Note: Descriptions are shown in the official language in which they were submitted.
10~683~
This is a division of Canadian Patent Application
Serial No. 273,653 filed lO March, 1977.
Bac~round o f the 1: nve n t ion
The invention is a regenerator that is more
particularly described as a regenerator for a fiber optic
transmission system.
Exploratory research and development of optical
devices and components have progressed sufficiently during
recent years that a fiber optic digital transmission system
now can be built and operated. ~ultimode and single
mode glass fibers can transmit optical sighals over long
distances. In new glass fibers, light losses are less than
5 decibels per kilometer for wavelengths between 800 nano-
meters and llO0 nanometers. Some single mode borosilicate
fibers have a minimum loss of 2.2 decibels per kilometer
at 850 and 1020 nanometers and a dispersion of approximately
0.4 nanoseconds per kilometer. Fibers with the above
characteristics are useful for digital communications
systems. A moderate pulse rate optical system can be
used economically for telephone exchange trunking in
congested metropolitan areas having a large and growing
traffic cross section.
In the glass flbers,optical pulses are attenuated
as they proceed through any fiber. Along any fairly long
communications route, it is necessary to regenerate the
optical pulse stream for insuring accurate communication
from one end of the system to the other.
In the prior art, most of the basic optical
regenerator circuitry is similar to the circuitry employed
3~ in commercial pulse code modulation systems which transmit
~OB6835
electrical pulses over copper wire pairs or coaxial cables.
In the optical regenerators, optical elments such as
avalanche photodiodes and lasers have been inserted for
converting optical power to electrical signals and
vice versa.
Also in the prior art, phase-lock loops have
been used to extract timing information from an input pulse
stream; however, those phase-lock loops use narrow range
acquisition requiring expensive crystal control.
Additionally in the prior art, an injection
laser is operated in response to a feedback control circuit
which senses the output optical power from the laser and
attempts to maintain a constant average optical output
power. Occasionally, many time slots pass wherein no
pulses are applied to the input of such a laser. The
prebias current rises to maintain constant output power.
Subsequently irreversible damage can occur to the laser
when pulses are ap~lied again.
Therefore, it is an object to provide an improved
rec~e~ tor for an optical transmission system.
It is another object to provide a regenerator
for a moderately high pulse rate and low error rate optical
transmission system.
It is a further object to provide an optical
regenerator having a wide dynamic range, i.e., operable
for a wide range of input optical powers.
;~ It is a still further object to provide an
optical regenerator for producing substantially constant
amplitude output pulses of light throughout a long
lifetime.
~0~683S
Summary of the Invention
In accordance with an aspect of the invention there is
provided an optical pulse stream regenerator comprising an
optical receiver responsive to an incident optical data
pulse stream for producing an electrical data pulse stream
representing the incident optical data pulse stream, the
optical receiver including an amplifier stage providing
variable gain in response to a first range of an automatic
gain control signal and having an avalanche photodiode pro-
viding variable gain in response to a second range of theautomatic gain control signal thereby maintaining the
electrical data pulses at substantially the same magnitude
regardless of fluctuations in the magnitude of pulses in
the incident optical data pulse stream; means responsive to
the electrical data pulse stream from the optical receiver
for recovering a clock pulse stream synchronized with the
incident optical data pulse stream; means responsive to the
clock pulse stream and to the electrical data pulse stream
from the optical receiver for regenerating an electrical
data pulse stream representing the incident optical data
pulse stream; and an optical transmitter responsive to the
: regenerated electrical data pulse stream for regenerating
an optical output data pulse stream representing the incident
optical data pulse stream.
According to another aspect of the invention there is
provided an optical pulse stream receiver comprising means
for receiving an incident optical pulse stream, the receiving
means including an amplifier stage providing variable gain in
response to a first range of an automatic gain control signal
and having an avalanche photodiode providing variable gain
in response to a second range of the automatic gain control
- 2a -
~0~6835
signal thereby maintaining the electrical data pulses at
substantially the same magnitude regardless of fluctuations
in magnitude of pulses in the incident optical data pulse
stream, means responsive to the electrical data pulse
stream from the receiving means for recovering a clock pulse
stream synchronized with the incident optical data pulse
stream, and means responsive to the clock pulse stream and
to the electrical data pulse stream from the receiving means
for regenerating an electrical data pulse stream represent-
ing the incident optical data pulse stream.
