Femtosecond Fiber-Optic Laser

Much of the revolution in fiber-optic communications, has been driven by new discoveries and inventions by optical physicists. New laser materials, and new designs and configurations, have created whole generations -- even new species -- of light sources, particularly in the last ten or fifteen years. Many of these developments have opened new technical possibilities and prospects for commercial application.

At the root of all this technical progress lies progress in understanding and discovery, much of it driven not by programmatic needs but by a pressing personal need to figure out how things fit together into a picture. This experiment is very well suited to that kind of curiosity and tinkering.

Simple lasers Fiber lasers are among the simplest lasers: they don't have transverse modes, typically, and they're solid-state, with few adjustable parameters. In addition, they're made from components whose standards in uniformity and reliability have been established by the requirements of the telecommunications industry.
At the same time, the optical nonlinearities of moderately intense ultrashort pulses (~100 fs) in these lasers make them an extremely rich place to discover fascinating and complex nonlinear physics. The experiments based on this laser will let you explore several regimes of nonlinear optical and laser physics.

Useful background You'll find it an advantage, but not essential, to have already done the experiments on the He-Ne laser (including modelocking) and on fiber-optics (how they work as waveguides, including single-mode waveguides, and the transverse distribution of fields). A little more important is the acoustic-waveguide experiment in this same lab-room, which leads you through pulse spreading due to group-velocity dispersion. If you haven't done the experiments, it may be useful to read the experimental guide sheets.

Advanced or specialist students may be interested to read the review paper on ultrashort-pulse fiber lasers by Nelson et al. 1997. Excellent books include Derickson 1998, Agrawal 2001 and Boyd 1992; the first two are available through the equipment wicket, and the last is available in the Physics Library.

Experimental guides for investigating the physics of this ultrafast fiber-optic laser are detailed below. For all of these, you should first read the following primer, which describes all the components of the laser, introduces several optical and nonlinear physics issues central to understanding this laser, and gives a good but fairly heuristic description of how the laser works.

Draft of Primer for Femtosecond Fiber-optic laser

Fiber-laser basics: a few things before you start...

Personal Safety with this Laser

The first issue is your safety: the fiber laser has a power comparable to a small HeNe laser and poses no perceptible danger to your skin, etc. Its wavelength and power is also eyesafe under normal conditions, but the wavelength is not visible, for two reasons. This wavelength is stopped in the cornea or lens of the eye, and does not go to the retina. As is the case with many lab HeNe lasers, you must not focus the beam to your eye with a lens, like a microscope or telescope.

To track the laser beam, there is a small infrared-laser revealing card, which will show a faint orange spot, barely visible under room lights.

You are encouraged to learn about laser safety, through the University of Toronto and other laser safety sites. One direct link to eye-safety is here.

When used with the fiber amplifier (advanced) the fiber laser is capable of 45 mW of output power, which is significant and can be dangerous to your eyes when operated in femtosecond mode and focused. There are no dangerous voltages in the laser.

Safety of the Laser Equipment

The second issue is the laser's safety: there is very little you can do to damage anything in the equipment, but:

Here's how to turn on the laser:

  1. ensure that the 9VDC power supply to the thermoelectric cooler is plugged in at the wall, and connected to the laser. There should be a green indicator light (LED) showing that it is plugged in.
  2. turn on the digital multimeter connected to the fiber laser. This monitors a thermistor measuring the temperature of the 980nm diode pump laser. It should read about 1 volt at all times; if the value differs more than 10% from this, do not run the laser, and if it's already running, shut it down immediately.
  3. turn on the Tektronix TDS210 oscilloscope near the left side of the laser. You'll be looking eventually for a signal 50-200 mV, with pulses at about 25 MHz. Start with triggering 'auto'
  4. turn on the bias supply voltage on the InGaAs photodetector attached to the oscilloscope
  5. turn on the main power push-button at the lower left of the diode laser controller. Near the middle of the panel is a selector that will let you cycle over four different settings for reading the controller and the pump-laser output. When turned on, the first value showing will be I-limit, the maximum current permitted to be sent to the pump. This supply is set for a current-limit of about 650 mA, which will prevent you from doing anything seriously wrong with the apparatus; this setting must not be changed or complete destruction of the diode laser may result.
  6. press the tactile membrane-switch at upper right, marked Enable. The pump laser is now on, and the driver current can be adjusted using the main knob. If you cycle through the display settings, you can monitor (in order): current limit, drive current, output power, and signal current from the built-in photodiode monitor.

