Integrated RF Front End with Stacked Transistor Switch

Abstract

A monolithic integrated circuit (IC), and method of manufacturing same, that includes all RF front end or transceiver elements for a portable communication device, including a power amplifier (PA), a matching, coupling and filtering network, and an antenna switch to couple the conditioned PA signal to an antenna. An output signal sensor senses at least a voltage amplitude of the signal switched by the antenna switch, and signals a PA control circuit to limit PA output power in response to excessive values of sensed output. Stacks of multiple FETs in series to operate as a switching device may be used for implementation of the RF front end, and the method and apparatus of such stacks are claimed as subcombinations. An iClass PA architecture is described that dissipatively terminates unwanted harmonics of the PA output signal. A preferred embodiment of the RF transceiver IC includes two distinct PA circuits, two distinct receive signal amplifier circuits, and a four-way antenna switch to selectably couple a single antenna connection to any one of the four circuits.

Description

BACKGROUND OF THE INVENTION

1. Field of the Invention

This invention relates broadly to integrated electronic circuits, and more specifically to RF transceiver circuitry.

2. Related Art

Wireless communications devices, especially handheld devices, are undergoing sustained development. Perhaps more than any other factor, the extreme popularity of cellular mobile telephones has motivated improvements in efficiency, speed, size and cost-effectiveness for RF transmission circuits in handheld devices. Enhancing the efficiency of such circuits is highly desirable so that the size of the required batteries may be reduced, while their life is extended. Cost-effectiveness is clearly always desirable for consumer products, particularly when such products require periodic replacement to stay abreast of changes in the technology. The steady advance of functionality in cellular telephones, combined with consumer preferences for light and small devices, puts a premium on reducing the volume required for RF transmission circuits. Additionally, transmitters must meet stringent emission limits, which have been established in order to facilitate high communication density at minimal power levels.

Most wireless communication units, such as cellular telephones, comprise at least one RF transceiver. A communication device, such as a cellular telephone, may comprise a multiplicity of RF (radio frequency) front end circuits, which are of primary interest herein. RF front end circuits (or subcircuits) typically include an RF transmit signal amplifier, a Power Amplifier (PA), a matching and filtering section, an antenna switch, and may include a received signal amplifier. A complete transceiver generally also includes a low-noise amplifier for the received signal. Of these circuits, the PA subcircuit is typically the most power-consuming portion of such transmitters, and, also typically, is the source of the most significant unintended or “spurious” emissions. In order to extend battery life, to meet stringent spurious emissions standards, and to minimize the cost of these high-volume consumer items, there is a need to improve the speed and efficiency, while reducing spurious emissions and manufacturing costs, for such PA subcircuits. Due to their need to handle high power, the PA and antenna switch subcircuits consume the most integrated circuit area. Manufacturing costs for integrated circuits are strongly dependent on the amount of device area required for each circuit. Consequently, substantial reductions in the area required for the various RF transceiver subsections will generally lead to commensurate reductions in manufacturing costs for transceiver circuits.

A range of PA topologies have been developed, each having different advantages. For example, PAs of class A, B, C, D, E and F are well known in the art. The primary amplifying devices in PAs of classes A-C are designed to operate in an “active” region of their operating range, thus intentionally conducting current while voltage is present across the device.

PAs of classes D, E and F attempt to reduce the power loss caused by such linear operation by employing amplifier devices as switches that minimize operation in active regions, rather than as linear amplifiers. However, the pulse-type outputs from such amplifiers generally require extensive filtering in order to establish a narrow-band sinusoidal output, as is typically required. While normal operation of PAs in classes D-F does not intentionally cause drive element devices to conduct while voltage is present across the devices, even switched devices consume real power due to current flowing while voltage is present during finite switching periods. Moreover, compared to drive devices in analog PAs operating at the same transmission center frequency, drive devices in class D-F switching circuits must often operate at much higher frequencies. The higher frequency signals include significant energy at undesired frequencies, and such undesired signal energies not only consume circuit power, but also require filtering to meet emission limits.

Integration of devices is generally desirable in order to improve various features of the resulting product, such as operating frequency and reliability, and may also reduce overall manufacturing costs, as well as likely reducing the volume occupied by the circuits. Field Effect Transistors (FETs) are extremely popular for both linear amplification and switching purposes in integrated circuits. However, integrated circuit (IC) FETs have a limited capability to withstand voltage between any two nodes, including gate-source, gate-drain, and drain-source node pairs. Such voltage withstand limitations may particularly impair the usefulness of IC FETs in high power switching circuits, in which inductive voltages may greatly exceed the supply voltage. As a particular example, the transmission output power capability of an RF PA is highly dependent upon the amplitude of the output voltage. One of the difficulties with existing PA technologies is that many otherwise desirably high-speed devices are fabricated using processes that tend to yield FETs having relatively low breakdown voltages. It is very desirable to solve this problem ant thereby provide a wider voltage range while retaining other desirable integrated device features. Such a solution enables integration on monolithic integrated circuits of power and control features that previously required separate processing, such as PA features and RF switch features. Integration of interacting circuits that were previously discrete will enhance yield and predictability, due to the process matching that is inherent in monolithic integration.

Methods and circuits are described herein that facilitate the fabrication of all of the transceiver RF circuits of a dual-band transceiver onto a single integrated circuit, thereby solving the problems and gaining the benefits noted above. Many of the benefits are achieved by integrating even the front-end portions of transceivers that do not necessarily include dual-band operation. One or more alternatives are described for each of numerous subcircuits (or corresponding methods), and a fully integrated RF front end, or an integrated RF transceiver, may be fabricated by using any compatible one of such alternatives for each section of the transceiver. Moreover, several of the subcircuits (or corresponding methods) that permit an integrated RF transceiver to be realized are also useful in other contexts, often independently of other RF transceiver subcircuits. Thus, various subcombinations of features described herein constitute useful inventions in their own right. Combined, various aspects of these subcombinations together achieve an integrated dual-band RF transceiver having all of the benefits noted above. Particularly notable among the independently useful subcircuits are stacked-FET RF switches and particular PA circuit topologies. Finally, the integration of certain RF transceiver subsections permits efficiencies in manufacturing without compromising safety and reliability of the final product.

SUMMARY

A combination of methods and/or circuits is described that enables the fabrication of a self-protected monolithic integrated circuit including all of the RF front-end sections of a communications transceiver. Such self-protected RF front-end circuits particularly include those sections, from a Power Amplifier (PA) through an antenna connection, that permit efficient internal protection from overload due to an improper, missing or damaged antenna.

Several subcombinations of the self-protected monolithic integrated RF front-end circuits have independent importance. One such subcombination is an integrated stacked-FET switch. One embodiment of this subcombination is a circuit including a multiplicity of FETs in a stack having drain-source channels coupled in series to control conductivity between nodes in a circuit. A control signal is coupled to a first FET to cause changes in conductivity of the first FET, and conductivity of the remaining FETs is enslaved to the conductivity of the first FET. A voltage withstand capability across the series combination of the FET stack may be substantially equal to a sum of drain-source voltage withstand capabilities of the individual FETs of the stack. A gate of each FET other than the first FET may be capacitively coupled to a common voltage.

Another subcombination is an RF Power Amplifier (PA) that may be referred to as an integrated iClass PA. One embodiment of this subcombination includes an input controlling an RF switch whose output is coupled to a supply source via an RF choke, and which operates with a characteristic drive output impedance at an operating frequency f₀. The drive output is coupled to an antenna connection having an expected antenna impedance via a coupling circuit that matches the drive output impedance to the expected antenna impedance, and also includes a circuit that dissipatively terminates signals at one or more harmonics of the operating frequency f₀. The iClass RF PA may further include a shunt filter configured to provide local minimum impedances between the drive output and the reference at a plurality of frequencies, including an even harmonic of f₀ and a non-unity odd harmonic of f₀. These minimum impedances may be approximately equal to the characteristic drive impedance.

One embodiment of the self-protected front-end circuit is an integrated circuit that includes a PA having an output amplitude regulator circuit that is controlled by an output limiting controller. The embodiment further includes an antenna switch that selectably couples an antenna connection to either the PA, via a matching and coupling circuit, or to a receive signal amplifier. The embodiment also includes an antenna connection sensor configured to sense current through, and/or voltage at, the antenna connection, and circuitry coupling an output of the antenna connection sensor to the output limiting controller, which is configured to affect the output amplitude regulator circuit to prevent the existence of currents or voltages in excess of design limits at the antenna connection.

