Amplifier linearization by learned linear behavior

Amplifiers – Hum or noise or distortion bucking introduced into signal...

Reexamination Certificate

Rate now

  [ 0.00 ] – not rated yet Voters 0   Comments 0

Details

C330S136000

Reexamination Certificate

active

06396344

ABSTRACT:

BACKGROUND OF THE INVENTION
The recent advances in communication have give rise to applications where in a multiplicity of individual channels, each of relatively narrow bandwidth, are combined for transmission over a broadband channel. Thus 100 10 kHz might be combined for transmission over a T1 channel. Or 100,000 such channels might be combined for transmission over a fiber optic channel. Each of the individual channels might be assigned to various subscribers on a demand basis. Subscribers will come and go according to their requirements so that some channels occupancy will vary with random like rates. In addition subscribers might use different modulation techniques within their bandwidth. When such a channel is amplified by an amplifier which is non-linear in some characteristics, intermodulation products will be generated. Some of these products will fall within the bandwidth of other subscribers where they may degrade the integrity of that channel. The problem is not serious in older, analog telephone systems because of a listeners ability to hear conversation through the noise. With the advent of digital systems, data errors may occur which will require reducing data rates to accommodate error detection/correcting codes or multiple transmissions. It is an object of this invention to provide a means of controlling errors in amplifier characteristics so as to minimize the nonlinearities which cause intermodulation and increase the usable bandwidth of the channel. It should be noted that the term amplifier is used repeatedly in the description of the invention which follows but the application of this invention extends beyond amplifiers to virtually any non linear process where predistortion is an appropriate linearizing modality.
DESCRIPTION PRIOR ART
The traditional prior art for the sensing and control of amplifier distortions is feedback. Such a feedback system is shown in FIG.
1
. Referring to
FIG. 1
an Amplifier
106
, with a gain of −G is to be controlled by feedback so that the gain of Amplifier System
100
gain is controlled to a prescribed value −G1. A portion of the output of Amplifier
106
is fed back through Attenuator
124
, and subtracted from the Input in Error Detector
123
. Attenuator
124
, is set to the inverse of the prescribed gain of the Amplifier, here 1/G1. The difference signal from error detector
123
, is Vout/G1−Vin, the amplification error, the difference between the input signal and the attenuated output signal. The detected error in the amplifier, is inverted in the Amplifier and subtracted from the output reducing the error. An error will exist so long as the Amplifier output does not meet the prescribed condition. Feedback is effective 1) so long as the amplifier gain is sufficient to reduce the error to nearly zero and 2) the feedback signal arrives at the amplifier input with an insignificant time delay.


Gain



of



an



amplifier



with



feedback
_



Output
=
-
G
*
(
Input
+
Output
G1
)
where
:
-
G



is



the



gain



of



amplifier



106
G1



is



the



prescribed



gain



of



the



amplifier
Gs



is



the



gain



of



the



amplifier



system
Gs
=
Output
Input
=
-
G
1
+
G
G1
.
(
1
)
The basic equation for the system gain Gs is given in Equation 1 below. Note that as the amplifier gain G increases the system gain Gs approaches the prescribed gain G1. Time delay in the amplifier is inevitable and, with increasing frequency, the period of the input signal approaches the amplifier delay and this system reaches its limitation.
The basic feed forward system is best understood with reference to
FIG. 2
In
FIG. 2
, the function of the Input
101
, Amplifier
106
, Attenuator
124
, and Error Detector
123
, remain unchanged. Delay
103
, has been added to balance, to the extent possible, the delay of the Amplifier
106
so that the attenuated output from the Amplifier and the delayed input signal arrive simultaneously at Error Detector
123
. The output of the error detector is not fed back to the input of the Amplifier
106
but rather is input to Error Amplifier
132
. The amplified error signal is added to output of Amplifier
106
, after Delay
131
, which balances the delay of Amplifier
132
, so as to arrive simultaneously at Error Corrector
105
and correct the error. Since error correction flows forward this system is referred to as feedforward. In
FIG. 1
the error correction flowed backward, hence the terminology feedback.
Since the signal error is generally a fraction of the amplifier output, the error amplifier can be smaller than the amplifier itself. Feedforward requires that the gain and delay characteristics of the amplifier and the error amplifier are precise and stable. Any error or drift in these characteristics will result in a failure to accurately correct the detects in the performance of the main amplifier. Feedforward, while achieving the required performance, is a complex and difficult technology to manufacture because the components must be manufactured to extremely tight tolerances to enable the complete feedforward system to meet its performance requirements. Any system component drift with temperature or will degrade the system performance and require realignment to achieve the required performance goals.
Conventional feedback and feedforward are thus ineffective in reducing distortion in broad band amplifiers. Feedback arrives too late and feedforward, being open loop requires precise balance of gains between Amplifiers
106
and
132
and that Amplifier
132
be distortion free.
The major drawback in the use of feedforward is that there is a very narrow range where the error amplifier is not overloaded by the desired signals. The typical feedforward system might include a 100 Watt main amplifier and a 20 Watt error amplifier coupled to the main amplifier through 10 dB directional couplers at the output and the input. In normal operation with a continuous two tone input the main amplifier and the error amplifier might have intermodulation products suppressed by 30 dBc, a typical value. The distortion output produced by the main amplifier is 100 mW. In order to cancel this product, the error amplifier, since it is coupled to the main amplifier through a 10 dB coupler must generate 1 W. The error amplifier has adequate power capability to cancel these distortion products. Suppose now that the input signals are changing in amplitude as in AM modulation, driving the main amplifier into mild compression so that its output is less than expected by 0.5 dB. The output should have been 113.6 W but is compressed to 100 W. In order to correct this amplitude distortion of 13.6 W., the error amplifier must product 136 W. into the 10 dB coupler. Of course, the error amplifier, capable of only 20 W. will be saturated and the feedforward loop's ability to cancel distortion will be severely compromised. If the feedforward system is operated with multiple FM (constant envelope) tones and there are greater than eight tones, the frequency at which these events occur is rare and the average power of the distortion products is acceptable.
The newer types of modulations in use today, CDMA, TDMA, TDMA burst mode, etc., contain considerable amounts of amplitude components so that amplitude limits of the main amplifier are frequently exceeded to limit the bandwidth of the modulation mask. Feedforward systems can have considerable problems with these modulation formats and may be forced to operate at reduced power to achieve satisfactory intermodulation performance.
There are several other techniques in the prior art to improve the performance of feedforward amplifiers operated with AM type signals. Pass

LandOfFree

Say what you really think

Search LandOfFree.com for the USA inventors and patents. Rate them and share your experience with other people.

Rating

Amplifier linearization by learned linear behavior does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Amplifier linearization by learned linear behavior, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Amplifier linearization by learned linear behavior will most certainly appreciate the feedback.

Rate now

     

Profile ID: LFUS-PAI-O-2880552

  Search
All data on this website is collected from public sources. Our data reflects the most accurate information available at the time of publication.