WO2004059832A2

WO2004059832A2 - Bias-management system and method for programmable rf power amplifier

Info

Publication number: WO2004059832A2
Application number: PCT/US2003/041637
Authority: WO
Inventors: David J. Hedberg; James B. Turner
Original assignee: Globespan Virata, Inc.
Priority date: 2002-12-23
Filing date: 2003-12-22
Publication date: 2004-07-15
Also published as: AU2003303446A1; WO2004059832A3; US6831517B1; AU2003303446A8

Abstract

A system (100) and method (200) for adaptively managing bias of an RF power amplifier (102) is provided. The system (100) incorporates a controller (116) configured to select a radio operating mode. A current-mirror circuit (114) is coupled to the controller (116) and configured to produce a reference current (IRef) as a function of the radio operating mode. A bias regulator (104) is coupled to the controller (116) and the current-mirror circuit (114) and configured to produce a driver-stage bias current (Ib1) and an output-stage bias current (Ib2) for the power amplifier (102) in response to the reference current (IRef). The system (100) also incorporates a DC-to-DC converter (118) coupled to the controller (116) and configured to provide a supply voltage (Vcc) for the power amplifier (102) in response to the radio operating mode. The system (100) also incorporates an envelope detector (120) configured to produce an envelope current (IEnv) in response to an RF input signal (126). The system (100) causes the reference current (IRef) to vary as a function of the envelope current (IEnv).

Description

BIAS-MANAGEMENT SYSTEM AND METHOD FOR PROGRAMMABLE RF POWER AMPLIFIER

TECHNICAL FIELD The present invention relates to the field of transmitter power amplifiers. More specifically, the present invention relates to the field of energy conservation and linearity improvements in transmitter power amplifiers for wireless communications systems.

BACKGROUND ART The demands for low cost and power efficient wireless local-area network (WLAN) radios capable of providing a wide range of data rates and operating modes have recently accelerated. This has put more emphasis on achieving low cost radios with low power consumption. Particularly with requirements for integrated radios that operate with both direct- sequence spread spectrum (DSSS) coded waveforms and orthogonal frequency-division multiplexing (OFDM) waveforms as found in the IEEE 802.11a, b, g, and h standards, the transmitter power amplifier should be capable of efficient operation in several transmission modes and at several transmit power levels. The DSSS PSK and complementary code keying (CCK) waveforms of the IEEE 802.11b standard typically have lower peak to average ratio envelope distributions than the OFDM waveforms required in IEEE 802.11a and g standards. Moreover, efficient transmissions of OFDM wavefoπns at different data rates may utilize different amounts of power-amplifier backoff from the compression point to maintain optimum output power and signal fidelity. Transmitted OFDM waveforms for IEEE 802.11a and g modes can have peak-to-average ratios of 12 dB or more in the highest data rate modes. The later modes typically require a greater amount of linearity for a given output power than for the lower data rate modes.

Historically, RF power amplifiers have had to accommodate a smaller range of requirements. This allowed them to be implemented with fixed voltage and current bias techniques. Typical IEEE 802.1 lb "Wireless Fidelity" (WiFi) radios utilize a fixed-bias class-A or class-AB biased RF power amplifier. Similarly, for IEEE 802.1 la OFDM radios, a fixed bias class A or highly biased class-AB RF power amplifier is common. Due to the linearity demands of OFDM transmission at higher data rates, if fixed bias is used, the power amplifier output capability must be sized larger in order to transmit at similar power levels as with the less demanding IEEE 802.1 lb modulation modes.

A commonly applied technique to improve linearity in power amplifiers is to increase the quiescent current so that the output transistor operates in a more linear portion of the I/V curve. However, higher quiescent current also results in greater power consumption during times when the power amplifier is transmitting. For maximum efficiency in a class-AB biased amplifier, the bias current and voltage should be selected carefully for transmit mode and output power level required. Improved efficiency results if the quiescent current and power-amplifier supply voltages are varied dynamically with the instantaneous power (envelope) of the transmitted signal. But optimizing the bias current and voltage for each signal level is difficult in a highly integrated radio product that is constrained by integrated-circuit (IC) technology cost and power efficiency considerations.

Improved power amplifier characteristics (e.g., linearity and maximum power output) and efficiency generally accrue when the IC process technology is well matched to the frequency range and circuit requirements of the power amplifier. Unfortunately, highest power amplifier performance levels can best be achieved with processes that also tend to be more costly and limited in circuit integration capabilities, e.g., gallium-arsenide (GaAs) metal semiconductor field-effect transistor (MESFET) and heterojunction bipolar transistor (HBT) technologies. This has inhibited the use of sophisticated biasing and bias management circuits within the RF power amplifier ICs in the highest performance applications, which could benefit from such circuits, because they could not be integrated easily due to device and cost constraints of the RF power amplifier IC processes.

On the other hand, the RF transceiver and radio ICs are implemented today as complex mixed signal circuits. These systems are desirably highly integrated in minimum cost IC processes, e.g., complementary metal-oxide semiconductor (CMOS) or bipolar complementary metal-oxide semiconductor (BiCMOS) IC processes.

DISCLOSURE OF INVENTION Accordingly, it is an advantage of the present invention that a bias-management system and method for a programmable RF power amplifier is provided.

It is another advantage of the present invention that a programmable bias-management system is provided that optimizes the performance of an RF power amplifier for multiple radio operating modes. It is another advantage of the present invention that a bias-management system is provided in which a reference transistor of a bias regulator and an output transistor of a power amplifier are fabricated upon a common integrated-circuit substrate to provide close thermal coupling between the bias regulator and the power amplifier. It is another advantage of the present invention that a bias-management system is provided in which a bias current for a power amplifier is dynamically adjusted as a function of an envelope of an input signal.

It is another advantage of the present invention that a bias-management system is provided in which a bias current for a power amplifier is dynamically adjusted to compensate for signal rectification within an output stage of the power amplifier.