A related embodiment is a method of making a monolithically integrated PA with protection from excessive output values caused by high Voltage Standing Wave Ratios (VSWRs) that are due to improper antenna impedance. The embodiment includes fabricating an RF PA on an integrated circuit chip to receive a transmit signal, and providing an output power limiting circuit for the PA. It also includes fabricating coupling, matching and filtering circuits on the same integrated circuit to condition a PA output signal having a PA output impedance to a different impedance desired for a connecting element that is connected to the integrated circuit to couple the conditioned signal to an antenna. The embodiment further includes disposing an antenna switch on the integrated circuit between the PA and the connecting element whereby the connecting element may be controllably coupled to either the conditioned signal from the PA, or to a receive amplifying circuit disposed on the integrated circuit. The embodiment includes providing a sensing circuit to sense a parameter of the signal delivered to the connecting element, and a PA control circuit to reduce power of the PA output signal in response to a value of the sensed parameter that is deemed excessive.

An embodiment of a further subcombination is a method of amplifying RF signals, and includes providing a plural-FET stack to control conduction between an output drive node and a reference node to effect a characteristic impedance at an operating frequency f₀. The embodiment further includes disposing, between the output drive node and the reference node, a shunt filter configured to dissipatively terminate a harmonic frequency of f₀. The shunt filter may include local minimum impedances at an even harmonic of f₀ and at a non-unity odd harmonic of f₀, and the local minimum impedances may be approximately equal to the characteristic impedance of the FET stack.

BRIEF DESCRIPTION OF THE DRAWINGS

Embodiments of the present invention will be more readily understood by reference to the following figures, in which like reference numbers and designations indicate like elements.

FIG. 1 is a block schematic diagram representative of some types of RF power amplifier (PA) circuitry.

FIG. 2 is a generalized schematic diagram of a stacked-FET control circuit useable with a PA such as shown in FIG. 1.

FIG. 3 is a simplified schematic diagram of an impedance matching and coupling bandpass filter useable in a PA such as shown in FIG. 1.

FIG. 4 is a simplified schematic diagram of a shunt filter useable in a PA such as shown in FIG. 1.

FIG. 5 is a simplified schematic diagram of a shunt power control circuit useable in a PA such as shown in FIG. 1.

FIG. 6 is a simplified schematic diagram of an output filter for Class F operation of a PA such as shown in FIG. 1.

FIG. 7 is a schematic diagram illustrating alternative biasing features for a stacked-FET amplifying circuit.

FIG. 8 is a simplified schematic diagram illustrating an alternative method of biasing FETs of a FET stack.

FIG. 9 is a simplified schematic diagram of a Class D PA employing complementary stacked-FET drive elements.

FIG. 10 is a schematic diagram of exemplary output filtering for a Class D PA such as represented in FIG. 9.

FIG. 11 is a simplified schematic diagram of a harmonic termination shunt filter useable in a PA such as shown in FIG. 1.

FIG. 12 is a simplified schematic diagram of an integrated dual-band RF transceiver.

FIG. 13 is a schematic diagram of an exemplary PA output power supervisor for directing output power levels in the transceiver of FIG. 12.

FIG. 14 is a schematic diagram of an exemplary series output regulator for the transceiver of FIG. 12.

FIG. 15 is a schematic diagram of an exemplary pulse width controller for the transceiver of FIG. 12.

FIG. 16 is a schematic diagram of an exemplary output voltage detector for the transceiver of FIG. 12.

FIG. 17 is a schematic diagram of an exemplary dual-band antenna switch for the transceiver of FIG. 12.

DETAILED DESCRIPTION

I. Power Amplifier Overview

FIG. 1 is a block diagram of an RF power amplifier (PA). The illustrated RF PA is quite general, in that varying the biasing of devices in a driver elements block 200, and/or varying details of the other blocks, will permit the RF PA illustrated in FIG. 1 to operate in any of amplifier classes A, B, C, E, F, or, as described further herein, as an iClass amplifier. FIGS. 2-5 each show an exemplary circuit to implement one of the blocks shown in FIG. 1.

An input 102 is provided to the PA with respect to a circuit reference, or common, 104. The input 102 generally comprises a properly biased signal at a center drive frequency, f₀. In response to the input 102, the driver elements block 200 controls conduction between a drive output node 106 and the circuit common 104. The driver elements block 200, in conjunction with current from V_DD via an RF choke (RFC) L_S 108, provides a signal having a particular impedance Zdrive. Zdrive may vary with frequency, but will refer to the drive impedance at the center operating frequency f₀, unless otherwise indicated. A shunt filter 400 may be coupled between the drive output node 106 and the circuit common 104. Numerous different filtering arrangements may be used, some examples of which are described subsequently herein.

An antenna 110 has a characteristic impedance Z_OUT, generally 50Ω (at the center frequency f₀ unless otherwise indicated). A block 300 is typically required to provide matching and coupling between the drive node 106 (at Zdrive) and the output at Z_OUT. Following the matching and coupling, an output filter section such as the combination of L_O 116 and C_O 118 may typically be disposed in the signal path before an RF switch, S_RF 120, which appropriately couples the output to the antenna 110. Because the PA circuit is integrated on a semiconductor device, and the antenna 110 is typically external to the IC comprising the PA, the antenna 110 often operates with a different reference voltage, for example a chassis ground 112, which has a non-zero impedance to the circuit common 104. Accordingly, the matching-coupling block 300, as well as the filter section 116-118, has an output that is referenced to chassis ground 112.

Power control may optionally be provided. One example employs a shunt power control block 500, which may provide a voltage offset between chassis ground 112 and circuit common 104 to reduce the amplitude of signals received by the antenna 110. A series regulator circuit, such as items 1400-1401 in FIG. 12, is probably used more commonly.

The monolithically integrated RF PAs, RF front ends, and RF transceivers described herein may be fabricated to operate at relatively high frequencies of at least 900 MHz and/or 2.4 GHz, and at moderate power levels. These designs are useful for transceivers having transmit power maximums of at least 0.5 W, 1 W, or 1.5 W RMS of RF output power delivered to the antenna connection when it is properly coupled to a matched antenna.

II. Stacked-FET Drivers

FIG. 2 is a simplified schematic diagram of a stacked-FET circuit that may be used for the driver elements block 200 in the RF PA of FIG. 1, for controlling conduction between the drive output node 106 and the circuit common 104. The stack includes two or more FETs of the same polarity, i.e., all FETs in a stack are N-channel FETs, or all are P-channel FETs, or at least all FETs in a stack operate substantially similarly as each other.

The FET stack 200 of FIG. 2 is configured to control conduction between two nodes of an integrated circuit. A terminal Vdrive_REF 202 is connected to one of the two nodes (e.g., circuit common 104 in FIG. 1), while a terminal Vdrive 224 is connected to the other node (e.g., Vdrive 106 in FIG. 1). For N-channel FETs (N-FETs) as illustrated in FIG. 2, Vdrive_REF 202 will be connected to the more negative of the two nodes, for example to circuit common 104 in FIG. 1. The terminal Vdrive_REF 202 is coupled to the source of a first FET of the stack 200, M₁ 204.

The FET stack 200 is controlled by means of an input signal, relative to terminal Vdrive_REF 202, that is coupled to the gate of the signal-input FET M₁ 204 via an input terminal 206. The drain of M₁ 204 is coupled to the source of a second FET M₂ 208. The gate of M₂ 208 is provided with a bias voltage VB₂ 210 via a bias resistor RB₂ 212, and is decoupled to Vdrive_Ru 202 via CG₂ 214. In some embodiments, these two FETs are sufficient, when properly configured to divide applied voltages so as to avoid exceeding breakdown limits of either device, to serve as a conduction controlling circuit to handle increased voltages in a circuit such as a PA or a quad mixer.

