The present invention relates to an amplifier circuit for providing a radio frequency output signal having a variable signal envelope. The present invention furthermore relates to a method of providing a power amplified radio frequency signal having a variable signal envelope.

In the field of mobile communication technology, the power efficiency of radio frequency signal amplifiers deployed in mobile communication terminals is an important requirement. The power efficiency of the RF amplifier will have a strong impact on the amount of electrical power drawn by the mobile station from the battery when transmitting RF signals and thus influences how long the mobile terminal can be operated without connection to a battery charger.

RF amplifiers output stages operating in class B or class C are known to be quite power efficient. Output stages

operating in class B or C essentially use a half wave or less than a half wave of the RF input signal to drive an amplifier device having a load impedance. For RF signals having an essentially constant signal envelope, the operating point of this kind of output stage can be set such that sufficiently low power dissipation in the amplifier device and thus, high power efficiency is achieved. In some designs, maximum power efficiency is achieved when the amplifier device is driven at such a level that the RF voltage level across the amplifier device is as large as possible. At lower driving levels however, the power efficiency will then be less than the optimal value.

RF signals having a variable signal envelope therefore impose additional challenges on the design of power efficient RF amplifiers. If the signal envelope of the RF signal is not constant but variable, this results in that the output, stage is dri en at variable power levels in accordance with the instantaneous signal envelope and thus at the average below levels at which the amplifier can operate with high

efficiency. This results in a reduced average power

efficiency of the amplifier.

The average power efficiency of the RF amplifier for RF signals with varying signal envelope can be improved if instead o a single amplifier device , a main amplifier device and one or more auxiliary amplifiers are used to provide the RF output signal through a combiner netv/ork . A main amplifier device is designed for example such that it. reaches its operating point of optimum power efficiency already at a point below the upper limit of the dynamic range of the input signal. If the input signal level reaches beyond this point, an auxiliary amplifier device contributes to the output signal through the combiner network. The Doherty type radio frequency power amplifiers use a main amplifier and an auxiliary amplifier in this manner such that they are

efficient for RF signals having a signal envelope that is variable over time.

The combiner network in an RF power amplifier having a main amplifier device and at least an auxiliary amplifier device will inherently exhibit some kind of band limited frequency response. For example, a Doherty type amplifier uses a quarter wavelength transmission line or an equivalent

reactive network for combining the RF power contributions of the main amplifier device and the auxiliary amplifier device.

Moreover, at least the level at which the auxiliary amplifier is driven, shows a strongly non- linear functional dependency from the signal envelope of the RF signal. The auxiliary amplifier has to be driven by a signal that is obtained by applying this non- linear function on the time varying signal envelope of the RF signal . This causes a widening of the bandwidth of the signal driving the auxiliary amplifier. As long as the signal envelope varies slowly, this bandwidth widening would not be critical for the operation of the combiner network or of the auxiliary amplifier device.

However , if the signal envelope of the R? signal becomes more wideband, the band widening effect of the non-linearity of the functional dependenc o the driving level of the.

auxiliary amplifier from the signal envelope can result in that the bandwidth of at least some of the components of the RF power amplifier may no longer be sufficient to keep the power efficiency of the amplifier at an acceptable level.

A power amplifier according to an embodiment of the present invention for providing a radio frequency output signal having a variable signal envelope comprises an amplifier circuit for providing a radio frequency output signal having a variable signal envelope, comprising a main amplifier device having an input and an output for outputting a first radio frequency output signal; an auxiliary amplifier device having an input and an output for outputting a second radio freque cy ou ut signal; a combi er network for combining at. least a component of said first radio frequency output signal from said first amplifier device and at least a component, of said second radio frequency output signal from said second amplifier device to provide a combined output signal of variable signal envelope to a load; a signal processing circuit having an input for receiving an input signal, and a first out ut, coupled to the input of the main amplifier device for providing a first, radio frequency signal to be amplified by said first amplifier device, and a second output coupled to the input of the second amplifier device for providing a second radio frequency signal to be amplified by said second amplifier device; wherein said signal processing circuit comprises a non-linear processing section coupled to the input of the signal processing circuit and adapted to provide at least said second radio frequency signal with a signal, envelope that has a non- linear dependency from an amplitude characteristic of the input signal ; wherein he. non-linear processing section is adapted such that the degree of non-linearity of the non- linear dependency varies

dependent on the amount of change per time unit of the amplitude characteristic of the input signal.