The fiber laser will now be operational. The way in which it operates, and all its characteristics, depend on adjustments made to the pump power and to elements of the fiber laser itself.

Exercise 1: Measuring the slope efficiency in cw mode:

First exercise: Measure the cw laser output power as a function of pump power

  1. the orange fiber-optic cable (about 1 m long) from the laser output should be connected to the 50/50 optical splitter, into the common port. The two output ports should go to the photodetector on the TDS210 oscilloscope, and to the Optical Spectrum Analyzer (Ando).
  2. set the fiber laser to operate in 'vanilla' cw mode, by loosening the stainless-steel thumbscrews on the pressure-plates of both polarization controllers, with your fingers. Do not unscrew them completely or they will fall off, which is fussy and annoying.
  3. you should see a flat signal of about 50 mV on the oscilloscope. Pivot the upper-left polarization controller while watching the output power on the oscilloscope. You may find the fiber laser already modelocking, but as you unscrew the pressure-plate you will go to cw mode. If when you pivot the controller the power on the oscilloscope changes, then there is still stress-birefringence in the fiber -- carefully unscrew the thumbscrew a little further, until the oscilloscope signal is no longer sensitive to the pivoting of that plate
  4. repeat for the lower-right polarization controller
  5. move the ouput fiber to the Exfo power meter. That meter has several settings for different types of laser -- you may have to press the button for 'wavelength' to cycle among different pre-set setups, to get the setting for 1550nm. With the pump laser drive current set around 600 mW, you should see 1-2 mW output.
  6. find the behaviour of the laser output power as a function of pump power. To track the diode-laser pump-power, you can use the built-in photodiode-monitor from the diode-laser driver, or you can use the calibration curve of drive current vs. pump-laser output, provided by JDS Uniphase, the manufacturer.
  7. plot your results in Kaleidagraph, on the lab computers; find the functional relation between pump power and fiber-laser output power. Explain each part of your observations -- how do you understand the features of what you see? With the understanding that you have formed, always see if you can test your ideas by changing something.

Variation: the fiber-ring can have a measure of residual birefringence, from stresses of being coiled up. Try small amounts of pressure from the thumbscrew of the upper left polarization controller, and different pivot-positions, to see if you can compensate, and maximize the cw power of the fiber laser. Then repeat the power measurements -- do you expect to see a difference?

Exercise 2: Obtaining modelocking

The femtosecond fiber laser Primer does a good job of sketching how mode locking works, and how to begin to make it modelock. In summary, you can:

  1. set the diode-laser driver current to 600mA
  2. as you monitor the cw power on the oscilloscope, gently begin to screw in the thumbscrew of the polarization controller, while also periodically pivoting the centre body. Initially, pivoting will not affect the laser cw power, but as the pressure plate begins to stress the glass fiber, you 'll begin to see that the cw power is affected by the orientation of the pivoting section, rolling about the axis of the fiber. This is the best indicator of when the glass is being stressed, because even very sensitive fingers will not feel much.
  3. after seeing the beginning of such an effect, one full turn of the pressure-screw is usually all that's needed
  4. it often happens that there is sufficient stress in the fiber (due to being gathered up into loops) to act as a second polarization waveplate. In that case, you will often get modelocking immediately after fiddling with the first polarization controller. It will appear on the oscilloscope as a train of pulses like this at about 25 MHz repetition rate (40 ns pulse-separation).
  5. if adjusting the first polarization controller does not spontaneously produce modelocking, repeat the first step above for the second polarization controller, rightmost and closer to you than the first. When both polarization controllers affect the cw intensity, as they are pivoted, you are within one turn of the thumbscrews of a proper pressure. You have made them sufficiently birefringent.
  6. you will find that the laser will modelock, or run cw, depending on the pivoting of the controllers -- that is, depending on the orientation of the stress-induced waveplate. Make modest, systematic changes to the pivoting of both controllers. On the small chance that this produces nothing, though the cw power rises and falls, try first using less pressure on the controllers (unscrew the thumbscrew slightly). If that fails, try a little more pressure than you had just now, before loosening.
  7. when you have modelocking, turn on the Ando Optical Spectrum Analyzer (OSA), and see what the spectrum looks like. At its best, it can appear as a smooth near-gaussian spectrum, with a width up to 40nm. More typically, it's fine if it's not quite smooth, and not quite gaussian, like this.
  8. starting with this 600mA drive current, and decreasing, repeat your earlier cw measurements of slope efficiency