In other embodiments, however, one or more additional FETs of the same polarity are connected in series with M₁ 204 and M₂ 208. Such additional FETs are represented in FIG. 2 by an Nth FET, M_N 216. As for each additional FET of the stack, the source of M_N 216 is coupled to the drain of the preceding FET of the stack, i.e., to the drain of FET M_N−1 (not shown, though if N=3 then M_N is M₂ 208). The drain of the last FET of the stack, M_N 216, is coupled to the output terminal Vdrive 224. Associated with each additional FET is a biasing voltage VB_N 218, which is coupled to the gate of the FET via a bias impedance such as RB_N 220, and a capacitor CG_N 222 for coupling the gate to a voltage such that the FET is enslaved to conduction by the signal-input FET (here, M₁ 204). As shown, enslaving may be effected by coupling the gate of each additional FET to Vdrive_REF 202.

FET stacks with at least nine FETs in series have been fabricated or simulated, and stacks of even more series FETs are certainly possible. Note that physical circuit couplings generally include finite capacitance, inductance, and resistance. For many purposes it is preferred that the FETs of the FET stack 200 be coupled with minimal impedance in series, drain to source. However, impedance may be intentionally added to such couplings. For example, it may be desirable to more closely control a drive impedance, and to dissipate heat in specific resistive series coupling elements rather than within the FETs themselves. It may also be desirable to add impedance between the FETs of the FET stack 200 so as to tune the conductance of the drive circuit.

II.A. FET Stack Biasing

In some embodiments, the FETs of a FET stack may all have substantially similar voltage withstand capabilities, such as breakdown voltages V_GS(br), V_DS(br), and V_DG(br). For some integrated circuit fabrication processes, these values will be similar from FET to FET. Moreover, for some integrated circuit fabrication processes, the breakdown voltages V_GS(br), V_DS(br), and V_DG(br) may be approximately equal to each other. Proper biasing will usefully ensure that none of these breakdown voltages is exceeded during normal operation of the circuit. In some embodiments, with proper biasing, voltage excursions between Vdrive_REF 202 and Vdrive 224 may be permitted to approach a sum of V_DS breakdown voltages for each constituent FET of the stack.

Biasing and coupling the FETs of a FET stack as described below may prevent voltages from exceeding any maximum allowable node to node voltage for any FET of the stack, even when the total voltage impressed from Vdrive 224 to Vdrive_REF 202 is nearly equal to the sum of the maximum allowable V_DS for the individual FETs of the stack. Unless otherwise noted, the maximum allowable voltage between any two nodes of the FETs (i.e., V_GS, V_DS, and V_DG) are generally assumed to be substantially equal, both for the various nodes of each FET, and from FET to FET, which accords with an exemplary semiconductor fabrication processes. However, the skilled person may readily extend the principles set forth below to encompass situations in which these maximum allowable voltages are not equal. Also, the calculations set forth below for N-channel FET stacks may be applied to P-channel FET stacks with appropriate inversions of polarities and references.

The impedance of the gate drive of M₁ may be selected according to ordinary transistor driving principles. In this exemplary embodiment, V_DS(max) is the same for all FETs of the stack. V_DS for M₁ will therefore approximate (Vdrive-Vdrive_REF)/N. For each FET M_“X”, for X values from 2 to N, the effective value of each biasing resistor RB_X is selected to control a time constant, τ_GX, of the gate coupling. τ_GX is, approximately, the sum of effective capacitances of the gate coupling capacitor CG_X plus the parasitic gate capacitances C_GPX, multiplied by the series impedance to a biasing voltage. Such series impedance is typically resistive, and will be designated RB_X(equiv). It may be desirable for τ_GX to be much longer than the period 1/f₀ of the center drive frequency, preferably 5-20 times as long. Thus, a good design center goal is:

RB _X(equiv)(C_GX+C_GPX)=10/f₀  (Eqn. 1).

With respect to Vdrive_REF, and for Vpeak that is the maximum expected value of Vdrive, one proper bias voltage value is simply a proportional portion of ½ of Vpeak:

VB _X =X(Vpeak)/2N  (Eqn. 2)

Thus, an example in which N=4 yields: VB₂=Vpeak/4, VB₃=3(Vpeak)/8, and VB₄=Vpeak/2.

FET Stack Gate Signal Coupling

In FIGS. 2 and 7, each gate node (V_Gx) is coupled via the capacitor C_GX to the reference voltage Vdrive_REF. Each gate node is also coupled to a DC bias voltage via a bias resistor. In this configuration, the effective drive voltage V_GSX for each FET M_X of the FET stack depends upon the voltage excursion of its source, V_SX, in conjunction with the impedance from the source to the gate node, and from the gate node to AC ground. These impedances are dominated by the gate-source capacitance and the coupling capacitor C_GX. Appropriate values for C_GX may be determined as follows.

In the exemplary embodiment, the maximum voltage between each node pair of each FET is the same. The voltage excursions of the source of FET M₂ must therefore not exceed the maximum V_DS for M₁. As such, the value of CG₂ is unlimited, and desirably large, for effecting AC grounding of the gate of M₂ and thereby providing the largest common-gate drive signal to M₂. V_GS (max) will not be exceeded for M₂ if the (DC) voltage on the gate is maintained within the range of the source voltage excursions. However, if (contrary to the assumptions above) the maximum V_DS1 exceeds the maximum V_GS2, then CG₂ values may need to be limited in a manner analogous to that described below for C_GX for X from 2 to N.

The voltage excursion of the source of each FET M_X with respect to Vdrive_REF, ΔV_SX, will be equal to the drain voltage excursion for M_(X−1), ΔV_D(X−1). This voltage, presuming equal division between the various FETs, is X(Vpeak−Vmin)/N. For Vmin=0, this is simply X(Vpeak)/N, and ΔV_SX=(X−1)(Vpeak)/N.

The parasitic gate-source capacitance C_GS of a FET increases, when V_GS=V_GS(on), to C_OX, the oxide capacitance. C_OX for a particular FET M_X is designated C_OXX. Because C_GX is coupled to the reference voltage Vdrive_REF, the net V_GSX will be capacitively divided between C_GX and C_OXX. Thus, the gate-source excursion ΔV_GSX=(ΔV_SX)/(1+C_OXX/CG_SX). Presuming equal maximums for V_GS and V_DS, it is desired to limit ΔV_GSX≤Vpeak/N. Thus, substituting for ΔV_GSX and ΔV_SX, Vpeak/N≥[(X−1)(Vpeak)/N]/[1+C_OXX/CG_SX]. Appropriate consolidation yields:

C _GX ≤C _OXX/(X−2)  (Eqn. 3)

For X=2, C_GX≤infinity, as expected. Also as expected, excessive values for C_GX will tend to cause excessive gate-source voltage excursions (ΔV_GSX). The inequality of Eqn. 3 may prevent excessive voltages between nodes of the devices. However, C_GX may desirably be as large as is allowable so as to provide the largest allowable drive levels without exceeding breakdown voltages. Accordingly, the inequality of equation 3 may be treated as an approximate equality.

The result set forth above may not apply when it is desired to divide voltage differently between different FETs of the stack, or when maximum gate-source voltages differ from maximum drain-source voltages. However, the skilled person will have no difficulty determining desirable values for C_GX for such various circumstances by calculations corresponding to those set forth above, with appropriately modified assumptions. Because the capacitors C_GX must sustain voltages exceeding the bias voltage of the corresponding FET M_X, a metal-insulator-metal (MIM) capacitor is a good choice. Moreover, the capacitance of both (parasitic) oxide capacitors and MIM capacitors is a direct function of geometry. Certain fabrication variables, such as lithographic variables, therefore tend to have similar effects on both types of capacitances, leaving the ratio of such capacitances relatively immune to such variables.

FIG. 3 shows an exemplary matching, coupling and filtering block 300, which, as shown in FIG. 1, may be disposed between the drive output node 106 and the antenna 110. The matching function transforms the typically 50Ω characteristic impedance of the antenna, Z_OUT, to the characteristic impedance of the drive output node 106 (both at the operating frequency f₀) in a manner well known to those of skill in the art. The coupling capacitor Cc 302 blocks DC from the drive output node 106, and may be selected to have an impedance at f₀ that is less, and desirably much less, than the impedance to chassis ground 112, or to circuit common 104, from either side of the capacitor 302. The matching circuit 300 includes an “A” filter element comprising an inductor L_A 304 and a capacitor C_A 306, which may be fabricated as part of the integrated circuit. The matching circuit 300 also includes a “B” filter element comprising an inductor L_B 308 and a capacitor C_B 310 to chassis ground 112 (to which the antenna 110 of FIG. 1 is referenced). The coupling capacitor C_C 302, as well as the inductor L_B 308 and the capacitor C_B 310 may be fabricated on an integrated circuit with the PA, but some of these devices are typically external to the integrated circuit.

