CATEGORÍAS TEORÍAS IMPLÍCITAS El saludo, el orden y el Rutina: Son acciones que Saludo Teoría dependiente ,
1.7.4. Análisis categorial y textual
The fibre laser incorporating the semiconductor amplifier modulator of figure 9.1 was
operated in the modelocked regime by applying a radio-frequency component to the drive current of the modulator. Since the resonant frequency of the modulator was known to be a multiple of ~312MHz the only other criterion requiring consideration was the length of the erbium fibre cavity section. By applying a 312MHz RP signal onto a subthreshold DC bias current the position of the 63% reflecting output mirror was translated until a pulsed output was observed. An InGaAs PIN photodetector/sampling oscilloscope arrangement was employed to monitor the output from the laser during the alignment procedure. The duration of the pulses were minimised by adjustments of the electrical drive signals and the relative positions of both cavity mirrors such that the pulses collided within the semiconductor amplifier which enhanced the degree of gain saturation produced. A typical example of the output pulses produced from this laser is illustrated by the oscillograms in figure 9.5.
HBniHflnBflH
Figure 9.5 Output pulses from the modelocked laser at 1^=312.6624MHz
It was evident that the resultant pulse durations were significantly less than the resolution limit of the sampling oscilloscope and so a synchronously operating steak camera system was employed for complementary pulse analysis. In figure 9.6 the intensity profile recorded using the
streak camera is reproduced for typical modelocked pulses from this laser configuration. The measured pulse duration was 6.9ps and the average output power was usually around 4mW (for
approximately lOOmW launched pump power at 532nm). The peak power of the pulses were thus consistently in;the region of 1.4-1,8W. The fact that no subpulsing is evident in this laser is
due to the enhanced peak power of the intracavity pulses which can sufficiently saturate the available gain present in the semiconductor amplifier such that no gain recovery or re-emission is possible.
lOOps
I--- --- -H
T=8.5ps (6.9ps)
Figme 9.6 Streak intensity profiles of modelocked laser pulses.
The laser performance varied slightly with modulation frequency. With negative detuning
the pulse peak diminished until a signal resembling the modulation response of the amplifier was produced. Witli positive frequency detuning the pulsewidth slowly increased until at 312.76MHz (that is lOOkHz above the optimum frequency) the output suddenly behaved as indicated in figure 9.7. Although this output is difficult to interpret from the sampling oscilloscope data, it appears
that the laser is pulsing with a duration somewhat greater than lOOps, but this is also rapidly turning on and off. As no other diagnostic was available to give additional direct insights the following explanation is put forward. When the intracavity pulse steadily phase-lags the detuned modulation signal applied to the semiconductor amplifier a situation arises where the gain thus created has dissipated due to fluorescent emission or CW oscillation. The pulse then experiences loss on propagation through the amplifier and is effectively extinguished. Another pulse can then build up, and will be extinguished after a characteristic time, and the whole process of creation and anniliilation of pulses can then be repeated.
Figure 9.7 Output pulse profile characteristic at ~100kHz positive frequency detuning.
For these results the DC component of the bias signal was kept small, and in some
instances was reduced to zero. It was found that the output pulses were not particularly sensitive
to this parameter, but it became apparent that the best quality pulses were generated when the measured direct current level (as monitored by a milliammeter) was at a specific value. The current indicated by the ammeter gives a measure both of the direct current applied from the power supply and of the rectified component of the RF modulation signal during its negative excursion. For the amplifier chips used in this work this value was between 50-55mA and varied little with modulation frequency or the pumping conditions of the erbium-doped fibre. As the modulation frequency was increased, and the RF power coupled to the amplifier decreased (due to imperfect impedance matching) so the applied DC current had to be increased accordingly to
sustain the current criterion.
The modulation frequency applied to the modulator was then doubled to 625MHz, and again clean output pulses were produced with no evidence of the subpulsing associated with the passive fibre cavity configurations discussed in chapter 3 (see figure 9.8). From streak camera measurements the pulse duration was found to be 8ps as illustrated in figure 9.9. On deconvolution of the temporal resolution of the streak camera system a value of 6.2ps was
obtained. Under these conditions the output power was 3.5mW and the peak pulse power was calculated to be 0.9W. The spectrum of the laser was measured to have a FWHM bandwidth of l.Onm. For the corresponding duration-bandwidth product of -0.8 this implies that appreciable chirp existed across the pulse envelope. This is not too surprising given that the primary pulse
shaping effect is gain saturation within the semiconductor amplifier which causes a frequency chirp due to the carrier density dependant refractive index of the medium. And indeed the AvAt product is typical of conventional modelocked semiconductor laser systems. The peaks on the laser spectrum are separated by ~0.66nm which is commensurate with the value obtained for the Fabry-Perot resonances of the amplifier subcavity. The modulation depth is low due to the use of the angled-ridge amplifier, although weak reflections from the surface of the microlensed fibres would act to exacerbate the spectral modulation.
v > 5 r r V : S C Co S
E l
I t ■ !4
Figure 9.8 Oscillograms of modelocked output pulse sequence at a 625MHz repetition frequency. I--- 1
Inm lOOps
T=8ps (6.2ps)
Figure 9.9 Streak intensity profiles of modelocked laser pulses at a 624.79MHz repetition frequency. (Insert: spectrum of the laser output).
These optimised pulses were again directed onto the InGaAs PIN photodetector and as illustrated in figure 9.10 the pulse duration was sufficiently short that the oscillogram effectively
shows the impulse response of this measurement equipment. Again with a positive modulation frequency detuning, which in this case was llOkHz, a sustained dropout mode of operation could be accessed (see figure 9.11). Two states could be clearly resolved; one being of a single pulse of ~75ps duration and the other was of low intensity with multiple peaks. This observation agrees with the explanation previously proposed, and the clarity of the two states suggests that the transition from one to the other is very fast although no technique was available to further examine this effect.
Figure 9.10 Impulse response of the InGaAs PIN photodetector/sampling oscilloscope arrangement obtained on illumination with optimised laser pulses.
Figure 9.11 ‘Dropout’ mode for 1 lOkHz positive detuning from the optimum frequency
After initial laser alignment and subsequent coupling to the streak camera system an output state as shown in figure 9.12 was frequently observed. This streak record indicates an
output consisting of two pulses of unequal intensity and separated by '-ISOps. In all observations
the trailing pulse intensity never exceeded that of the leading, or main pulse.
loop s lOOps
Figure 9.12 (a) Double pulse output, (b) Optimised output pulse.
The existence of a double pulse output was a feature of the long modelocked erbium fibre lasers investigated by Schlager et al^^^. They found that the two pulses were in orthogonal polarisation states and the time delay was simply due to the polarisation dispersion of the fibre cavity. In most instances the trailing pulse could be extinguished by affecting the polarisation
state within the laser cavity. In very long (>lkm) laser cavities the two pulses were of roughly equal intensity and a single pulse output could not be produced. In the laser detailed in this work the secondary pulse only existed when the cavity alignment was poor, and by optimisation of the pulse duration the trailing pulse could be easily extinguished. This was mainly due to the different gains of the amplifier for the two polarisation states, and also because as the first pulse becomes sufficiently intense the amplifier gain is depleted and any secondary pulses present will be
severely attenuated.