The above and other advantages of the present invention are carried out in one form by an RF power-amplifier bias-management system incorporating a digitally programmed controller configured to according to a selected radio operating mode, a current-mirror circuit coupled to the controller and configured to produce a reference current as a function of the radio operating mode, a bias regulator coupled to the controller and the current-mirror circuit and configured to respond to the reference current, and a power-amplifier output stage coupled to the bias regulator and having a bias current determined by the bias regulator.

The above and other advantages of the present invention are carried out in another form by a method of adaptively controlling operation of an RF power amplifier. The method incorporates selecting a radio operating mode, producing a reference current in response to the selecting activity, establishing a bias current for an output stage of the power amplifier in response to the producing activity, and determining a supply voltage for the power amplifier in response to the selecting activity.

BRIEF DESCRIPTION OF DRAWINGS

A more complete understanding of the present invention may be derived by referring to the detailed description and claims when considered in connection with the Figures, wherein like reference numbers refer to similar items throughout the Figures, and: FIG. 1 depicts a schematic of an RF power-amplifier bias-management system in accordance with a preferred embodiment of the present invention;

FIG. 2 depicts a flowchart of a process to adaptively control operation of an RF power amplifier in accordance with a preferred embodiment of the present invention;

FIG. 3 depicts a flowchart of a subprocess to fabricate thermally linked components for a bias-management system in accordance with a preferred embodiment of the present invention;

FIG. 4 depicts a flowchart of a subprocess to establish quiescent bias currents for an RF power amplifier in accordance with a preferred embodiment of the present invention;

FIG. 5 depicts a graph of gain versus output power for multiple quiescent bias currents in accordance with a preferred embodiment of the present invention; FIG. 6 depicts a flowchart of a subprocess to compensate quiescent bias currents for thermal variations in an RF power amplifier in accordance with a preferred embodiment of the present invention;

FIG. 7 depicts a flowchart of a subprocess to adaptively adjust bias currents for an RF power amplifier in response to an envelope of an input signal in accordance with a preferred embodiment of the present invention;

FIG. 8 depicts a flowchart of a subprocess to adaptively adjust bias currents for an RF power amplifier in response to signal rectification in an output stage of the power amplifier in accordance with a preferred embodiment of the present invention; FIG. 9 depicts a graph of gain versus output power for quiescent and dynamic bias currents in accordance with a preferred embodiment of the present invention; and

FIG. 10 depicts a graph of average collector current versus output power for quiescent and dynamic bias currents in accordance with a preferred embodiment of the present invention.

BEST MODE FOR CARRYING OUT THE INVENTION

FIG. 1 depicts a schematic of bias-management system 100 for an RF power-amplifier 102, and FIG. 2 depicts a flowchart of a process 200 to adaptively control operation of RF power amplifier 102 in accordance with a preferred embodiment of the present invention. The following discussion refers to FIGs. 1 and 2. In the preferred embodiment of FIG. 1, power amplifier 102 is typical of a power amplifier configured to meet the requirements of direct-sequence spread spectrum (DSSS) coded waveforms and orthogonal frequency-division multiplexing (OFDM) transmissions as found in the IEEE 802.11a, b, g, and h standards for a wireless local-area network (WLAN) radio (not shown). As such, RF power amplifier 102 should be capable of efficient operation in several transmission modes and at several transmit-power levels. For example, the DSSS phase- shift keying (PSK) and complementary code keying (CCK) waveforms of the IEEE 802.11b standard typically have lower peak-to-average ratio envelope distributions than the OFDM waveforms of the IEEE 802.1 la and g standards. Thus, the PSK and CCK mode waveforms of IEEE 802.1 lb can be transmitted at a higher output power level using less output power back- off than for the OFDM mode waveforms that have higher peak-to-average power ratios.

Those skilled in the art will appreciate that bias-management system 100 as depicted in FIG. 1 is exemplary, and that other embodiments of RF power-amplifier 102 and/or other components of system 100 may be implemented without departing from the spirit of the present invention. Bias-management system 100 includes three primary sections: RF power amplifier 102; a bias regulator 104; and a bias reference current preprocessor 106. RF power amplifier 102 contains an output stage 108 and a gain or driver stage 110. Preprocessor 106 contains a digital- to-analog converter (DAC) 112 and a current-mirror circuit 114. Also included in the preferred embodiment of system 100 are a system controller 116, a dc-to-dc converter 118, and an envelope detector 120. Each of these basic components of system 100 is discussed in detail hereinafter.

FIG. 3 depicts a flowchart of a subprocess 202 to fabricate thermally linked components for bias-management system 100 in accordance with a preferred embodiment of the present invention. The following discussion refers to FIGs. 1-2, and 3.

Conventionally, the components of an RF power amplifier are fabricated upon a first substrate using a first integrated circuit (IC) technology suited to power-amplifier^' applications, i.e., with a process technology capable of providing the necessary power at the desired transmission frequencies. Similarly, the components of other portions of a WLAN radio are conventionally fabricated upon a second substrate using a second IC technology suited to the support circuitry, i.e., having the lowest cost at the desired frequencies. This is true of the present invention as well, where RF power amplifier 102 is fabricated upon a first substrate 122 and preprocessor 106 is fabricated upon a second substrate 124.

In the present invention, however, it is highly desirable that an output transistor Q2 of RF power amplifier 102 be thermally coupled to a reference transistor Q3 of bias regulator 104. This is accomplished during the manufacture of physical components of a WLAN radio by fabricating specific components upon specific substrates in a subprocess 202 of operation process 200. Those skilled in the art will appreciate that the term "operation" in this discussion is taken to mean not only the operation of a finished and complete WLAN radio, but also those parts of the manufacturing process that affect the operation of the finished radio.

For simplicity, only those components of bias-management system 100 are discussed in fabrication subprocess 202. It is to be assumed that all other components, including those neither depicted in the Figures nor discussed herein, are also fabricated in manners well known to those skilled in the art. In fabrication subprocess 202, a "hypertask" (i.e., a task of tasks) 204 fabricates all those components that are to be fabricated upon first substrate 122 using a first IC technology. In a task 206 of hypertask 204, that portion of output stage 108 of power amplifier 102 containing output transistor Q2 is fabricated. Similarly, in a task 208 of hypertask 204, that portion of bias regulator 204 containing reference transistor Q3 within bias regulator 104 is fabricated. By being fabricated upon a common substrate 122, output transistor Q2 and reference transistor Q3 are thermally coupled such that a change in the characteristics of output transistor Q2 due to thermal variations produces a corresponding change in the characteristics of reference transistor Q3. In addition, fabrication on a common substrate results in correlated process-dependent characteristics such as Vbe match, which ensures the current ratio in these transistors are determined by size ratio and not process characteristic differences.