According to a non- limiting example embodiment of the present invention, the degree of non-linearity of the non- linear dependency is higher if the amplitude characteristic varies at: a first , slow rate than if he amplitude characteristic varies at a faster race . In this manner the power amplifier circuit according to this embodiment of the present invention dynamically restricts the bandwidth expansion of the signal driving the auxiliary amplifier device .

If the non-linear dependency is denoted as f (x) and f is approximated in terms of a polynomial expression as a sum from i ( integer} = 0 to N>0 of a-j_*x ¹ , the degree of non- linearity would be N . The present invention is, however , not limited to any polynomial representation of the non-linearity f. Rather, the term 'degree of non-linearity' is intended to comprise any measure that allows distinguishing between strongly non- linear dependencies and less strongly non- linear dependencies . According to a non-limiting example, at a rate of variation of zero or close to zero, the non-linearity exhibits two or more sections of di fferent yet positive slopes joined by a sharp bend. With an increasing rate of variation, the bend changes into a progressively softer curve. The non-linear dependency can be provided in any sui able manner including digital signal processing,

overshoot extraction circuits, adders/subtractors and filters for frequency selective filtering of the extracted overshoot of the input signal. The non- linear dependency can be

implemented using analogue or digital circuits or a

combination of both. For example, the amplitude

characteristic subject to non- linear processing of the input signal can be processed in a digital signal processing means to obtain a control signal that controls the signal envelope of a radio frequency signal driving the auxiliary amplifier device. According to another non-limiting example embodiment, the signal processing circuit provides a suitable signal directly at the radio frequency or at a suitable intermediate frequency .

Some embodiments of the present invention achieve high efficiency and low bandwidth expansion by the dynamic nonlinear dependency of the signal envelope of at least the signal driving the auxiliary ampl ifier device from the input signal. The power efficiency of the power amplifier circuit can remain high particularly when the signal envelope of the RF output signal varies slowly, that is at a low rate, such that an advantageous average power efficiency of the

amplifier circui can be achieved . Advantageously but not necessarily, also the signal driving the main amplifier device has a signal envelope that shows a non- linear dynamic dependency from an amplitude characteristic of the input signal, which will, however, be different from and preferably complementary to the non- linear dependency used for driving the first amplifier device.

The term 'envelope' of a signal denotes the difference bet een any local maximum and the adjacent local minimum of the signal. For example, i the local maxima (the local positive peaks) of the RF signal are joined by an upper line and all local minima (the local negative peaks) of the RF signal are joined by a lower line, the vertical extension between the upper and the lower line yields the signal envelope . An RF signal carrying an amplitude modulation has a signal envelope that is variable over time. The signal envelope may or may not: correspond to the amplitude modulated signal, depending on the kind of amplitude modulation that is used . Also, while amplitude modulated signals are useful examples of signals having a variable signal envelope, an RF signal having a variable signal envelope does not always have to carry an amplitude modulation. A very simple yet non- limiting example of obtaining a measure of the envelope of a signal is rectifying the signal and smoothing the result by a capacitor .

According to a non- limiting example embodiment of the resent invention, the RF power amplifier circuit can be designed such that the input signal to the power amplifier circuit can be an amplitude modulated signal at radio frequency or at a suitable intermediate frequency, such that the above

mentioned amplitude characteristic of the input signal is the envelope of the input signal. According to other embodiments of the present invention, the input signal can be a base band signal such that the amplitude charac eristic of the input signal would then be the instantaneous i put signal

amplitude . According to yet another embodiment, the input signal is ime and amplitude discrete and re resents in digital form an amplitude characteristic to control the envelope of the output signal of the RF power amplifier.

The present invention also relates to a mobile communication device comprising an RF power amplifier as claimed. Such mobile communication devices include mobile radio telephones, smart phones, portable computers like tablet PCs and the like. The present invention furthermore relates to a base station for a mobile communication network, the base station comprising an RF power amplifier as claimed. The term base station includes Base Transceiver Stations BTS , Node B, eNode B, Radio Network Controllers and the like .