III. Shunt Filtering

FIG. 4 illustrates a shunt filter 400 that may be employed in an iClass PA such as illustrated in FIG. 1. A node 402 of shunt filter 400 may be connected to the drive output node 106 of FIG. 1, and an opposite node 404 may be connected to circuit common 104 of FIG. 1. The shunt filter 400 may provide local minimum impedances at each of one or more particular frequencies. The minimum impedances may be matched to the drive circuit impedance (as established, for example, by the drive elements 200 and the RF choke L_S 108). The shunt filter elements may be fabricated as part of the integrated circuit that includes the drive elements 200, thus reducing loop areas of currents passing through the elements of the shunt filter 400.

A shunt filter 400 for FIG. 1 may be a “transmission line filter” fabricated of reactive elements that are substantially distributed over an appropriate length, for example ¼ wavelength at f₀. Such a transmission line may be coupled to circuit common via a resonant circuit having a maximum impedance at f₀, such as a filter 600 as illustrated in FIG. 6 and described in more detail hereinbelow. Such a configuration for the shunt filter 400 provides local impedance minimums (approximately zero) at each even harmonic of f₀, and local maximum impedances at each odd harmonic of f₀. Stated more concisely, such a configuration may typically be said to reflect all odd harmonics, and to short all even harmonics, thus permitting operation as a Class F PA.

However, the shunt filter 400 illustrated in FIG. 4 for use in an iClass PA generally differs from such a transmission line filter. First, the shunt filter 400 may employ lumped, rather than distributed, elements. Consequently, local minimum impedances may occur at selected frequencies rather than at all odd, or at all even, harmonics of a resonant frequency (e.g., f₀). Second, the filter may employ series resistive elements to intentionally establish a non-zero value of local minimum impedance. It may be useful, for example, to control the local impedance minimum values so as to match an impedance of the drive circuit at the corresponding frequency (or, alternatively, at f₀). As a result of such differences between the circuit illustrated in FIG. 4 and a conventional transmission line filter, the magnitudes of currents in current loops may be reduced. Moreover, drive element power dissipation may be reduced at the frequencies corresponding to the selected minimum impedances.

In FIG. 4, a first shunt filter element includes L_SF1 406, R_SF1 408, and C_SF1 410. These components establish a local minimum impedance at a particular frequency, with the impedance increasing for both higher and lower frequencies. A switch S1412, in conjunction with an additional capacitor C_SF3 414, represents an optional circuit for adjusting the frequency of the minimum impedance of the first filter element. As shown, the effective value of the series capacitor of the first filter element is increased when S1 is closed and C_SF3 is disposed in parallel with C_SF1 410.

Of course, such frequency adjustability may be effected in numerous different manners. For example, S1412 may be a FET for electronically switching the frequency. Additionally or alternatively, C_SF1 410, as well as optional C_SF3 414, may be varactors (with the corresponding addition of an appropriate control circuit for the DC voltages on such varactors). Moreover, the capacitor C_SF3 414 may be disposed in series connection, rather than parallel connection, with C_SF1 410, in which event the switch S1 may be configured to bypass the capacitor C_SF3 414. Yet further, analogous techniques may be employed to vary inductance, rather than capacitance. For example, the switch S1412 may selectably bypass a second inductive element, so as to vary the effective inductance of the first shunt filter element.

The second shunt filter element comprises an inductor L_SF2 416, a resistive element R_SF2 418, and a capacitor C_SF2 420. The resonant frequency of the second filter element (or, indeed, of any further filter element) of the shunt filter 400 may be varied by similar techniques as outlined above with respect to the first filter element. It may be useful to have a minimum impedance that is substantially resistive, and/or that is non-zero. In one embodiment, the first and second filter elements are designed to provide local minimum impedances, at a second harmonic and a third harmonic of the operating frequency f₀ respectively, which are approximately equal to the drive circuit impedance. Though only two filter elements are illustrated, additional harmonics may desirably be treated with additional filter elements (not shown).

FIG. 11 is a schematic diagram of a circuit alternative for shunt filter 400 together with matching and coupling circuit 300. Zdrive and Z_OUT of FIG. 11 are connected as shown in FIG. 1. C_C 302 is substantially the same as in FIG. 3, providing DC isolation of the PA from the antenna output. The shunt filter includes a parallel resonant circuit primarily consisting of C_SF 980 and L_SF 982, which together function as a tank circuit that is resonant at f₀. All integer harmonic frequencies of f₀ are coupled through Rdrive 984, which is preferably selected to be approximately equal to the characteristic drive impedance of the PA switching circuit. Thereby, all harmonic frequencies of f₀ are terminated at the drive impedance. In some embodiments, alternative filtering, such as two parallel tank circuits each resonant near f₀, may be disposed in series above Rdrive 984. Dual tank circuits may be configured either to resonate at substantially identical frequencies, thereby increasing the impedance at f₀ and reducing power loss at f₀, or to resonate at slightly different frequencies, thereby broadening the range of frequencies at which the circuit has high impedance to ease manufacturing tolerances. Following the shunt filter, a matching and filtering network 990 may be as illustrated in FIG. 3, except for two differences: first, coupling capacitor Cc 302 is omitted, and second, C_A 306, being on the antenna side of coupling capacitor Cc 302, is coupled to earth ground 112 rather than circuit common 104. The two common references may be made effectively identical in some integrated circuit layouts.

IV. PA Output Power Control

FIG. 5 illustrates elements of one possible shunt power control circuit 500 for the PA of FIG. 1. A power control input 502 may establish a bias with respect to chassis ground 112. An inductive impedance Lg1504 (which may reflect, for example, the inductance of a bond wire) is illustrated between chassis ground and the source of a power control FET M_PC 506. A connection 508 may be coupled to circuit common 104 in FIG. 1. An inductance Lg2510 typically exists in series with a bypass capacitor for the power control circuit, C_PC 512. Assuming that the DC voltage V_DD is with respect to chassis ground 112, the substantially DC voltage established across C_PC 512 will reduce the effective supply voltage with respect to circuit common 104.

Other techniques may also be used to control the output power for a circuit as shown in FIG. 1, particularly when operated in a Class F configuration (rectangular wave input control). Such other techniques may be used either in addition to, or instead of, shunt power control as described above with respect to FIG. 5.

As a first example, referring also to FIG. 2, the bias voltages on the FETs M₂ 208 . . . M_N 216 may be adjusted. Efficiency will decrease, but power output will decrease more rapidly. As noted above, bias may generally be set such that VB_X=X(Vpeak)/2N. However, if VB2 is decreased well below the calculated value, the output voltage Vdrive (in iClass operation with harmonic termination) will also decline. Thus, for example, a circuit may be configured as shown in FIG. 7, except that the effective average voltage at the gate of M₂ 208 may be controllably reduced. This may be accomplished by making the value of RB₁ 708 variable (e.g., by means of a parallel FET). Alternatively, the value of RB₁ 708 may be reduced, and RB₁ 708 may be coupled to a variable voltage source rather than to reference 202. Reducing the bias voltage will cause corresponding reductions in drive output voltages. As described below in more detail, the self adjusting bias supply circuit of FIG. 7 will permit the bias on RB_N to gradually follow the reductions in Vdrive(peak) that are caused by varying the bias voltage on the gate of M₂.

PA output power may also be controlled by varying the amplitude of the drive signal. The conduction impedance of the drive elements will be higher when driven with a lower amplitude rectangular wave, resulting in a smaller effective drive voltage. The efficiency of this technique is typically comparable to the efficiency of varying bias voltages.

As discussed below with respect to FIGS. 12 and 14, a series regulator circuit may be used to control PA output power either alone, or in conjunction with one or more other power control techniques.