To a lesser degree, it is desirable that that portion of driver stage 110 of power amplifier 102 containing driver transistor Ql also be thermally matched and characteristically matched to output transistor Q2. This is effected by a in a task 210 of hypertask 204, wherein driver transistor Ql is fabricated. Desirably, the IC technology used for the fabrication of transistors Ql, Q2, and Q3 (along with various other components) is a technology that is suitable for use with RF power amplifier

102. In the preferred embodiment, this technology is a heterojunction bipolar transistor (HBT) technology, though it will be understood that this is not a requirement of the present invention.

Those skilled in the art will appreciate that the process flow discussed herein in conjunction with hypertask 204 of subprocess 202 does not indicate sequence. For example, that tasks 206, 208, and 210 of hypertask 204 would be effected substantially simultaneously. That is, output, reference, and driver transistors Q2, Q3, and Ql would be substantially simultaneously fabricated upon substrate 122.

Also in fabrication subprocess 202, a hypertask 212 fabricates preprocessor 106 and other components that are to be fabricated upon second substrate 124 using a second IC technology. In a task 214, DAC 112 is fabricated, and in a task 216, current-mirror circuit 114 is fabricated.

Desirably, the IC technology used for the fabrication of DAC 112 and current-mirror circuits 114 is a technology that is suitable for use with large scale integration of radio transceiver circuits or a mixed signal system-on-a-chip IC. In the preferred embodiment, this technology is complementary metal-oxide semiconductor (CMOS) or bipolar complementary metal-oxide semiconductor (BiCMOS) technology, though it will be understood that this is not a requirement of the present invention.

Those skilled in the art will again appreciate that the process flow discussed herein in conjunction with hypertask 212 of subprocess 202 does not indicate sequence. For example, tasks 214 and 216 of hypertask 212 would be effected substantially simultaneously.

It may be seen in FIG. 1 that bias regulator 104 also contains a source transistor Q4. In the preferred embodiment, source transistor Q4 need not be thermally coupled to output transistor Q2. Therefore, though not shown in FIG. 3, source transistor Q4 may be fabricated on first substrate 122 by hypertask 204 using the first IC technology, or on second substrate 124 by hypertask 212 using the second IC technology. In practice, source transistor Q4 would normally be fabricated on whichever one of substrates 122 or 124 is more cost effective. Alternatively, source transistor Q4 could be implemented as a low-cost discrete part if neither substrate 122 nor 124 can provide the needed device characteristics.

In the present invention, bias currents Ibl and Ib2 for driver and output transistors Ql and Q2, respectively, are produced and controlled by bias regulator 104. Each of bias currents Ibl and Ib2 has a quiescent component and a dynamic component. The quiescent component is a DC (no-signal) bias-current value. The dynamic component is a dynamic variation superimposed upon the quiescent component in response to the signal. Alternatively, driver- transistor bias current Ibl can be connected to a fixed source, independent of bias regulator 104. Before the dynamic component can be superimposed upon the quiescent component, the quiescent components of bias currents Ibl and Ib2, must first be established. To this end, bias- management system 100 is momentarily assumed to be simplified by the omission of controller 116, DC-to-DC converter 118, preprocessor 106 (i.e., DAC 112 and current mirror circuit 114), and envelope detector 120. As a result, the power source for transistors Ql, Q2, and Q4 are momentarily presumed to be system primary voltage Vdd.

Quiescent DC bias current Ib2 for output transistor Q2 is set by setting a bias reference current IRef. Typically, a ratio N of a size of output transistor Q2 to a size of reference transistor Q3, and a corresponding ratio in the resistances of their base resistors Rb2 and Rb3, will be set so that a quiescent DC collector current Ic2 in output transistor Q2 will be a factor of N times greater than bias reference current IRef. The resultant DC base-emitter voltages Vbe2 and Vbe3 of output transistor Q2 and reference transistor Q3 will be substantially equal. A typical relationship is given in the following equations:

Ic2 = lRef/ N; (1)

Rb3 = N ^• Rb2; and (2)

Vbe3 £ Vbe2. (3)

Since, as discussed hereinbefore, output transistor Q2 and reference transistor Q3 are formed upon common substrate 122, base-emitter voltage Vbe2 and Vbe3 will tend to track with temperature and process variations. This results in a substantially constant ratio between output-transistor collector current Ic2 and reference current IRef.

Reference transistor Q3 is fabricated on a common substrate 122 with output transistor Q2. As discussed hereinbefore, driver transistor Ql is also fabricated upon common substrate 122. Therefore, quiescent DC bias current Ibl for driver transistor Ql is also set by reference current IRef, as determined by a ratio M of a size of driver transistor Ql to the size of reference transistor Q3, and a corresponding ratio in the resistances of their base resistors Rbl and Rb3. The resultant DC base-emitter voltages Vbel, Vbe2, and Vbe3 of driver transistor Ql, output transistor Q2, and reference transistor Q3 will all be substantially equal, and will all tend to track with temperature and process variations. A typical relationship is given in the following equations:

Icl = lRef/ M; (4)

Rb3 = N - Rbl; and (5)

Vbe3 S Vbe2 ≥ Vbel. (6)

The matching of the electrical and thermal characteristics of transistors Ql, Q2, and Q3 is advantageous. Because of this, transistors Ql, Q2, and Q3 are fabricated on common substrate 122. Bias-regulator source transistor Q4 sources the bias (base) currents Ibl, Ib2, and Ib3 of driver transistor Ql, output transistor Q2, and reference transistor Q3. Source transistor Q4 does not require matched characteristics, and therefore may be fabricated elsewhere than on common substrate 122, if desired. In practice, source transistor Q4 may be fabricated upon power- amplifier substrate 122 or preprocessor substrate 124, whichever realizes the most effective fiscal solution.

