Cn and Ln forming the resonant circuit for frequency fl, etc.). In many practical cases, the space occupied by these resonant networks would be much smaller than the space required by the single large capacitor C₂.

In a still further embodiment, the low impedance in the power supply connection node for some of the most important frequency components may be created by employing a virtual ground node by using two identical amplifiers that share the same power supply connection node, as is shown in Figure 5. When the amplifier inputs are driven by a 50% duty cycle signal with one of the amplifier inputs 180 degrees out of phase, all internal waveforms are identical but shifted in the time domain by a half period of the operating frequency in the two amplifiers. Even though the time domain waveforms are not half wave symmetric, the fundamental and all odd harmonic frequency components of the waveforms will be out of phase. Consequently, the two amplifiers will cancel each others' effects on the current drained from the power supply node at these frequencies. This embodiment simplifies the structure of the bypass network, since the fundamental frequency and all odd harmonics are suppressed without the assistance of an additional bypassing network.

The output signals of these two amplifiers must be combined in a suitable balun. Since the DC voltages of the two outputs are identical, it is not necessary to use a DC blocking capacitor between them. The balun may be used for DC blocking and impedance matching of the load. An additional advantage of this configuration is that some of the components, like Cn and C₂ι of Figure 5, may be combined due to the symmetry of the circuit.

In the embodiments described above the load R_L is connected to the end of the inductor L₂, opposite the connection to the power supply and is therefore connected to the first node, where the switch is connected, through the serial inductor L₁. However, this is not essential. Since the current is circulating in the resonant circuit, the output for the load R_L can be taken out from almost any part of the resonant circuit, by means of a suitable coupling circuit. In other words, it is possible to split the resonator into several pieces, and take the output from another node.

Hence, in a modified embodiment, as illustrated in Figure 6, the capacitor Ci of Figure 3 is split into two series connected capacitors, denoted Ci and C₃, and the load R_L is connected to the node between these capacitors. In this circuit, the capacitors provide the required DC blocking and convert the impedance level down at the same time. In cases where the preferred resistive load across the resonator is higher than the actual load impedance (which is usually 50 Ohms), this embodiment is effective in providing simultaneous impedance matching and DC blocking

Similarly, it is possible to split the resonator inductor L₂ of Figure 3 into two series connected inductors L₂ and L₃, as shown in Figure 7. The output node for the load R_L can then be provided by the middle node between these two inductors. This would also convert the impedance level down. Yet another alternative is shown in Figure 8. In this circuit the inductive component of the resonant circuit (the inductor L₂ in Figure 3) is replaced with a transformer Tj and the output for the load R_L is taken from the secondary winding. This arrangement works well at low frequencies, and also provides appropriate impedance matching and DC blocking. It will be appreciated that certain features of the various preferred embodiments described above could be combined if required. For example, the inductor L₂ of could be replaced by an alternative inductive component such as a transformer primary winding or a short segment of a transmission line that provides inductive reactance and a DC path.

Claims

CLAIMS:

1. A switching power amplifier comprising an electronic switch in parallel with a resonant network, wherein the electronic switch is DC-connected to a first node of the resonant network, and a power supply is DC-connected to a second node of the resonant network.

2. A switching power amplifier as claimed in claim 1, wherein the first and second nodes are DC-connected to each other through one of the resonant network components.

3. A switching power amplifier as claimed in claim 1 or 2, wherein the load is connected to the first node, preferably through a DC blocking component or an impedance matching network.

4. A switching power amplifier as claimed in claim 1 , 2 or 3, wherein the power supply connection node presents high impedance in DC when the electronic switch is open, the power supply is disconnected, and the output load is removed or DC blocked.

5. A switching power amplifier as claimed in any preceding claim, wherein the input of the amplifier is to the electronic switch and the output is across the two output nodes of the resonant network.

6. A switching power amplifier as claimed in any preceding claim, wherein there is no AC blocking inductor between the second node and the supply voltage source, and/or there is no DC blocking element between the resonant network and the switching device.

7. A switching power amplifier as claimed in any preceding claim, wherein the electronic switch is a transistor.

8. A switching power amplifier as claimed in any preceding claim, wherein the supply current flows through the inductive component in the resonant network.

9. A switching power amplifier as claimed in claim 8, wherein the end of the inductive component that connects to the supply is kept in low impedance by the internal source impedance of the power supply, assisted by a bypass capacitor and a set of series resonators.

10. A switching power amplifier as claimed in claim 9, wherein the load is connected to the other end of the inductive component, which is also DC- connected to the switch through a second inductive component, and the common node of the two inductive components is used as the output node of the resonant network.

11. A switching power amplifier as claimed in any preceding claim, wherein an inductive component is connected in series with the electronic switch.

12. A switching power amplifier as claimed in any preceding claim, wherein the resonant network comprises a first capacitor, a second capacitor and an inductor, wherein the second capacitor and the inductor form a series LC circuit having the second capacitor connected to ground, and the first capacitor is connected in parallel with this series LC circuit.

13. A switching power amplifier as claimed in claim 12, wherein the power supply is connected at the node between the second capacitor and the inductor, and the switching element is connected at the node common to the inductor and first capacitor.

14. A switching power amplifier as claimed in claim 13, wherein the second capacitor has a much larger capacitance than the first capacitor.

15. A switching power amplifier as claimed in any of claims 1 to 11, wherein the resonant network comprises a first capacitor, a plurality of tuned series resonance circuits connected in parallel and an inductor, wherein the plurality of parallel tuned series resonance circuits and the inductor form a series LC circuit having the plurality of tuned series resonance circuits connected to ground, and the first capacitor is connected in parallel with this series LC circuit.

16. A switching power amplifier as claimed in claim 15, wherein the power supply is connected at the node between the plurality of tuned series resonance circuits and the inductor, and the switching element is connected at the node common to the inductor and first capacitor.

17. A switching power amplifier as claimed in claim 1, 2 or 3, comprising two amplifier circuits connected to the same power supply connection node in order to form a virtual ground node, with the output signals of the two amplifiers being combined in a balun.

18. A switching power amplifier as claimed in claim 1 or 2, wherein the load is connected to a third node of the resonant network.

19. A switching power amplifier as claimed in claim 18, wherein the third node is provided by splitting a capacitor or an inductor of the resonant network into two series components.

20. A switching power amplifier substantially as hereinbefore described with reference to Figure 3, 4 or 5 of the accompanying drawings.