 |
Description  |
|
|
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates generally to communication systems employing
amplification devices. More particularly, the invention pertains to a
non-linear predistortion or postdistortion generator for coupling in-line
with an optical receiver, optical laser transmitter or an amplifier to
minimize second and third order distortion caused by the signal
processing.
2. Description of the Related Art
Analog intensity modulation of a distribution feedback (DFB) laser is a
widely used technique to transmit analog signals, such as sound or video
signals and data, on optical fibers over a long distance. Optical detector
also is widely used in fiber optic link. The performance of DFB lasers and
optical detectors are limited by their distortion performance. Improving
second order and third order distortion performance can greatly improve
the entire system performance and increase the entire system dynamic
range.
Amplifiers are also widely used in many types of communication
applications. Although it is preferable to keep amplifiers within their
linear range of operation, it has been increasingly necessary to extend
the operation of amplifiers into high power and high frequency regions of
operation. Typically, the output power of an amplifier is limited by the
non-linearity of the active devices, including bipolar transistors and
FETs. These non-linearities result in distortions which are impressed upon
the signal being amplified. Reducing the non-linear distortions of an
amplifier results in increases of the output power, the system dynamic
range and the carrier-to-noise ratio. Accordingly, minimizing distortions
and achieving linear frequency response is paramount to efficient
amplifier operation.
Minimizing distortion is particularly important when a series of amplifiers
is cascaded over a signal transmission path, such as a series of RF
amplifiers in a CATV transmission system. Disposed throughout a CATV
transmission system are RF amplifiers that periodically amplify the
transmitted signals to counteract cable attenuation and attenuation caused
by passive CATV components, such as signal splitters and equalizers. The
RF amplifiers are also employed to maintain the desired carrier-to-noise
ratio. Due to the number of RF amplifiers employed in a given CATV
transmission system, each RF amplifier must provide minimum degradation to
the transmitted signal.
Many amplifiers are subject to a wide range of ambient operating
temperatures. These temperature changes may affect the operating
characteristics of certain electronic components within the amplifier,
thereby inducing additional distortions. A temperature range of
-40.degree. C. to +85.degree. C. is not uncommon for many amplifier
applications in a communication environment. To ensure consistent
performance over the operating bandwidth, and to minimize resulting
distortions, an amplifier must be designed for a broad range of ambient
operating temperatures.
The distortions created by an amplifier which are of primary concern are
second (even) and third (odd) order harmonic intermodulation and
distortions. Prior art amplifier designs have attempted to ameliorate the
effects of even order distortions, such as composite second order (CSO)
distortion, by employing push-pull amplifier topologies, since the maximum
second order cancellation occurs when equal amplitude and 180.degree.
phase relationship is maintained over the entire bandwidth. This is
achieved through equal gain in both push-pull halves by matching the
operating characteristics of the active devices. In some cases, second
order correction is still needed in order to get good CSO performance.
Many prior art designs include the use of a separate second order
distortion circuit to provide such the correction for CSO.
However, odd-order distortion is difficult to remedy. Odd-order distortion
characteristics of an amplifier are manifest as cross modulation (X-mod)
and composite triple beat (CTB) distortions on the signal being amplified.
X-mod occurs when the modulated contents of one channel being transmitted
interferes with and becomes part of an adjacent or non-adjacent channel.
CTB results from the combination of three frequencies of carriers
occurring in the proximity of each carrier since the carriers are
typically equally spaced across the frequency bandwidth. Of the two noted
distortions, CTB becomes more problematic when increasing the number of
channels on a given CATV system. While X-mod distortion also increases in
proportion to the number of channels, the possibility of CTB is more
dramatic due to the increased number of available combinations from among
the total number of transmitted channels. As the number of channels
transmitted by a communication system increases, or the channels reside
close together, the odd-order distortion becomes a limiting factor of
amplifier performance.
There are three basic ways of correcting distortion created by a non-linear
device (NLD): 1) reduce the signal power level; 2) use a feed forward
technique; and 3) use a predistortion or postdistortion technique. The
first method reduces the signal power level such that the NLD is operating
in its linear region. However, in the case of an RF amplifier this results
in very high power consumption for low RFoutput power.