- 2b -
~086835
These and other objects of the invention are
realized by an illustrative optical regenerator includ:ing
an optical receiver, a decision and timing circuit and
an optical transmitter. In response to an incident
optical data pulse stream, a variable gain receiver
section produces a constant amplitude electrical data
pulse stream representing the incident optical data pulse
stream. The receiver section is stabilized because an
amplifier varies gain in response to a first range of an
automatic gain control signal and an avalanche photodiode
varies gain in response to a second range of the automatic
gain control signal. In response to the receiver electrical
data pulse stream, a pnase-lock frequency-lock circuit in
t'ne timing circuit recovers a clock pulse stream that
is synchronized with the incident optical data pulse
stream, and tne decision circuit regenerates an electrical
data pulse stream represen-ting the incident optical data
pulse stream. In the transmitter section, an injection
laser is controlled by the regenerated electrical data
pulse stream and by a prebias current that is established
in response to the difference between a signal representing
the optical output power of the laser and a signal repre-
senting the regenerated electrical data pulse stream. A
resulting regenerated output optical data pulse stream
represents the incident optical data pulse stream.
A feature of the invention is an optical receiver
section wnerein an amplifier varies gain in response to a
first range of an automatic gain control signal and an ava-
lanche photodiode varies gain in response to a second range
of the automatic gain control signal.
1086835
Another feature is an optical regenerator
including a phase-lock frequency-lock circuit responsive
to the electrical data pulse stream from the receiver
section for recovering a clock pulse streal~ synchronized
with the incident optical data pulse stream.
Another feature is a clock recovery circuit
including a frequency difference detector responsive to
a baud component signal from the data pulse stream
for producing a series of pulses having a polarity to
reduce the difference between the frequency of a controlled
clock pulse oscillator and the baud of the data pulse
stream and havillg a rate proportional to the frequency
difference for forcing the frequency of the controlled
oscillator into the capture range of a phase-lock loop
responsive to the baud component signal for forcing
the frequency and the phase of the controlled clock
pulse oscillator into synchronization with the received
optical data pulse stream.
A further feature is an injection laser control
circuit that supplies to the laser a preb~as slgnal tnat
is established in response to the difference between a
signal representing the optical output power of the laser
and a signal representing the regenerated electrical data
pulse stream.
Brief Description of the Drawings
The present invention taken in conjunction with the
invention disclosed in copending Canadian Patent Application
Serial No. 273,653 which was filed 10 March 1977, will be
described in detail hereinbelow with the aid of the accompany-
ing drawings, wherein:
FIG. 1 is a block diagram of an illustrative optical
pulse stream regenerator;
', ;
- ' ~
1~36835
FIG. 2 is a scihematic diayram of an illustrative
receiver section for the optical regenerator shown in
FIG. l;
FIG. 3 is a block diayram of an illustrative
decision and timing circuit for the optical regenerator
shown in FIG. l; and
FIG. 4 is a schematic diagraM of an illustrative
transmitter section for the optical regenerator shown
in FIG. 1.
Detailed Descrip-tion
.
Optical Regenerator
Referring now to FIG. 1, there is shown an
optical regenerator for a fiber optic digital transmission
system. Although the whole system is not shown, it is noted
at the outset that optical signals in the system are
innerently unipolar in nature. Light sources transmit
full optical power for a "1" and are off for a "O". It
is noted that for a typical laser source, as much as
-` 5-10 percent of the power of a "1" is transmitted for a "O".
The regenerator includes a high gain optical
receiver section 200 having a variable gain avalanche
pnotodiode detector and preamplifier 202 for receiving
the input unipolar optical data pulse stream 201 and
converting it into a stream of electrical signals. The
`~ electrical signals are processed through a variable gain
~; amplifier 203, a fixed gain amplifier 205, and a filter/
equalizer 206 for producing a constant amplitude electrical
data pulse stream that appears on a lead 207.
A pulse stream similar to the electrical data
pulse stream on le-ad 207 is fed back through lead 208, a
dc level restorer, peak detector and control circuit 210.
-- 5 --
101~6835
Automatic c3ain control signals from circuit 210 are
produced on a lead 213.
In t;~e receiver section 200, different variable
gain devices respond to separate ranges of the automatic
gain control signal. Both the variable gain amplifier 203
and the avalanche photodiode detector 202 are variable
gain devices which are biased for minimum gain in response
to hig'n level input optical pulses.
