Fiber Bragg Gratings Filter WDM Signals
Fiber Bragg gratings are efficient tools for multiplexing and demultiplexing WDM signals and for chromatic-dispersion compensation.
Fiber Bragg gratings are versatile wavelength filters for multiplexing
and demultiplexing wavelength division multiplexing (WDM) signals. They also can
compensate for chromatic dispersion that can degrade the quality of the WDM signal
in an optical fiber.
Such a grating is, in fact, a fiber with a core
that has a sinusoidal variation of refractive index (Figure 1). This variation affects
the phase of light passing through the grating so that some wavelengths interfere
constructively and some destructively.
Figure 1. A fiber Bragg grating
is most efficient as a WDM signal filter when it reflects a single wavelength and
transmits the other wavelengths.
In this regard, a fiber Bragg grating
resembles a thin-film coating, but there are significant differences. In a thin-film
filter, the changes in refractive index are discontinuous and on the order of 10—1.
In the grating, the variation in refractive index is continuous and much smaller.
Moreover, a thin-film filter deals with waves in free space, while the grating’s
waves are confined to a waveguide. Thus, the fiber Bragg grating is most efficient
as a filter when it reflects a single wavelength and transmits the rest, while a
thin-film filter reflects all wavelengths except the one filtered out.
Because the selected wavelength reflects from
the grating, it must work in conjunction with a circulator or other isolating component.
A circulator, for instance, is a three-port device designed so that light entering
the first port emerges from the second port, and light entering the second port
emerges from the third. An optical add/drop multiplexer is an example of a fiber
Bragg grating used to multiplex and demultiplex a WDM signal (Figure 2). A signal
composed of many wavelengths enters Port 1 of the first circulator, emerges from
Port 2 and enters the grating, where the selected wavelength, λ2, is reflected and
all others are transmitted. The reflected light enters Port 2 and emerges from Port
3 in a process that demultiplexes it from all the other wavelengths.
Figure 2. In an optical add/drop multiplexer, a WDM signal with many wavelengths enters Port
1 of the first circulator, emerges from Port 2 and enters the fiber Bragg grating,
where the selected wavelength, l2, is reflected and all the others transmitted.
The reflected light enters Port 2 of the first circulator and emerges from Port
3, essentially being dropped.
The “add” signal at λ2
enters Port 1 of the second circulator and emerges from Port 2. It enters the grating,
reflects from it and joins the other wavelengths transmitted through the grating.
It has been multiplexed into the WDM signal.
In the early years of WDM, almost all
multiplexing/demultiplexing was based on thin-film technology, and thin films are
still the best solution when the separation between adjacent channels is 100 GHz
or more. But as channel spacing drops below 100 GHz, thin-film multiplexers and
demultiplexers become increasingly complex and expensive. Fiber Bragg gratings,
on the other hand, can easily handle channel spacing of 25 GHz and perhaps even
less. One drawback is their thermal sensitivity, which, although an order of magnitude
greater than that of a thin-film filter, can be reduced by athermal packaging. This,
however, adds a layer of complexity and expense. An advantage of the gratings is
that they are fabricated in the fiber itself. There is no optical loss through coupling
out of the fiber into another device.
Chromatic dispersion in optical fiber results
from both the bulk dispersion of the glass and the waveguide dispersion of the fiber.
The signal degrades as it travels because dispersion causes pulses to widen as they
propagate down the fiber, eventually making it difficult to distinguish a 1 from
a 0 in a signal. In the extreme case, the pulses level out to a flat, continuous
Pulses widen because they are not perfectly
monochromatic. Each pulse has a bandwidth that, according to the Fourier theorem,
is inversely proportional to its temporal width. As the pulse travels down the fiber,
the short-wavelength components get ahead of the long-wavelength components, causing
the entire pulse to widen. This is analogous to a gym class taking a half-mile run;
at the starting line the runners are all close together, but by the time they’ve
gone half a mile, they’re spread out over many yards.
Interestingly, the problem of signal
degradation due to chromatic dispersion scales as the square of the pulse frequency.
The problem is 16 times worse at 10 GHz than it is at 2.5 GHz. Part of the reason
is that, as mentioned above, a 10-GHz pulse has four times the bandwidth of a 2.5-GHz
pulse, which means that the pulse widening is four times greater. But the 10-GHz
pulse train has one-fourth the space between pulses that the 2.5-GHz pulse train
has, so it has one-fourth the tolerance for pulse widening. Pulse widening up by
a factor of four combined with a reduced tolerance for widening by a factor of four
degrades the signal 16 times. Hence, as systems move to higher pulse frequencies,
the importance of chromatic-dispersion compensation will increase dramatically.
For chromatic dispersion compensation
applications, the grating in the fiber is not linearly sinusoidal as it is in a
multiplexing or demultiplexing device. Rather, it is chirped, with spacing varying
over the length of the grating (Figure 3).
Figure 3. A fiber Bragg grating (FBG) designed for chromatic-dispersion compensation
differs from one designed for multiplexing and demultiplexingin that the fiber is
not linearly sinusoidal. Instead, it is chirped, with varied spacing over the length
of the grating.
In one example, the shorter wavelengths
will reflect from the right-hand side of the grating, and longer wavelengths from
the left-hand side. In a pulse that has traveled through a considerable length of
fiber, the short wavelengths have moved ahead of the long wavelengths. Inject the
pulse into this grating, and the short wavelengths must travel a greater distance
before reflection than the long wavelengths. Both wavelengths emerge from the grating
at relatively the same time, and the pulse has been compressed, (almost) to its
In the runner analogy, suppose that
each runner must touch a different post before returning to the starting line. If
the fastest runner’s post is farthest from the starting line, and everybody
else’s post is scaled according to his or her speed, all should return to
the starting line at the same time.
Figure 4. Networks rely on a circulator
to couple a pulse of a multiwavelength WDM signal into a fiber Bragg grating.
In practice, networks will use a circulator
to couple the pulse into the grating (Figure 4). Because each channel in a WDM signal
requires its own fiber Bragg grating, the signal must be demultiplexed and fed into
a bank of gratings (Figure 5).
At present, there are drawbacks to
such a solution. First, it works only with a particular set of channels. Adding
channels or changing their wavelengths requires redesign of the dispersion compensator.
Second, redesign of the compensator is necessary if the system’s dispersion
changes (because a portion of the fiber is being replaced, for example).
Figure 5. Each channel in a WDM signal requires its own fiber Bragg
grating, so the signal must be demultiplexed and fed into a bank of gratings.
Research continues to resolve such
issues. Several R&D projects are working on making very long fiber Bragg gratings
to allow a single device to compensate for dispersion across the entire C-band without
demultiplexing the signal. Scientists also are exploring combining fiber Bragg grating
compensators with heaters, so that the compensation can be thermally tuned to make
up for small, short-term dispersion changes in the fiber.
Meet the author
Breck Hitz is executive director of the Laser
and Electro-Optics Manufacturers’ Association (LEOMA). The material in this
article is taken from LEOMA’s short course, “Understanding Fiberoptics
MORE FROM PHOTONICS MEDIA