The present invention also relates to a method of providing a power amplified radio frequency signal having a variable signal envelope, comprising providing a first radio frequency output signal by a main amplifier device; providing a second radio frequency output signal by an auxiliary amplifier device; combining at least a component of said first radio frequency output signal from said first amplifier device and at least a component, of said second radio frequency output signal from said second amplifier device in a combiner network to provide a combined output signal of variable signal envelope to a load; signal processing an input signal in a signal processing circuit to provide a first output signal to the input of the main amplifier device for

providing a first radio frequency signal to be amplified by said first amplifier device and a second output signal to the input of the second amplifier device for providing a second radio frequency signal to be amplified by said second amplifier device; said signal processing comprising nonlinear processing of said input signal to provide a nonlinear dependency of the envelope of at least said second radio frequency signal from an amplitude characteristic of the input signal; wherein the degree of non-linearity of the non- linear dependency varies dependent on the amount of change per time unit of the amplitude characteristic of the input signal .

In the following, advantageous example embodiments of the present invention will be described with reference to the accompanying figures. It has to be noted that everything in the figures and the accompanying description is meant to be purely illustrative and shall not be construed to limit the scope of the present invention.

In the figures,

Fig 1 shows a block diagram of an exemplary embodiment of a radio frequency power amplifier according to the present invention ;

Fig. 2A shows an example embodiment of a non- linear

dependency f of the envelope of the second radio frequency signal from an amplitude characteristic of the input signal; Fig. 2B shows an example diagram for illustrating dynamic power efficiencies of a radio frequency power amplifier;

Fig. 3 shows an example block diagram of the signal processing section 3 shown in Fig, 1 ;

Fig. 4 shows another example block diagram of the signal processing section 3 shown in Fig. 1 ;

Fig. 5 shows an example block diagram of a dynamic nonlinear processing section;

Fig. 6 shows another example block diagram of a dynamic non- linear processing section; and

Fig. 7 shows an example f lev; chart for illustrating in exemplary fashion a method of providing a power amplified radio frequency signal.

Fig. 1 shows a block diagram of an exemplary embodiment of a radio frequency power amplifier according to the present invention. In this figure, 1 and 2 denote a main amplifier device and an auxiliary amplifier device, respectively, la and 2a denote inputs of the respective amplifier devices, lb and 2b denote outputs of the respective amplifier devices, 3 denotes a signal processing circuit having an input for receiving a signal In as well as outputs Outl and Out2

respectively coupled, to the inputs la and 2a of the main and auxiliary amplifier devices 1 and, resp. In the exemplary embodiment shown in fig. 1, the amplifier devices 1 and 2 each include a solid state device and a load impedance, Zl and Z2, in the drain circuit, resp. , such that the respective amplifier device can operate for example in class B or class C by virtue of a suitable bias network which is well known as such and, therefore, not shown. 4 denotes a combiner network for combining RF output signals from the main and auxiliary amplifier devices at the output of the RF power amplifier 10 into a load ZL, for example an antenna network.

For efficient operation of an RF power amplifier, the output netwo k that connects the transistor output terminals to a common load needs to be designed according to certain rules . In case of a Doherty amplifier it can be advantageous to use a λ/4 transmission line as the combiner network 4 or any- other suitable reactive network. Furthermore, the output signals (RF currents and voltages) need to be shaped to include both linear and nonlinear components, the nonlinear components being nonli ear functions of the desired RF output signal envelope. This invention deals with this shaping o these nonlinear parts, or a combination of nonlinear and linear parts, of the output signals.

The nonlinear parts of the output signals can be seen as extra voltages ( compared to a regular amplifier) that push up the transistor output voltages so they stay close to the maximum allowed voltage for the upper part of the amplitude range . They can alternatively be seen as the extra currents that push down (reduce) the RF currents so that their sum stays close to a parabolic curve {vs. composite amplifier output voltage) . (An entirely parabolic sum current gives constant efficiency under constant supply voltage.)

The nonlinear parts of the output currents or the output voltages are, because they are nonlinear functions of the (envelope of the) RF signal, more wideband than the output- signal ,

The wideband nonlinear components in the outputs of the transistors (or at least one of them) have to have high precision i order to achieve he linearity and efficiency required in many applications. This means that they often in practice have to be created in the digital domain. When they are created digitally, they must each fit inside the

bandwidth of a Digital - to-Analog Converter (DAC) which can be comprised, for example , in the signal processing circui {3} shown in Fig. 1, or at any other suitable location. A

wideband DAC is more expensive than a narrowband one. For wideband systems the large bandwidth expansion factor

combined with the large output signal bandwidth may mean that the appropriate DAC is prohibitively expensive, or even impossible to manufacture.

