Title of Invention	A METHOD FOR CONTROLLING ALIGNMENT OF A FEED FORWARD AMPLIFIER AND A FEED FORWARD AMPLIFIER
Abstract	A pilot system and method is disclosed that increases the rate of convergence of the second loop alignment control in a feed forward amplifier. Both a pilot generation and detection system and search algorithm controlling the alignment are disclosed. By measuring the fi'equency of the generated pilot, phase information regarding the second loop cancellation transfer fiinction can be inferred. Changes in the pilot fi-equency as the search algorithm makes steps in the second loop alignment indicate errors in the direction of the search. Using this pilot frequency measurement along with the existing log-power measurement of the residual pilot power improves the convergence speed because fewer steps will be made to reach the optimal alignment setting.

Title of Invention

A METHOD FOR CONTROLLING ALIGNMENT OF A FEED FORWARD AMPLIFIER AND A FEED FORWARD AMPLIFIER

Abstract

A pilot system and method is disclosed that increases the rate of convergence of the second loop alignment control in a feed forward amplifier. Both a pilot generation and detection system and search algorithm controlling the alignment are disclosed. By measuring the fi'equency of the generated pilot, phase information regarding the second loop cancellation transfer fiinction can be inferred. Changes in the pilot fi-equency as the search algorithm makes steps in the second loop alignment indicate errors in the direction of the search. Using this pilot frequency measurement along with the existing log-power measurement of the residual pilot power improves the convergence speed because fewer steps will be made to reach the optimal alignment setting.