The second method is the feed forward technique. Using this technique, the
input signal of the main amplification circuit is sampled and compared to
the output signal to determine the difference between the signals. From
this difference, the distortion component is extracted. This distortion
component is then amplified by an auxiliary amplification circuit and
combined with the output of the main amplification circuit such that the
two distortion components cancel each other. Although this improves the
distortion characteristics of the amplifier, the power consumed by the
auxiliary amplification circuit is comparable to that consumed by the main
amplification circuit. This circuitry is also complex and very temperature
sensitive.
The third method is the predistortion or postdistortion technique.
Depending upon whether the compensating distortion signal is generated
before the non-linear device or after, the respective term predistortion
or postdistortion is used. In this technique, a distortion signal equal in
amplitude but opposite in phase to the distortion component generated by
the amplifier circuit is estimated and generated. This is used to cancel
the distortion at the input (for predistortion) or output (for
postdistortion) of the amplifier, thereby improving the operating
characteristics of the amplifier.
One such distortion design, as disclosed in U.S. Pat. No. 5,703,530 and
shown in FIG. 1, relies upon a traditional .pi.-attenuation network and a
delay line for gain compensation; and a diode pair coupled with a delay
line for distortion and phase compensation. This circuit generates a
distortion that is equal in amplitude but opposite in phase to the
distortion introduced by the amplifier. Plots of the distortions
contributed by the distortion generator and the distortions manifest by
the amplifier are shown in FIGS. 2 and 3. As shown, the distortion signal
compensates for the distortions generated by the amplifier. However, the
use of delay lines in such a manner is impractical since delay lines are
physically large, are difficult to adjust and the results are inconsistent
across a wide frequency range. Additionally, both amplitude and phase
information are required for correct compensation. The '530 patent also
states that the system disclosed therein is not ideal for certain
application, such as predistortion for CATV RF amplifiers, due to the
excessive losses introduced by the distortion circuit.
An inline predistortion design, as disclosed in U.S. Pat. No. 5,798,854,
provides compensation for NLDs by applying a predistorted signal equal in
magnitude but opposite in phase to the distortion produced by the NLD.
However, the circuitry disclosed therein is not matched to the NLD.
Additionally, the '854 patent presents a design that is typical of the
prior art in the use of a high resistance bias for the diodes. This will
reduce the correction efficiency and increase the effects of temperature
upon the circuit.
Prior art designs also use separate correction circuits to correct for
second and third order distortions if both types of corrections are
required. This increases the cost of the overall circuit design and also
generates more circuit losses.
Accordingly, there exists a need for a simple distortion generator which
counteracts the distortion created by an NLD. The circuit should not
introduce additional signal delay and should operate over a wide frequency
bandwidth and wide ambient temperature range.
SUMMARY OF THE INVENTION
The present invention is an in-line predistortion or postdistortion
generator for coupling in-line with an NLD to produce an output signal of
useful amplitude, but with low composite second order, composite triple
beat and cross modulation distortions. The distortion generator comprises
an instant controlled non-linear attenuator which utilizes the non-linear
current flowing through a pair of diodes to provide the proper amount of
signal attenuation over the entire frequency bandwidth. The distortion
generator circuitry is always matched to the NLD, thereby ensuring a
frequency response that is predictable and predefined. The distortion
generator permits selective adjustment of the non-linear current flowing
through the diodes to create a second order distortion. The distortion
generator also includes a temperature compensation circuit to ensure
consistent operation throughout a wide temperature range.
Accordingly, it is an object of the present invention to provide a
temperature compensated distortion generator which minimizes composite
second order, cross modulation and composite triple beat distortions
manifested by an NLD such as an RF amplifier, a laser diode or a
photodetector.
Other objects and advantages of the of the present invention will become
apparent to those skilled in the art after reading a detailed description
of the preferred embodiment.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 is a schematic diagram of a prior art distortion generator.
FIG. 2 is a combination plot of the effect of using the outputs from the
prior art distortion generator shown in FIG. 1 with an RF amplifier.
FIG. 3 is a combination plot of the effect of using the outputs from the
prior art distortion generator shown in FIG. 1 with an RF amplifier.
FIG. 4 is schematic diagram of a .pi. attenuator.
FIG. 5 is a signal diagram of the diode non-linear current caused by the
input voltage.
FIG. 6 is a schematic diagram of the preferred embodiment of the second and
third order distortion generator of the present invention.
FIG. 7 is a schematic diagram of the temperature compensation circuit.