After being converted into analog form, the wideband

nonlinear signals must be translated to the ight frequency, filtered and amplified. These processing stages become more difficult the more wideband the signals are.

A high order f nction follows more closely the ideal

nonlinear ( sharp-kneed) curve than does a low-order nonlinear function. The difference between the ideal and actual transistor voltage amplitudes is therefore less than for the low-order function, resulting in higher efficiency.

However, the bandwidth of the high-order function is larger than that of the low-order function.

Scate-of -the art static nonlinear functions for use for example in Doner y amplifiers are thus either wideband (high order) and give high efficiency, or are more narrowband (for example low order polynomials) but give low efficiency.

Requirements for high efficiency and low bandwidth expansion are met simultaneously by using dynamic nonlinear functions. This removes the drawbacks of both high- order and low-order static nonlinear functions by, in effect, being low order (and therefore having low bandwidth expansion) for fast. (i.e. wideband) signal variations, and being high order (and having high efficiency) for slow (i.e. narrowband) signal

variations .

As shown in Fig. 2A, a dynamic nonlinear function at an average (but not ail the time) stays closer to the ideal, sharp-kneed, function, especially in the knee region. This beha iour gives a higher average efficiency of the amplifier. The corresponding efficiency curves are shown in Fig. 2B.

Fig. 3 shows an example block diagram of the signal

processing section 3 shown in Fig. 1. In Fig. 3, 31 denotes a non- linear processing section that provides a dynamically non- linear dependency fl from an amplitude characteristic of the input signal In input into the RF power amplifier circuit 10 shown in Fig. 1. Which particular amplitude characteristic will be used, depends on whether the input signal In is an RF signal with varying signal envelope, or whether according to another embodiment the signal In is a base band

representation of the signal envelope of the RF signal to be generated by the power amplifier circuit 10, or whether the input signal In is a digital representation of the signal envelope to be to be generated by the RF power amplifier 10 according to yet another embodiment . Of course , the present invention is not limited to any of these particular

embodiments. 32 denotes a delay element to provide a delayed version InD of the amplitude characteristic of the input signal In to the RF power amplifier circuit as shown in Fig. 1. 33 denotes a subtracter for subtracting the dynamically non- linearly processed input signal from the delayed version of the input signal In. The delayed version of the input signal and the output of the subtractor 33 are further processed in Up--Converters 34 and 35 , respectively, i order to provide the RF signals to be amplified by the main

amplifier deveice and auxiliary amplifier device, resp.

Fig. 4 shows another example block diagram of the signal processing section 3 shown in Fig. 1. In Fig. 4, the same reference numerals are used as in Fig . 3 to denote the same or corresponding parts, such that these parts need not be described again . 36 denotes an adder for adding a filtered version of the output from the dynamically no - linear

processing section 31 to the delayed version InD of the amplitude characteristic of the input signal In. H12 and K22 denote filter circuits for cross coupling the branch signals in the main amplifier branch (Outl) and the auxiliary amplifier branch (Out2 ) . In this example embodiment, also th main amplifier signal branch includes a portion of the dynamic non- linear function. The filters H12 and H22 are designed such that the dynamic non- linear function cancels a the output of the RF power amplifier 10.

Instead of being static functions of the input signal, the nonlinear signals of this invention are dynamic functions, which are by way of non- limiting example embodiments created by nonlinear filtering operations. In a preferred example embodiment shown in Fig . 5 and in another example embodiment shown in Fig. 6, the nonlinear filtering is carried out with respect to one or more boundaries . The boundaries can for example be constant amplitude limits or functions of the input signal amplitude or input signal level.

An exemplary embodiment of a dynamic non- linear processing section shown in Fig. 5 uses a cascade of stages 3 Hi, i = 1 .. n, each extracting in section 320 overshooting parts (with respect to an amplitude boundary) of the output signal In 31 Out31 of the previous stage, the extracted overshoots being filtered in section 321 to the desired bandwidth (and placement) before reinsertion to the signal by means of an adder 323.

The bandwidth- restricted nonlinear signals can in many cases be created and successfully applied without detailed

knowledge of the output network's frequency dependences . A more optimal design may include the output network's

frequency dependence.

In each stage shown in Fig, 5, the overshooting {above a certain level or outside an amplitude-dependent boundary) part of the signal coming from a previous stage passes through a filter that restricts its bandwidth. The result is then subtracted from the signal from the previous stage. In this way, a band-limited and amplitude-shaped signal is built up stage-by- stage (from generally smaller and smaller contributions) .