Full Text	FEED FORWARD AMPLIFIER SYSTEM AND METHOD USING THE PILOT FREQUENCY FROM A POSmVE FEEDBACK PILOT GENERATION AND DETECTION CIRCUrr TO IMPROVE SECOND LOOP CONVERGENCE RELATED APPUCATION INFORMATION The present application claims priority to U.S. provisional application serial no. 60/668,363 filed April 5, 2005, and of U.S. provisional application serial no. 60/670^08 filed April 13, 2005, Hie disclosures of wbich are incorporated herein hy ref^ence in iSassx entirety. BACKGROUND OF THE INVENTION 1.. Field of the Invention The present invention relates to RF pOAver anq)lifiers and amplification methods. More paxticidarly, the present invention relates to feed forward poorer anxplifiers and methods of using a pilot to align the loops of a &ed forward ampUfier. 2. Description of the Prior Art and Related Information A primary goal of RF power amplifier design is linearity over the range of power operation. Linearity is simply the ahility to amplify witiiout distortion. This requiremeat is critical for modem whreless communication systems but it is increasmgly difficult to achieve. This is due primarily to the bandwidth requirements of modem wireless communication systems vMoh axe placing increasing demands on amplifier linearity. Feed forward compensation is a well known approach applied to amplifies to improve linearity by estimating and canceling distortioiL In feed forward RF power amplifiers an etror amplifier is employed to amplify only distortion components w^iich are tiien combmed witii the main amplifier ou^nit to cancel the main amplifier distortion compon^it Figure 1 illustrates a conventiotial feed forward amplifier design having a main amplifier 1 and an error amplifier 2. The hasic elements also include delays 3, 4 in the main and error path, respectively, and main to error path coi^lers 5, 6, 7 and 8. Additional elements not shown are also typically present in a conventional feed forward architecture as is well known to those skilled in the art. The delays, coi^lers and error amplifier are designed to retract distortion components from the main path and iigect out of phase distortion conqxments fix>m the error path into the main amplifier ou^ut at coxq>ler 8 to substantially eliminate the distortion component in the main amplifier path. The performance of a feed forward amplifier may typically be analyzed based on two cancellation loops. Loopl, called Ihe earner cancellation loop, ideally provides a signal at the output of coiq>le3: 7 ^th the input RF earner component cancelled and only a distortion component remaining. Loop 2 is referred to as the error cancellation loop or auxiliary path loop. In loop 2 the distortion component provided firom coiq)ler 7 is amplified by the error amplifier 2 and injected at coiq)ler 8 to cancel the distortion component in the main path and ideally provide a distortion firee signal at the output The quality of the distortion estimate (carrier cancellation) is determined by the alignment of the first loop in terms of gain and phase. The distortion cancellation in turn is determined by the alignment of the second loop in terms of gain and phase. In prior art systems, a pilot 9 is injected into the main amplifi^ path of tiie first loop, acting, like a known distoMon signal. The pilot signal is detected at the feed forward amplifier output by a pilot detector 10 and used to aid tide alignment process for tiie second loop. When the second loop is aligned, the pilot is cancelled. If the second loop is misaligned, residual pilot power will be detected at the output of the feed forward amplifier. The degree of the misalignmCTt is estimated fix>m tiie measured power of the residual pilot The alignment of the second loop is adjusted in an iterative manner with tiae goal of reducing the residual pilot power. Generally, it is desirable to have the feed forward anq>lifier control system adapt to tixe optimal settings as &st as possible to minimize the amount of time the amplifier operates at a less than optimal setting. One difficulty witfa alignment control algorilihms iised to adjust the alignment settings (gain and phase) from any initial setting to that ^A^iich results in the best measured alignment is the difficulty in finding the coirect diiecdon of adjustment in Hxt two dimensional ^D) gain-phase space. Prior alignment control algorithms typically rely on dther the '^steepest descent" or the "coordinate descent" algorithms. The steepest descent algori&m adjusts the alignment settings in a direction of the gradient within the 2D gain-phase space. Dithering Ihe alignment in orthogonal directions and measuring Ihe changes in measured misalignment provides an estimate of the gradient. The coordinate descent algorithm performs two separate ID searches along pre-defined orthogonal directions (usually the gain and phase axes). The alignments are dithered to determine -which direction along the respective coordinates reduces measured misalignment Both &ese approaches have disadvantages in practical systems ^^4lich employ control processors with limited processing power and \^iere fEist loop alignment is desbred. As a result the desked &st and accurate loop convergrace has not been achieved in practical adaptive feed forward systems. Accordingly, a need presraxtly exists for a system and method for more rapid loop alignment control in a feed forward amplifier system. SUMMARY OF THE INVENTION In a first aspect the present invention provides a method for controlling alignment of a control loop in an amplifier system comprising g^oerating a variable firequency pilot signal, injecting the pilot signal into the amplifier system, and detecting any uncanceled pilot signal at an output of the control loop. The method fiuiher comprises detecting the fi:equency of the generated pilot signal, adjusting one or more parameters of the control loop, detecting a firequency change m the variable firequency pilot signal and controlling the adjusting based on the detected fiequency change. In a preferred ranbodunent of the method for controlling alignment of a control loop in an amplifier system generating the variable firequency pilot signal comprises using feedback firom the output of the amplifier system to generate the pilot signal. Adjusting one or more parameters preferably comprises ac^usfing the gain and phase of a signal path in the control loop using gain and phase adjusters, respectively, and the direction of flie gain and phase adjustment is changed based on the detected fiequency change in the pilot signal. In a preferred ^nbodiment of the meftod the generated pilot signal is an RF signal. In one embodiment detecting the frequency of the generated pilot signal comi>rises detecting the RF frequency. Alternatively, the generated pilot signal is an RF signal generated by iq) converting an IF signal and detecting frie frequency of tiie generated pilot signal conq^irises detecting ttie IF frequency of tiie IF signal. According to another aspect tiie present invention provides a method for controlling alignment of a feed forward amplifier system comprising an input for receivbig an input signal, a first carrier cancellation control loop coupled to the iiq>ut and having a main amplifier, a second error cancellation control loop coupled to the first control loop and having an error amplifier and a gain adjusts and a phase adjuster, and an output coupled to the second control loop and providing an output signal. The method conxpzises sampling the output signal, generating a variable fi:equency pilot signal fiom the sanq>led output signal, injecting it into the first control loop, and detecting the frequency of the generated pDot signal. The method fruther comprises adjusting the settii^ of the gain and phase adjusters in the second control loop from a first adjustment setting to a second adjustmeait setting using an alignment direction, detecting the frequency of the generated pilot signal after the adjusting, detecting the difference in the frequency of the generated pilot signal between the first and second adjustment settings, altering tiie alignm^it direction using tiie frequency difference between the first and second adjustment settings. The method further coniprise adjusting the setting of &€ gain and phase adjusters in the second control loop from the second setting to a third setting using the altered alignment direction. In a preferred embodiment of the method for controlluxg alignment of a feed forward amplifier system, altering tlie alignment direction using the frequency difference between the first and second adjustment settings comprises multiplying tiie firequency difference by a direction chaise parameter. The method further comprises determining if tiie direction change parameter is too great or too small, and decreasing or increasing the direction change parameter if necessary. In one embodiment of the method the generated pilot signal is an RF signal and detecting the frequency of the generated pilot signal conq)rises detecting the RF frequency. In ano&er embodiment of lixe method fiie generated pilot signal is an RF signal generated by up converting an IP signal and detecting the frequency of the generated pilot signal comprises detecting the IF frequency of Ifae IF signal. According to another aspect the present invention provides a feed forward amplifier comprising an RF input for receiving an RF signal, and a cazrier cancellation loop comprisiag a main amplifier receiving and amplifying the RF ^gnal, a main amplifier output sampling coupler, a first delay coupled to tiie RF input and providing a delayed RF signal, and a carrier cancellation combiner coiq}ling tiie delayed RF signal to ibs sampled output fix>m the main amplifier. The feed forward amplifier frirther conq>rises an ^ror cancellation loop comprising an error amplifier receiving and amplifymg the output of tiie carrier cancellation combmer, a gain adjuster and a phase adjuster coupled between tiie carrier cancellafion combiner and error amplifier and respectively receiving gain and phase adjustment control signals, a second delay co\q)led to the output of the main amplifier, and an error injection coupler combining the output from the error amplifier and tiie delayed main amplifier output from the second delay so as to cancel distortion introduced by the main amplifier. The feed forward amplifier fiirtiier comprises an RF ou^ut coupled to tiie error iqection coi^ler output and providing an amplified RF signal, an output sampling cox^ler for providing a sampled output of the amplified RF signal, a and positive feedback pilot generator chxniit for generating a pUot signal from the sampled output of the amplified RF sigrud and providing the pilot signal to the input of the main amplifier. The positive feedback pilot generator circuit includes a fi:equency detector for detecting the firequency of the generated pilot signal and provides a pilot frequency signal. A controller programmed with a loop control algoritimi is coi5)led to receive the pilot frequency signal and outputs the gain and phase adjustment control signals to tiie gain adjuster and phase adjusts. The controller adjusts the direction of change of the gain and phase adjustment control signals provided to the gain adjuster and phase adjuster based on changes in the pilot frequency signal. In a preferred embodiment of the feed forward amplifier the positive feedback pilot ^nerator circuit further comprises means for providing a detected pilot power signal fit>m the sampled output of the amplified RF signal which varies with Ihe str^igth of the uncancelled distortion from the error cancellation loop and the controller is coupled to receive the detected pilot power signal. In a preferred embodiment the positive feedback pilot generator circuit comprises means for generating an intemiediate frequency pilot signal fcom the sampled outpTit of the amplified RF signal, a local oscillator providing a fixed firequency signal, and a mixer receiving the intemiediate fiequ^icy pilot signal aini fixed firequency signal and outputting the pilot signal at an KF fi:equency. In one embodim^it, the positive feedback pilot generator circuit fiirther comprises a sampling coiq)ler, coupled to the output of the mixer and providing the sampled KF frequency pilot signal to the firequraicy detector, and the firequency detector detects the KF frequency of the pilot signal and provides the pilot firequency signal correspondii^ thereto to said controller, hi another embodiment the positive feedback pilot generator circuit fiirther comprises a sampling coi^ler, coupled to the ou^ut of the means for generating an intermediate firequency pilot signal, the sampling coiq>ler providing tiie sampled intermediate frequency pilot signal to the firequency detector, and the firequency detector detects the firequency of the intermediate firequency pilot signal and provides tiie pilot fiequency signal corresponding thereto to the controller. The means for generating an intermediate firequency pilot signal firom the sampled ouQnit of the amplified KF signal may comprise a second mixer cottpled to the local oscillator and receiAong the sampled output of the amplified RF signal and providing an intermediate frequency sampled output signal and a band limiter for providing a band limited signal corresponding to uncancelled pilot signal in the sampled output In a preferred embodiment the pilot frequency signal is a voltage corresponding to the detected firequency. In a preferred embodiment the control algorithm iteratively adjusts ihe aligrmient direction to minimize the detected firequency change. In a preferred embodiment the control algorithm also adjusts the amount of alignment direction change based on successive increases or decreases in the detected firequency change, Fur&er features and advantages will be appreciated from the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS Figure 1 is a block schematic drawing of a jnior art feed forward power aiiq>lifier. Figure 2 is a block schematic drawing of a feed forward power amplifier employing a positive feedback pilot generation and detection circuit with measurement of pDot frequency ia accordance with the pres^ invention. Figure 3 is a block schematic drawing of a first embodiment of the positive feedback pilot generation and detection circuit employed in the feed forward power anq)I]fier of Figure 2 with measurement of pilot frequency at RF. Figure 4 is a block schematic drawing of a second embodiment of frie positive feedback pilot generation and detection circuit employed in the feed forward pov^^er amplifier of Figure 2 with measurement of pilot frequency at IF. Figure 5 is a contour plot illustrating no phase change, (S4»i = 0), in response to an alignment step (a2-auP2-Pi)> vftuch is in the direction of Ihe optimal alignment (cXo,Po)> Figure 6 is a contour plot illustrating phase change, 5^u hi response to an alignment step (oa-oLu^-P\\ which is not in the direction of the optimal alignment (Oo^Po)- Figure 7 is a contour plot illustrating change in step direction based on pilot fi:equency change, ScDpHot,!* when ke is too small. Figure 8 is a contour plot illustrating change in step direction based on pilot frequency change, 5mpaot,is ^^tienke is too large. Figure 9 is a flow diagram illustrating an algorithm for selecting Ihe step direction in the loop 2 alignment search in accordance with tte piesent invention. DETAILED DESCRffTION OF THE INVENTION The present invention provides a feed forward amplifier system and method which uses the pilot frequency fix>m a positive feedback pilot generation and detection drcmt to improve second loop convergence. A positive feedback pilot generation system is disclosed in U.S. patent application no. 10/838,985 filed May 5, 2004, the disclosure of yAAdk is mcorporated hwrdn by reference in its entirety. The pilot generation and detection systrai of the above-noted 10/838,985 application operates using an intennediate frequency (IF) detection circuit and positive feedback. It is used in a feed forward power amplifier to assist tiie automatic control of the second loop alignment. The pilot system generates a pilot tone vAien the second loop of a feed forward amplifier is misaligned. The pilot system also detects llie residual pilot at the output of the feed.forward an^lifie^, after the second loop cancellation. A voltage proportional to Ihe log of the detected power is provided fit>m tiie pilot system to an adaptive alignment controller. The adaptive controller adjusts the alignment of the second loop to minimize the detector volt^e Qog of the residual pilot power). In the present invention tiie pilot system also detects tiie fi:equency of the generated pilot tone and the system controUea: uses the frequency information to control the direction of the alignment adjustment steps to improve the convergeiice speed of second loop alignment Figure 2 is a block schematic drawing of a feed forward power amplifier employing a positive feedback pilot generation and detection circuit with measurement of pilot frequency in accordance with the present inventioiL Positive feedback pilot generation and detection circuits in accordance with two embodiments of the invention are shown in Figure 3 and Figure 4. Figure 3 shows an embodiment employix^ frequency measurement performed on tiie RF pilot Figure 4 shows an embodiment employing firequency measurement performed on the IF pilot. Both embodiments are useful because the search algorithm uses the fi-equfflcy diSerence of tiie pilot before and after a step in the alignment setting. It should also be appreciated tiiat frequency measurements can be obtair}ed fi:om other positions within the RF and IF circuits (and such implementations are equally within fee scope of the present invention). The feed forward an:q)lifier of the present invention may also incorporate known features other than the novel aspects described in detail herein and such known features will not be described in detail. For example, additional features of a feed forward amplifier architecture and control system are described in US patent application serial no. 10/365,111 filed Februaay 12, 2003, U.S. patent no. 6,794,933, the disclosure of which is incorporated hearein by reference in its entirety. Referring to Figure 2, the feed forward amplifier includes an input 12 ^^ch receives an input KF signal to be amplified and an output 14 which outputs the amplified RF signal. The RF signal may be a hi^ bandwidb signal such as a CDMA (Code Division Multiple Access) spread spectrum communication signal or WCDMA (Wide Code Division Multiple Access) signal or other RF signal. The iiiput RF signal is split into a main amplifier signal path and an error ampUfier signal path at ir^ coiipler 30 in accordance wilh well known feed forward amplifier design. The main an:q)lifier signal path includes main amplifier 16. The main amplifier signal path fiuther mcludes input and pre-distortion circuitry 20. The input circuitry may inuciude conventional preamplifier and group delay circuitry (not shown), and gain and phase control circuitry 50,52, respectively, implemented in accordance with conventional feed forward design. The pre-distortion circuitry 48 in turn pre-distorts the input signal to reduce IMDs introduced by main amplifier 16 and may be optional In some implementations. Ir^ut and predistortion circuitry 20 is controlled by loop 1 control signals 44 provided firom controller 24. In particular, these control signals include predistortion control signals 49, gain adjuster settings 51 and phase adjuster settings 53. A positive feedback pilot generation circuit 22 (described in detail in relation to Figures 3 and 4 below) provides a pilot signal 58 which is iiyected into the main amplifier iixput at pilot injection coupler 23 as illustrated and is used to control loop 2 aligmnent (as described below). The positive feedback pilot gen^Bfion circuit 22 also provides a signal corresponding to Ihe frequency of the generated pilot signal along line 61 to controller 24 winch is izsed to improve the rate of convergence of the loop 2 alignment control (as described in more detail below). The pilot signal is extracted at the amplifier output by pilot sampliog coupler 25 and detected by circuit 22 and Ihe detected pilot signal 60 is used by controller 24 to provide the loop control to minimiy^ the pilot signal in the output signal and thereby minimize distortion in the ou^nxt signal (as described m more detail below). The main amplifier signal path finlher includes a main amplifier output sampling coiipler 26 and delay 28, generally in accordance vAQx conventional feed forward design. Still referring to figure 2, the oror amplifier signal patii includes input signal coi^ler 30 wiiich san:q)les the RF input signal and provides it to the error amplifier 34 via delay 32, carrier cancellation combiner 36 and pre-error inpizt circuitry 38. More specifically, delay 32 and carrier cancellation combiner 36 operate as in a conventional feed forward amplifier such that the sanapled output of Ifae main amplifier 16 is att^iuated by attenuator 40 and combined with the delayed input signal at cairi^ cancellation combiner 36 to substantially cancel all but the distortion component of the sampled signal fix)m the main signal pafti. This carrier cancellation completes loop 1 of the feed forward amplifier. Tlie output of carrier cancellation combiner 36 is sampled by coupler 37 and the sampled signal is provided to carrier cancellation detector 39. The detected carrier cancelled signal 41 is provided to controller 24 v^ch uses the detected signal to control Ifae loop 1 gain and phase adjuster settings SI, 53 to minimize the detected carrier. In some applications and implementations it may be advantageous to control the loop 1 cancellation at combiner 36 to retain some RF carrier component in the resulting signal and the resulting signal is not purely the distortion component of tiie main amplifier. Nonetheless, for the purposes of the present application the resulting signal will be referred to as the distortion conq)onmt and it should be understood some carrier component may be included. This distortion component of the signal is provided to pre-error input curcuitry 38. Pre-error input ckcuitry 38 may include conventional preamplifier and group delay cucuitry (not shown), and gain and phase cozrtrol circuitry 54, 56. Controller 24 provides loop 2 control signals 46, comprising gain adjuster settings (a) on line 55 and phase adjuster settings (P) on line 57, to minimize the detected pilot fix)m pilot detector 22. UnUke the main path a predistortion circuit is typically not required in the error path due to the more linear nature of the error amplifier operation. The output of circuitry 38 is provided to error amplifier 34 which restores the magnitude of the sampled distortion components (DvlDs) to that in the main signal path. The aniplified distortion component output fi-om error amplifier 34 is combined out of phase with the delayed main amplifier output at error injection coiq}ler 42 to cancel the distortion component in the main signal path. This earor cancellation completes loop 2 of fhe amplifier. A substantially distortion fitee amplified signal is liien provided to fhe output 14. A san^le of the ou^ut sigoal 18 is provided by coupler 25 to pilot detector and generator circuit 22. Any residual pilot signal in the output is detected by the pilot detector circuitry 22 and provided as a pilot power signal 60. The pilot power 60 is used by the controller 24, along with the carrier cancelled signal 41, to provide control signals 44 and 46. The two controls 44, 46 may be essentially independent and may be viewed as separate control of the two loops; loopl comprismg circuitry 20, main amplifier 16, main amplifier output sampling coupler 26, attenuator 40, input signal coi^ler 30, groxsp delay 32 and carrier cancellation combiner 36; and loop 2 comprisii^ main amplifier sampling coupler 26, attenuator 40, carri^ cancellation combiner 36, pre-error circuit 38, error amplifier 34, delay 28 and error injection coupler 42. Loop 1 control by controller 24 employs signal 41 to adjust gain and phase adjusters SO, 52 to minimize the detected carrier 41 at the output of Loop 1. Loop 2 control by controller 24 employs the detected pilot pow^ 60 to adjust the gain and phase adjusters 54, 56 to minimize the detected pilot pow^ 60 and the detected pilot fi:equency 61 to select the adjustment direction in the two dimensional gain/phase space to minimize the number of adjustment steps needed to reach the optimal adjustment setting, as described in more detail below. Referring to figure 3, a preferred embodiment of Ihe positive feedback pilot generator 22 is illustrated in a block schematic drawing. As shown the circuit