As the magnitude of the input optical pulses
decreases causing the automatic gain control signal to
increase through a first range, the control signal is
applied by way of the lead 213 to the variable gain
amplifier 203 and to a variable voltage supply 215. In
this first range, only the variable gain amplifier 203
responds to the control signal by increasing gain as the
magnitude of t~e input optical pulses decreases.
After gain of tne variable amplifier 203 is
increased to its full value, the control signal continues
to increase and be applied to the amplifier 203 and the
20 voltage supply 215. Only the variable voltage supply 215
responds to the control signal in this range causing gain
of the avalanche photodiode to increase as the level of
the optical pulses decreases further.
The electrical data pulse stream on lead 207 is
applied to a decision and timing circuit 300 for regener-
ating on a lead 302 an electrical data pulse stream
representing tne incident optical data pulse stream. A
phase-lock freqeuncy-lock loop 303 produces, on a lead 304,
a clock pulse stream that is recovered from information
contained in the electrical data pulse stream from the
receiver on the lead 207. A decision and retiming
~; _
6835
circuit 306 produces a regenerated electrical data pulse
stream on the lead 302 in response to the clock pulse
stream on tlle lead 304 and the electrical data pulse
stream on the lead 207.
An optical transmitter section 400 regenerates
an optical data pulse stream 402 in response to tne
regenerated electrical data pulse stream on the lead 302.
The transmitter section 400 includes an injection laser 404
that is controlled by the regenerated electrical data
pulse stream on lead 302 and by a prebias signal produced
by a prebias, monitor and control circuit 406 on a lead 403.
The prebias signal is established in response to the
difference between a signal representing a sample 407 of
the optical output power of the laser and a signal repre-
senting the regenerated electrical data pulse stream on
the lead 302. The optical output data pulse stream 402
represents the incident optical data pulse stream.
Receiver Section
Referring now to FIG. 2, there is shown a
detailed schematic diagram of the high gain optical receiver
section 200 which detects input optical pulses of a data
pulse stream received through an optical fiber 201 by
converting those pulses into a stream of electrical data
pulses on the lead 207.
The incident optical data pulse stream, which
is applied by way of the optical fiber 201, impinges on a
varlable gain avalanche photodiode 220. A silicon
avalanche photodiode is a semiconductor device that is
normally operated with a reverse bias creating within the
device a high electric field region. Varying the
magnitude of the incident light varies current conducted
101~835
by the avalanche photodiode and therefore the slgnal
applied to a following preamplifier stage.
As a result of the incident optical data pulse
stream, pairs of holes and electrons are generated within
the avalanche photodiode. These holes and electrons drift
in opposite directions in the device under influence of
the electrical field. As the holes and electrons flow
in opposite dirctions in the photodiode 220, they form
pulses of current which are removed from the photodiode
by the preampllfier circuitry connected to the photodiode.
The avalanche photodiode is a square law detector
that produces a current that varies linearly in response
to variations of incident light power. The photodiode 220
is arranged to provide variable gain to input optical
signals in response to high voltage reverse bias variations
generated by the voltage supply 215 in response to the
control signal on lead 213, as described by P. K. Runge,
IEEE 1974 International Conference on Communications, 17B.
. .
Lower power input light pulses cause greater bias voltage
and a resulting maximum gain of approximately 40 dB.
Higher power input light pulses cause a lower bias voltage
and a resulting minimum gain of approximately 12 dB,
determined by the characteristic~ of the avalanche photo-
diode. The arrangement for varying the bias voltage by
the control signal on the lead 213 will be explained
subsequently.
The pulses from the photodiode are coupled ~ s
through a capacitor 221 to the base input of a common-
emitter connected transistor 222 in the preamplifier 202.
Output from the collector of the transistor 222 is coupled
through an emitter-follower connected transistor 223 and
.~, ~ - . . . .
1086835
is applied to the input of an output driver transistor 224
of the preamplifier 202. A feedback loop through a
resistor 226 couples the output of the transistor 223 to
the transistor 222 for stabiliziny operation of tne
preamplifier 202. Preamplifier output signals produced
at the collector of the transistor 224 are coupled through
a capacitor 227 to the input of the variable gain amplifier
203 at a first gate electrode 230 of a dual gate MOS
device 231.
The MOS device 231 is an n-channel enhancement-
mode device arranged to present a high input impedance
and to provide variable gain in response to the control
signal applied by way of a second gate electrode 232.