The DC base bias currents in Ql, Q2, and Q3 depend directly on a bias source voltage VBias from the emitter of source transistor Q4:

VBias = Vbel + Rbl ^• Ibl = Vbe2 + Rb2 ^■ Ib2 = Vbe3 + Rb3 ^■ Ib3. (7)

FIG. 4 depicts a flowchart of a subprocess 218 to establish quiescent bias currents Ql and Q2 for RF power amplifier 102 in accordance with a preferred embodiment of the present invention. The following discussion refers to FIGs. 1, 2, and 4. FIG. 1 is hereinafter considered in toto.

In subprocess 218, a task 220 selects an operating mode for the radio. Bias-management system 100 incorporates system controller 116. In the preferred embodiment, system controller

116 provides a typical media-access controller (MAC) function or the like. Those skilled in the art will appreciate that in some system-on-a-chip implementations, system controller 116 and preprocessor circuit 106 may be integrated on a common substrate. The implementation of system controller 116 and preprocessor circuit 106 on a common substrate does not depart from the spirit of the present invention.

System controller 116 is configured to select an operating mode for the radio in task 220. For example, a multi-mode WLAN radio may desirably operate as either an IEEE 802.1 lb or an

IEEE 802.1 lg radio. The transmission demands of the DSSS waveforms of an IEEE 802.11b radio are different than the transmission demands of the OFDM waveforms of an IEEE 802.1 lg radio.

Those skilled in the art will appreciate that, when bias-management system 100 is used with a single-mode WLAN radio, then system controller 116 may be hardwired, i.e., effectively replaced by connections providing the appropriate signals to preprocessor 106, etc. The omission of controller 116 does not depart from the spirit of the present invention.

Controller 116 allows the operator (either human or computer) to select the radio operating mode and establish an appropriate set of conditions for power amplifier 102. This set of conditions may include a supply voltage Vcc and driver- and output-transistor bias currents Ibl and Ib2.

In a task 222 of subprocess 218, a suitable mode-specific supply voltage Vcc is determined for power amplifier 102. An output of system controller 116 is passed to preprocessor 106 to allow establishment of specific quiescent bias currents Ibl and Ib2 to optimize power amplifier 102 for a given mode of operation (discussed hereinafter). This leads to the realization that, in some modes of operation, power amplifier 102 is better optimized with a specific power-amplifier supply voltage Vcc, rather than system primary voltage Vdd. Therefore, it is desirable that power amplifier 102 be provided with mode-specific supply voltage Vcc Mode-specific supply voltage Vcc is derived from system primary voltage Vdd by high- efficiency switching DC-to-DC converter 118 under control of system controller 116. Supply voltage Vcc is fed to the collectors of driver and output transistors Ql and Q2 through collector inductors Lcl and Lc2, respectively. In this manner, controller 116 optimizes power-amplifier supply voltage Vcc for each mode of operation in a multi-mode WLAN radio. The inclusion of DC-to-DC converter 118 allows the achievement of greater efficiencies of power amplifier 102 that are possible otherwise.

Those skilled in the art will appreciate that some embodiments of the present invention may not require the use of DC-to-DC converter 118. For example, where power-amplifier supply voltage Vcc has the same optimal value for all modes of operation peculiar to a specific WLAN radio, where the WLAN radio is a single-mode radio, or where a simpler >„ implementation of bias-management system 100 is desired. In such a case, power-amplifier supply voltage Vcc may be derived directly from system primary voltage Vdd. It will be appreciated that the omission of DC-to-DC converter does not depart from the spirit of the present invention. In a task 224 of subprocess 218, mode-specific quiescent reference current IRef is produced. In task 220, discussed hereinbefore, system controller 116 selects the radio operating mode. This is passed to preprocessor 106 as operating-mode control signals 126. Control signals 126 pass to DAC 112 and to current-mirror circuit 114. Cuιτent-mirror circuit 114 produces a programmable dynamic bias current, which is summed with an output of DAC 112 to form a reference control current IDAC Collector current IC4 of transistor Q4 is supplied from a positive supply voltage Vcca, which is typically also the supply voltage for preprocessor circuit 106.

A small fixed bias source current lb is provided by preprocessor circuit 106 in order to provide a base current Ib4 for source transistor Q4. In this circuit, fixed bias source current lb is substantially greater than a required source-transistor base current Ib4. A reference current IRef is therefore the sum of fixed bias source current lb (minus source-transistor base current Ib4) and reference control current IDAC.

Reference current IRef is much greater than source-transistor base current Ib4 (which is also a collector current IC3 of reference transistor Q3). Reference current IRef is therefore approximated by the following equations:

lRef= Ib + lDAC - Ib4, (8)

lb > Ib4, and (9)

iRef≤ Ib + lDAC. (10)

In a task 226 of subprocess 218, quiescent bias current Ib2 for output stage 108 of RF power amplifier 102 is produced by bias regulator 104. Reference current IRef passes through reference transistor. The value of reference current IRef is reflected in base-emitter voltage Vbe3.

Base current Ib3 is a function of bias voltage VBias, base-emitter voltage Vbe3, and the value of base resistor Rb3:

Ib3 = (VBias - Vbe3) / Rb3. (11) This serves to establish bias source voltage VBias.

Since, as discussed hereinbefore, output and reference transistors Q2 and Q3 are fabricated upon the same substrate with a predetermined size ratio N, which ratio N is also present in base resistors Rb2 and Rb3, output-stage bias current (i.e., output-transistor base current) Ib2 is also determined:

Vbe3 = Vbe2, (12)

Rb2 = N - Rb3, (13)

Ib2 = (VBias - Vbe2) / Rb2 = N - Ib3. (14)

Similarly, in a task 228 of subprocess 218, quiescent bias current Ibl for driver stage 110 of RF power amplifier 102 is produced by bias regulator 104:

Ibl = (VBias - Vbel) / Rbl = M - Ib3. (15)

By properly adjusting current mirror circuit 114 under the control of system controller 116, subprocess 218 may produce any reasonable value for reference current IRef. This allows any reasonable value for output-stage bias current Ib2, and directly affects the performance of power amplifier 102. FIG. 5 depicts a graph 300 of normalized power amplifier gain 302 versus output power 304 for multiple quiescent output-transistor bias currents Ib2 in accordance with a preferred embodiment of the present invention. The gain curves are normalized with respect to their low level, small signal gains to better show relative behavior. The following discussion refers to FIGs. 1, 2, and 5.

