Coherent combining of the output of two semiconductor lasers using optical phase-lock loops.

Abstract

We have experimentally demonstrated current-injection optical phase-lock loops (OPLLs) based on commercial single-section semiconductor distributed-feedback (DFB) lasers. Using two parallel OPLLs, we have obtained 87% efficient coherent power combining of the two DFB lasers. The rms differential phase error between the two lasers is about 30 degrees .

Full text

Page 1

Coherent combining of the output of two
semiconductor lasers using
optical phase-lock loops
Wei Liang and Amnon Yariv
Department of Applied Physics, MS 128-95, California Institute of Technology, Pasadena, California 91125, USA
Anthony Kewitsch and George Rakuljic
Telaris Inc., 2118 Wilshire Boulevard No. 238, Santa Monica, California 90403, USA
Received August 29, 2006; revised November 9, 2006; accepted November 10, 2006;
posted November 14, 2006 (Doc. ID 74481); published January 26, 2007
We have experimentally demonstrated current-injection optical phase-lock loops (OPLLs) based on commer-
cial single-section semiconductor distributed-feedback (DFB) lasers. Using two parallel OPLLs, we have ob-
tained 87% efficient coherent power combining of the two DFB lasers. The rms differential phase error be-
tween the two lasers is about 30°. © 2007 Optical Society of America
OCIS codes: 140.2010, 030.1640, 140.3490.
High-power high-brightness laser systems made by
coherently combining a large number of light emit-
ters are of great interest to a range of applications
that includes material processing, remote sensing,
and precompensation and adaptation in atomo-
spheric beam propagation.1,2Various technologies
have been explored to achieve the goal by using semi-
conductor (SC) lasers and fiber lasers.1,3–8In this Let-
ter we report what we believe to be the first coherent
power addition of two commercial grade SC lasers by
using current-injection optical phase-lock loops9–13
(OPLLs). Compared with the previous coherent
power-combining techniques, including the optical in-
jection OPLLs, the current-injection OPLLs tech-
nique can provide individual electronic control of the
frequency and phase of each laser in an array with-
out external phase modulators and thus can lead to a
big coherent emitting aperture that can be controlled,
focused, distortion corrected, scanned in space, and
pulsed by using pure electronic control. Throughout
the rest of this Letter we will use “OPLLs” to refer to
the current injection OPLLs only.
In an OPLL, the phase and frequency of the slave
laser are forced to track those of the master laser by
a negative feedback control loop. Ever since the early
pioneering work and demonstration of the first
OPLL,9different types of lasers have been used to
build OPLLs.10–13Compared with fiber lasers and
solid state lasers, SC lasers are especially attractive
for applications requiring high power and high effi-
ciency. In addition, the frequency and phase control
of SC lasers can be conveniently realized by control of
the injection current. Implementing OPLLs based on
SC lasers, however, has proved to be a challenging
task, since, because of the large laser linewidth, it re-
quires wideband feedback electronics and small feed-
back loop delays. Unfortunately, the achievable loop
bandwidth is seriously limited, as single-section SC
lasers typically exhibit a phase reversal in their FM
response in the frequency range of 0.1–10 MHz due
to the competition between thermal and electronic
effects.14,15This makes it very difficult, if not impos-
sible, to lock single-section SC lasers with a line-
width of above a few megahertz. Recently, with im-
provements in the design and the fabrication process,
the linewidth of SC lasers has been reduced to sub-
megahertz, which makes it possible to phase lock
them with just a few megahertz of loop bandwidth. In
our experiment, we successfully phase locked com-
mercial distributed-feedback (DFB) lasers of frac-
tional megahertz linewidth.
To determine the current frequency response of the
free-running SC laser, we employed a fiber Mach–
Zehnder interferometer as a frequency discriminator
and used an Agilent 4395A network analyzer. The de-
tails of the measurement procedure are similar to
those described in Ref. 16. In our measurement we
biased the DFB laser at 400 mA, yielding an output
power of 16 dBm and a wavelength of 1538 nm. The
measured FM response of the laser is plotted in Fig.
1. Near 5 MHz, the amplitude of the FM response
manifests a dip, and the phase of the FM response
exhibits a 90° shift. Added to the 90° shift due to the
inherent integration by the laser,2and the 180° shift
Fig. 1.
section DFB laser.
(Color online) Measured FM response of a single-
370
OPTICS LETTERS / Vol. 32, No. 4 / February 15, 2007
0146-9592/07/040370-3/$15.00© 2007 Optical Society of America