The source and drain electrodes are connected through
resistors to sources of negative and positive potential,
respectively. A diode 233 limits the control signal
range which affects the gain of the MOS device 231. Gain
can vary over a range from a minimum of -14 dB to a
maximum of approximately 6 dB. A significant reduction
in thermal noise is achievd by using tne IGFET device
rather than a bipolar transistor. The output from the
drain of the MOS device 231 is coupled by way of a
capacitor 234 to a cascade of two variable gain emitter-
coupled palrs having emitter-followers in their outputs.
In the first emitter-coupled pair, signals from
the MOS device 231 are applied to the base electrode of
a transistor 235. Output signals generated at a collector
electrode of another transistor 236 are coupled through
an emitter-follower connected transistor 237 and a
capacitor 238 to the second emitter-coupled pair. Gain
of the first emitter-coupled pair is variable and is
g _
101~1~5835
controlled by a signal applied to the base electrode of
an emitter current source transistor 239. Maximum gain
of approximately 13 ds occurs when sufficient emitter
current is conducted to achieve minimur,l re in the transistors
235 and 236. A minimum gain of approximately 3 dB is
achieved by reducing emitter current and thereby increasing
re of the transistors 235 and 236.
In the second emitter-coupled pair, the signals
coupled through the capacitor 238 are applied to the base
input electrode of a transistor 240. Outpu-t signals,
produced on the collector electrode of a transistor 241
are coupled through an emitter-follower transistor 242
and a capacitor 2g3 to the input of the fixed gain staye
205. Gain of the second emitter-coupled pair is controlled
by a signal applied to the base electrode of the transistor
244. The gain of the second emitter-coupled pair also
varies from a maximum of approximately 13 dB to a minimum
of approximately 3 dB in response to changes of ernitter
current supplied through the emitter current source
; 20 transistor 244.
A pair of diodes 246 and 247 shifts the control
voltage in order to limit the signal range which affects
the gain of the first and second emitter-coupled pairs.
The gain varies as the control signal varies from a
maximum negative value to approximately ground potential.
Full gain of the variable gain amplifier 203 is
a design parameter which is selected to establish a
secondary reference for determining the gain of the
avalanche photodiode 220. The full gain of the amplifier
203 is selected so that the avalancne photodiode operates
hear its optimum gain when the smallest useful optical
-- 10 --
.
10~835 .
signals are received Fu11 gain of the amplifier 203 is
adjustec' for operation by initially adjusting a resistor 249
in the drain circuit of the I~IOS device 231.
The fixed gain stage 205 includes a pair of
direct coupled common-er,litter connected transistors 250
and 251 having a signal feedback path from the collector
electrode of the transistor 251 throuyh a capacitor 252
and a resistor 253 to the emitter electrode of the
transistor 250. The signàl feedback path stabilizes gain
of the stage 205 over the useful band of the amplifier and
provides a low output impedance. A direct current feedback
path from the emitter electrode of the transistor 251
through a resistor 254 to the base electrode of the
transistor 250 stabilizes bias of the fixed gain stage 205.
The gain of amplifier stage 205 is approximately 26 dB.
Output signals from the fixed gain stage 205 are produced
at the collector of transistor 251 and are coupled through
a capacitor 256 and a resistor 257 to the low-pass filter
206. The resistor 257 and the low output impedance of
the stage 205 are selected to provide an optimum driving
source impedance for the low-pass filter 206. A loss of
about 6 dB occurs due to the resistor 257.
The preamplifier 202, the variable gain stages
203, and the fixed gain amplifier 205 include several
capacitors inserted for suppressing power supply noise.
Any selected power supply may be sufficiently free of
noise that those capacitors become superfluous to operation
of the receiver section 200.
The low-pass filter 206 is arranged to optimize
an eye pattern for an ideal rectangular input pulse stream.
The filter 206 is designed to shape pulses SUC;l that,
~086835
for any useful input optical pul.se stream and the actual
frequency characteristic of the amplifiers 202 and 203,
the pulses have a raised cosine shape or any other
desirable shape for good regenerator performance. The
filtered pulse stream is coupled through a capacitor 262,
an emitter-follower connected transistor 263 and a lead
208 to the dc level restorer and peak detection circuit
210. Another output from the filter 206 is produced on
the lead 207 for couplin~ the output pulse stream from
the receiver to the decision and timing circuit 300, shown
in FIG. 1. The termination for the filter 206 is located
in the decision and ti.ning circuit 300.