FIG. 5 shows exemplary values of normalized output-stage bias current Ib2 for use with varying power-amplifier operating modes. For each of these curves, Vcc is 3.0 volts, source resistances are nominally 50 Ω, and mid-frequency transistor current gains are nominally 80 dB.

Curve 306 depicts normalized gain 302 versus output power 304 for an exemplary power amplifier with near class-A operation with a bias current Ic2 of 145 mA. At this bias, a -1 dB gain 312 results in an approximately +21.6 dBm output power 314. Curves 308 and 310 depict the results of lowering bias current Ic2 to operate more deeply towards class-AB operation. With curves 308 and 310, output power 304 at 1 dB gain compression has fallen substantially due to the effects of output current clipping with reduced quiescent bias. Curve 308 depicts nonualized gain 302 versus output power 304 for class-AB operation with a bias current Ic2 of 68 mA. At this bias, a -1 dB gain 312 results in an approximately +19.1 dBm output power 316. Similarly, curve 310 depicts normalized gain 302 versus output power 304 for class-AB operation with a bias current Ic2 of 45 rrιA,r At this bias, a -1 dB gain 312 results in an approximately +17.9 dBm output power 318. Note, however, that the linearity at lower output power levels 304 is not degraded significantly as quiescent bias current Ic2 is reduced. In fact, although not shown explicitly in chart 300, the linearity in some class -AB amplifiers for quiescent currents Ic2 of 45 mA and below may improve slightly due to the effects of a gain extension characteristic as average collector current increases at increased output power levels 304 due to RF rectification. Those skilled in the art will appreciate that curves 306, 308, and 310 are normalized to the same gain at low power. However, in reality, it is recognized that amplifier gain will drop slightly when DC quiescent bias is reduced.

As depicted in chart 300, power amplifier 102 operates with near full class-A quiescent bias in order to achieve adequate linearity at high output power 304 (on the order of +20 dBm). Chart 300 also shows that quiescent bias current Ib2 may be decreased when lower output operating power is acceptable. With lowered bias current Ib2, power amplifier 102 retains the desired linearity. Thus, efficiency can be improved at these lower operating power levels.

FIG. 6 depicts a flowchart of a subprocess 230 to compensate quiescent bias currents Ibl ad Ib2 for thermal variations in RF power amplifier 102 in accordance with a preferred embodiment of the present invention. The following discussion refers to FIGs. 1, 2, and 6.

As discussed hereinbefore, reference transistor Q3 is fabricated upon common substrate 122 with output and driver transistors Q2 and Ql of power amplifier 102. This causes transistors Ql, Q2, and Q3 to be thermally coupled. It is this thermal coupling that allows subprocess 230 to function. Once subprocess 218 has established quiescent bias currents Ibl and Ib2 for RF power amplifier 102, then subprocess 230 of process 200 compensates quiescent bias currents Ibl and Ib2 for thermal variations in power-amplifier substrate 122. In a pseudotask 232 of subprocess 230, driver, output and reference transistors Ql, Q2, and Q3 are thermally coupled. Since, in subprocess 202 discussed hereinbefore, transistors Ql, Q2, and Q3 were fabricated upon common substrate 122, transistors Ql, Q2, and Q3 are inherently thermally coupled. Pseudotask 232, therefore, is not actually performed. Rather, pseudotask 232 is a confirmation of subprocess 202.

Power-amplifier substrate 122 is subject to thermal variations. These variations may be produced either as a result of environmental conditions or as a result of power consumption within power amplifier 102. In either case, thermal variations in substrate 122 lead to variations in the conductivity of the base-emitter junction of output transistor Q2. This change in conductivity is detected in a task 234 as a bias offset (not shown) effecting a change in base- emitter voltage Vbe2 of output transistor Q2.

Since, due to the geometries of transistors Q2 and Q3, base-emitter voltages Vbe2 and Vbe3 of are substantially equal, the bias offset is also induced at reference transistor Q3. As a result, reference-transistor base-emitter voltage Vbe3 changes to track output-transistor base-emitter voltage Vbe2. A task 236 then adjusts reference-transistor bias current Ib2 for the change in base-emitter voltage Vbe3.

In a task 238, reference transistor Q3 responds to the change in bias and adjusts source transistor Q4 to effect a change in reference current IRef and bias source voltage VBias.

The change in reference current IRef causes tasks 240 and 242 to adjust the output-stage and driver-stage bias currents Ib2 and Ibl of output and driver transistors Q2 and Ql of power amplifier 102. The action of subprocess 230 therefore results in reasonable temperature variations having only a small secondary effect upon the performance of RF power amplifier 102.

Those skilled in the art will appreciate that the process flow discussed herein in conjunction with subprocess 230 does not indicate sequence. Sequence is depicted only for clarity of description of the context of the process. For example, tasks 240 and 242 would be effected substantially simultaneously. FIG. 7 depicts a flowchart of a subprocess to adaptively adjust bias currents for an RF power amplifier in response to an envelope of an input signal in accordance with a prefeiτed embodiment of the present invention. The following discussion refers to FIGs. 1, 2, and 7.

Bias-management system 100 offers adaptive dynamic bias control in addition to the quiescent and thermal bias control discussed hereinbefore. In the preferred embodiment, one form of dynamic bias control is achieved tlirough envelope detector 120. This form of dynamic bias control utilizes envelope detector 120 in conjunction with current-mirror circuit 114 to cause bias regulator 104 to dynamically vary the bias currents Ibl and Ib2 of power amplifier 102.

In the absence of an RF input signal 126, envelope detector 120 produces a static envelope current lEnv. Static envelope current lEnv is compensated for in current-mirror circuit

114 and has no significant impact upon power-amplifier quiescent bias currents Ibl and Ib2.