comprises a detection signal path 62 and a pilot generation signal path 64. The sampled RF output 18 of the feed forward amplifier is the ii^ut to Ihe detection path 62. (An alternative approach is to measure the output of a dynamic range extender (DRE), vMch provides the feed forward amplifier output with some carrier cancellation. Such a dynamic range extender is described in U.S. Patent No. 6,147,555 issued November 14, 2000, e.g., in Figures 14 and 15 thereof, the disclosure of which is incorporated herein by reference.) The detection portion 62 of the system preferably conqxrises a bandpass pow^ detector circuit, which detects uncancelled power in a relatively narrow bandwidth portion of the sampled amplifier output 18 at a fiequency outside of the RE carrier bandwidth. The bandpass power detector circuit preferably comprises a mixer 66, bandpass filter 72, and a power detector 76. IF gain stages 70, 74 may also be employed. depending on the signal strengfii of tiie sanq>led output 18. The RF iiq[)ut 18 to the detection path is down-converted to an IF fiequency by Local OsdUator (LO) 68 and mixer 66- The IF signal is then bandpass filtered by filter 72 to provide a relatively narrow bandwidth signal mcluding the pilot signal fi^uency. The power of Ibis bandpass limited signal is Ihen detected by power detector 76. Power detector 76 may comprise a log detector or RMS detector, for example. The output 60 of the power detector 76 conesponds to the residual pilot power after the second loop cancellation. This pilot power output 60 is provided to the feed forward loop controller 24 (Figure 2). The pilot generation circuitry 64 is preferably tiie reverse line-up of the bandpass power detector circuit with the addition of a limiter before the bandpass filter. The pilot generation circuit 64 preferably comprises a limiter 82^ bandpass filter 84, mixer 88, and IF gain stages 80, 86. Additional or fewer IF gain stages may be employed, depending on signal strength. Ttie pilot generation circuit 64 uses Ihe bandpass filtered IF signal 78 firom the detection patii 62 as an input The signal 78 is anqilified by IF gain stage 80 then passed through limiting circuit 82 that clips tiie amplitude of the signal when the signal is above a threshold level. The limited signal is banc^ass filtered by filter 84 then iq>-coiiverted to RF by mixer 88 and LO 68, after a second IF gain stage 86 (if necessary). ' The above-mentioned limiter 82 limits the amplitude of the pilot. The limiter 82 may be a device that reduces the amplitude of a signal exceeding a threshold or a nonlinear device that saturates vjbsn driven by a M^ level signal. Saturation, or gain reduction with increasing signal level, occurring in olher parts within the pilot generator 64, such as the second multiplier 88 or IF gain stages 80,86, may also provide a means of limiting. The same LO 68 fiequency is preferably used for botii the pilot detection down-conversion at mixer 66 and tiae pilot generation i^MJonversion at mixer 88. The frequency of LO 68 is chosen to place the pilot signal outside of the bandwidth of tiie RF carrier of tiie input signal to the feed forward amplifier and to fecilitate detection of the signal m circuit 62. Also, a suitable choice of LO fiequency may allow a relatively inesqpensive IF filter 72 to be employed. For example, a IX) fiequency of about 85 MHz fiequeiicy sihifi from the earner band will allow an me3q>ensive SAW filter to be used, e.g. wife, a 5 ME[z pass band. Various other choices of LO fi:equency and jBlter passband are also possible, howev^. As fiirther shown in Figure 3 Ihe pilot signal output line 58 of pilot generation circuitry 64 is sampled by sampling coiq>ler 90 and the sampled output (pilot signal) is provided to fiequency measurement circuit 92. Frequency measurement circuit 92 detects the RF fiequency of tiie sampled pilot signal and provides a corresponding voltage signal along line 61 to controller 24 (Figure 2). An alternate embodiment of the pilot detection and generation circuit 22 is shovm. in Figure 4. This embodiment is identical to Figure 3 ^th the exception that tiie output pilot signal is measured at IF instead of RF, More specifically, as shown the IF pilot signal ou^ut from IF gain circuit 86 is sampled by sanq>Iing coiqpl^ 94 and the sampled IF output (IF pilot signal) is provided to firequency measurement circuit 96. Frequency measurement circuit 96 detects the IF frequency of the sampled pilot signal and provides a corresponding voltage signal along line 61 to controller 24 (Figure 2). In operation, the pilot detection and generation circuit 22 creates a narrow bandwidth, positive feedback loop through the main amplifier 16 and tiie second loop of tiie feed forward amplifiier (Figure 2). When combined with tiie limiting drcuit 82, a limit-cycle oscillation will develop usii^ noise presort in the feed forward amplifier and the pilot system, assuming that the loop has sufficient gain. The cancellation of tiie second loop affects the gain and phase of the positive feedback loop. As a result, good alignment of the second loop will 5iq)press the limit-cycle oscillation. The degree of alignment required to siqjpress the limit cycle is selectable based on the amount of IF gain provided by the IF gain stages preceding the limiter 82 or by adjustmg the clipping threshold of limiter 82. A number of modifications of the illustrated implementation of the positive feed back pilot generation circuit 22 are possible. For example, Ibe circuit may employ an automatic level control circuit v^th related modifications in Ihe overall circuit design, as described in application serial no. 11/369,529 filed TsAsxcb. 7, 2006, the disclosure of vAich is incorporated herein by refermce in its entirety. Also, an inq)lementation of Hie bandpass po^ver detector circuit 62 may employ an RF filter which is placed before the mixer 66 to reject image frequencies. In such an approach, a similar RF filter is preferably included wi&in the pilot generation path 64 afier the mixer 88. Also, it is possible to eliminate the bandpass filter 84 within the pilot generation path 64. However, such an implemetrtalion without filter 84 may not be preferred since it will waste pilot energy by producing signal components that are not detectable by the bandpass power detector circuit 62. These additional spectral components will be attenuated by the second loop cancellation as part of the feed forward compensatioiL Also, as noted above, the number of IF gain stages, the threshold of limiter 82, the LO firequency and the filter passband bandwidtti may aU be varied in accordance with the particular implementation and the particular RF carrier being amplified. Next the use of the pilot fi«quency for improved loop 2 convergence will be described in more detail. By measuring the firequency of the generated pilot, phase information regarding the second loop cancellation transfer function can be inferred. Changes in the pilot fi-equency as the search algorithm makes steps in the second loop aligiuneat indicate errors in Ihe direction of the search. The cancellation transfo function of the second loop is determined by gain and phase alignment adjusters (54, 56, respectively, in Figure 2). Assume that the alignment adjuster is modeled as where Oopt and Popt are the optimal gain and phase alignment settbigs, respectively, and Aoopt and Apopt are the misalignment in the gain and phase adjusters, respectively. Assuming &e jAoopti and \|Apopt\| are small, the output of Ihe pilot detector can be approximated as \)sdiere \|k\| and Pmb are constants. It can be seen from (Eq. 2) that when Hie detected volt^e is plotted as a function of the gam and phase adjuster settuxgs, Ifae resulting contours are concentric ellipses surrounding the optimal alignment setting (see Figure S and Figure 6). The phase shift of the second loop cancellation transfer function is It can be seen from CBq. 3) that a step in the alignment setting that keeps the ratio APopt/Aocopt constant will not alter the cancellation phase. This corresponds to making an alignment step that is in the direction of the optimal alignmmt (see Figure 5). If a step in the alignment setting is not in the direction of tiie optimal alignment, the phase will change (see Figure 6). To illustrate the phase change, assume tiie initial alignment settmg is (ai,Pi) and the alignment after the step (Aai,Api) is (a2,P2) - (ai+Aai,pi+APi). The phases before and after the step are The frequency of the generated pilot generated is a natural mode of the positive feedback. It must be wititin the passband of the pilot system and create a loop phase that is a multiple of 2?! radians. That is. Die pilot frequency must satisfy where copuot is Ihe pilot firequeQcy, Aioop is the total loop delay, and ^o is a phase of&et Changes in the phase of the cancellation transfer function of the second loop affect (Eq. 7). As a result, to preserve ^Bq. 7), the pilot frequency must chai^ as well. That is, the change in pilot frequency due to a phase change induced by alignment stq) 1 is Thxis, a change in the pilot frequency, measured at eidier IF or RF, will indicate changes in the phase of the second loop cancellation transfix frmctioa The manner in which the frequency change mformation is used to select the next step direction is next described. The first step direction is AA&ere ke is a constant Figure 7 and Figure 8 illustrate the effect tiiat &e selection of ke has on the search trajectory- In Figure 7, the value of ke is too small making "flie new search direction, 02, too small of a change fix)m the first search direction, Bj. As a result, the pilot frequency wiU continue to increase. Successive increases (or decreases) in the pilot taeqpsacy suggest an increase in ke is needed. In Figure 8, the value of ke is too large makii^ the new search dhection, 62, too large of a change frt)m the first search direction, Gi. As a result, the change in pilot fiequency will alternate directions each step. Alternating chants in the pilot firequency suggest a decrease in ke is needed. A preferred embodiment of the algorithm for selecting the step direction 6ZH-I in the aligmnent search is sho^Mi in Figure 9. As shown Ihe algorithm initiates at 100 and at 102 an initial aligmnent stq) direction is selected, which initial direction may be arbitrary. Next at 104 the algorithm proceeds to measure the pilot fiequency based on the pilot frequency signal provided to the controller along line 61 (Figure 2). At 106 a counter is initialized to begm a series of aligmnent steps using measurements of the pilot frequency in order to optimize the step direction. More specifically, at 108, die algorithm initiates an alignment step (Aai,Api) in the initml aligmnent direction by incrementing the gain and phase adjuster settmgs corresponding to the selected direction. Next at 110 the algorithm proceeds to measure the pilot firequency at the new settings using the pilot frequency signal provided to the controller along line 61. Next at 112 tiie algorithm proceeds to compute the difference in the pilot frequency between the initial setting and the new setting. Next at 114 the difference in pilot firequency, determined at 112, is used to alter the alignment st^ direction, multiplying the drff^ence in frequency by a constant value ke defining the amount of change in step direction (i.e. the size of the angle of direction change in 2D gain phase space). Next at 116 it is determined if the value of the constant ka is too large or too small and if necessary the value of the constant ke is increased or decreased (as described above in relation to Figures 7 and 8). Next at 118 the counter is incremented and Ihe alignment adjustment step direction processing flow, 108,110,112,114 and 116 is repeated. This iterative process flow continues as long as it is converg^g, vMch is indicated by a decreasing level of Ihe detected pUot power 60. The detected pilot power 60, denoted by Vdet» is measured at 104 and 110, and the difference, AVdet, is conqnited at 112. The search is convergmg as desired when AVdet For the case ^ere the alignment adjustment step (Aa,AP) causes the iterative process to divCTge, as indicated by AVdet > 0, the alignment adjustment direction is reversed by adding TI radians to (Eq. 10) and 114, and the step size is reduced before repeating 108. The algorithm for selecting the step size used at 108 may be the same as tiie power minimization approaches described in U.S. patent aH)lication 10/733,498 filed December 11, 2003, U.S. patent no. 7,002,407, and U.S. patent application 11/018,216 filed December 21,2004, the disclosures of wbich are incorporated herein by referrace in their entirety. The algorithm of Figure 9 and additional aspects of alignment control processing described in tixe above noted applications and patents may be unplemented in controller 24 n^ng a suitably progranmied microprocessor (additional details are described in the above noted patent applications incorporated herein by reference). It is worth noting that the frequency change, ScOpAob can be large due to the 2TC radian multiple in (Eq. 7). When the frequency shifts near the edge of the passband of the pilot S3^stem, and a discrete frequency change of 27m / Aioop may occur to ferce the pilot frequency to remain within the passband. When large changes in frequency are detected, the measured value of ScDpiiot should not be used in 114. Within 114, Sa> should be constrained, using instead a modified 6piiot I- Alternatively, the search algorithm in Figure 9 can be restarted afrer detecting large fr:equency changes. Large frequency changes also occur when tiie convergence of the iterative process is nearly complete. The pilot amplitude drops rapidly when the alignment is near an optimal setting because the loop gain of the positive feedback is no longer sufSdent to maintain the pilot oscillation making the detected pilot firequency measur^nent 61 unreliable. Such converged conditions are desirable and are indicated by Vdd reaching its minimimi value QAVdetl > 0 for aU possible alignm^it step directions). When this condition is detected, any search direction can be selected as long as it is varied over time. An alternative embodiment of the algorithm for selecting the step direction in the alignment search is desadbed below. Rather than selecting ke, it is possible to base tiie search direction on the sign of the difference in the pilot frequency (ScOpaoO- The search direction is updated using when the detected pilot power 60 is increasing (diverging, AVdet > 0). The search algorithm forces lateral movement in the trajectory relative to the direct pa& to the optimal settmg. Lateral movement changes Hxe angle 8^ (see Figure 6)» causing a frequency change, 5a>pUot. Note that 5^ and ScDpUot are proportional to the ratio of the step size and the distance to the optimal settmg. The algorithm also adjusts the step size to make the e7q)ected value of \|5\| constant. As an illustrative example, the step size can he increased by a fector of 1,4 when 15^\| 0.3. The step size can also be decreased by 0.7 when AVdet> 0, Since both S^) and AVdet are used to adjust the step size, the search is better damped than if it was based on the detected pilot power 60 only. The best thresholds and step adjustment &ctors are dependent on the feed forward amplifier system and can be obtained easily using experiments. The present invention has been described in relation to a presently preferred embodiment, however, it will be £^(>preciated by those skilled in the art "Biat a variety of modifications, too numerous to describe, may be made vMlo remaining wifiiin the scope of the present invention. Accordii^y, the above detailed description should be viewed as illustrative only and not limiting in nature. WHATISCSLAIMEDIS: 1. A method for controlling alignment of a control loop in an amplifier system, comprising: generating a variable frequency pilot signal and injecting the pilot signal into &e amplifier system; detecting any uncanceled pilot signal at an output of the control loop; detecting the frequency of the generated pilot signal; adjusting one or more parameters of the control loop; detecting a frequency change in the variable frequency pilot signal; and controlling said adjusting based on the detected frequency change. 2. A method for controlling alignment of a control loop in an amplifier system as set out in claim 1, wherein generating said variable frequency pilot signal comprises using feedback from the ou^ut of the amplifier system to generate the pilot signal. 3. A method for controlling alignment of a control loop in an amplifier syst^n as set out in claim 1, v^erein said adjusting one or more parameters comprises adjusting &e gain and phase of a signal path in the control loop using gain and phase adjusters, respectively. 4. A method for controlling alignment of a control loop in an amplifier system as set out in claim 3, v^erein the direction of the gain and phase adjustment is changed based on said detected fi:equency change in the pilot signal. 5. A metixod for controlling alignment of a control loop in an amplifier system as set out in claim 1, v^erein the gen^ated pilot signal is an RF signal and -herein detecting the frequency of the generated pilot signal comprise detecting the RP firequency. 6. A method for controlling alignment of a control loop in an amplifier system as set out in claim 1, wherein tiie generated pilot signal is an RF signal generated by vp converting an IF signal and whei^i detecting Hie frequency of ihe generated pilot signal con:q)rises detecting the IF firequency of said IF signal. 7. A method for controlling alignment of a feed forward amplifier system comprising an input for receiving an input signal, a first carrier cancellation control loop coupled to the mput and having a main axx^lifier, a second error cancellation control loop coupled to the first control loop and hamg an error amplifier and a gain adjuster and a phase adjuster, and an output coiq)led to the second control loop and providing an output signal, the method comprising: sampling Ibe output signal; generating a variable firequency pilot signal fiom the sampled output signal and injecting it intxj the first control loop; detecting the firequency of the generated pilot signal; adjusting the settmgs of the gain and phase adjusters in said second control loop fiom a first adjustment setting to a second adjustment setting using an alignment direction; detecting the firequency of the generated pilot signal after said adjusting; detecting the difiference in the firequency of Ihe generated pilot signal between said first and second adjustment settings; altering the alignment direction using the firequency difference between said first and second adjustment settings; and adjusting the settings of the gain and phase adjusters in said second control loop firom the second setting to a third setting using the altered alignment direction. 8. A method for controlling alignment of a feed fi)rward an^lifier system as set out in claim 7, wherein altering the alignment direction using the firequency difference between said first and second adjustment settings comprises multiplying the fiequency difference by a direction change parameter. 9. A method finr controUii^ alignment of a feed forward amplifier system as set out in claim 8, finrther conqnising determining if the direction change parameter is too great or too small, and decreasing or increasing tiie direction change parameter if necessary. 10. A method for controlling alignment of a feed forward amplifier system as set out in claim 7, wherein the generated pilot signal is an RF signal and "^lerdn detecting the frequency of the generated pilot signal comprises detecting the RF frequency. IL A mettiod for controlling alignment of a feed forward anq)lifier system as set out in cl^m 7, wherein the generated pilot signal is an RF signal g^ierated by up converting an IF signal and wherein detecting the frequency of the generated pilot signal con:q)iises detecting the IF fi:equency of said IF signal. 12. A feed forward amplifier, comprising: an RF iiqjut for receivii^ an RF signal; a carrier cancellation loop comprising a main amplifier receiving and amplifying said RF signal, a main amplifier output sampling coiq)ler, a first delay coupled to the RF irqait and providing a delayed RF signal, and a carrier cancellation combiner coipling the delayed RF signal to the sampled output from the main amplifier; an error cancellation loop comprising an error amplifier receiving and amplifying tbe output of the carrier cancellation combiner, a gain adjuster and a phase adjuster coiq)led between the carrier cancellation combiner and error amplifier and respectively receiving gain and phase adjustment control signals, a second delay coupled to the ou^ut of the main amplifier, and an error injection coi^ler combining tiie output from Ihe error amplifier and the delayed main amplifier output from the second delay so as to cancel distortion introduced by the main amplifier; an RF ou^ut coi^led to the error injection coupler outpfot and providing an amplified RF signal; an output sampling coupler for providing a sampled output of the amplified RF signal; a positive feedback pilot generator circuit for generating a jnlot signal fit>m tiie sampled output of the amplified RF signal and providing the pilot signal to the ii^ut of the main amplifier, ibt positive feedback pilot generator circuit including a firequency detector for detectu^ the frequency of the generated pilot signal and providing a pilot frequency signal; and a controller programmed vdtii a loop control algorithm, the controller coiq)led to receive Hbs pilot frequency signal and oxd^ntting said gain axui phase adjustment control signals to said gain adjuster and phase adjuster, the controller adjusting the direction of change of tiie gain and phase adjustment control signals provided to said gain adjuster and phase adjuster based on changes in the pilot frequency signal. 13. A feed forward amplifier as set out in claim 12, wh^ein said positive feedback pilot generator circuit further comprises means for providing a detected pilot power signal from the sampled oidput of the amplified RF signal vMch. varies with the strength of the uncancelled distortion firom the error cancellation loop and wherein said controller is coupled to receive the detected pilot power signal. 14. A feed forward amplifier as set out in claim 12, wherein said positive feedback pilot generator circuit comprises means for g^iera&ig an uitermediate frequency pilot signal from the sampled output of Ihe amplified RF signal, a local oscillator providing a fixed frequency signal, and a mixer receiving Has intermediate firequency pilot signal and fixed firequency signal and outputting the pilot signal at anRF firequency. 15. A feed forward amplifier as set out in claim 14, wherein said positive feedback pilot generator circuit further comprises a sampling coiqpler coi^>led to the ou:^ut of said mixer and providing the sampled RF firequency pilot signal to said firequency detector, and wherein said firequency detector detects the RF frequency of said pilot signal and provides said pilot firequency signal corresponding thereto to said controller. 16. A feed forward amplifier as set out in claim 14, wherein said positive feedback pilot generator circuit further comprises a sampling coupler, coupled to the output of said means for generating an intermediate fi^umcy pilot signal, said sampling coupler providing the sampled iixtermediate frequency pilot signal to said frequency detector, and wherein said firequ^icy detector detects the frequency of said intermediate firequency pilot signal and provides ^d pilot frequency signal corresponding thereto to said controller. 17. A feed forward amplifier as set out in claim 14, wherein said means for generating an intermediate frequency pilot signal £rom the sampled output of the amplified RF signal comprises a second mixer coiqiled to the local oscillator and receiving the sampled output of tiie amplified RF signal and providing an intermediate firequency sampled output signal and a band limiter for providing a band limited signal corresponding to uncancelled pilot signal in the sampled output. 18. A feed forward amplifier as set out in claim 12, vrfieiein said pilot fiiequency signal is a voltage corresponding to the detected frequency. 19. A feed forward amplifier as set out in claim 12, wherein said control algorithm iteratively adjusts Hxt aligmnent direction to minimize the detected frequency change. 20. A feed forward anq>lifier as set out in claim 19, v^rein said control algoriflmi adjusts the amount of alignment direction change based on successive increases or decreases in the detected frequency chai^.