An additional filter at the input can be used to obtain a desired frequency response, for example one that gives maximum output power or maximal efficiency in the whole bandwidth, at the output node of the main amplifier.

For Doherty amplifiers in which one can disregard the

frequency dependence in the output network, the nonlinear signal can be shaped independently using a single amplitude function as boundary. This translates into a cylindrical boundary in the 3 -dimensional (complex signal as tv/o

dimensions plus the signal's amplitude as the third) space. In the Doherty case, the amplitude limit is constant, usually set to half the maximum output level. The resulting signal can be subtracted from a scaled version of the original signal, or a scaled version of the original signal can be subtracted from it. These schemes result in two canonical nonlinear signal versions that can be applied in Doherty amplifiers. The signals that are input to the power

transistors are made from combinations of scaled linear and nonlinear signals (made with the present method) . The scaling can ( suboptimaily} be made frequency dependent, with

generally different frequency dependencies for the different components and for the same component in the different combinations .

Fig. 6 shows another example block diagram of a dynamic nonlinear processing section for dynamic non- linear signal shaping .

Here , the peak (auxiliary) amplifier drive signal is built up from zero by successively adding filtered, rectified

differences between the signal under construction and the desired (without bandwidth restrictions} nonlinear signal. In correctly tuned Dcherty amplifiers in which one cannot disregard the frequency dependence in the output combiner network, it is more optimal to shape the nonlinear signals using information about the output network. This can be for example in the form of Z-parameters , describing the voltage responses at nodes in the output network in response to current excitations. In Dcherty- 1 ike amplifiers only the "main" amplifier output node may need to be precisely emulated, since only the voltage at this node directly needs to be nonii early shaped using a boundary function.

Signal shaping for a Doherty amplifier with frequency

dependence in the output network can thus be achieved with the structure shown in Fig. 6, with the input signal

optionally pre- filtered to achieve flat frequency response at the output node (by Zli/Zol, Z being Z- arameters of the output network and subscript o referring to output node) .

Another method for generating the dynamic nonlinear function of Doherty and similar amplifiers includes generating the optimal drive signals without bandwidth limitations. The overshooting parts {outside of boundaries) of bandwidth- limited versions of these signals are then extracted, recalculated into reinsertion signals, and filtered with the bandwidth- limiting filter in subsequent stages. By storing the derivatives of the efficiency with respect to the drive signals, the optimal direction for the reinsertion signal can be calculated. This works for moderate bandlimiting , since the filtered signals are close to the optimal ones.

Yet another method is similar to the previous one, except that the optimal direction of reinsertion is assumed to be in the direction of the optimal drive signals without bandwidth limitations. This way, the derivatives do not have to be stored. The starting material is the drive signal that would give the highest efficiency without voltage limitations.

Usually this means that a single transistor is driven. Fig. 7 shows an embodiment of a method of providing a pov/er amplified radio frequency signal having a variable signal amplitude. In his flow, the operations S12 and S22 relate to processing signals for the auxiliary amplifier branch 2 shown in Fig . 1. The operations Sll and S21 relate to processing signals for the main amplifier 1 branch shown in Fig. 1. In operation Sll, the input signal In to the RF power amplifier 10 shown in Fig, 1 is processed to obtain the signal Out 1 for driving the main amplifier 1. I operation S12 , the input signal In is processed with dynamic non-linearity to obtain the signal Out2 for driving the auxiliary amplifier 2. In the operations S21 and S22, resp . , the main amplifier device and the auxiliary amplifier device are driven in accordance with the signals Out 1 and Out2 , res . In operation S3, the outputs of the main amplifier device and of the auxiliary amplifier device are combined together using the combiner network. 4 shown in Fig. 1 in order to feed the combined output to a load ZL, as shown in S4.

Advantages of the Invention

The invention allows operation with high efficiency, under restricted bandwidth conditions. The bandwidth expansion may be considerably reduced . Consequently, the invention also allows operation with increased signal bandwidth while retaining an acceptable efficiency.

The embodiments described above are to be understood as a few illustrative examples of the present invention . I v/iil be understood by those skilled in the ar that various

modifications , combinations and changes may be made to the embodiments without departing from the scope of the present invention which is defined by the appended claims.