V. Alternative PA Embodiments

FIG. 6 is a simplified schematic diagram of a filter circuit 600 that may be employed in a manner similar to the shunt filter 400 in FIG. 1 to form a versatile PA architecture. The drive output node 106 of FIG. 1 may be coupled to a Zdrive node as an input to 600. The Zdrive node may be coupled via a ¼ wavelength transmission line 602 and a coupling capacitor C_C1 612 to an output filter section. The output filter section may comprise a parallel combination of L_OF1 610 and C_OF1 608, resonant at the operating frequency f₀. Unlike some embodiments of shunt filters 400, the output filter section of FIG. 6 is typically not part of the PA integrated circuit, and thus is referenced to chassis ground 112 rather than to circuit common. The impedance of this bandpass filter to ground 112 falls rapidly as the frequency deviates from f₀, and, therefore, the harmonics of the operating frequency are effectively shorted to ground at the output filter end of the transmission line 602. The standing waves of the properly tuned ¼ wavelength transmission line therefore provide a high impedance at each odd harmonic, and a low impedance at each even harmonic, as seen at the Zdrive node. A Z_OUT node 604 may be coupled to a further output filter section 116-118, an RF switch 120, and antenna 110 as shown in FIG. 1. A matching network (not shown) may also be required, similar to that illustrated in FIG. 3 with the coupling capacitor Cc 302 omitted. Such further filtering and matching circuits, or a transmission line coupled thereto, will ideally appear to the filter circuit 600 as a resistive impedance R_OUTequiv 606 at the operating frequency f₀.

Modified as described above, the circuit of FIG. 1 may be operated as an RF PA of Class A, Class B, Class C, Class E or Class F. For Class A operation, the input signal 102 is sinusoidal and does not cause the current through M₁ 202 of FIG. 2 to go to zero. For Class B operation, the input signal 102 is sinusoidal but M₁ 202 conducts only 50% of the time (conduction angle 180 degrees). Operation may be Class C, with a conduction angle less than 180 degrees, which yields some efficiency improvement as compared to the Class B operation. In each case the FETs M₂ to M_N are enslaved to M₁, and the FET stack of FIG. 2 functions substantially as a single device. The circuit of FIG. 1 may also be operated as an iClass PA in a configuration related to Class F but having dissipative termination for harmonics of the operating frequency.

The circuit of FIG. 1, configured as described immediately above, may also be operated as a Class F RF PA. For Class F operation the input signal is preferably a square wave having a duty cycle that causes the circuit 200 to conduct at precisely a 50% duty cycle. The output voltages resulting from Class F operation generally increase substantially when the conduction duty cycle deviates from 50%. Unfortunately, ordinary manufacturing component variations tend to cause the duty cycle to deviate from 50%, and consequently the circuit may not readily be able to safely and reliably utilize the full voltage withstand capability of the drive element(s).

VI. Alternative Bias and Slaving

Embodiments of a FET stack, as described herein, may include a signal-input FET that receives a drive signal coupled to its gate with respect to a reference voltage that is coupled to its source. The remaining FETs of the stack may be enslaved to the signal-input FET, such that they conduct under the control of conduction in the signal-input FET. The method by which the other FETs of a FET stack are enslaved to the signal-input FET must cooperate with the method employed to properly bias the FETs. Accordingly, enslavement and biasing are addressed together.

In RF PAs generally according to FIG. 1, the peak voltage of drive output node 106 (with respect to circuit common 104) will often exceed twice the available supply voltage V_DD 114. As such, bias voltages as required for the driver elements of FIG. 2 may not be readily available. This lack may be remedied by recourse, for example, to a charge pump. A charge pump that is preferred from the standpoint of minimal noise generation is described in commonly owned and copending U.S. patent application Ser. No. 10/658,154, “Low-Noise Charge Pump Method and Apparatus,” which is hereby incorporated in its entirety by reference. As described therein in detail, a low-current voltage source may be readily established at any desired voltage. Such voltage source may be provided, as needed, to any of the bias voltage inputs VB₂ 210 to VB_N 218 in FIG. 2.

FIG. 7 illustrates a self-adjusting bias supply that may be employed to bias the FETs of a FET stack. As in FIG. 2, a signal input 206 is coupled to the gate of a signal-input FET M₁ 204. The source of the FET M₁ 204 is coupled to Vd_REF 202, while its drain is coupled in series with each subsequent FET of the stack, including M₂ 208 . . . M_N 216. The drain of the last FET of the stack, M_N 216, is coupled to Vdrive 224. To provide a bias voltage that reflects Vdrive 224, a diode (or equivalent) D_B 702 charges a bias supply capacitor C_B 704 to Vbias 706. Vbias will charge to approximately Vpeak, the peak value of Vdrive 224 with respect to Vd_REF 202. If a time constant associated with C_B 704 is sufficiently long, then Vbias will remain substantially at this value. The time constant is the product of the capacitance of C_B 704 multiplied by the resistance, to Vd_REF 202, of the resistive voltage divider having N resistors including RB₁ 708, RB₂ 710, . . . , and RB_N 712. The total resistance of this voltage divider may be designated R_Bsum.

With respect to Equations 1, 2 and 3 that are set forth above, “X” represents the position of the particular FET in a stack, and N represents the total number of FETs in such stack. Assuming that all FETs are approximately identical, it may be seen that:

RB ₁ =RB ₂ ==RB _(N−1)  (Eqn. 4), and, accordingly,

RB _N=(N−1)RB₁  (Eqn. 5).

In view of equations 1-5, it may be seen that, for the last FET of the stack (X=N),

(C_GX+C_OXX)=C_OX(N−1)/(N−2)  (Eqn. 6),

RB _X(equiv) =RB ₁(N−1)/2  (Eqn. 7), and

RB ₁≥20(N−2)/[C_OX(N−1)²f₀]  (Eqn. 8).

Thus, for N=3, RB₁≥5/C_OX/f₀, and RB₁ declines monotonically as N increases (for given values of C_OX & f₀).

The total resistance R_Bsum of the resistive divider described above, in which the lower (N−1) resistors are RB₁ and the top (or Nth) resistor is the sum of the lower resistors, is simply 2(N−1)RB₁. The ripple on Vbias 706 may be acceptably low if the time constant CB(RBsum)≥10/f₀. Coupling that criteria with Eqn. 8 yields

C _B ≥C _OX(N−1)/(N−2)/4  (Eqn. 9).

Thus, for N=3, C_B≥C_OX/2. As N increases, smaller values of C_B (with respect to C_OX) will be required to achieve the same ripple voltage.

A significant ripple voltage is not necessarily a problem, and C_B may desirably assume even smaller values if rapid self-adjustment response is required. Indeed, in view of the filtering effected by each gate bypass capacitor CG_X in conjunction with RB_X(equiv), an average value is the main consideration for Vbias. However, if the average value of Vbias is permitted to decline significantly below Vpeak for any reason, including the presence of substantial ripple on C_B, the skilled person will understand that the resistive divider values should be adjusted accordingly.

FIG. 8 illustrates an alternative for providing both bias and gate coupling for FETs M₃ and above (X≥3). A reference 802 is coupled to the source of a signal-input FET M₁ 804, the gate of which is coupled to an input signal 806. The drain of M₁ 804 is coupled to the source of a second FET M₂ 808. A bias voltage is applied to a bias input 810, which is coupled via a bias resistance RB 812 to the gate of M₂ 808, and to a relatively large capacitance CG₂ 814. The drain of M₂ 808 is coupled to the source of a third FET of the stack, M₃ 816. The drain of M₃ 816 may be coupled to a further FET stage, if present. However, the drain of the FET of the last stage, M₃ 816 as shown in FIG. 8, is coupled to an output node Vdrive 818.

The gate 820 of FET M₃ 816 may be coupled to the base of the preceding stage FET M₂ 808 via a zener diode DZ 822. DZ 822 may have a conduction threshold knee at approximately the maximum desired value for V_DS of M₃ 816. (A circuit operating similarly to a zener diode may be used instead of DZ 822.) Additional FET stages designated by subscripts “Y” may be added. For such additional stages, corresponding additional zener diodes may be employed in like manner as DZ 822, i.e., anode to the gate of additional FET My, and cathode to the gate of M_(Y−1).