Full Text

FEED FORWARD AMPLIFIER SYSTEM AND METHOD USING THE PILOT
FREQUENCY FROM A POSmVE FEEDBACK PILOT GENERATION AND
DETECTION CIRCUrr TO IMPROVE SECOND LOOP CONVERGENCE
RELATED APPUCATION INFORMATION
The present application claims priority to U.S. provisional application serial no. 60/668,363 filed April 5, 2005, and of U.S. provisional application serial no. 60/670^08 filed April 13, 2005, Hie disclosures of wbich are incorporated herein hy ref^ence in iSassx entirety.
BACKGROUND OF THE INVENTION
1.. Field of the Invention
The present invention relates to RF pOAver anq)lifiers and amplification methods. More paxticidarly, the present invention relates to feed forward poorer anxplifiers and methods of using a pilot to align the loops of a &ed forward ampUfier.
2. Description of the Prior Art and Related Information
A primary goal of RF power amplifier design is linearity over the range of power operation. Linearity is simply the ahility to amplify witiiout distortion. This requiremeat is critical for modem whreless communication systems but it is increasmgly difficult to achieve. This is due primarily to the bandwidth requirements of modem wireless communication systems vMoh axe placing increasing demands on amplifier linearity. Feed forward compensation is a well known approach applied to amplifies to improve linearity by estimating and canceling distortioiL In feed forward RF power amplifiers an etror amplifier is employed to amplify only distortion components w^iich are tiien combmed witii the main amplifier ou^nit to cancel the main amplifier distortion compon^it