In the circuit 210, a direct current component
is restored to the signal by that portion of the circuit
including a capacitor 270 and a diode 271. The diode
clamps the signal so that negative polarity pulses are
forwarded. A pair of diodes 272 and 273 superimpose
a fixed bias on the clamped signals to overcome voltage
~drop across the diode 271 and restore the direct current
potential at a value slightly below ground potential.
Tne peak of the resulting waveform is thereafter
determined by a peak detection arrangement including a for-
ward biased diode 276, a pair of resistors 277 and 278, and :
a capacitor 279. Current flowing through the diode 276
leaves on the capacitor 279 a charge that is negative with
respect to ground. Output of the peak detection circuit
is compared with a reference voltage determined by tne
source of negative potential 281 and a voltage divider .:
including resistors 282, 283 and 284. The comparison is
30 performed by an operational amplifier 285 with feedback :
resistor 290 and resistor 284 setting the gain at 40 dB and
- 12 -
. , : ~ ' '' - , ;
~0~36835
capacitor 291 providing stability. The operational
amplifier produces on its output lead 213 the automatic
gain control sic3nal that varies as a function of the
difference between the detected signal peaks and the
reference voltage. The automatic gain control signal
falls to a negative potential of approximately 3 volts
when the intensity of incident liyht is at a maximur,l and
rises to a positive potential of approximately 4 volts
when the intensity of incident light is at a minimum.
A resistor 294 and a capacitor 295 determine
the cutoff frequency of the feedback loop. By choice this
cutoff is established at approximately 0.25 Hz.
Noise is a limiting factor in the operation of
-the subject regenerator. There are three significant
types of noise involved in the operation of the regenerator.
'l'he~ al noise arises in the IIOS amplifier stage. Shot
noise arises from tne Poisson distribution of the primary
electron current in the avalanche photodiode 220. Also,
in the photodiode, excess multiplication noise occurs as a
result of the gain process. The magnitude of the excess
multiplication noise is significantly larger than the
shot noise, is an increasing function of avalanche gain,
and does not have a Gaussian amplitude distribution.
Because the excess multiplication noise is gain dependent,
there is an optimum value of avalanche gain. Because the
noise is non-Gaussian, a new approach is used in the
design of the receiver section.
The automatic gain control signal produced by
the operational amplifier 285 on lead 213 is coupled to
the variable voltage supply 215 for controlling the
magnitude of the bias voltage applied to the avalanche
- 13 -
-
~o86835
photodiode 220. This varying bias voltage controls the
gain of the avalanci~e photodiode. The gain control
signal from the operational amplifier 285 also is applied
to the variable gain stage 203 for controlling the gain
thereof. This gain control signal causes changes of gain
to compensate for any fluctuations in the magnitude of
the output pulses in response to variations of the input
optical pulses, aging of devices, or variation of ambient
conditions.
As mentioned previously, the magnitude of the
automatic gain control signal can vary widely. Different
ranges of the gain control signal affect different ones
of the aforementioned variable gain stages.
When the input optical pulses are at a high level,
the bias voltage produced by the variable voltage supply 215
is held at a minimum value of approximately 150 volts so
that the avalanche photodiode 220 operates near its
minimum gain of approximately 12 dB. At the same time,
the field effect transistor stage is operated at its
minimum gain of approximately -14 dB. The emitter-coupled
stages, in the variable gain stage 203, are each operated ~ -
at a minimum gain of approximately 3 dB. -
In the variable gain stage 203, the emitter-
coupled pairs are arranged to be'~the first stages responding
to any decrease in the level of the input optical pulses.
Such a reduction in the level of the input optical pulses
causes the gain control signal on lead 213 to raise the
potential on the bases of the transistors 239 and 244 from the
minimum of -3 volts. In response to this change in the
automatic gain control signal on lead 213, the transis-
tors 239 and 244 supply more current to the emitter-coupled
10~335
pairs. ~s a result r o~ the transistors 235, 236, 240
and 241 is reduced and their gain is increased in a first
range of the automatic gain control signal until minimum
re occurs. The automatic gain control sicJnal has no
more effect on the emitter-coupled pairs when the control
signal rises sufficiently to reverse bias tie diodes 246
and 247.
As the automatie gain control signal swings
more and more positively, it also increases the gain of
the i50S device 231. The positive swing of tne gain control
signal that is applied to the device 231 is limited to
one diode drop above ground potential by the limiting diode
233. The arrangement of the second gate of the ~50S
device 231 and the limiting diode 233 cause the gain of
the ~50S device to increase as the automatie gain eontrol
signal on lead 213 inereases through a second range.