Page 2

due to the negative feedback loop, the resulting 2?
phase shift near 5 MHz combined with sufficient loop
gain can cause oscillation and instability in the feed-
back loop. In this simple analysis we neglect the ef-
fect of the feedback time delay, an assumption that
will be justified later.
Figure 2(a) is a schematic plot of one of the two
OPLLs. We used an Agilent 81640A tunable laser
with a submegahertz linewidth as the master laser.
The optical signals from both lasers are combined by
using a 2?2 optical fiber coupler. Half of the com-
bined signal is fed to a HP 11982A photodetector
(PD), and the resulting beat signal is fed to a HP
8565E rf spectrum analyzer. The beat signal output
of the second PD (New Focus 1544B) is mixed with
the 1.5 GHz offset rf signal generated by a HP 8350A
signal generator. The mixer acts as a phase detector.
The phase error signal passes through a loop filter
and is then injected into the DFB laser to complete
the control loop. In our experiment we used a first-
order lead-lag filter, which can increase the signal-to-
noise ratio of the OPLL. We estimate the total loop
delay time as ?5 ns. At 5 MHz this corresponds to a
9° phase delay, which is not significant in limiting the
loop bandwidth. The acquisition bandwidth is mea-
sured to be about 10 MHz; the frequency difference
between the two lasers needs to be manually tuned to
the vicinity of the offset frequency 1.5 GHz for lock-
ing to occur. In our next design this function will be
performed automatically. The power spectrum of the
beat signal between the master and the slave lasers
is plotted in Fig. 2(b). The strength of the locked sig-
nal is more than 30 dB above the noise pedestal.
From the spectrum, we can calculate the locked sig-
nal power and the noise power within the 50 MHz
sweep bandwidth. We estimate 90% optical power of
the DFB laser is phase locked to the master laser.
The residual rms differential phase noise between
the two lasers is estimated to be 19°.
In a second experiment we phase locked two DFB
slave lasers separately to a single (Agilent 81640A)
master laser and combined their output. The sche-
matic of the experimental setup is illustrated in Fig.
3(a). The two OPLLs are offset by the same 1.5 GHz
rf signal. The optical signals from the two OPLLs are
combined by using a 2?2 optical fiber coupler. One
output is used to simultaneously monitor the locking
status of the two slave lasers to the master laser in
the frequency domain by using the HP 8565E spec-
trum analyzer. The second output is displayed “as is”
in the time domain using a Tektronix TDS3052B os-
cilloscope. The latter beat signal is shown in Fig. 3(b).
If we ignore the slight power difference between the
combined optical signals, the combined power at the
PD is
P = 2I?1 + cos??? t + ?? + ?n??,
?1?
where I stands for the power of each of the two opti-
cal signals, and ??, ??, and ?nare respectively the
frequency difference, phase difference, and differen-
tial phase noise between the combined two optical
Fig. 2.
sured power spectrum of a 1.5 GHz locked beat signal.
(Color online) (a) Schematic of an OPLL. (b) Mea-
Fig. 3.
power addition of two SC DFB lasers locked to a common
reference laser. (b) Time domain measurement of combined
power. Blue dots, measured data; red solid curve, smoothed
data.
(Color online) (a) Experimental setup for coherent
February 15, 2007 / Vol. 32, No. 4 / OPTICS LETTERS
371