Figure 1 illustrates a conventiotial feed forward amplifier design having a main amplifier 1 and an error amplifier 2. The hasic elements also include delays 3, 4 in the main and error path, respectively, and main to error path coi^lers 5, 6, 7 and 8. Additional elements not shown are also typically present in a conventional feed forward architecture as is well known to those skilled in the art. The delays, coi^lers and error amplifier are designed to retract distortion components from the main path and iigect out of phase distortion conqxments fix>m the error path into the main amplifier ou^ut at coxq>ler 8 to substantially eliminate the distortion component in the main amplifier path.
The performance of a feed forward amplifier may typically be analyzed based on two cancellation loops. Loopl, called Ihe earner cancellation loop, ideally provides a signal at the output of coiq>le3: 7 ^th the input RF earner component cancelled and only a distortion component remaining. Loop 2 is referred to as the error cancellation loop or auxiliary path loop. In loop 2 the distortion component provided firom coiq)ler 7 is amplified by the error amplifier 2 and injected at coiq)ler 8 to cancel the distortion component in the main path and ideally provide a distortion firee signal at the output
The quality of the distortion estimate (carrier cancellation) is determined by the alignment of the first loop in terms of gain and phase. The distortion cancellation in turn is determined by the alignment of the second loop in terms of gain and phase. In prior art systems, a pilot 9 is injected into the main amplifi^ path of tiie first loop, acting, like a known distoMon signal. The pilot signal is detected at the feed forward amplifier output by a pilot detector 10 and used to aid tide alignment process for tiie second loop. When the second loop is aligned, the pilot is cancelled. If the second loop is misaligned, residual pilot power will be detected at the output of the feed forward amplifier. The degree of the misalignmCTt is estimated fix>m tiie measured power of the residual pilot The alignment of the second loop is adjusted in an iterative manner with tiae goal of reducing the residual pilot power. Generally, it is desirable to have the feed forward anq>lifier control system adapt to tixe optimal settings as &st as possible to minimize the amount of time the amplifier operates at a less than optimal setting.

One difficulty witfa alignment control algorilihms iised to adjust the alignment settings (gain and phase) from any initial setting to that ^A^iich results in the best measured alignment is the difficulty in finding the coirect diiecdon of adjustment in Hxt two dimensional ^D) gain-phase space. Prior alignment control algorithms typically rely on dther the '^steepest descent" or the "coordinate descent" algorithms. The steepest descent algori&m adjusts the alignment settings in a direction of the gradient within the 2D gain-phase space. Dithering Ihe alignment in orthogonal directions and measuring Ihe changes in measured misalignment provides an estimate of the gradient. The coordinate descent algorithm performs two separate ID searches along pre-defined orthogonal directions (usually the gain and phase axes). The alignments are dithered to determine -which direction along the respective coordinates reduces measured misalignment Both &ese approaches have disadvantages in practical systems ^^4lich employ control processors with limited processing power and \^iere fEist loop alignment is desbred. As a result the desked &st and accurate loop convergrace has not been achieved in practical adaptive feed forward systems.
Accordingly, a need presraxtly exists for a system and method for more rapid loop alignment control in a feed forward amplifier system.
SUMMARY OF THE INVENTION
In a first aspect the present invention provides a method for controlling alignment of a control loop in an amplifier system comprising g^oerating a variable firequency pilot signal, injecting the pilot signal into the amplifier system, and detecting any uncanceled pilot signal at an output of the control loop. The method fiuiher comprises detecting the fi:equency of the generated pilot signal, adjusting one or more parameters of the control loop, detecting a firequency change m the variable firequency pilot signal and controlling the adjusting based on the detected fiequency change.
In a preferred ranbodunent of the method for controlling alignment of a control loop in an amplifier system generating the variable firequency pilot signal comprises using feedback firom the output of the amplifier system to generate the pilot signal. Adjusting one or more

parameters preferably comprises ac^usfing the gain and phase of a signal path in the control loop using gain and phase adjusters, respectively, and the direction of flie gain and phase adjustment is changed based on the detected fiequency change in the pilot signal. In a preferred ^nbodiment of the meftod the generated pilot signal is an RF signal. In one embodiment detecting the frequency of the generated pilot signal comi>rises detecting the RF frequency. Alternatively, the generated pilot signal is an RF signal generated by iq) converting an IF signal and detecting frie frequency of tiie generated pilot signal conq^irises detecting ttie IF frequency of tiie IF signal.
According to another aspect tiie present invention provides a method for controlling alignment of a feed forward amplifier system comprising an input for receivbig an input signal, a first carrier cancellation control loop coupled to the iiq>ut and having a main amplifier, a second error cancellation control loop coupled to the first control loop and having an error amplifier and a gain adjusts and a phase adjuster, and an output coupled to the second control loop and providing an output signal. The method conxpzises sampling the output signal, generating a variable fi:equency pilot signal fi*om the sanq>led output signal, injecting it into the first control loop, and detecting the frequency of the generated pDot signal. The method fruther comprises adjusting the settii^ of the gain and phase adjusters in the second control loop from a first adjustment setting to a second adjustmeait setting using an alignment direction, detecting the frequency of the generated pilot signal after the adjusting, detecting the difference in the frequency of the generated pilot signal between the first and second adjustment settings, altering tiie alignm^it direction using tiie frequency difference between the first and second adjustment settings. The method further coniprise adjusting the setting of *&€ gain and phase adjusters in the second control loop from the second setting to a third setting using the altered alignment direction.
In a preferred embodiment of the method for controlluxg alignment of a feed forward amplifier system, altering tlie alignment direction using the frequency difference between the first and second adjustment settings comprises multiplying tiie firequency difference by a direction chaise parameter. The method further comprises determining if tiie direction change parameter is too great or too small, and decreasing or increasing the direction change

parameter if necessary. In one embodiment of the method the generated pilot signal is an RF signal and detecting the frequency of the generated pilot signal conq)rises detecting the RF frequency. In ano&er embodiment of lixe method fiie generated pilot signal is an RF signal generated by up converting an IP signal and detecting the frequency of the generated pilot signal comprises detecting the IF frequency of Ifae IF signal.
According to another aspect the present invention provides a feed forward amplifier comprising an RF input for receiving an RF signal, and a cazrier cancellation loop comprisiag a main amplifier receiving and amplifying the RF ^gnal, a main amplifier output sampling coupler, a first delay coupled to tiie RF input and providing a delayed RF signal, and a carrier cancellation combiner coiq}ling tiie delayed RF signal to ibs sampled output fix>m the main amplifier. The feed forward amplifier frirther conq>rises an ^ror cancellation loop comprising an error amplifier receiving and amplifymg the output of tiie carrier cancellation combmer, a gain adjuster and a phase adjuster coupled between tiie carrier cancellafion combiner and error amplifier and respectively receiving gain and phase adjustment control signals, a second delay co\q)led to the output of the main amplifier, and an error injection coupler combining the output from the error amplifier and tiie delayed main amplifier output from the second delay so as to cancel distortion introduced by the main amplifier. The feed forward amplifier fiirtiier comprises an RF ou^ut coupled to tiie error iqection coi^ler output and providing an amplified RF signal, an output sampling cox^ler for providing a sampled output of the amplified RF signal, a and positive feedback pilot generator chxniit for generating a pUot signal from the sampled output of the amplified RF sigrud and providing the pilot signal to the input of the main amplifier. The positive feedback pilot generator circuit includes a fi:equency detector for detecting the firequency of the generated pilot signal and provides a pilot frequency signal. A controller programmed with a loop control algoritimi is coi5)led to receive the pilot frequency signal and outputs the gain and phase adjustment control signals to tiie gain adjuster and phase adjusts. The controller adjusts the direction of change of the gain and phase adjustment control signals provided to the gain adjuster and phase adjuster based on changes in the pilot frequency signal.

In a preferred embodiment of the feed forward amplifier the positive feedback pilot ^nerator circuit further comprises means for providing a detected pilot power signal fit>m the sampled output of the amplified RF signal which varies with Ihe str^igth of the uncancelled distortion from the error cancellation loop and the controller is coupled to receive the detected pilot power signal. In a preferred embodiment the positive feedback pilot generator circuit comprises means for generating an intemiediate frequency pilot signal fcom the sampled outpTit of the amplified RF signal, a local oscillator providing a fixed firequency signal, and a mixer receiving the intemiediate fiequ^icy pilot signal aini fixed firequency signal and outputting the pilot signal at an KF fi:equency. In one embodim^it, the positive feedback pilot generator circuit fiirther comprises a sampling coiq)ler, coupled to the output of the mixer and providing the sampled KF frequency pilot signal to the firequraicy detector, and the firequency detector detects the KF frequency of the pilot signal and provides the pilot firequency signal correspondii^ thereto to said controller, hi another embodiment the positive feedback pilot generator circuit fiirther comprises a sampling coi^ler, coupled to the ou^ut of the means for generating an intermediate firequency pilot signal, the sampling coiq>ler providing tiie sampled intermediate frequency pilot signal to the firequency detector, and the firequency detector detects the firequency of the intermediate firequency pilot signal and provides tiie pilot fiequency signal corresponding thereto to the controller. The means for generating an intermediate firequency pilot signal firom the sampled ouQnit of the amplified KF signal may comprise a second mixer cottpled to the local oscillator and receiAong the sampled output of the amplified RF signal and providing an intermediate frequency sampled output signal and a band limiter for providing a band limited signal corresponding to uncancelled pilot signal in the sampled output In a preferred embodiment the pilot frequency signal is a voltage corresponding to the detected firequency. In a preferred embodiment the control algorithm iteratively adjusts ihe aligrmient direction to minimize the detected firequency change. In a preferred embodiment the control algorithm also adjusts the amount of alignment direction change based on successive increases or decreases in the detected firequency change,
Fur&er features and advantages will be appreciated from the following detailed description.