Envelope detector circuit 120 is typically designed to have very small quiescent output current when an RF input signal 128 near zero.

When RF input signal 128 is present, RF input signal 126 enters power amplifier 102 and is passed to envelope detector 120. In a task 246 of subprocess 244, envelope detector 120 detects (e.g., rapidly rectifies) RF input signal 126 to dynamically vary envelope current lEnv in response to an envelope (not shown) of RF input signal 126. Envelope current lEnv is assessed by current-mirror circuit 114 and produces variations in reference control current IDAC This effects, in a task 248, a dynamic adjustment of reference current IRef, and consequently bias currents Ibl and Ib2. Reference current IRef thus varies in response to the envelope of RF input signal 126. That is, dynamic components are added to the static DAC reference current to modulate the reference current IRef, based on the instantaneous RF envelope current lEnv.

The dynamic components of reference current IRef are passed to bias voltage VBias by reference transistor Q3. These dynamic components are the passed to output transistor Q2 in a task 250 and driver transistor Ql in a task 251. That is, dynamic components of driver and output bias currents Ibl and Ib2 are developed. This results in dynamic bias control of driver stage 110 and output stage 108 of power amplifier 102 in response to RF input signal 126.

In the preferred embodiment, both the driver stage and output stage biases are controlled together. Those skilled in the art will appreciate that a suitable alternative is that only the output stage be so controlled, with the driver stage bias operated at a constant bias level. The use of this or other potential alternatives does not depart from the spirit of the present invention.

Reference transistor Q3 and regulator transistor Q4 of bias regulator 104 should respond at speeds that are fast relative to the bandwidth of the envelope of RF input signal 126. This results in a response bandwidth of at least 50 MHz for IEEE 802.11 signals. This dynamic bias control is achieved with current-mirror circuit 114 in preprocessor 106. In current-mirror circuit 114, the RF signal level represented in lEnv, and the output device base-currents, represented in collector current Ic4, can be combined to generate a desirable dynamic correction profile for reference current IRef.

In addition, the preferred embodiment of the present invention includes provisions for DAC 112 to produce reference control current IDAC to control the amount of static DC current in reference current IRef. This in turn sets the small-signal quiescent DC current of output transistor Q2. Desirably, both the quiescent DC current and the amount of RF-envelope dependent correction can be programmed digitally by controller 116 to optimize the efficiency and linearity of output transistor Q2 for a given radio operation mode. In essence, the preferred embodiment of the present invention detects RF input signal 128 at an appropriate point 120 in the signal path. Additionally, a base current demand of transistors Ql and Q2 changes in response to RF input signal 126. This change in base current demand is sensed in a collector current Ic4 of source transistor Q4 and is approximated by the equation:

Ic4 ≤ Ib2 + Ibl, (16) where it is assumed that base current Ib3 of reference transistor Q3 is much smaller than the sum of base currents Ibl plus Ib2.

Reference current IRef (i.e., the collector current Ic3 of reference transistor Q3) is generated as the sum of control currents from DAC 112 and current-mirror circuit 114 as follows:

IRef = kl ^• lEnv + k2 ^• Ic4 + IDAC, (17)

where kl and k2 are selectable gain-ratio constants of current-mirror circuit 114. Reference control current IDAC and current-mirror gain ratio constants kl and k2 may be programmed by system controller 116.

It should be noted that dynamic biasing can still be achieved if either of the two signal- dependent terms lEnv and Ic4 in equation (17) is set equal to zero by setting either constant kl or k2 to zero. In some embodiments, efficiency and linearity improvement may be adequate with only one control mechanism present in the design.

FIG. 8 depicts a flowchart 254 of a subprocess to adaptively adjust bias currents Ibl and Ib2 for RF power amplifier 102 in response to signal rectification in output stage 108 in accordance with a preferred embodiment of the present invention. The following discussion refers to FIGs. 1, 2, and 7.

The preferred embodiment of the present invention compensates for compression induced by an increase in signal strength when output transistor Q2 is biased for class-AB or class-B operation. When the RF signal level increases to large levels, the average base current demand of output transistor Q2 goes up, due to rectification of the RF transistor current at the base. Since the current can only be supplied through Rb2 from VBias, the average base-emitter DC voltage bias Vbe2 of output transistor Q2 decreases. This de-biasing effect results in further compression of the output characteristics with increasing power levels. This mechanism contributes to output power compression in all similar fixed current-source-biased power amplifier output stages and is well known to those skilled in the art. When average bias voltage Vbe2 is decreased with an increase in signal level, the compression characteristic of a power amplifier becomes softer and results in a notable increase in distortion at the higher power levels.

Ideally, bias current Ib2 should increase with an increase in signal level to offset compression. This is accomplished in subprocess 254 passing the bias offset generated by signal rectification in output stage 108 to bias controller 104. Bias controller 104 in turn adjusts bias currents Ibl and Ib2 of power amplifier 102 to compensate for the bias offset.

RF input signal 128 enters driver stage 110 of RF power amplifier 102. Within driver stage 110, RF input signal 128 is amplified to become a RF intermediate signal 130. RF intermediate signal 130 passes from driver stage 110 to output stage 108 of RF power amplifier

102. Within output stage 108, RF intermediate signal 130 is amplified to become an RF output signal 132.

In a task 256, RF intermediate signal 130 is rectified by the base-emitter junction of output transistor Q2 during amplification. This rectification, which is most evident when output transistor Q2 is operating in class-AB or class-B mode, produces a bias offset (not shown). In a task 258, the bias offset is summed with base-emitter voltage Vbe2 and bias current Ib2 is dynamically lowered. Since base-emitter voltages Vbe2 and Vbe3 of output and reference transistors Q2 and Q3 are always substantially equal, a task 260 detects the presence of the bias offset and adjusts the base-emitter voltage Vbe3 of reference transistor Q3 accordingly. In a task 262, the base current Ib2 of reference transistor Q3 is dynamically compensated. This results in a dynamic adjustment of reference current IRef in a task 264. Dynamically adjusted reference current IRef in turn results in dynamically adjusted output-stage and driver-stage bias currents Ib2 and Ibl in tasks 266 and 268.