Additionally, the automatic gain control signal
affects the variable voltage supply 215. When tne c~ontrol
signal is at a low negative potential, a p-n-p transistor
296, which is arranged in a common-emitter configuration,
conducts heavily shunting current away from a pair of
zener diodes 297 and 298 in the voltage supply 215. At
such a time the voltage output of the supply 215 is held
at a minimum voltage of approximately 150 volts, as
determined by another zener diode 299. After the gain
control signal rises to about 1.5 volts positive with
respect to ground, the transistor 296 conduets less and
less. As the transistor is thus turned off, the zener
diodes 297 and 298 eonduet more and more eurrent. The
high voltage is inereased gradually from the minimum of
about 150 volts to a maximum of about 425 volts eausing
6835
the gain of the avalanche photodiode 220 to increase from
12 d~ to 40 dB. The zener diodes 297, 29~ and 299 assure
that the reverse bias never exceeds tne maximum allowable
bias of the photodiode 220. The third zener diode 299
in the high voltage supply is included therein to assure
that the high voltage output never decreases below the
desired rninimum of 1;0 volts, the minimum bias for
operating the photodiode 220.
It is advantageous to control both the
avalanche gain and the electrical gain in a multiple ranye
control loop. sy tnis multiple range gain control loop,
the avalanche gain is held at its relatively noiseless
low gain when medium to high intensity input optical
signals are received. The gain of the MOS device also
is kept at its relatively noiseless low gain when hign
intensity input optical signals are received. Only the
emitter-coupled pairs, which have insufficient gain to
produce noise in the output, are adjusted in gain to ~ .
compensate for variations of intensity in the high range
of input optical signals. By separating the control of
gain in different variable gain devices to separate
ranges of the automatic gain conrol signal on lead 213,
loop stability is maintained while better noise performance
is achieved.
Decision and Timing Circuit
Referring now to FIG. 3, there is shown the
decision and timing circuit 300 including a phase-lock
frequency-lock loop timing recovery circuit 303 and a
decision and retiming circuit 306. The timing recovery
circuit 303 advantageously may be arranged like the
circuit described in detail in Canadian Patent No. 1,051,527,
16 -
101~6835
which issued in the name of Jules ~. sellisio on 27 ~arch
1979. The timing recovery circuit 303 is arranyed for receiving
the stream of data pulses occurring on the lead 207 and for
producing on its ~utput lead 304 a perio~ical low jitter
timing signal. This timing signal forms a clock pulse stream
that is synchronized with the incident optical data pulse stream.
The decision and retiming circuit 306 generates on the lead 302,
a regenerated electrical data pulse stream synchronized with
the input optical pulse stream. The pulses have well
defined high and low states, low timing jitter and a low
error rate.
The baseband data signals on the lead 207 con-
tain some information which characterizes the bit rate
and the phase of the optical pulse stream. The charac-
teristics of the bit rate and the phase, together with
their statistical variation, are described by W. R. Bennett
in the Bell System Technical Journal, Vol. 37, No. 6,
November 195~ (pages 1501 through 1542). A baud extractor
310 includes a high-pass filter with a nonlinear charac-
teristic for extracting from the data pulse stream onlead 207 both frequency and phase information of the data
pulse stream.
The timing recovery circuit 303 includes in a
frequency-lock loop a frequency difference detector
arrangement 311 for producing error signals to reduce any
difference between the baud of the data pulse stream on
lead 207 and the frequency of a controlled oscillator 316
except when the amplitude of a baud component signal
on a lead 314 falls below a predetermined value. In the
frequency difference detector arrangement 311, there are
- 17 -
1~i36835
first and sccon~l bran~hes eacn including one of a pair of
multipliers 312 and 313 which are arranged to multiply
the extracted baud compollent signal on leads 314 with
periodical signals produced by the controlled oscilla-tor
316.
Tne controlled oscillator 316 is an adjustable
frequency source that changes frequency of its output
signals on lead 317 in response to control signals applied
by way of a lead 318. A phase shifter 320 receives the
output signals from t'ne controlled oscillator 316 and
produces on leads 321 and 322 output signals having the
same frequency as -the controlled oscillator but having
phases which differ from each other. By multiplying the
two different phase components of the periodic signals
from the controlled oscillator 316 with the extracted
baud component signal on leads 314, periodic waves
including both frequency sum and frequency difference
component signals are produced on leads 323 and 324 in
the two branches.