BRIEF DESCRIPTION OF THE DRAWINGS
Figure 1 is a block schematic drawing of a jnior art feed forward power aiiq>lifier.
Figure 2 is a block schematic drawing of a feed forward power amplifier employing a positive feedback pilot generation and detection circuit with measurement of pDot frequency ia accordance with the pres^ invention.
Figure 3 is a block schematic drawing of a first embodiment of the positive feedback pilot generation and detection circuit employed in the feed forward power anq)I]fier of Figure 2 with measurement of pilot frequency at RF.
Figure 4 is a block schematic drawing of a second embodiment of frie positive feedback pilot generation and detection circuit employed in the feed forward pov^^er amplifier of Figure 2 with measurement of pilot frequency at IF.
Figure 5 is a contour plot illustrating no phase change, (S4»i = 0), in response to an alignment step (a2-auP2-Pi)> vftuch is in the direction of Ihe optimal alignment (cXo,Po)>
Figure 6 is a contour plot illustrating phase change, 5^u hi response to an alignment step (oa-oLu^-P\\ which is not in the direction of the optimal alignment (Oo^Po)-
Figure 7 is a contour plot illustrating change in step direction based on pilot fi:equency change, ScDpHot,!* when ke is too small.
Figure 8 is a contour plot illustrating change in step direction based on pilot frequency change, 5mpaot,is ^^tienke is too large.
Figure 9 is a flow diagram illustrating an algorithm for selecting Ihe step direction in the loop 2 alignment search in accordance with tte piesent invention.

DETAILED DESCRffTION OF THE INVENTION
The present invention provides a feed forward amplifier system and method which uses the pilot frequency fix>m a positive feedback pilot generation and detection drcmt to improve second loop convergence. A positive feedback pilot generation system is disclosed in U.S. patent application no. 10/838,985 filed May 5, 2004, the disclosure of yAAdk is mcorporated hwrdn by reference in its entirety. The pilot generation and detection systrai of the above-noted 10/838,985 application operates using an intennediate frequency (IF) detection circuit and positive feedback. It is used in a feed forward power amplifier to assist tiie automatic control of the second loop alignment. The pilot system generates a pilot tone vAien the second loop of a feed forward amplifier is misaligned. The pilot system also detects llie residual pilot at the output of the feed.forward an^lifie^, after the second loop cancellation. A voltage proportional to Ihe log of the detected power is provided fit>m tiie pilot system to an adaptive alignment controller. The adaptive controller adjusts the alignment of the second loop to minimize the detector volt^e Qog of the residual pilot power). In the present invention tiie pilot system also detects tiie fi:equency of the generated pilot tone and the system controUea: uses the frequency information to control the direction of the alignment adjustment steps to improve the convergeiice speed of second loop alignment
Figure 2 is a block schematic drawing of a feed forward power amplifier employing a positive feedback pilot generation and detection circuit with measurement of pilot frequency in accordance with the present inventioiL Positive feedback pilot generation and detection circuits in accordance with two embodiments of the invention are shown in Figure 3 and Figure 4. Figure 3 shows an embodiment employix^ frequency measurement performed on tiie RF pilot Figure 4 shows an embodiment employing firequency measurement performed on the IF pilot. Both embodiments are useful because the search algorithm uses the fi-equfflcy diSerence of tiie pilot before and after a step in the alignment setting. It should also be appreciated tiiat frequency measurements can be obtair}ed fi:om other positions within the RF and IF circuits (and such implementations are equally within fee scope of the present invention). The feed forward an:q)lifier of the present invention may also incorporate known features other than the novel aspects described in detail herein and such known features will

not be described in detail. For example, additional features of a feed forward amplifier architecture and control system are described in US patent application serial no. 10/365,111 filed Februaay 12, 2003, U.S. patent no. 6,794,933, the disclosure of which is incorporated hearein by reference in its entirety.
Referring to Figure 2, the feed forward amplifier includes an input 12 ^^ch receives an input KF signal to be amplified and an output 14 which outputs the amplified RF signal. The RF signal may be a hi^ bandwidb signal such as a CDMA (Code Division Multiple Access) spread spectrum communication signal or WCDMA (Wide Code Division Multiple Access) signal or other RF signal. The iiiput RF signal is split into a main amplifier signal path and an error ampUfier signal path at ir^ coiipler 30 in accordance wilh well known feed forward amplifier design. The main an:q)lifier signal path includes main amplifier 16. The main amplifier signal path fiuther mcludes input and pre-distortion circuitry 20. The input circuitry may inuciude conventional preamplifier and group delay circuitry (not shown), and gain and phase control circuitry 50,52, respectively, implemented in accordance with conventional feed forward design. The pre-distortion circuitry 48 in turn pre-distorts the input signal to reduce IMDs introduced by main amplifier 16 and may be optional In some implementations. Ir^ut and predistortion circuitry 20 is controlled by loop 1 control signals 44 provided firom controller 24. In particular, these control signals include predistortion control signals 49, gain adjuster settings 51 and phase adjuster settings 53.
A positive feedback pilot generation circuit 22 (described in detail in relation to Figures 3 and 4 below) provides a pilot signal 58 which is iiyected into the main amplifier iixput at pilot injection coupler 23 as illustrated and is used to control loop 2 aligmnent (as described below). The positive feedback pilot gen^Bfion circuit 22 also provides a signal corresponding to Ihe frequency of the generated pilot signal along line 61 to controller 24 winch is izsed to improve the rate of convergence of the loop 2 alignment control (as described in more detail below). The pilot signal is extracted at the amplifier output by pilot sampliog coupler 25 and detected by circuit 22 and Ihe detected pilot signal 60 is used by controller 24 to provide the loop control to minimiy^ the pilot signal in the output signal and thereby minimize distortion in the ou^nxt signal (as described m more detail below). The main amplifier signal path finlher

includes a main amplifier output sampling coiipler 26 and delay 28, generally in accordance vAQx conventional feed forward design.
Still referring to figure 2, the oror amplifier signal patii includes input signal coi^ler 30 wiiich san:q)les the RF input signal and provides it to the error amplifier 34 via delay 32, carrier cancellation combiner 36 and pre-error inpizt circuitry 38. More specifically, delay 32 and carrier cancellation combiner 36 operate as in a conventional feed forward amplifier such that the sanapled output of Ifae main amplifier 16 is att^iuated by attenuator 40 and combined with the delayed input signal at cairi^ cancellation combiner 36 to substantially cancel all but the distortion component of the sampled signal fix)m the main signal pafti. This carrier cancellation completes loop 1 of the feed forward amplifier. Tlie output of carrier cancellation combiner 36 is sampled by coupler 37 and the sampled signal is provided to carrier cancellation detector 39. The detected carrier cancelled signal 41 is provided to controller 24 v^ch uses the detected signal to control Ifae loop 1 gain and phase adjuster settings SI, 53 to minimize the detected carrier. In some applications and implementations it may be advantageous to control the loop 1 cancellation at combiner 36 to retain some RF carrier component in the resulting signal and the resulting signal is not purely the distortion component of tiie main amplifier. Nonetheless, for the purposes of the present application the resulting signal will be referred to as the distortion conq)onmt and it should be understood some carrier component may be included. This distortion component of the signal is provided to pre-error input curcuitry 38. Pre-error input ckcuitry 38 may include conventional preamplifier and group delay cucuitry (not shown), and gain and phase cozrtrol circuitry 54, 56. Controller 24 provides loop 2 control signals 46, comprising gain adjuster settings (a) on line 55 and phase adjuster settings (P) on line 57, to minimize the detected pilot fix)m pilot detector 22. UnUke the main path a predistortion circuit is typically not required in the error path due to the more linear nature of the error amplifier operation. The output of circuitry 38 is provided to error amplifier 34 which restores the magnitude of the sampled distortion components (DvlDs) to that in the main signal path. The aniplified distortion component output fi-om error amplifier 34 is combined out of phase with the delayed main amplifier output at error injection coiq}ler 42 to cancel the distortion component in the main signal path. This

earor cancellation completes loop 2 of fhe amplifier. A substantially distortion fitee amplified signal is liien provided to fhe output 14.
A san^le of the ou^ut sigoal 18 is provided by coupler 25 to pilot detector and generator circuit 22. Any residual pilot signal in the output is detected by the pilot detector circuitry 22 and provided as a pilot power signal 60. The pilot power 60 is used by the controller 24, along with the carrier cancelled signal 41, to provide control signals 44 and 46. The two controls 44, 46 may be essentially independent and may be viewed as separate control of the two loops; loopl comprismg circuitry 20, main amplifier 16, main amplifier output sampling coupler 26, attenuator 40, input signal coi^ler 30, groxsp delay 32 and carrier cancellation combiner 36; and loop 2 comprisii^ main amplifier sampling coupler 26, attenuator 40, carri^ cancellation combiner 36, pre-error circuit 38, error amplifier 34, delay 28 and error injection coupler 42. Loop 1 control by controller 24 employs signal 41 to adjust gain and phase adjusters SO, 52 to minimize the detected carrier 41 at the output of Loop 1. Loop 2 control by controller 24 employs the detected pilot pow^ 60 to adjust the gain and phase adjusters 54, 56 to minimize the detected pilot pow^ 60 and the detected pilot fi:equency 61 to select the adjustment direction in the two dimensional gain/phase space to minimize the number of adjustment steps needed to reach the optimal adjustment setting, as described in more detail below.
Referring to figure 3, a preferred embodiment of Ihe positive feedback pilot generator 22 is illustrated in a block schematic drawing. As shown the circuit comprises a detection signal path 62 and a pilot generation signal path 64. The sampled RF output 18 of the feed forward amplifier is the ii^ut to Ihe detection path 62. (An alternative approach is to measure the output of a dynamic range extender (DRE), vMch provides the feed forward amplifier output with some carrier cancellation. Such a dynamic range extender is described in U.S. Patent No. 6,147,555 issued November 14, 2000, e.g., in Figures 14 and 15 thereof, the disclosure of which is incorporated herein by reference.) The detection portion 62 of the system preferably conqxrises a bandpass pow^ detector circuit, which detects uncancelled power in a relatively narrow bandwidth portion of the sampled amplifier output 18 at a fiequency outside of the RE carrier bandwidth. The bandpass power detector circuit preferably comprises a mixer 66, bandpass filter 72, and a power detector 76. IF gain stages 70, 74 may also be employed.

