Fiber Amplifiers

Fiber amplifier systems often provide a very high gain of several tens of decibels. This implies that different parts of the active fiber(s) see very different amounts of optical powers or pulse energies.

There are then several reasons why a large mode area is needed for the last stage (a power amplifier stage):

• Nonlinear effects would be excessive for too low mode areas.
• When using a double-clad fiber for high average powers, a large mode area deceases the cladding/core area ratio and thus improves pump absorption, so that a shorter fiber length can be used; that further reduces nonlinear effects.
• Further, one avoids or reduces problems with excessive gain saturation (and resulting pulse shape distortions) by high-energy pulses. Finally, one also avoids problems resulting from excessive gain (e.g., strong amplified spontaneous emission, see part 4) if a high energy needs to be stored in the fiber.

On the other hand, a smaller mode area remains preferable for a low-power preamplifier:

• We like to have a high gain efficiency in order to obtain a high amplifier gain while using a small pump power.
• The power conversion efficiency gets higher in the low average power regime.
• We can use strictly single-mode fibers with robust guiding, allowing tight coiling for a compact setup while keeping the beam shape stable.
• The above mentioned reasons for higher mode areas do not apply for a low-power preamplifier: there are no (or at least smaller) issues with nonlinearities, gain saturation or ASE.

Typically, the power amplifier stage has a substantially lower gain but provides the largest part of the output power.

When using two different fibers as explained above, one may want to minimize the pump power for the preamplifier by pumping that fiber directly into the core. At a low power level, this is easy to do, e.g. using a fiber-coupled diode laser and a dichroic fiber coupler. That option, however, may not be suitable for high powers as needed at the end; there, one wants to do cladding pumping, i.e., inject pump light into the pump cladding of a double-clad fiber (see part 6). Obviously, that is at least easier when using two different amplifier stages with accordingly different fibers.

With fibers, we are sometimes short of ends. More fibers help by having more ends. In a two-stage amplifier, for example, we already have four fiber ends into which we can inject pump power. This is welcome, for example, if we couldn't get enough pump power launched with two pump diodes only.

In order to avoid trouble with excessive amplified spontaneous emission (ASE, see part 4), a first step is to design for a gain which is not higher than needed. In particular, the gain efficiency should not be made higher than necessary. Sometimes, however, we just need a very high gain of 60 dB, for example. That in a single amplifier stage would then lead to excessive ASE; more precisely, before we reaching that gain we would convert much of the pump power into ASE power.

Multi-stage amplifiers allow one to solve that problem by ASE elimination between the stages. Essentially, one strongly attenuates ASE there, so that it has to built up “from scratch” in the next stage. There are different options for that:

• If only narrowband signals or pulses need to be amplified, one can simply use a narrowband optical bandpass filter which transmits the useful radiation (signals or pulses) while rejecting most of the ASE. If one can use an interference filter with a 1-nm filter bandwidth, for example, that may reduce the unwanted ASE powers already by roughly an order of magnitude. For femtosecond pulses, however, which are intrinsically broadband, this is a less effective method, as a larger filter bandwidth is required.
• A Faraday isolator allows one to strongly suppress ASE in backward direction. That is particularly important since a preamplifier will usually be more sensitive to ASE than a power amplifier stage.
• For pulsed operation, one may also use time gating with an optical switch between the stages, which is opened only for a short time interval around each pulse to be amplified. (For example, an acousto-optic modulator or an electro-optic modulator can be used.) That method does not avoid possible problems with ASE directly around the pulses, which may disturb some application, but it reduces the ASE power loss in the system.
• If the signals are well polarized throughout the device, one may also use polarizers between the stages to suppress ASE in the unused polarized direction. However, that measure is less common, as it achieves less than the others and the signals are often unpolarized.

In some cases, one has to combine two or more of these options in order to achieve sufficient ASE suppression.

A low noise figure of an amplifier (see part 9) may be difficult to achieve with a high-power double-clad fiber amplifier. This is because such devices usually operate with a fairly low excitation density at the input end, particularly when being backward-pumped. However, if one combines such a power amplifier with a core-pumped preamplifier, the overall device can have a very good noise figure.

A Faraday circulator (i.e., a Faraday isolator with an additional output port) allows one to extract any signal light coming back from the power amplifier, for example as a result of back-reflection from the application. In that way, one may avoid damage of the preamplifier or the seed source. Note that back-reflected light would be further amplified there. And a Faraday isolator at the output may not work since it would have to withstand high optical powers. (Of course, it feedback gets strong enough, it may also well kill a circulator between the stages.)

Also, one may monitor power levels between amplifier stages, e.g. in order to switch off the device when detecting bad operation conditions, such as too strong optical feedback.

For some applications such as optical fiber communications with wavelength division multiplexing, one requires amplifiers with a “flat” gain, i.e. with low variations of gain within some spectral window. This is often done by inserting some gain-flattening filter, which provides higher optical losses in wavelength regions where the gain would be too high.

But where to put such a filter? If we put it in front of the amplifier, it will spoil the noise figure (see part 9). If we put it behind, it degrades the power conversion efficiency. Both problems can be avoided by putting the filter between two amplifier stages, where neither noise issues nor power losses are critical.

Gain flattening may also be facilitated by combining two fibers with different chemical compositions of the fiber core, leading to different gain spectra and an overall broader gain spectrum.

Here you can submit questions and comments. As far as they get accepted by the author, they will appear above this paragraph together with the author’s answer. The author will decide on acceptance based on certain criteria. Essentially, the issue must be of sufficiently broad interest.

Please do not enter personal data here; we would otherwise delete it soon. (See also our privacy declaration.) If you wish to receive personal feedback or consultancy from the author, please contact him e.g. via e-mail.