DESCRIPTION OF THE PREFERRED EMBODIMENTS
The preferred embodiment of the present invention will be described with
reference to the drawing figures where like numerals represent like
elements throughout. Although the preferred embodiment of the present
invention will be described, for simplicity of explanation, as being
coupled with an RF amplifier, those skilled in the art would clearly
recognize that such a distortion generator could also be utilized to
compensate for distortion in laser transmitters, optical detectors, and
other electronic components which operate over a wide range of
frequencies. The description herein is not intended to be limiting, rather
it is intended to be illustrative.
The present invention will be described with reference to FIG. 4, whereby a
.pi. attenuator network 20 is shown. The network 20 comprises a selected
configuration of resistors Z.sub.1, R.sub.1, R.sub.2, R.sub.3, Z.sub.0,
R.sub.p. The signal source is input at signal input 30 and the output of
the attenuator network 20 is seen across the output 95. Z.sub.1 is the
source of internal impedance which should be equal to the system impedance
Z.sub.0, which is seen across the output 95. In an embodiment of the
invention for use with a CATV system, the impedance values Z.sub.1 and
Z.sub.0 are equal to 75 Ohms. Three of the resistors R.sub.1, R.sub.2,
R.sub.3 form a .pi. attenuator configuration. Preferably, the values (Y)
of resistors R.sub.2 and R.sub.3 are equal, and substantially larger than
the value (X) of resistor R.sub.1. Resistor R.sub.p is connected in
parallel with resistor R.sub.1.
As one skilled in the art would clearly recognize, when the following
condition is satisfied:
X=2Z.sub.0.sup.2 Y/(Y.sup.2 -Z.sub.0.sup.2) Equation (1)
the attenuator network 20 is matched at input and output, from DC to very
high frequencies. For one example of the attenuator when X=7.5 and Y=1.5K,
the power attenuation A for this attenuator network 20 is:
##EQU1##
Under the condition when Z.sub.0 <<Y, (as is the case when X=7.5 and
Y=1.5K):
A.congruent.(2Z.sub.0 /(2Z.sub.0 +X)).sup.2 Equation (3)
A(dB)=10 lg A Equation (4)
When X=7.5 and Y=1.5k, A (dB).congruent.0.42 dB. This means the attenuator
network 20 has very low insertion losses and a good frequency response.
When X has a small variation due to the parallel of R.sub.p, shown in FIG.
4, from Equation (3)
##EQU2##
From Equation (6):
##EQU3##
For example, If R.sub.p =375 ohms then:
##EQU4##
Equation (8) shows that when R.sub.p (375 ohms) is in parallel with R.sub.1
(7.5 ohms), the attenuation will be reduced by 0.00868 dB. This amount of
attenuation change is needed for non-linear compensation for an amplifier.
This example also shows that when the value of R.sub.p >>R.sub.1,
(i.e., when R.sub.p is 50 times larger than R.sub.1), adding R.sub.p
parallel with R.sub.1 has almost no effect on the impedance match, and the
voltage drop over the R.sub.p is mainly determined by the value of
R.sub.1.
However, if a linear resistor R.sub.p is used in the attenuator network 20,
there will be no distortion signal produced. The attenuator network 20 as
shown is a linear device. In order for a distortion circuit to operate
effectively, diodes are used to create a non-linear resistance.
Preferably, Schottky diodes are utilized. At small current, diode current
is exponentially proportional to the voltage across over the diode. Thus
diodes can be used as a non-linear resistance. For non-linear
applications, the amount of attenuation can be calculated as:
##EQU5##
Where I.sub.p is the current flow through R.sub.p, (the non-linear
resistance). I.sub.1 is the current flow through R.sub.1. Equation 9
provides the relationship of the attenuation change due to the current
change in I.sub.p. This equation is accurate over a broad frequency range.
The relationship between the delta attenuation and a change in current is
still valid when the resistance is a non-linear resistor. Accordingly,
Equation 9 provides a good estimation of how much non-linear current is
required for predistortion or postdistortion purposes.