depending on the signal strengfii of tiie sanq>led output 18. The RF iiq[)ut 18 to the detection path is down-converted to an IF fiequency by Local OsdUator (LO) 68 and mixer 66- The IF signal is then bandpass filtered by filter 72 to provide a relatively narrow bandwidth signal mcluding the pilot signal fi^uency. The power of Ibis bandpass limited signal is Ihen detected by power detector 76. Power detector 76 may comprise a log detector or RMS detector, for example. The output 60 of the power detector 76 conesponds to the residual pilot power after the second loop cancellation. This pilot power output 60 is provided to the feed forward loop controller 24 (Figure 2).
The pilot generation circuitry 64 is preferably tiie reverse line-up of the bandpass power detector circuit with the addition of a limiter before the bandpass filter. The pilot generation circuit 64 preferably comprises a limiter 82^ bandpass filter 84, mixer 88, and IF gain stages 80, 86. Additional or fewer IF gain stages may be employed, depending on signal strength. Ttie pilot generation circuit 64 uses Ihe bandpass filtered IF signal 78 firom the detection patii 62 as an input The signal 78 is anqilified by IF gain stage 80 then passed through limiting circuit 82 that clips tiie amplitude of the signal when the signal is above a threshold level. The limited signal is banc^ass filtered by filter 84 then iq>-coiiverted to RF by mixer 88 and LO 68, after a second IF gain stage 86 (if necessary). '
The above-mentioned limiter 82 limits the amplitude of the pilot. The limiter 82 may be a device that reduces the amplitude of a signal exceeding a threshold or a nonlinear device that saturates vjbsn driven by a M^ level signal. Saturation, or gain reduction with increasing signal level, occurring in olher parts within the pilot generator 64, such as the second multiplier 88 or IF gain stages 80,86, may also provide a means of limiting.
The same LO 68 fiequency is preferably used for botii the pilot detection down-conversion at mixer 66 and tiae pilot generation i^MJonversion at mixer 88. The frequency of LO 68 is chosen to place the pilot signal outside of the bandwidth of tiie RF carrier of tiie input signal to the feed forward amplifier and to fecilitate detection of the signal m circuit 62. Also, a suitable choice of LO fiequency may allow a relatively inesqpensive IF filter 72 to be employed. For example, a IX) fiequency of about 85 MHz fiequeiicy sihifi from the earner band will allow an

me3q>ensive SAW filter to be used, e.g. wife, a 5 ME[z pass band. Various other choices of LO fi:equency and jBlter passband are also possible, howev^.
As fiirther shown in Figure 3 Ihe pilot signal output line 58 of pilot generation circuitry 64 is sampled by sampling coiq>ler 90 and the sampled output (pilot signal) is provided to fiequency measurement circuit 92. Frequency measurement circuit 92 detects the RF fiequency of tiie sampled pilot signal and provides a corresponding voltage signal along line 61 to controller 24
(Figure 2).
An alternate embodiment of the pilot detection and generation circuit 22 is shovm. in Figure 4. This embodiment is identical to Figure 3 ^th the exception that tiie output pilot signal is measured at IF instead of RF, More specifically, as shown the IF pilot signal ou^ut from IF gain circuit 86 is sampled by sanq>Iing coiqpl^ 94 and the sampled IF output (IF pilot signal) is provided to firequency measurement circuit 96. Frequency measurement circuit 96 detects the IF frequency of the sampled pilot signal and provides a corresponding voltage signal along line 61 to controller 24 (Figure 2).
In operation, the pilot detection and generation circuit 22 creates a narrow bandwidth, positive feedback loop through the main amplifier 16 and tiie second loop of tiie feed forward amplifiier (Figure 2). When combined with tiie limiting drcuit 82, a limit-cycle oscillation will develop usii^ noise presort in the feed forward amplifier and the pilot system, assuming that the loop has sufficient gain. The cancellation of tiie second loop affects the gain and phase of the positive feedback loop. As a result, good alignment of the second loop will 5iq)press the limit-cycle oscillation. The degree of alignment required to siqjpress the limit cycle is selectable based on the amount of IF gain provided by the IF gain stages preceding the limiter 82 or by adjustmg the clipping threshold of limiter 82.
A number of modifications of the illustrated implementation of the positive feed back pilot generation circuit 22 are possible. For example, Ibe circuit may employ an automatic level control circuit v^th related modifications in Ihe overall circuit design, as described in application serial no. 11/369,529 filed TsAsxcb. 7, 2006, the disclosure of vAich is incorporated

herein by refermce in its entirety. Also, an inq)lementation of Hie bandpass po^ver detector circuit 62 may employ an RF filter which is placed before the mixer 66 to reject image frequencies. In such an approach, a similar RF filter is preferably included wi&in the pilot generation path 64 afier the mixer 88. Also, it is possible to eliminate the bandpass filter 84 within the pilot generation path 64. However, such an implemetrtalion without filter 84 may not be preferred since it will waste pilot energy by producing signal components that are not detectable by the bandpass power detector circuit 62. These additional spectral components will be attenuated by the second loop cancellation as part of the feed forward compensatioiL Also, as noted above, the number of IF gain stages, the threshold of limiter 82, the LO firequency and the filter passband bandwidtti may aU be varied in accordance with the particular implementation and the particular RF carrier being amplified.
Next the use of the pilot fi«quency for improved loop 2 convergence will be described in more detail. By measuring the firequency of the generated pilot, phase information regarding the second loop cancellation transfer function can be inferred. Changes in the pilot fi-equency as the search algorithm makes steps in the second loop aligiuneat indicate errors in Ihe direction of the search. The cancellation transfo function of the second loop is determined by gain and phase alignment adjusters (54, 56, respectively, in Figure 2). Assume that the alignment adjuster is modeled as

where Oopt and Popt are the optimal gain and phase alignment settbigs, respectively, and Aoopt and Apopt are the misalignment in the gain and phase adjusters, respectively. Assuming &e jAoopti and |Apopt| are small, the output of Ihe pilot detector can be approximated as

\)sdiere |k| and Pmb are constants. It can be seen from (Eq. 2) that when Hie detected volt^e is plotted as a function of the gam and phase adjuster settuxgs, Ifae resulting contours are concentric ellipses surrounding the optimal alignment setting (see Figure S and Figure 6).
The phase shift of the second loop cancellation transfer function is

It can be seen from CBq. 3) that a step in the alignment setting that keeps the ratio APopt/Aocopt constant will not alter the cancellation phase. This corresponds to making an alignment step that is in the direction of the optimal alignmmt (see Figure 5). If a step in the alignment setting is not in the direction of tiie optimal alignment, the phase will change (see Figure 6). To illustrate the phase change, assume tiie initial alignment settmg is (ai,Pi) and the alignment after the step (Aai,Api) is (a2,P2) - (ai+Aai,pi+APi). The phases before and after the step are

The frequency of the generated pilot generated is a natural mode of the positive feedback. It must be wititin the passband of the pilot system and create a loop phase that is a multiple of 2?! radians. That is. Die pilot frequency must satisfy

where copuot is Ihe pilot firequeQcy, Aioop is the total loop delay, and ^o is a phase of&et Changes in the phase of the cancellation transfer function of the second loop affect (Eq. 7). As a result, to preserve ^Bq. 7), the pilot frequency must chai^ as well. That is, the change in pilot frequency due to a phase change induced by alignment stq) 1 is

Thxis, a change in the pilot frequency, measured at eidier IF or RF, will indicate changes in the phase of the second loop cancellation transfix frmctioa
The manner in which the frequency change mformation is used to select the next step direction is next described. The first step direction is

AA&ere ke is a constant Figure 7 and Figure 8 illustrate the effect tiiat &e selection of ke has on the search trajectory- In Figure 7, the value of ke is too small making "flie new search direction, 02, too small of a change fix)m the first search direction, Bj. As a result, the pilot frequency wiU continue to increase. Successive increases (or decreases) in the pilot taeqpsacy suggest an increase in ke is needed. In Figure 8, the value of ke is too large makii^ the new search dhection, 62, too large of a change frt)m the first search direction, Gi. As a result, the change in

pilot fiequency will alternate directions each step. Alternating chants in the pilot firequency suggest a decrease in ke is needed.
A preferred embodiment of the algorithm for selecting the step direction 6ZH-I in the aligmnent search is sho^Mi in Figure 9. As shown Ihe algorithm initiates at 100 and at 102 an initial aligmnent stq) direction is selected, which initial direction may be arbitrary. Next at 104 the algorithm proceeds to measure the pilot fiequency based on the pilot frequency signal provided to the controller along line 61 (Figure 2). At 106 a counter is initialized to begm a series of aligmnent steps using measurements of the pilot frequency in order to optimize the step direction. More specifically, at 108, die algorithm initiates an alignment step (Aai,Api) in the initml aligmnent direction by incrementing the gain and phase adjuster settmgs corresponding to the selected direction. Next at 110 the algorithm proceeds to measure the pilot firequency at the new settings using the pilot frequency signal provided to the controller along line 61. Next at 112 tiie algorithm proceeds to compute the difference in the pilot frequency between the initial setting and the new setting. Next at 114 the difference in pilot firequency, determined at 112, is used to alter the alignment st^ direction, multiplying the drff^ence in frequency by a constant value ke defining the amount of change in step direction (i.e. the size of the angle of direction change in 2D gain phase space). Next at 116 it is determined if the value of the constant ka is too large or too small and if necessary the value of the constant ke is increased or decreased (as described above in relation to Figures 7 and 8). Next at 118 the counter is incremented and Ihe alignment adjustment step direction processing flow, 108,110,112,114 and 116 is repeated. This iterative process flow continues as long as it is converg^g, vMch is indicated by a decreasing level of Ihe detected pUot power 60. The detected pilot power 60, denoted by Vdet» is measured at 104 and 110, and the difference, AVdet, is conqnited at 112. The search is convergmg as desired when AVdet For the case ^ere the alignment adjustment step (Aa,AP) causes the iterative process to divCTge, as indicated by AVdet > 0, the alignment adjustment direction is reversed by adding TI radians to (Eq. 10) and 114, and the step size is reduced before repeating 108. The algorithm for selecting the step size used at 108 may be the same as tiie power minimization approaches