If you see other sorts of behaviour, you're quite welcome to explore settings of pump power, and settings of the polarization controller. Some very interesting things can happen! One that is studied as a later, more advanced, exercise is shown here. However, you should get the 'standard' behaviour above before proceeding to the next exercise.

Exercise 3: Measuring pulse duration

Start by reading about autocorrelation as a way to measure pulse duration for ultrafast laser pulses. The autocorrelator we have in the lab is described in detail in Using the Interferometric Autocorrelator, which you should also read.

  1. use the moving-mirror configuration first
  2. find the pulse duration of the pulses of any modelocked pulse you can output; use only the standard ~1m orange MetroCor fiber on the output, to start with
  3. while monitoring the autocorrelation 'live' in real time, make changes to the polarization controllers, and see what happens to the output as you adjust them, particularly as you roll them over until the modelocking stops. Note any changes in pulsewidth, shape, or stability
  4. click the RUN/STOP button and estimate the autocorrelation width; from this find the pulse duration

If you see other sorts of behaviour, you're quite welcome to explore settings of pump power, and settings of the polarization controller. Some very interesting things can happen! One that is studied as a later, more advanced, exercise is shown here. However, you should get the 'standard' behaviour above before proceeding to the next exercise.

Exercise 4: Measuring fiber dispersion

As in the acoustic-waveguide experiment, pulses will disperse in fibers, stretching longer and chirping as they propagate. Use different lengths of fiber, and measure the pulse duration at the output to figure out what the dispersion of the fiber is. Be careful and gentle with the fiber cables, since many of them are hand-made and all of them are fragile with respect to scratches, crushing forces, and too-tight loops (no bends tighter than 50 mm diameter please!)

  1. use the moving-mirror configuration first
  2. set the fiber laser up for stable operation with fairly short output pulse durations
  3. keep the 1m orange (MetroCor) fiber cable still attached to the autocorrelator input, and add different lengths of different fibers between the fiber laser output connector and the 'permanent 1m orange fiber cable. The signals may change on the autocorrelator, but if you do not change the 1m orange fiber to the autocorrelator, you will not need to adjust the autocorrelator alignment (and it won't help)
  4. after making several relatively quick assessments, and a plan for your measurements, switch to the long-timebase configuration of the autocorrelator. Make your first measurement the one of using just the original 1m orange MetroCor fiber
  5. find pulse durations as a function of length of fiber, for each fiber; bear in mind that you will measure the intensity autocorrelation, not the fringes here (the fringes show always the coherence time, now!); explain what you see
  6. find the dispersion of each fiber, in units of ps nm^-1 m^-1 (i.e., ps/(nm*m) )

Exercise 5: Advanced experiments

If you've succeeded at the above experiments, you're welcome to try one or more of these advanced experiments. Discuss your goals, and your methods, with a knowledgeable TA or supervising professor, before you go too far. Feel free to contact me (Professor Robin Marjoribanks) to discuss any special questions.

  1. effects of variable gain on pulse duration
  2. can the right fiber recompress a pulse that has been stretched by the gain above, to make a much more powerful femtosecond pulse?
  3. multiple-pulse output of the fiber-laser: conditions for causing; effects on spectrum; effects on autocorrelation
  4. encoding time-information (e.g., semiconductor reflectivity) on a frequency-chirped pulse and reading it out with a spectrograph (reference)
  5. nonlinear optics

Supporting pages

Ultrafast fiber laser components

Instruments & tools

Principles of autocorrelation

Using the autocorrelator

Interactive software and illustrations

Optics/Photonics Lab links

Reference Texts

Scientific Paper References

Instructor's links

Laser Safety

Laser sites, including laser safety information

Central Laser Facility of the Rutherford Appleton Laboratory (UK)

Lawrence Livermore National Laboratory

Laboratory for Laser Energetics (University of Rochester)

Last revised: 16 March 2003 -- rsm