- Active Alignment Is Here to Stay
Adrian C. Goding
As long as light sources, amplifiers and regenerators remain expensive,
the benefits of active alignment’s lower insertion losses
will continue to outweigh its costs. And are its disadvantages really so bad?
Active alignment is a ubiquitous process in fiber optic component assembly that sets the
photonics industry apart from the semiconductor and electronics industries with
which it is most often compared. For any company accustomed to applying the highly
flexible and scalable electronics and semiconductor automation processes to photonics,
the necessary evil of active alignment is certainly a frustrating bottleneck. At
first glance, it seems that replacing active alignment with a passive method would
represent the Holy Grail of fiber optics assembly technology.
The benefits of active alignment continue to outweigh its costs. If this gradient index lens
were aligned and bonded using a state-of-the-art passive system, its insertion loss
would be 10 times higher than if it were bonded actively.
The problems with active alignment are well-known:
It requires source and/or detector coupling, prealignment, a capital investment
in nanopositioning tools that are sufficiently robust for the factory-floor environment
and much more time than is tolerated in semiconductor and electronics assembly.
Finally, although we use the term “active
alignment,” the task is actually active alignment and bonding. It is
really not so difficult to actively align two parts. The challenge is to epoxy,
solder, fuse or laser weld those parts so that they still will be aligned after
they have been shipped to the customer.
For these reasons, pigtailing via active
alignment has historically been the most labor-intensive step in the fiber optic
assembly process, and it remains the most difficult to automate.
But is active alignment really so problematic?
We think it is getting a bad rap. The
truth is that active alignment has its advantages. It offers a tremendous cost
savings because it compensates for all the sloppy tolerances that originally made
it possible to fabricate the piece parts at low cost. A gradient index lens in which
the optical and mechanical axes coincide within better than ±0.5 μm would
be prohibitively expensive. And imagine the price tag on a pair of fiber collimators
whose optical axes could be aligned to within a few microradians simply by slipping
them into a precise tube.
As for its drawbacks, it is now possible
to perform source and detector coupling with semiautomated or fully automated modules.
The cost of the nanoalignment tools has fallen dramatically because of competitive
forces and advances in technology. Alignment time is now measured in seconds instead
of in minutes. And, as challenging as it may continue to be, we have made significant
advances in maintaining alignment through the bonding process.
How much of a difference is there between active and passive alignment?
Simple and cheap
Consider the case of a fiber optic collimator,
in which a gradient index lens must be aligned and bonded to a fiber capillary.
Assuming that a state-of-the-art passive alignment system is used to align these
two parts within 1 μm, the typical insertion loss would be as high as —1.5
dB. Actively performing the alignment and bonding, however, yields a typical loss
of —0.1 to —0.2 dB.
Insertion loss is heavily dependent on alignment. Until light sources, amplifiers
and regenerators become more economical, active alignment’s lower insertion
losses will make it the preferred solution.
Active alignment enables a manufacturer
to build this collimator to specifications using two relatively imprecise (and thus
inexpensive) piece parts. This is in contrast to building the same collimator by
passively aligning a set of optomechanically perfect — and outrageously expensive
— piece parts. Moreover, building this collimator by passive alignment assumes
the technology to locate, place and bond cylindrical glass components within these
±1-μm tolerances. This is a tall order even for state-of-the-art passive
assembly systems because of the position feedback requirement.
By comparison, active alignment feedback
is extremely simple. The feedback generated at the detector or detectors is often
a single number directly related to the misalignment error — and as many as
six degrees of freedom may be contributing to this error — as well as to
the overall product performance (total decibel loss of the end product).
One hope for eliminating the need for
position feedback in passive alignment is the use of silicon structures such as
V-grooves and wells that passively mate with micron precision. However, this only
shifts the burden to metrology and testing. The silicon V-groove assemblies must
be tested to verify the interfiber spacing, and the final product still must be
tested for insertion loss.
Of course, efforts are under way to redesign and
re-engineer optical components so that they may be aligned with looser tolerances.
New manufacturing methods will improve the precision of piece parts without significantly
affecting their cost. Silicon optical benches will continue to be deployed in increasing
numbers. There is even work focused on installing simple microelectromechanical
devices or piezo actuators inside the optical components, which would greatly relax
the placement tolerances and offer the additional benefit of enabling realignment
over the lifetime of the component.
The driver for precise alignment is
minimizing light loss, stemming from the relatively high cost of sources, amplifiers
and regenerators, as well as the fiber’s inability to carry higher powers.
Perhaps the introduction of extremely low cost, mass-produced amplifiers, such as
semiconductor optical amplifiers, will allow for “lossy” components
that are easier to align passively.
But into the foreseeable future, minimizing
light loss will continue to be very important. The comparative performance of components
is measured in tenths of a decibel, and the benefit of lower insertion loss continues
to outweigh the costs of active alignment.
Nevertheless, as component makers seek
to eliminate active alignment steps in their processes, manufacturers of precision
alignment and assembly systems will adapt — offering active, passive and hybrid
Meet the author
Adrian C. Goding is a director of technical sales
at Adept Technology Inc. in San Jose, Calif. He holds a BS and an MS in optics from
the University of Rochester in New York.
MORE FROM PHOTONICS MEDIA