described in U.S. patent aH)lication 10/733,498 filed December 11, 2003, U.S. patent no. 7,002,407, and U.S. patent application 11/018,216 filed December 21,2004, the disclosures of wbich are incorporated herein by referrace in their entirety. The algorithm of Figure 9 and additional aspects of alignment control processing described in tixe above noted applications and patents may be unplemented in controller 24 n^ng a suitably progranmied microprocessor (additional details are described in the above noted patent applications incorporated herein by reference).
It is worth noting that the frequency change, ScOpAob can be large due to the 2TC radian multiple in (Eq. 7). When the frequency shifts near the edge of the passband of the pilot S3^stem, and a discrete frequency change of 27m / Aioop may occur to ferce the pilot frequency to remain within the passband. When large changes in frequency are detected, the measured value of ScDpiiot should not be used in 114. Within 114, Sa> should be constrained, using instead a modified 6piiot I- Alternatively, the search algorithm in Figure 9 can be restarted afrer detecting large fr:equency changes.
Large frequency changes also occur when tiie convergence of the iterative process is nearly complete. The pilot amplitude drops rapidly when the alignment is near an optimal setting because the loop gain of the positive feedback is no longer sufSdent to maintain the pilot oscillation making the detected pilot firequency measur^nent 61 unreliable. Such converged conditions are desirable and are indicated by Vdd reaching its minimimi value QAVdetl > 0 for aU possible alignm^it step directions). When this condition is detected, any search direction can be selected as long as it is varied over time.
An alternative embodiment of the algorithm for selecting the step direction in the alignment search is desadbed below. Rather than selecting ke, it is possible to base tiie search direction on the sign of the difference in the pilot frequency (ScOpaoO- The search direction is updated using

when the detected pilot power 60 is increasing (diverging, AVdet > 0). The search algorithm forces lateral movement in the trajectory relative to the direct pa& to the optimal settmg. Lateral movement changes Hxe angle 8^ (see Figure 6)» causing a frequency change, 5a>pUot. Note that 5^ and ScDpUot are proportional to the ratio of the step size and the distance to the
optimal settmg.
The algorithm also adjusts the step size to make the e7q)ected value of |5| constant. As an illustrative example, the step size can he increased by a fector of 1,4 when 15^| 0.3. The step size can also be decreased by 0.7 when AVdet> 0, Since both S^) and AVdet are used to adjust the step size, the search is better damped than if it was based on the detected pilot power 60 only. The best thresholds and step adjustment &ctors are dependent on the feed forward amplifier system and can be obtained easily using experiments.
The present invention has been described in relation to a presently preferred embodiment, however, it will be £^(>preciated by those skilled in the art "Biat a variety of modifications, too numerous to describe, may be made vMlo remaining wifiiin the scope of the present invention. Accordii^y, the above detailed description should be viewed as illustrative only and not limiting in nature.

WHATISCSLAIMEDIS:
1. A method for controlling alignment of a control loop in an amplifier system,
comprising:
generating a variable frequency pilot signal and injecting the pilot signal into &e amplifier system;
detecting any uncanceled pilot signal at an output of the control loop;
detecting the frequency of the generated pilot signal;
adjusting one or more parameters of the control loop;
detecting a frequency change in the variable frequency pilot signal; and controlling said adjusting based on the detected frequency change.
2. A method for controlling alignment of a control loop in an amplifier system as set out in claim 1, wherein generating said variable frequency pilot signal comprises using feedback from the ou^ut of the amplifier system to generate the pilot signal.
3. A method for controlling alignment of a control loop in an amplifier syst^n as set out in claim 1, v^erein said adjusting one or more parameters comprises adjusting &e gain and phase of a signal path in the control loop using gain and phase adjusters, respectively.
4. A method for controlling alignment of a control loop in an amplifier system as set out in claim 3, v^erein the direction of the gain and phase adjustment is changed based on said detected fi:equency change in the pilot signal.
5. A metixod for controlling alignment of a control loop in an amplifier system as set out in claim 1, v^erein the gen^ated pilot signal is an RF signal and -herein detecting the frequency of the generated pilot signal comprise detecting the RP firequency.
6. A method for controlling alignment of a control loop in an amplifier system as set out
in claim 1, wherein tiie generated pilot signal is an RF signal generated by vp converting an IF

signal and whei^i detecting Hie frequency of ihe generated pilot signal con:q)rises detecting the IF firequency of said IF signal.
7. A method for controlling alignment of a feed forward amplifier system comprising an input for receiving an input signal, a first carrier cancellation control loop coupled to the mput and having a main axx^lifier, a second error cancellation control loop coupled to the first control loop and hamg an error amplifier and a gain adjuster and a phase adjuster, and an output coiq)led to the second control loop and providing an output signal, the method comprising:
sampling Ibe output signal;
generating a variable firequency pilot signal fiom the sampled output signal and injecting it intxj the first control loop;
detecting the firequency of the generated pilot signal;
adjusting the settmgs of the gain and phase adjusters in said second control loop fiom a first adjustment setting to a second adjustment setting using an alignment direction;
detecting the firequency of the generated pilot signal after said adjusting;
detecting the difiference in the firequency of Ihe generated pilot signal between said first and second adjustment settings;
altering the alignment direction using the firequency difference between said first and second adjustment settings; and
adjusting the settings of the gain and phase adjusters in said second control loop firom the second setting to a third setting using the altered alignment direction.
8. A method for controlling alignment of a feed fi)rward an^lifier system as set out in claim 7, wherein altering the alignment direction using the firequency difference between said first and second adjustment settings comprises multiplying the fiequency difference by a direction change parameter.
9. A method finr controUii^ alignment of a feed forward amplifier system as set out in claim 8, finrther conqnising determining if the direction change parameter is too great or too small, and decreasing or increasing tiie direction change parameter if necessary.

10. A method for controlling alignment of a feed forward amplifier system as set out in claim 7, wherein the generated pilot signal is an RF signal and "^lerdn detecting the frequency of the generated pilot signal comprises detecting the RF frequency.
IL A mettiod for controlling alignment of a feed forward anq)lifier system as set out in cl^m 7, wherein the generated pilot signal is an RF signal g^ierated by up converting an IF signal and wherein detecting the frequency of the generated pilot signal con:q)iises detecting the IF fi:equency of said IF signal.
12. A feed forward amplifier, comprising:
an RF iiqjut for receivii^ an RF signal;
a carrier cancellation loop comprising a main amplifier receiving and amplifying said RF signal, a main amplifier output sampling coiq)ler, a first delay coupled to the RF irqait and providing a delayed RF signal, and a carrier cancellation combiner coipling the delayed RF signal to the sampled output from the main amplifier;
an error cancellation loop comprising an error amplifier receiving and amplifying tbe output of the carrier cancellation combiner, a gain adjuster and a phase adjuster coiq)led between the carrier cancellation combiner and error amplifier and respectively receiving gain and phase adjustment control signals, a second delay coupled to the ou^ut of the main amplifier, and an error injection coi^ler combining tiie output from Ihe error amplifier and the delayed main amplifier output from the second delay so as to cancel distortion introduced by the main amplifier;
an RF ou^ut coi^led to the error injection coupler outpfot and providing an amplified RF signal;
an output sampling coupler for providing a sampled output of the amplified RF signal;
a positive feedback pilot generator circuit for generating a jnlot signal fit>m tiie sampled output of the amplified RF signal and providing the pilot signal to the ii^ut of the main amplifier, ibt positive feedback pilot generator circuit including a firequency detector for detectu^ the frequency of the generated pilot signal and providing a pilot frequency signal; and

a controller programmed vdtii a loop control algorithm, the controller coiq)led to receive Hbs pilot frequency signal and oxd^ntting said gain axui phase adjustment control signals to said gain adjuster and phase adjuster, the controller adjusting the direction of change of tiie gain and phase adjustment control signals provided to said gain adjuster and phase adjuster based on changes in the pilot frequency signal.
13. A feed forward amplifier as set out in claim 12, wh^ein said positive feedback pilot generator circuit further comprises means for providing a detected pilot power signal from the sampled oidput of the amplified RF signal vMch. varies with the strength of the uncancelled distortion firom the error cancellation loop and wherein said controller is coupled to receive the detected pilot power signal.
14. A feed forward amplifier as set out in claim 12, wherein said positive feedback pilot generator circuit comprises means for g^iera&ig an uitermediate frequency pilot signal from the sampled output of Ihe amplified RF signal, a local oscillator providing a fixed frequency signal, and a mixer receiving Has intermediate firequency pilot signal and fixed firequency signal and outputting the pilot signal at anRF firequency.

15. A feed forward amplifier as set out in claim 14, wherein said positive feedback pilot generator circuit further comprises a sampling coiqpler coi^>led to the ou:^ut of said mixer and providing the sampled RF firequency pilot signal to said firequency detector, and wherein said firequency detector detects the RF frequency of said pilot signal and provides said pilot firequency signal corresponding thereto to said controller.
16. A feed forward amplifier as set out in claim 14, wherein said positive feedback pilot generator circuit further comprises a sampling coupler, coupled to the output of said means for generating an intermediate fi^umcy pilot signal, said sampling coupler providing the sampled iixtermediate frequency pilot signal to said frequency detector, and wherein said firequ^icy detector detects the frequency of said intermediate firequency pilot signal and provides ^d pilot frequency signal corresponding thereto to said controller.

17. A feed forward amplifier as set out in claim 14, wherein said means for generating an
intermediate frequency pilot signal £rom the sampled output of the amplified RF signal
comprises a second mixer coiqiled to the local oscillator and receiving the sampled output of
tiie amplified RF signal and providing an intermediate firequency sampled output signal and a
band limiter for providing a band limited signal corresponding to uncancelled pilot signal in
the sampled output.
18. A feed forward amplifier as set out in claim 12, vrfieiein said pilot fiiequency signal is a
voltage corresponding to the detected frequency.
19. A feed forward amplifier as set out in claim 12, wherein said control algorithm
iteratively adjusts Hxt aligmnent direction to minimize the detected frequency change.
20. A feed forward anq>lifier as set out in claim 19, v^rein said control algoriflmi adjusts
the amount of alignment direction change based on successive increases or decreases in the
detected frequency chai^.

Documents:

http://ipindiaonline.gov.in/patentsearch/GrantedSearch/viewdoc.aspx?id=jlNDiwSegrwMbvvhrdvXcQ==&loc=egcICQiyoj82NGgGrC5ChA==

« Previous Patent

Next Patent »

Patent Number

278156

Indian Patent Application Number

4971/CHENP/2007

PG Journal Number

52/2016

Publication Date

16-Dec-2016

Grant Date

15-Dec-2016

Date of Filing

05-Nov-2007

Name of Patentee

POWERWAVE TECHNOLOGIES, INC

Applicant Address

1801 E ST ANDREW PLACE SANTA ANA CALIFORNIA 92705

Inventors:

#	Inventor's Name	Inventor's Address
1	BRAITHWAITE, RICHARD, NEIL	1677 N YUROK STREET ORANGE CA 92867

PCT International Classification Number

H04L25/49

PCT International Application Number

PCT/US06/12202

PCT International Filing date

2006-04-03

PCT Conventions:

#	PCT Application Number	Date of Convention	Priority Country
1	60/668,363	2005-04-05	U.S.A.
2	11/392,170	2006-03-29	U.S.A.
3	60/670,908	2005-04-13	U.S.A.