VI.A. Alternative Stacked FET Switch Configurations and Extensions

The FET stacks described above with respect to FIGS. 1-8 employ N-channel FETs (N-FETs). P-channel FET (P-FET) stacks may be fabricated analogously, by reversing the polarity of each voltage and polarized component associated with the stack. The P-FET stack reference voltage will generally be coupled to the source of a first, signal-input FET M_PI. Such inverted circuits will operate according to substantially the same principles as the N-FET stack circuits described above. For example, Vdrive 818 may be negative with respect to reference 802 in FIG. 8 if all FETs are P-channel, and the zener DZ 822 connection is reversed (anode and cathode exchanged).

FIG. 9 is an exemplary circuit that employs both an N-channel FET stack comprising N-channel FETs M_N1 902, M_N2 904 and M_N3 906, plus a P-channel FET stack comprising P-channel FETs M_P1 908, M_P2 910 and M_P3 912. For Class D operation, an input square wave may be provided with respect to common 914 at the input 916 to the N-FET stack, and coupled to an input for the P-FET stack on the gate of M_P3 912 via a capacitor C_GP1 918. A bias voltage, set for example to one half of V_GS(on) below the P-FET stack reference V_DD 930, may be provided for M_N 908 via a bias resistor R_BP1 920. Alternatively, the capacitor C_GP1 918 and the bias resistor R_BP1 920 may be deleted, and the input 916 and the gate of M_P1 908 may each be driven, instead, by means of an appropriate non-overlap clock generator (not shown).

Control of the N-FETs M_N2 904 and M_N3 906 is substantially as described with respect to FIG. 2 (for N=3). The gate of M_N2 904 is coupled to common (i.e., decoupled) via a capacitor C_GN2 922 having a relatively large value, and may be biased to about (V_DD/3) volts via a bias resistor R_BN2 924. The gate of M_N3 906 is decoupled to common via a capacitor C_GN3 926 having a value calculated as described with respect to FIG. 2, and may be biased to (V_DD/2) volts via a bias resistor R_BN3 928.

The P-FET stack is controlled analogously as the N-FET stack. The polarities of the bias voltages are inverted, and referenced to the “reference voltage” of the P-FET stack, which in this case is V_DD 930. For purposes of capacitively decoupling the P-FET gates, the fact that the P-FET reference voltage is V_DD 930 is likely to make little difference, because V_DD is typically closely coupled to the circuit common 914 that is the reference for the N-FETs. Therefore, decoupling capacitors 932 and 936 may alternatively be connected to circuit common 914. As shown, however, the gate of M_P2 910 is decoupled to V_DD via a relatively large capacitor C_GP2 932, and biased to about ⅔ V_DD via a bias resistor R_BP2 934. The gate of M_P3 912 is decoupled to V_DD via a capacitor C_GP3 936. The value of C_GP3 936 may be calculated as described with respect to FIG. 2 for X=3 and N=3. The gate of M_P3 912 is biased to about V_DD/2 via a bias resistor R_BP3 938.

An output voltage Vdrive 940 will be driven between common and V_DD, according to whether the N-FET stack is conducting or the P-FET stack is conducting. The output Vdrive 940 may be shunt filtered by a shunt filter 950, and may be processed by a matching and coupling circuit 960, as described below in more detail with respect to FIG. 10. From the matching and coupling circuit 960 the signal may proceed to an antenna 942, typically via a transmission line, one or more further filter sections, and an RF switch (not shown).

The shunt filter 950 of FIG. 9 may be similar to that shown in FIG. 4, or that shown in FIG. 6. The matching and coupling circuit 960 of FIG. 9 may, for example, be similar to that shown in FIG. 3. However, FIG. 10 illustrates filtering that may be employed for both blocks 950 and 960 in the circuit of FIG. 9. The capacitor Cs 952 may serve as the shunt filter 950. The remainder of FIG. 10 may function as the matching and coupling circuit 960 of FIG. 9. An inductor L_C 954 may comprise a physical coupling connection. A coupling capacitor Cc 962 serves to block DC. L_A 964, C_A 966, L_B 968 and C_B 970 may be configured for matching to the output impedance Z_OUT, which is typically 50 ohms.

VII. Monolithically Integrated, Medium Power Dual-Band RF Transceiver

An RF transceiver, such as the dual-band RF transceiver represented in FIG. 12, typically includes a received-signal amplifier such as items 1226 or 1256 of FIG. 12. Such a received-signal amplifier is typically a low noise amplifier (LNA), and is employed to condition signals received from the antenna. An RF front end may be considered to be an RF transceiver circuit that does not necessarily include an LNA.

In most RF transceivers, discrete integrated circuits must be combined in a module to fabricate a complete RF front-end section. Typically, at least the antenna switch will be fabricated on a different, separate integrated circuit (IC) from the PA, and often many more discrete integrated circuits must be connected via off-chip wiring to fabricate an RF front end module. Each such discrete integrated circuit must be defined by particular performance requirements which ensure that the module functions properly even when the discrete integrated circuits which it comprises are from different lots, or have been designed and manufactured differently from other integrated circuits that perform the same tasks. Such performance requirements, which are thus developed to achieve mix-and-match flexibility and reliability, may well exact a cost for the discrete ICs that are combined in these devices.

PAs in multiple-IC transceiver modules typically produce a signal of substantial power on demand An antenna switch unit couples an antenna (more precisely, an antenna connection) to either a transmit signal matched to the expected antenna impedance (e.g., 50 ohms), or to a receive signal input. However, damage to the antenna connection or the antenna may cause the impedance reflected to the antenna connection point from the antenna connecting line to vary drastically from its expected value. In such event, a large voltage standing wave (VSW) may be caused by the resulting mismatch between that reflected impedance, and the expected value to which the transmit signal has been matched. Voltage excursions much larger than those expected during normal operation may be generated as a consequence of such mismatch-induced VSWs. Voltage withstand requirements for antenna switches are typically set much higher than normal peak operating voltages to avoid damage under such mismatch conditions.

The IC area occupied by switching devices (such as FETs) in a power-switching circuit, such as an antenna switch, may increase as the square of the voltage they are capable of withstanding. Thus, halving the required withstand voltage may reduce the switch device area to one fourth. Moreover, because these devices dominate the IC area used by an antenna switching circuit, a very substantial saving in IC area (and thus in manufacturing cost) may be realized by reducing their required withstand voltage. Such reduction may not be practical when discrete ICs must be coupled to fabricate an entire transceiver. However, a single IC that includes all devices from a PA, through an antenna switch, and to an antenna connection, may take advantage of reliable internal coupling and close device matching to protect against high mismatch-induced VSWs. Due to these advantages of integration, substantial savings in device area can be realized as compared to combining discrete ICs to fabricate a comparably-performing transceiver.

FIG. 12 is a simplified block schematic diagram of the primary RF sections of a dual-band transceiver that is configured to benefit from such internal protection. A low-level signal at a first operating frequency f_O1 is coupled to an input node 1202 from a source which may (but need not) be on the same IC chip. The signal is amplified through any suitable amplifier, as indicated by amplifier 1204. The signal produced by the amplifier 1204 may deviate considerably from a preferred rectangular shape if a pulse adjustment circuit 1500 is provided to improve rectangularity, and preferably to also adjust the duty cycle, of the waveform that is ultimately coupled into a power amplifier (PA) 1206.

The output of the pulse adjustment circuit 1500 is the input to the PA 1206, which draws power from a supply V_DD 114 via supply conditioning elements, including a series regulator 1400 and an RF choke (RFC) L_S 108, to generate a PA output signal. The PA output signal has a characteristic impedance resulting from the input signal, the PA circuit elements, and the supply conditioning elements, and generally differs from the impedance expected at the antenna node 1214. A coupling, matching and filtering network may be needed, for example as represented by a block 1210. Such a network may couple the PA output signal to the antenna switch while blocking DC current, and may transform the PA output impedance to the desired antenna node impedance (e.g., 50 ohms). It may also filter undesirable signal components, such as harmonics of f_O1, from the PA output signal before coupling it to an “A” input of an antenna switch 1700. If separate grounds are maintained as a matter of design preference, then the output of the coupling, matching and filtering block 1210 may be referenced to a ground reference 112, which may be distinguishable from a circuit common reference 104 used elsewhere in the circuit. The antenna switch 1700 selectably couples the signal to the antenna node 1214, from whence it may be coupled, for example by transmission line, to an antenna that may be separated from the IC chip.