FIG. 9 depicts a graph 320 of normalized gain 302 versus output power 304 for quiescent and dynamic output-stage bias currents Ib2 and Ibl in accordance with a preferred embodiment of the present invention. The following discussion refers to FIGs. 1 and 9.

Curve 306 depicts normalized gain 302 versus output power 304 for near class-A operation with a fixed bias current Ib2 of 145 mA. At this bias, a -1 dB gain 312 results in an approximately +21.6 dBm output power 314. Curve 322 depicts normalized gain 302 versus output power 304 for near class-A operation with a dynamic bias current Ib2 of 65-130 mA under control of bias-management system 100. At this bias, a -1 dB gain 312 results in an approximately +23 dBm output power 324. This represents a significant increase in output power 304 while average quiescent bias current Ib2 has decreased. Moreover, dynamic-bias curve 322 demonstrates that gain 302 exhibits a slight expansion as the DC-bias level is increased over the range. After taking into account compression characteristics of power-amplifier driver stage 110, linearity has improved substantially over the range of output power 304 as compared to a fixed-bias scheme.

In the preferred embodiment, with dynamic bias regulator 104 configured as shown in FIG. 1, quiescent output bias current Ib2 will automatically drop to approximately 65 mA when the envelope level is small. Thus, with OFDM signals, where the peak to average envelope level can vary as much as 12 dB, the average quiescent bias current will be much lower than 130 mA and closer to 75 mA. Since linearity is also improved, the operating output power back-off level relative to amplifier compression that is consistent with full signal fidelity will also be reduced, resulting in higher output power. Accordingly, subprocess 254 achieves operating power efficiency improvement with active dynamic bias, relative to fixed bias, provided the RF signal is amplitude modulated. Also, an increase of 1-2 dBm in output power 304 is possible for these wavefoπns with active dynamic biasing under the constraint of a fixed power-amplifier collector voltage Vcc.

FIG. 10 depicts a graph 326 of average collector cuπent 328 versus output power 304 for quiescent and dynamic output-stage bias cuπents Ib2 in accordance with a prefeπed embodiment of the present invention. The following discussion refers to FIGs. 1 and 10.

Curve 330 depicts average collector cuπent 328 with quiescent bias of 145 mA for a fixed bias cuπent Ib2. Curve 322 depicts average collector cuπent 328 with quiescent bias varied between 65 mA and 130 mA as a function of output envelope power level for a variable dynamic bias cuπent Ib2. Note that for fixed-bias curve 330, average quiescent bias cuπent Ib2 falls off at higher powers when power amplifier 102 enters compression. This is due to finite DC base resistance Rb2 and the fact that at higher power levels the average demand on base cuπent Ib2 increases due to saturation and rectification of the RF signal.

Operating power consumed from the power supply (and hence the efficiency of power amplifier 102) will be determined by a weighted average based on the signal envelope output power distribution.

The adaptive algorithms of the present invention determine an efficient operating mode for power amplifier 102 that is compatible with both output level and radio modulation mode linearity requirements. Then the parameters of bias regulator 104 are programmed digitally, typically based on a pre-programmed look-up-table in a manner well lαiown to those skilled in the art, to optimize performance and efficiency of power amplifier 102 for the selected radio operating mode.

In summary, the present invention teaches a bias-management system 100 and method 200 for a programmable RF power amplifier 102. Bias-management system 100 optimizes the performance of RF power amplifier 102 for multiple radio operating modes. A reference transistor Q3 of a bias regulator 104 and an output transistor Q2 of power amplifier 102 are integrated upon a common substrate 122 to provide close thermal coupling between bias regulator 104 and power amplifier 102. A bias cuπent Ib2 for power amplifier 102 is dynamically adjusted as a function of an envelope of an RF input signal 126 and to compensate for signal rectification within an output stage 108 of power amplifier 102. Although the prefeπed embodiments of the invention have been illustrated and described in detail, it will be readily apparent to those skilled in the art that various modifications may be made therein without departing from the spirit of the invention or from the scope of the appended claims.

Claims

CLAIMSWhat is claimed is:

1. An RF power-amplifier bias-management system (lOO)comprising: a controller (116) configured to select a radio operating mode; a cuπent-miπor circuit (114) coupled to said controller (116) and configured to produce a reference cuπent (IRef) as a function of said radio operating mode; a bias regulator (104) coupled to said controller (116), coupled to said cuπent-miπor circuit (114), and configured to respond to said reference cuπent (IRef); and a power-amplifier output stage (108) coupled to said bias regulator (104) and having a bias cuπent (Ib2) determined by said bias regulator (104).

2. A system (100) as claimed in claim 1 additionally comprising a power-amplifier driver stage (110) coupled to said bias regulator (104) and having a bias cuπent (Ibl) determined by said bias regulator (104).

3. A system (100) as claimed in claim 2 additionally comprising a DC-to-DC converter (118) coupled to said controller (116) and configured to provide a supply voltage (Vcc) for said output stage (108) and said driver stage (110) in response to said radio operating mode.

4. A system (100) as claimed in claim 1 additionally comprising a DC-to-DC converter (118) coupled to said controller (116) and configured to provide a supply voltage (Vcc) for said output stage (108) in response to said radio operating mode.

5. A system (100) as claimed in claim 1 wherein: said bias regulator (104) comprises a reference transistor (Q3); said output stage (108) comprises an output transistor (Q2); and said reference transistor (Q3) and said output transistor (Q2) are formed upon a common substrate (122).

6. A system (100) as claimed in claim 5 wherein: said reference transistor (Q3) is thermally coupled to said output transistor (Q2) through said common substrate (122); and said bias regulator (104) is configured so that changes in conductivity of said output transistor (Q2) due to thermal variations are compensated for by proportionate changes in conductivity in said reference transistor (Q3).

