Amplifiers – Hum or noise or distortion bucking introduced into signal...
Reexamination Certificate
1999-02-12
2001-05-15
Lee, Benny (Department: 2817)
Amplifiers
Hum or noise or distortion bucking introduced into signal...
C375S296000
Reexamination Certificate
active
06232835
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to a system and method for linearizing the gain error of a power amplifier. More particularly, the invention relates to directional derivative and gradient methods for estimating and correcting complex gain error.
BACKGROUND OF THE INVENTION
Radio frequency power amplifiers have a nonlinear power transfer function. In other words, both the amplitude and phase components of the amplifier's gain depend on the power of the amplifier's input signal. This amplifier non-linearity is undesirable because it distorts the output waveform, broadens the output waveform's spectrum, and generates interference within, adjacent channels.
Generally, power amplifier linearization is sought using one of three techniques: vector feedback, adaptive predistortion, and adaptive feedforward compensation. In each of the three techniques, the amplifier's complex gain error must be estimated, preferably in real-time. It is toward this estimation step that the present invention is directed.
Conventionally, the gain error is estimated by comparing the amplifier's output signal with a reference signal, usually the amplifier's input signal. This comparison measures quantities such as the correlation or the error power of the two signals.
Unfortunately, conventional correlation circuits introduce DC offsets and common-mode feedthrough errors. Also problematic, conventional error power measurement circuits use a search process that disturbs the gain along the amplification path and yields a signal convergence that is too slow for vector feedback systems.
It would be desirable to have a gain error estimator that does not suffer from these problems.
SUMMARY OF THE INVENTION
The illustrated embodiments avoid conventional common-mode feedthrough problems by measuring the error signal across a single square-law detector in a single processing path.
Instead of conventionally varying the gain along the amplifier path to search for the minimum error, the embodiments vary the amplitude and/or phase along a reference path. In other words, differential measurements needed to estimate the gain error gradient are obtained by amplitude and/or phase modulating the reference signal. The modulated error power measurement is later demodulated, so that the amplifier gain is unaffected by the search process.
Although modulating the reference signal is only a one-dimensional search of the amplitude phase (&dgr;a-&dgr;ø) space, by implementing two concurrent (and preferably orthogonal) searches, one can resolve the gain error gradient from the directional derivatives obtained.
Thus, according to one embodiment of the invention, there is provided a method of linearizing the vector gain of a power amplifier including generating a directional derivative signal representing a phase-amplitude space directional derivative of an error signal of the power amplifier and modulating the gain of the power amplifier in response to the directional derivative signal. Generating a directional derivative signal might include locating in the phase-amplitude space of the error signal a vector that is substantially equal to the directional derivative. By extension, locating a vector that is substantially equal to the directional derivative might include searching for the best approximation of the directional derivative along a bounded one-dimensional path in the phase-amplitude space of the error signal.
Alternatively, locating a vector that is substantially equal to the directional derivative might include: searching for the best approximation of the directional derivative along a bounded one-dimensional path in the phase-amplitude space of the error signal, the path being parallel to the phase axis of the phase-amplitude space; searching for the best approximation of the directional derivative along two bounded one-dimensional paths in the phase-amplitude space of the error signal; searching for the best approximation of the directional derivative along two substantially orthogonal bounded one-dimensional paths in the phase-amplitude space of the error signal; searching for the best approximation of the directional derivative alternately along each of two substantially orthogonal bounded one-dimensional paths in the phase-amplitude space of the error signal; searching for the best approximation of the directional derivative simultaneously along each of two substantially orthogonal bounded one-dimensional paths in the phase-amplitude space of the error signal and then combining into a weighted average the best approximation of the directional derivative along each of the two substantially orthogonal bounded one-dimensional paths; or searching at a first frequency for a first component of the best approximation of the directional derivative along a first bounded one-dimensional path in the phase amplitude space of the error signal and then searching at a second frequency for a second component of the best approximation of the directional derivative along a second bounded one-dimensional path in the phase amplitude space of the error signal.
Desirably, the method includes: receiving a feedback signal corresponding to an output signal output from the power amplifier, receiving a reference signal corresponding to an input signal input to the power amplifier, modulating the phase of the reference signal through a bounded range of phase shift angles, subtracting the feedback signal from the modulated reference signal to produce a difference signal, and rectifying the difference signal to produce a rectified difference signal that corresponds to the magnitude of a directional derivative oriented parallel to the phase-amplitude space phase axis.
Alternatively, the method might include generating a gradient signal representing a phase-amplitude space gradient of the error signal of the power amplifier. In this case, the method desirably further includes: receiving a feedback signal corresponding to an output signal output from the power amplifier, receiving a reference signal corresponding to an input signal input to the power amplifier, modulating the phase of the reference signal through a bounded range of phase shift angles, modulating the amplitude of the reference signal through a bounded range of amplitude shift levels, subtracting the feedback signal from the modulated reference signal to produce a difference signal, and rectifying the difference signal to produce a rectified difference signal that corresponds to the magnitude and phase of the gradient signal.
Still alternatively, generating the gradient signal might include: receiving a feedback signal corresponding to an output signal output from the power amplifier, receiving a reference signal corresponding to an input signal input to the power amplifier, subtracting the magnitude of the feedback signal from the magnitude of the reference signal to produce an amplitude component of the gain error gradient, modulating the phase of the reference signal through a bounded range of phase shift angles, subtracting the feedback signal from the modulated reference signal to produce a difference signal, rectifying the difference signal to produce a rectified difference signal that corresponds to a phase component of the gradient signal.
It should be added that either generating a directional derivative signal or generating a gradient signal might further include; attenuating the output signal by the nominal gain of the amplifier, time delaying the input signal by the loop delay of the amplifier, demodulating the rectified difference signal to produce a demodulated difference signal, lowpass filtering the demodulated difference signal to produce a filtered difference signal, or normalizing the filtered difference signal with respect to the instantaneous power of the reference signal to produce a normalized signal.
According to another embodiment of the invention, there is provided means for linearizing the vector gain of a power amplifier including means for generating a directional derivative signal representing a phase-amp
Lee Benny
Nguyen Khanh Van
Nortel Networks Limited
LandOfFree
System and method of linearizing the gain error of a power... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with System and method of linearizing the gain error of a power..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and System and method of linearizing the gain error of a power... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2541129