Availability of the antenna connection of the antenna switch on the same IC chip as the PA (and all intervening circuitry) provides an opportunity to reliably limit the maximum electrical stress that must be endured by the antenna switch circuitry, by the PA, or by coupling, matching or filtering elements. An output sensor 1600 may be coupled to the antenna node 1214, sensing the electrical stress and providing a signal that will cause the PA to reduce its output if the electrical stresses are excessive. To this end, the output 1220 of output sensor 1600 is coupled to an input “B” of a PA control block 1300. An “A” input 1224 to the PA control block 1300 may receive an amplitude control signal to adjust the envelope amplitude of the PA output signal. This input may also be used to restrict, or even to terminate, output from the PA. Both the “A” and “B” inputs may affect an output “D” that is coupled from the PA control block 1300 to the series regulator block 1400. A “C” input 1222 to the PA control block 1300 may be provided with information, or a signal, that controls an “E” output from the block 1300. The “E” output may be coupled to the pulse adjustment circuit 1500 to control the duty cycle of the rectangular wave that is input to the PA 1206. Duty cycle control may, for example, provide another means to reduce the power level of the PA output signal. The signal path from 1202 may be tuned for a first band of operating frequencies, which include fol.

The antenna switch 1700 may selectably decouple the antenna node 1214 from the first-band transmit signal on input “A,” and couple the antenna node instead to output “B” so as to deliver a signal received from the antenna to a receive preamplifier 1226 for a first receive band. The receive preamplifier 1226 (as well as 1256) is preferably a low noise amplifier (LNA). LNAs are not necessarily included in integrated front ends as described herein, though they typically are included in complete transceiver circuits. The output from receive preamplifier 1226, if present, may be delivered to further first receive band circuitry either on or off the same IC chip. The antenna switch 1700 may similarly selectably couple the antenna node 1214 to a second receive band preamplifier 1256 to amplify a signal from the antenna to a second receive band output node 1258. That output may be delivered to further second receive band circuitry either on or off the IC chip.

Similarly as described above with respect to the first transmit band circuitry, a transmit signal at a second operating frequency foe in a second operating frequency band may be provided to an input 1232, and amplified by an amplifier 1234. The duty cycle and waveform of the signal output from the amplifier 1234 may be conditioned by a pulse adjustment 1501 under control of a PA control block 1301, and then delivered as an input to a second band PA 1207. The second band PA 1207 will generate a second-band PA output signal using power that is provided from V_DD 114, as limited by a series regulator 1401 under control of the PA control block 1301, via an RF choke 109. The second-band PA output will have a characteristic impedance, and will be coupled to the “D” input of the antenna switch 1700 via a block 1211 that couples the signal, matches the PA output and antenna node impedances, and filters the output signal. The antenna switch 1700 may be controlled to couple the “D” input to the antenna node 1214, from whence the signal will be delivered to the antenna 1216. The output 1220 of the output sensor 1600 may be coupled also to a “B” input to the second-band PA control block 1301, whereby excess output voltage will cause the second-band PA output signal to be reduced to safe levels. The second-band PA control block 1301 may also accept an envelope-control signal at an “A” input 1254, as well as a duty-cycle control signal at a “C” input 1252.

Though not shown, control circuitry is preferably enabled only when the associated PA is active. Exemplary circuitry for a PA control block, such as 1300 or 1301, is shown in FIG. 13. Exemplary circuitry for a series regulator, such as 1400 or 1401, is shown in FIG. 14. Exemplary circuitry for a pulse adjustment circuit, such as 1500 or 1501, is shown in FIG. 15. The PAs 1206 or 1207 may be fabricated as described for Driver Elements 200 of FIG. 1, together with a shunt filter, such as block 400 of FIG. 1 or the appropriate elements of FIG. 11. Coupling, matching and filtering circuits 1210 and 1211 may be fabricated as described above for FIG. 3, or FIG. 10, or in any other manner to obtain similar coupling, matching and filtering effects. Note that if a circuit as shown in FIG. 11 is to be employed, the coupling capacitor 302 will be disposed before the shunt filter, and the matching and filtering will be provided subsequently, as in block 990.

FIG. 13 illustrates exemplary circuitry for a PA control block 1300. An enable input “A” 1302 may be coupled directly to a FET 1304, such that if input 1302 is approximately ground potential, an output “D” 1306 can draw no current. Conduction into the output “D” 1306 may control the PA output power via a series regulator, such as shown in FIG. 14, such that when output 1306 conducts no current, no current will be provided to the PA, reducing output power to zero as discussed below with respect to FIG. 14. FET 1308 is biased by resistors 1310 and 1312, which may have equal values, a nominal value such as 30-50 kΩ being selected for engineering convenience. This configuration protects low-voltage FETs, ensuring that V_GD, V_GS, and V_DS for all of FETs 1304, 1308 and 1314 do not significantly exceed V_DD/2.

A power sense input “B” 1316 may be coupled to resistor 1318. Resistor 1318 may be about 30-50 kΩ, and reasonably equal to a resistor 1320 to establish unity gain for op amp 1322. A power set input “C” 1324 may be set, in one embodiment, from 0V to 2*Vth (FET threshold voltage), where Vth may be 0.4 to 0.7V, nominally about 0.5V, and is consistent within the circuit. The noninverting input of op amp 1322 is prevented from exceeding this voltage range by means of a resistor 1326 (e.g., 30-50 kΩ) together with diode-connected FETs 1328 and 1330, thus limiting the maximum power that may be selected. The skilled person may adjust circuit values, and circuit design, so as to achieve a selectable output power up to a fixed circuit design maximum. In particular, one or both diode-connected FETs 1328 and 1330 may be replaced by a network that includes a bandgap reference, for example to increase accuracy of power settings and output voltage limits. Many other techniques may be employed to achieve similar effects. When power sense input “B” 1316 exceeds a value established by the power set input voltage, FET 1314 will cease conducting, precluding conduction into output “D” 1306.

The PA control block 1300 also provides an output “E” 1512 to control the duty cycle adjustment effected by the pulse adjustment circuit 1500 of FIG. 12. A reference voltage, which may be adjustable according to factors such as fabrication process parameters, is provided at an input 1332. This voltage is doubled by an op amp 1334 under control of equal-valued resistors 1316 and 1318. Of course, in other embodiments the gain of circuitry such as shown in FIG. 13 will likely be different, and resistors setting such gains, for example 1316 and 1318, will accordingly differ in value. By reducing the duty cycle somewhat, the PA output power may be correspondingly reduced, and by reducing it to zero the PA output may be suppressed entirely. A reference voltage provided to an input 1322 of an amplifier 1320, the gain of which may be controlled in the usual manner by resistors 1324 and 1326, may serve to establish the voltage at output “E” 1512 to control duty cycle of the output of block 1500.

FIG. 14 illustrates an exemplary series regulator circuit 1400 for limiting the effective voltage provided to the PA, and thus limiting the PA output amplitude. The voltage provided to input 1306, as compared to V_DD, is divided via resistors 1406 and 1404 to control a P-channel FET 1402 and also protect FET 1402 from excessive voltage between any two nodes. A P-channel FET 1408 is biased by resistors 1410 and 1412 so as to divide maximum voltages that are generated between V_DD and an output 1414, somewhat equally between FETs 1402 and 1408. The output 1414 provides power to the PA via an RF choke. The FETs 1402 and 1408 may, for example, have V_GS threshold voltages of between −0.4V and −0.7V. Resistors 1404, 1406, 1410 and 1412 may all be substantially equal, with a magnitude selected for engineering convenience to be, for example, 30-50 kΩ. These exemplary values and relative values may be varied for engineering convenience.