7. A system (100) as claimed in claim 1 wherein: said system (100) additionally comprises an envelope detector (120) coupled to said cuπent-miπor circuit (114) and configured to produce an input-signal envelope cuπent (lEnv) in response to an RF input signal (128); and said cuπent-miπor circuit (114) is configured to produce said reference cuπent (IRef) as a function of said radio operating mode and said input-signal envelope cuπent (lEnv).

8. A bias-management system (100) for a transmitter power amplifier (102), said system (100) comprising: a cuπent-miπor circuit (114) configured to produce a quiescent reference cuπent (IRef) in response to a digital control signal (126); a bias regulator (104) coupled to said cuπent-miπor circuit (114) and configured to respond to said quiescent reference cuπent (IRef); and an output stage (108) coupled to said bias regulator (104) and having a quiescent bias cuπent (Ib2) in response to said bias regulator (104).

9. A system (100) as claimed in claim 8 additionally comprising a driver stage (110) coupled to said bias regulator (104) and having a quiescent bias cuπent (Ib2) in response to said bias regulator (104).

10. A system (100) as claimed in claim 8 wherein: said bias regulator (104) comprises a reference transistor (Q3); said output stage (108) comprises an output transistor (Q2); said reference transistor (Q3) is thermally coupled to said output transistor (Q2) through a common substrate (122); and said bias regulator (104) is configured so that changes in conductivity of said output transistor (Q2) due to thermal variations in said common substrate (122) are compensated for by proportionate changes in conductivity in said reference transistor (Q3).

11. A system (100) as claimed in claim 8 wherein: said system (100) additionally comprises an envelope detector (120) configured to produce an envelope cuπent (lEnv) in response to an RF input signal (128); said cuπent-miπor circuit (114) is configured to produce said quiescent reference cuπent (IRef) in response to said digital control signal (126), and is further configured to dynamically alter said quiescent reference cuπent (IRef) into a dynamic reference cuπent (IRef) in response to said envelope cuπent (lEnv); said bias regulator (104) is configured to respond to said dynamic reference cuπent (IRef); and said output stage (108) has a dynamic bias cuπent (Ib2) in response to said bias regulator (104).

12. A bias-management system (100) for a transmitter power amplifier (102), said system (100) comprising: a driver stage (110) having a driver-stage bias cuπent (Ib2) with quiescent and dynamic components; an output stage (108) coupled to said driver stage (110) and having an output-stage bias cuπent (Ib2) with quiescent and dynamic components; a bias regulator (104) coupled to said driver and output stages (108) and configured to provide said driver-stage and output-stage bias cuπents (Ibl,Ib2); and a cuπent-miπor circuit (114) coupled to said bias regulator (104) and configured to establish said quiescent components of said driver-stage and output-stage bias cuπents (Ibl,Ib2).

13. A system (100) as claimed in claim 12 wherein: said driver stage (110) is configured to amplify an input signal (128) into an intermediate signal (130), and to pass said intermediate signal (130) on to said output stage (108); said output stage (108) rectifies a portion of said intermediate signal (130) to produce a dynamic bias offset, and passes said dynamic bias offset to said bias regulator (104); and said bias regulator (104) establishes said dynamic components of said driver-stage and output-stage bias cuπents (Ibi,Ib2) in response to said dynamic bias offset.

14. A system (100) as claimed in claim 12 wherein: said system (100) additionally comprises an envelope detector (120) coupled to said driver stage (110) and to said cuπent-miπor circuit (114), and configured to produce an envelope cuπent (lEnv) in response to an RF input signal (128); and said cuπent-miπor circuit (114) is configured to produce both said quiescent and said dynamic components of said driver-stage and output-stage bias cuπents (Ibi,Ib2).

15. A system (100) as claimed in claim 12 wherein: said output stage (108) comprises an output transistor (Q2) fabricated upon a substrate (122); said bias regulator (104) comprises a reference transistor (Q3) fabricated upon said substrate (122); said output stage (108) develops a bias offset in response to a thermal variation of said substrate (122); and said bias regulator (104) compensates said output-stage bias cuπent (Ib2) for said bias < offset.

16. A system (100) as claimed in claim 12 wherein: a portion of said output stage (108) comprising an output transistor (Q2) is fabricated upon a first substrate (122); a portion of said bias regulator (104) comprising a reference transistor (Q3) is fabricated upon said first substrate (122); and said cuπent-miπor circuit (114) is fabricated upon a second substrate (124).

17. A method (200) of adaptively controlling operation of an RF power amplifier (102), said method (200) comprising: selecting (220) a radio operating mode; producing (224), within a cuπent-miπor circuit (114), a reference cuπent (IRef) in response to said selecting activity (220); establishing (226), within a bias regulator (104), a bias cuπent (Ib2) for an output stage (108) of said power amplifier (102) in response to said producing activity (224); and determining (222) a supply voltage (Vcc) for said power amplifier (102) in response to said selecting activity (220).

18. A method (200) as claimed in claim 17 additionally comprising establishing (228) a bias cuπent (Ibl) for a driver stage (110) of said power amplifier (102) in response to said producing activity (224).

19. A method (200) as claimed in claim 17 additionally comprising: a) fabricating (206) a portion of said power amplifier (102) comprising an output transistor (Q2) upon a first substrate (122); b) fabricating (210) a portion of said bias regulator (104) comprising a reference transistor (Q3) upon said first substrate (122); and c) fabricating (216) said cuπent-miπor circuit (114) upon a second substrate (124).

20. A method (200) as claimed in claim 19 additionally comprising: thermally coupling (232) said reference transistor (Q3) to said output transistor (Q2) through said first substrate (122); and compensating (230) for thermal changes in said output transistor (Q2) by effecting thermal changes in said reference transistor (Q3).

21. A method (200) as claimed in claim 17 wherein: said method (200) additionally comprises detecting (246) an envelope cuπent (lEnv) in response to an RF input signal (128); and said producing activity (224) produces said reference cuπent (IRef) in response to said detecting activity (246).

22. A method (200) as claimed in claim 17 wherein: said method (200) additionally comprises rectifying (256) a portion of an RF signal (132) to produce a bias offset; and said establishing activity (226) establishes said bias cuπent (Ib2) in response to said producing and rectifying activities (224,256).