PHTN1500: Advanced Laser Theory
Mathematical Models of Fiber Lasers (2016F)

Introduction

In this lab, the optical characteristics of an Erbium-Doped Fiber Laser (EDFL) will be examined. Tying together concepts from various courses, studetns will determine key parameters of the laser including saturation and optical loss of laser elements, then apply these to a model for the laser. The Rigrod model will be also adapted and applied to this ring laser.

The key concepts covered in this lab hence include:

Experiment:

In this experiment, models will be applied to an erbium-doped fiber ring laser. This lab will also serve to tie-together some concepts from classes on fiber-optics with those from the laser classes. This simple laser consists of a Nortel 17U EFDA, a JDS AC1100 splitter used as an output coupler (a 50%/50% splitter with exact test specs are available in the lab), and a JDS TB9 tunable grating filter.

The actual laser is built as a ring and lases unidirectionally given the Faraday isolators in the EDFA - while other configurations of fiber lasers are possible (some using a standing-wave configuration with a "traditional" HR and OC), unidirectional lasing means that only a ring configuration is possible in this case. There are, of course, ramifications of this to the calculation of gain saturation as well in the formulation of the Rigrod model for this laser.

Erbium Doped Fiber Laser
Actual laboratory erbium-doped fiber laser setup

Part 1: EDFL Characteristics

Output from the laser will follow, approximately, the gain curve of the amplifier. Connect the laser as per the diagram, set the grating wavelength to 1545nm, and turn the amplifier on. Observe the output power on the OSA - remember to add 15dBm to the observed output reading to compensate for the attenuator which protects the OSA input (which CAN handle amaximum input power of +10dBm (!).

As a reminder, the filter has FC/APC connectors and so connections to that unit are via the green-ended cables only to avoide excessively high losses.

Erbium Doped Fiber Laser
Setup to characterize the erbium-doped fiber ring laser

Now, vary the wavelength in 1nm increments in both directions observing the output power until lasing ceases. The laser will oscillate over a surprisingly large tuning range of about 60nm in total.

Homework: Plot the output from the laser in actual dBm vs. wavelength.

Part 2: Amplifier Saturation Power

In order to formulate a model for the laser (using either the simple or the Rigrod models), several parameters are required including the saturation power of the amplifier which must be measured.

FIRST, connect the output of the diode laser directly to the OSA to determine the exact wavelength and exact power of the largest output mode. Preheat the diode laser and EDFA for at least ten minutes before use (and leave the diode laser source on during this part of the experiment) as it has a TEC and is thermally stabilized. It will be around 1545nm and around -8dBm (but measure it precisely). This diode will exhibit one central mode considerably larger than all others.

Disconnect the input from the amplifier (currently connected to the grating) and connect it to the diode via an attenuator as shown below:

Erbium Doped Fiber Laser - Saturation Power
Setup to measure amplifier small-signal gain and saturation power (the laser diode has been replaced: see text)

As required, disconnect and reconnect cables which go directly to the amplifier input and output (which are PC type with black or white ends): these are behind an interlocked door and are difficult to connect regardless (unless your fingers are _really_ tiny) - be sure they are seated properly and the keyways are aligned before tightening.

The attenuator between the output of the laser and the OSA may be removed WHEN the output power will not exceed levels which will damage the OSA (+10dBm) - check the level with the attenuator installed and remove it if power levels are low enough to allow this.

Install various attenuators between the output apeture of the diode and the input fiber to the amplifier, or combinations of attenuators, to effect losses of approximately 5dB or 6dB steps. For example, addition of a 15dB and a 6dB attenuator to a laser diode output of -8dBm will yield and input power of -29dBm (Assuming -8dBm from the diode, plus another 21dB loss from the attenuators for a power output of -29dBm).

For each step, record the input power (calculated from the attenuators used), and the output power both in dBm. This must be at exacty the same wavelength as the peak recorded from the diode. Use the marker to assess this: In ther MARKER section press the MAIN marker then use the knob to select the required peak - it will be required to "fine tube" the wavelength of the marker to match the output peak for each and every measurement.

Gain may be computed by subtracting the input from the output - at low input powers gain is expected to be quite large (>>30dB) but will fall as input power increases and the amlifier saturates. The input signal range required is from -35dBm (below which the signal-to-noise ratio will be too low for accurate measurement) to about -8dBm (where the output will completely saturate to the maximum amplifier output of +17dBm regardless). In order to assess when a measurement is inaccurate, disconnect the input of the EDFA altogether and observe the output at 1545nm ... this will be the "noise floor" of the system and measurements close to this level will not be accurate!