The sum frequency components are filtered out
by low-pass series filters 327 and 328. Frequency
difference components pass through the filters 327 and
328 to the comparators 330, 331. Each of the comparators
quantizes the frequency difference signals.
A nonideal differentiator 332 produces an
output pulse for every transition in the waveform from
the comparator 330. The output pulse rate from the
differentiator is directly proportional to the frequency
difference between the baud component of the signal on
the leads 314 and the frequency of the controlled
oscillator signal on lead 321 except when the amplitude
- 18 -
1086835
of the Daud component falls below the value required to
produce a beat signal large enough to change the output
level of the comparator.
The outputs of the differentiator 332 and the
comparator 331 are multiplied together in a multiplier 333
which generates at node F a series of constant polarity
pulses. These pulses occur at a rate that is directly
proportional to the frequency difference between the baud
component of the slgnal on the leads 314 and the frequency
of the controlled oscillator signal on the lead 317.
Polarity of the output control signals depends upon tAe
sign of this frequency difference.
The frequency-lock loop includes a third brancn
wherein the output of the multiplier 333 is applied through
a series circuit including a filter 334, a summing circuit
336, a loop filter 337 and the lead 318 to the controlled
oscillator 316. The polarity of the control pulses
at node F is such that they cause the frequency difference
to decrease.
The phase-locked loop is a series circuit which
includes a multipller 340 which is coupled by way of a
low-pass filter 342 to a second input of the summing
circuit 336. It is further connected througn the loop
filter 337, the controlled oscillator 316, the phase
shifter 320 and a lead 345 to a second input of the multi-
plier 340. In the summing circuit 336, a control
component including phase error signals produced by the
multiplier 340 and the filter 342 is combined with tne
series of pulses from the multiplier 333.
Phase-lockiny is achieved through the phase-
lock loop when the frequency difference falls withi.l
-- 19 --
.
~`\
108~i835
the pull-in ran~e of the phase-lock loop. ~s the phase
of the oscillator 316 alic3ns with the phase of the input
o~tical pulse stream, pulses at node F cease allowiny the
control signal on the lead 31~ from the low-pass filter
337 to be a function of only the slowly varying phase-
error voltage at node G. i~agnitude of the phase-error
volta~e decreases until it reaches a small noise signal
near zero when the phases are fully aligned.
The frequency-lock loop assures wide range
pull-in regardless of the bandwidth of the phase-lock
loop because the frequency-lock loop generates the stream
of control pulses at node F in response to ti~e frequency
difference whenever the phases are not locked. In
response to the signal derived from the stream of control
pulses, the frequency of the oscillator 316 is swept
toward the baud of the input optical pulse stream.
Tne stream of clock pulses on lead 304 from
the oscillator 316 and the data pulse stream on lead 207
are applied concurrently to the decision and retiming
circuit 306. In the decision and timing circuit, a
comparator 350 samples the data pulses by comparing them
with a reference level voltage VR for determining whether
or not each input data pulse is a "l" or a "0". Because
the transitions of the data pulses on lead 207 are somewhat
irregular in timing, the output of the comparator 350 is
a signal which is well defined in level but not in timing.
This output signal is resampled by applying it to input D
of a master-slave flip-flop 352 for retiming and reshaping.
The clock pulse sequence on lead 304 from the
30 timing circuit 303 drives the master-slave flip-flop 35
for regenerating on the lead 302 the data pulse stream,
- 20 -
10~36835
as a stream of well-shaped electrical data pulses havirlg
nearly constant amplitude and consistent timing of
transitions. This output electrical pulse stream is
synchronized with and represents the incident optical
data pulse stream.
Transmitter Section
. _ . . .
Referring now to FIG. 4, there is shown a
schematic diagram of the optical transmitter section 400
which receives the regenerated electrical data pulse
stream on lead 302 and produces on a glass fiber 402 a
regenerated optical pulse stream representing the input
optical pulse stream. The optical transmitter section 400
advantageously may be arranged like the circuit described
in detail in U.S. Patent No. 4,009,385 which issued on 22
February 1977 in the name of Darrell D. Sell. The
regenerated electrical data pulse stream on lead 302 is first
applied to an inverter 401 for producing a negative polarity
pulse for each input electrical pulse. The regenerated optical
data pulse stream is produced by a Stripe-geometry AlGaAs
double hetero-structure injection laser diode 410, wnic.l is
connected to the collector output of a driving transistor 411
of an emitter-coupled pair including transistors 411 and
412. The characteristic curve of the laser diode 410, whicn
plots output power as a function of driving current,
includes a knee that causes a threshold in the characteris-
tic. This thresnold varies in response to changes of
tempera-ture of the laser and as a result of aging.