Referring to FIG. 5, when the input sinusoidal voltage wave changes from
V.sub.1 to V.sub.2 to V.sub.3, the output current changes from I.sub.1 to
I.sub.2 to I.sub.3 respectively. The non-linear current used for third
order correction is:
I.sub.non-linear.congruent.I.sub.1 -2I.sub.2 +I.sub.3 Equation (10)
From Equation 9, the non-linear current needed is:
##EQU6##
Only non-linear current will be useful for predistortion or postdistortion
purposes. Equation 11 can be rewritten in the form of:
##EQU7##
Accordingly, I.sub.non-linear eff in Equation 12 is the effective
non-linear current going to the output port 114 which is shown in FIG. 6.
I.sub.output in Equation 12 is the total current that goes to the output
port 114. Equation 12 also shows that it is the non-linear current flowing
through the diodes which causes the distortion correction. Any method
which increases the non-linear current may increase the correction
efficiency. Equation 13 shows that only a small part of the non-linear
diode current is effectively being used for correction.
The .pi. attenuator network 20 has low insertion loss and the voltage drop
of the input voltage on R.sub.1 (shown in FIG. 4) is proportional to the
input voltage. This voltage may be used to drive a pair of diodes to
produce non-linear current and provide third order correction. The
non-linear current flowing in the diodes will cause an attenuator to
provide less attenuation at larger RF amplitudes, (i.e. when the input
signal has a higher power). This may be used to compensate for the signal
compression caused by amplification. Because of the relatively high value
of the diode's non-linear resistance, the match of the attenuator network
is almost unchanged. This match will not be changed even over temperature.
Additionally, frequency response over multi-octave frequency bands is
favorable.
The mechanisms of the second order correction circuit is also clear. If the
DC bias on each of the two diodes is different, for every RF positive
circle and negative circle, I.sub.non-linear eff will be different.
Accordingly, instead of third order correction, this circuit will also
provide second order correction.
Referring to FIG. 6, the preferred embodiment of the attenuator 100 for
both second and third order predistortion and postdistortion is shown. The
attenuator 100 of the present invention includes several additional
components that modify a traditional .pi. attenuator to achieve
significantly better performance over a wide frequency and temperature
range. The attenuator 100 has an input port 101, an output port 114 and
two bias control points 116, 123. The attenuator 100 may be used in a
predistortion configuration with an amplifier or in a postdistortion
configuration. For a predistortion configuration, the output port 114 is
connected to the input of an amplifier. For the postdistortion
configuration as shown in FIG. 6, an output signal generated by an
amplifier, is applied to the input port 101. The attenuator 100 includes
resistors 105, 106, 107, 108, 112; capacitors 102, 103, 104, 111, 113,
115; diodes 109, 110, and an inductor 117.
In most prior art applications, an inductor is used as a phase control
element to change the correction signal phase. However, in the present
invention, the inductor 117 is used in series with the resistor 108 to
make a parallel resonance circuit with the forward biased diode capacitor.
The inductive reactance cancels the specific capacitive reactance of the
diodes. At the resonance frequency, the capacitance of the diodes 109, 110
will be compensated by the inductor 117 so that the impedance between
points 118 and 119 will be purely resistive and can be calculated as
follows:
R.sub.impedance between 118, 119 =L/(C*R); Equation (14)
where L is the inductance of 117 in Henrys; C is the total forward biased
capacitor in Farads; and R is the resistance 108 in Ohms. By carefully
controlling L and C, one may get the following:
R.sub.impedance between 118, 119 =R Equation (15)
This means the capacitive effect has been totally canceled and an ideal
pure resistive load over a very wide frequency range has been achieved.
In prior art systems, the capacitance associated with the diodes has not
been considered. In predistortion applications, Shottky diodes are forward
biased, which results in a greater capacitance. When an RF signal is input
across the diodes, the average capacitance increases. Even at a bias of 0
volts, the impedance introduced by the diodes' capacitance may not be
ignored since the capacitance in parallel with the PN junction of the
diodes will reduce the overall voltage drop on the diodes, thus reducing
the non-linear current produced by the diodes and the overall correction
effect. Compensating for the capacitance associated with the diodes 109,
110, the inductor 117 resonates with the capacitance of the diodes 109,
110 at higher RF frequencies, thus extending the overall frequency
response of the circuit.
The function of the resistors 105, 106, 107, 108, 112 and the-capacitors
102, 103, 104, 111, 113, 115 and inductance 117 is to form a modified n
attenuation network in comparison to the .pi. attenuation network 20 shown
in FIG. 4. The capacitors 102, 103, 104, 111, 113, and 115 are also used
for DC blocking | | |