Page 3

signals at the PD. When at least one of the two slave
lasers is not locked to the reference laser, the output
is an ac signal [the right part of Fig. 3(b)] with the
frequency of ??. The data show as a noisy scatter of
points since ?? is in the megahertz range, while the
time scale of the oscilloscope is set at seconds. When
both DFB lasers are locked ???=0?, the signals are
coherently added and the output of the PD consists
of, ideally, a dc signal that in our case varies slowly
on the time scale of seconds as can be seen in the left
part of Fig. 3(b) (blue dots). This slow variation re-
flects, as it is should, the change of the difference in
optical path lengths experienced by the two combined
optical signals due to slow variation of temperature
and environment. The variation of the optical path
lengths will be significantly reduced as we move to
free-space power combining later. Since the optical
phase of the slave laser output signal can be directly
controlled by the feedback current signal, the slow
phase variation can also be compensated with an ad-
ditional feedback loop made of optical phase detec-
tors and electric phase shifters. By calculating the
contrast,i.e.,
?Pmax−Pmin?/?Pmax+Pmin?,
smoothed data [red solid curve in Fig. 3(b)], we esti-
mate the beam combining efficiency is about 87%.
From the uncertainty of the dc signal (i.e., the scat-
tering of the data) we can estimate the rms phase er-
ror ?nbetween the two combined signals is about 30°.
In the first experiment described above we have esti-
mated from the power spectrum that the rms phase
error between the slave laser and the master laser in
each OPLL is about 19°; thus the rms phase error be-
tween the two slave lasers can be approximated as
?2?19=27°. Thus the rms phase errors calculated
from the frequency domain and that from the time
domain measurements agree with each other. In the
above measurement and analysis, the contribution of
the master laser in the combined output power is
negligible, which can be justified by the fact that its
output power is almost 16 dB smaller than those of
the slave lasers.
In conclusion, we have coherently added the optical
power of two SC DFB lasers by phase locking them to
of the
a single master laser by using OPLL technology. This
points the way to potential coherent combining of a
large array of SC lasers. Further improvement in the
locking efficiency requires either that the laser line-
width be further reduced or that the loop bandwidth
be increased. Efforts along these directions are in
progress.
The authors acknowledge the support of DARPA’s
Microsystems Technology Office office (M. Stickley).
W. Liang’s e-mail address is liangwei@its.caltech.edu.
References
1. A. Shirakawa, T. Saitou, T. Sekiguchi, and K. Ueda,
Opt. Express 10, 1167 (2002).
2. A. Yariv, Opt. Lett. 30, 2191 (2005).
3. N. W. Carlson, G. A. Evans, J. M. Hammer, M. Lurie,
S. L. Palfrey, and A. Dholakia, Appl. Phys. Lett. 50,
1301 (1987).
4. J. R. Leger, M. L. Scott, and W. B. Veldkamp, Appl.
Phys. Lett. 52, 1771 (1988).
5. W. Wang, K. Nakagawa, S. Sayama, and M. Ohtsu,
Opt. Lett. 17, 1593 (1992).
6. V. A. Kozlov, J. Hernandez-Cordero, and T. F. Morse,
Opt. Lett. 24, 1814 (1999).
7. L. Bartelt-Berger, U. Brauch, A. Giesen, H. Huegel,
and H. Opower, Appl. Opt. 38, 5752 (1999).
8. S. J. Augst, T. Y. Fan, and A. Sanchez, Opt. Lett. 29,
474 (2004).
9. L. H. Enloe and J. L. Rodda, Proc. IEEE 53, pp. 165
(1965).
10. W. R. Leeb, H. K. Philipp, A. L. Scholtz, and E. Bonek,
Appl. Phys. Lett. 41, 592 (1982).
11. R. C. Steele, Electron. Lett. 19, 69 (1983).
12. T. J. Kane and E. A. P. Cheng, Opt. Lett. 13, 970
(1988).
13. L. N. Langley, M. D. Elkin, C. Edge, M. J. Wale, U.
Gliese, X. Huang, and A. J. Seeds, IEEE Trans.
Microwave Theory Tech. 47, 1257 (1999).
14. S. Kobayashi, Y. Yamamoto, M. Ito, and T. Kimura,
IEEE J. Quantum Electron. 18, 582 (1982).
15. P. Correc, O. Girard, and I. F. de Faria, Jr., IEEE J.
Quantum Electron. 30, 2485 (1994).
16. W. V. Sorin, K. W. Chang, G. A. Conrad, and P. R.
Hernday, J. Lightwave Technol. 10, 787 (1992).
372
OPTICS LETTERS / Vol. 32, No. 4 / February 15, 2007