The regenerated electrical pulse stream on the
lead 302 is applied to the base input of the transistor 412
while a reference voltage Vx is applied to the base input
~086835
of the transistor 411. The laser diode 410 is turned "ON"
and "OFF" in response to the level of the pulses applied
to the transistor 41~ by way of the lead 302 producing the
output optical pulse stream 402 representing the input
optical pulse stream.
Since the laser diode 410 has a characteristic ~ -
including a threshold voltage, it is desirable to
operate the laser witn a prebias current, conducted through
the lead 403 and having a magnitude slightly less than the
value of the threshold current. Each negative polarity
pulse from the inverter 401 will cause the total current
conducted through the transistor 411 and the lead 403 to
exceed the threshold of the laser causing the laser to
emit a substantial optical pulse. ~ach low level signal
from the inverter 401 will cause the transistor 411 to cut
off and the laser to operate at the prebias current in tlle
lead 403. This current is below the threshold level,
causing a very low optical output from the laser 410.
The optical ouput on tne glass fiber 402 is
produced from a front surface of the laser diode 410.
The optical pulse pattern from the laser can be stabilized
against temperature fluctuations and aging by automatically
adjusting the prebias current to maintain a constant light
; pulse power.
Such adjustments are made by a feedback circuit
arrangement that derives a control signal from optical
pulse power, emitted from the back mirror of tile laser
diode 410 and detected by a slow speed photodiode 413
wllich does not have to resolve the output pulses. The
output of the photodiode 413 is proportlonal to the
laser peak output averaged over the time constant of the
1086835
photodiode. Current generated in the photodiode 413 is
applied to one input of a lligh yain difference amplifier
414. A second input of the difference amplifier 414 is
derived from the regenerated electrical data pulse
stream on the lead 302.
The input data pulse stream from the inverter
401 is applied to a base input of a transistor 416 ln an
emitter-coupled ~m~arator,including transistors 416
and 417. A reference voltage level Vy is applied to a
base input of the same comparator. The comparator output
signals taken from the collector of the transistor 417
are coupled to the second input of the difference
amplifier 414 as a reference voltage.
The inputs to the difference amplifier 414 are
coupled through identical low-pass filters 418 and 419
to average the signal peaks over several pulses. The
output of the difference amplifier 414 is a control
signal which varies as a function of tlle difference between
the feedback derived from the optical output and the
signal derived from the regenerated electrical pulse
stream on the lead 302.
Output signals from the difference amplifier 414
are coupled through an amplifier 422 to the laser drive
lead 425. The feedback control signal causes the
difference amplifier 414 and the amplifier 422 to establish
in lead 403 a prebias current which is conducted through
the laser diode 410. The magnitude of the prebias current
is adjusted by means of a potentiometer 430 so that the
laser diode 410 operates slightly ~elow its t'nreshold when
no pulses are applied over the lead 302.
Any fluctuation of the laser optical output caused by
temperature variation or aging of the device will cause a
- 23 -
1086835
corresponding chanqe in current conducted through thephotodiode 413. In response to this cllange of current,
the high gain difference amplifier 414 and the amplifier
422 will generate a compensating change in the prebias
current. The prebias current is thus compensated to main-
tain a constant optical output pulse level from the laser
diode 410.
The input pulse stream that is applied to the
prebias control circuit by way of the transistors 416
and 417 produces a variable reference level which is
applied to -the difference amplifier 414 for adjusting the
prebias current in accordance with variations in the input
pulse stream. This variable reference level further
maintains the constant output optical pulse level. By
referencing the prebias difference amplifier 414 to the
signal derived from the input pulse stream, laser diode
lifetime is prolonged over that of an arrangement wherein
a constant reference level is applied to the difference
amplifier 414. Such a constant reference level will cause
the prebias current to rise when a long series of zeros
occurs in the input pulse stream. A subsequent "1" or
series of "ls" can cause irreparable damage to the laser
diode.
The foregoing describes an embodiment of the
invention, and in view of that description, other embodi-
ments will be obvious to those skilled in the art. The
embodiment described herein and those additional embodiments
are considered to be within the scope of the invention.
- 24 -