FIG. 15 is a schematic representation of an exemplary signal conditioning circuit 1500. An input signal may be provided on an input node 1502, and coupled via a diode-connected FET 1504 to the input of an inverter 1506. When the input signal voltage, plus the V_DS threshold of FET 1504, is less than the threshold of the inverter 1506, the output of inverter 1508 will be low. However, when the voltage of the input signal rises above this value, the FET 1504 will cease conducting. Thereafter, even if the voltage of input 1502 is quite high (e.g., V_DD), current from the input will be limited by P-channel FET 1510 under control of an input voltage 1512. Such current through the FET 1510 must charge a capacitor 1514 (which may be a metal-insulator-metal or “MIM” capacitor of about 0.025 to 0.05 pF) until the input to inverter 1506 rises above its switching threshold. At that point the inverters 1506 and 1508 will change state rapidly due to positive feedback via the capacitor 1514, providing square edges at an output node 1516, rise and fall times being limited primarily by delays through the inverters 1506 and 1508. If the signal at the input 1502 is a roughly rectangular wave of about 50% duty cycle, as in a preferred embodiment, the voltage at the control input 1512 may be adjusted such that the output duty cycle is reduced from 50% to an arbitrarily lower value that may reach zero. The input signal may be configured to have a duty cycle exceeding 50% if a wider range of output duty cycle is desired.

FIG. 16 is a schematic of exemplary circuitry 1600 for sensing peak voltage at a sense node 1602, which may for example be connected directly to antenna node 1214 of FIG. 12. An input divider may be used as shown to sense relatively high voltages. Four roughly equal resistors 1604, 1606, 1608 and 1610 having a relatively low resistance, such as 1 kΩ, may be used. Diode-connected FET 1612 conducts when this voltage is high, providing current through a resistor 1614 of about 24 kΩ to a capacitor 1616 of about 1 pF. Many other values may be used, so long as the time constant established by resistor 1614 and capacitor 1616 is much smaller than the duration of any event that could cause the output voltage to rise (for a given level of PA output signal). For example, an antenna mechanical event that caused the antenna impedance to vary drastically from the design value, thereby causing a high voltage standing wave to appear, will take at least milliseconds to occur. The time constant of approximately 24 nS of the exemplary circuit 1600 is well below such an event duration. However, the corner frequency due to these components should generally be well below both the first and second band operating frequencies f_O1 and f_O2, in order to avoid circuit oscillations. If event durations may approach 1/f_O, then other common circuit design considerations may require a more complicated circuit to avoid oscillation while ensuring that the response is sufficiently fast to prevent excessive voltages.

FIG. 17 is a simplified schematic of exemplary circuitry 1700 for an antenna switch. Further details regarding design and fabrication of such an RF switch may be found in U.S. Pat. No. 6,804,502, issued Oct. 12, 2004 and entitled “Switch Circuit and Method of Switching Radio Frequency Signals.” Circuitry to provide control signals is not shown. Moreover, the control voltages should preferably be either “high,” nearly V_DD, or “low” at nearly −V_DD. To generate −V_DD control voltages, a negative voltage generator will be helpful, preferably a low-noise circuit such as described in copending, published U.S. patent application Ser. No. 10/658,154, filed Sep. 8, 2003 and entitled “Low Noise Charge Pump Method and Apparatus.” Such a low-noise charge pump is important for avoiding unintended emissions from the antenna.

A port node 1780 is the common connection of the switch 1700. In FIG. 12, the common connection of the switch 1700 is coupled to the antenna node 1214. The common connection is generally coupled to only one RF port (port A 1710, port B 1730, port C 1750, or port D 1770) at a time. Each RF port has a corresponding “+” control node and a corresponding “−” control node. For ports A, B, C and D, the “+” control nodes are nodes 1708, 1728, 1748 and 1768, respectively, while the “−” control nodes are nodes 1718, 1738, 1758 and 1778, respectively.

To couple an RF port to the common connection, a “high” voltage (−V_DD) is applied to the port's corresponding “+” control node, while a “low” voltage (˜−V_DD) is applied to the port's corresponding “−” control node. Meanwhile, a “low” voltage is applied to each “+” control node corresponding to another RF port, and a “high” voltage is applied to each “−” control node corresponding to another RF port. Thereby, a selected RF port will be coupled to the common connection, while every other RF port will be coupled to ground. Thus, to couple RF port A 1710 to common connection 1780, a “high” voltage is applied to control nodes 1708, 1738, 1758, and 1778, while a “low” voltage is applied to all other control nodes (1718, 1728, 1748 and 1768).

Every resistor will typically have the same value. In some embodiments, the value will be roughly 30-50 kΩ. The resistor is selected such that the time constant of the parasitic gate capacitance of a FET (e.g. M1_A 1701), in conjunction with the value of its corresponding gate resistor (e.g. 1704) is much greater than 1/f_O, where f_O is the lowest significant frequency of the RF signal being controlled. The illustrated configuration serves to divide the voltage appearing across FET stacks (such as the stack consisting of FETs M1_A, 1701, M1_B 1702 and M1_C 1703, the stack consisting of FETs M2_A, 1704, M2_B 1705 and M2_C 1706, and so on) uniformly, reducing compression effects. The FET stacks (such as FETs 1701, 1702 and 1703) that provide the switching functions may include more or less than the three devices that are shown for illustration; stacks of at least nine devices have been successfully fabricated. Due to the voltage stress distribution uniformity, a wide range of signal voltages and fabrication process parameters may be accommodated.

Integrated Circuit Fabrication and Design

Integrated circuit fabrication details are not provided in the above description. In some preferred embodiments, including some which have output powers in excess of 1 W at around 2.4 GHz, the integrated circuits may be fabricated in accordance with ultrathin silicon on sapphire processing as described in U.S. Pat. No. 5,663,570, issued Sep. 2, 1997 and entitled “High-Frequency Wireless Communication System on a Single Ultrathin Silicon On Sapphire Chip.” Other semiconductor-on-insulator (SOI) techniques may be used to fabricate a dual-band transceiver integrated circuit as described above, for at least some frequency bands and power levels.

The preferred integrated circuit fabrication techniques described above readily produce FETs having a rather low maximum VDs. Accordingly, various techniques are described for stacking FETs to achieve control of higher voltages while maintaining consistent processing. Using other manufacturing techniques, or lower voltages and impedances, a need for cascode or multiply-stacked FETs may be avoidable.

CONCLUSION

The foregoing description illustrates exemplary implementations, and novel features, of a method and apparatus that employs stacked transistors to control conduction between a pair of nodes in an integrated circuit. The skilled person will understand that various omissions, substitutions, and changes in the form and details of the methods and apparatus illustrated may be made without departing from the scope of the invention. Numerous alternative implementations have been described, but it is impractical to list all embodiments explicitly. As such, each practical combination of the apparatus or method alternatives that are set forth above, and/or are shown in the attached figures, constitutes a distinct alternative embodiment of the subject apparatus or methods. Each practical combination of equivalents of such apparatus or method alternatives also constitutes a distinct alternative embodiment of the subject apparatus or methods. Therefore, the scope of the presented invention should be determined only by reference to the appended claims, and is not to be limited by features illustrated in the foregoing description except insofar as such limitation is recited, or intentionally implicated, in an appended claim.

It will be understood that similar advantages of integration will accrue to circuits having other functional blocks. For example, mixers may be incorporated on such a device, enabling integration of more portions of transmission signal processing. Phase locked loops may further enhance the ability to generate the transmission signal on the same monolithic IC as the RF front end or transceiver. Additional types of filters may be useful, for either or both of receive and transmission processing.

All variations coming within the meaning and range of equivalency of the various claim elements are embraced within the scope of the corresponding claim. Each claim set forth below is intended to encompass any system or method that differs only insubstantially from the literal language of such claim, if such system or method is not an embodiment of the prior art. To this end, each described element in each claim should be construed as broadly as possible, and moreover should be understood to encompass any equivalent to such element insofar as possible without also encompassing the prior art.

Claims

1. A module comprising: at least one integrated RF Power Amplifier (PA) circuit; wherein the at least one integrated RF PA circuit is incorporated into the module and further includes: an input node to accept an input signal with respect to a reference voltage, the input node connected to a first gate of a first MOSFET, wherein a source of the first MOSFET is connected to the reference voltage;one or more MOSFETs connected in series with the first MOSFET to form a transistor stack, wherein the first MOSFET comprises a bottom transistor of the transistor stack, and the one or more MOSFETs comprise intermediate transistors and/or a top transistor of the transistor stack, wherein the transistor stack is configured to control conduction between the reference voltage and an output drive node, and wherein the output drive node is connected to a drain of the top transistor of the transistor stack; andone or more predominantly capacitive elements connected directly between a corresponding gate of the one or more MOSFETs and the reference voltage.