In the lab, before proceeding: quickly sketch, in your lab notebook, a plot the net gain (in dB) on the y-axis vs. output power (dBm) on the x-axis. Identify the maximum gain (g0) and the -3dB point (Psat) - The maximum gain will be evident as a "flat region" where gain is fairly constant over a range of powers (the gain will be higher than that at extremely small input powers but noise will also be much higher). If these are not readily obvious from the graph, or if anomalous data is present (e.g. a point does not fall within the expected trend) ask the professor and if necessary repeat that portion of the observations (occasionally, for example, a fiber is misaligned when an attenuator is installed ... the sketch will reveal this and the measurement may be repeated).

Homework: Plot the net gain (in dB) vs. output power (dBm) precisely. Identify the maximum gain (g0) and the -3dB point (Psat) and report both parameters in a paragraph below the graph as these parameters are required for the models.

Note that the gain measured is actually net gain (i.e. actual gain minus absorption loss).

Part 3: Measuring gth of the Laser

In order to measure gth, losses of optical elements in the loop must be determined including that of the grating filter. To measure the loss of the grating filter accurately, we will use a single mode from the diode which has a narrow bandwidth and so will accurately reflect the conditions expected in the laser: while the output from the diode might be rated at -8dBm, this includes ALL output modes ... most of which will never pass through the filter in the operating laser!

Use the same wavelength as determined in the previous part of the lab (IMPORTANT: Set the RESOLUTION of the OSA to 0.1nm first).

Erbium Doped Fiber Laser - Grating Loss Measurement
Setup to measure grating loss, and hence gth (the laser diode has been replaced: see text)

The attenuator between the output of the filter and the OSA may be eliminated since the output power will not exceed levels which will damage the OSA.

Now, reconfigure the fiber cables so that the input from the diode passes through the filter and to the OSA (you must use cables with one end PC (the diode and OSA ends) and APC (green) ends at the grating filter. Tune the grating filter carefully (by 0.01nm increments as required) to the single peak from the diode output previously recorded (the wavelength displayed on the grating filter might deviate by as much as 0.5nm from the actual wavelength determined via the OSA). This is a VERY TOUCHY operation and will take some time to do precisely. The wavelength of the displayed peak (as determined using the marker) should be identical to that determined above for the diode alone without the filter in place. Measure the peak power transmitted on that wavelength. Subtract the output from the input (both in dBm) to determine the loss of the grating filter - it will be well under 10dB, if it is not the filter was not tuned properly. Total loss is now that of the grating filter plus the loss from the output coupler (the splitter) which may be read from the datasheet provided in the lab which is specific to this coupler (be sure to record this).

Analysis:

Assignment

Hand In a WORD PROCESSED (not handwritten) lab assignment as follows. Following the submission format for PHTN1400, a title page is first required to identify the lab and the satudent. Next, put each question on a new page and ensure each page has a title "Question 1", "Question 2", etc. Also, please ensure the lab report is in a bound folder for submission (no loose pages, and no binders).

  1. Show a graph of EDFL output (in dBm) vs. wavelength with a graph of typical EDFA GAIN (in dB) vs. wavelength on the same graph (you should assume strong pumping of the amplifier). State, in a paragraph at the end of the question, the exact output power measured at 1543nm.
  2. Show a graph of gain (in dB) vs. output power (in dBm) including two lines to show max gain ("flat") and the -3dB point. Include a paragraph explaining these parameters and summarize the small-signal gain and the saturation power determined in the experiment.
  3. Outline the measurements and calculation of threshold gain, and finally the expected output power using both the simple model and the Rigrod model, both for this specific ring laser configuration. Show all calculations required. Summarize, at the end of the question, the output power expected by each model. (This question is worth considerably more marks than others in this lab)
  4. The small-signal gain measured was, in reality, gain minus absorption loss. Discuss how this affects both the Rigrod model (knowing that the basic Ridgrod model does not take loss into account - i.e. how would you do this regardless ?), and how does it affect the simple model. Obviously one model reflects the observed results better ... hopefully this, in part, explains this.
  5. In Part B you used a single mode from the diode output to measure the saturation power of the amplifier. Does the fact that other modes are allowed to be amplified change the behaviour of the amplifier? Consider the nature of homogeneous/inhomogeneous materials (see section 3.4 of Laser Modeling).