1. Introduction

Tunable laser sources enable a plethora of applications from spectroscopy [1–4] and sensing [5–7] to optical coherence tomography [8,9] and telecommunications [10–12]. For industrial and commercial applications, monolithic fiber based lasers are of particular interest as they make up an all-closed platform with no risk of misalignment as well as providing excellent beam quality while also allowing for small footprints, by appropriate coiling. These lasers commonly rely on fiber Bragg gratings (FBG) [13] for wavelength selectivity, which are tunable by applying stress [14–16] or changing their temperature [17]. Although easily implemented, these tuning schemes are rather slow and can be challenging when aiming for wide tuning ranges. A novel approach for generating tunable laser operation from an actively mode-locked laser with FBGs was demonstrated by Morton et al. [18] in 1993. In this work, they coupled a linearly chirped fiber Bragg grating (C-FBG) [13], i.e. a FBG with a linearly changing grating period, to a laser diode to form a linear cavity which was then driven by a radiofrequency signal. As the radio frequency was changed the mode-locked operation forced the laser to operate with a different cavity length and hence also with a different center wavelength, as a result of the distributed reflection from the C-FBG. The tuning range achieved in their work was limited to 0.3 nm due to the characteristics of the C-FBG. An adaptation of this approach to an Er-doped fiber laser was demonstrated in 1998 by Li and Chan [19]. Their laser was based on a C-FBG connected to a circulator in a ring laser –a configuration commonly referred to as a sigma cavity [3,20]. The laser was then mode-locked by an electro-optic modulator and allowed for a 7.2 nm tuning range, limited by the reflection bandwidth of the C-FBG, by, just as in [18], changing the modulation frequency. Further efforts to extend the tuning range have for instance been made by Burgoyne and Villeneuve by using multiple C-FBGs [20] and also in recent work employing a highly broadband (${\sim} $50 nm) C-FBG [3]. A different technique relying on an array of discrete fiber Bragg gratings was introduced by Tiess et al. in 2015 [21] and was shown to support a stepped tuning range up to 74 nm in an Yb-doped sigma cavity fiber laser. However, the sigma cavity’s inherent feature of having mismatched repetition rates for different wavelengths makes tuning in applications relying on synchronization with other sources more difficult. A novel cavity layout that solves this issue was demonstrated, with a stepped tuning range of 25 nm, by Tiess et al. in 2017 [22]. The design is based on a ring cavity employing two circulators with the mid-ports connected to the opposite ends of a FBG array, thus making the roundtrip time equal for all wavelengths. The theta-cavity configuration has also been implemented with a C-FBG to generate a lasing bandwidth of 62 nm, which was also discretized into separate lasing peaks by including an etalon [23].

In this work, we present a novel concept for designing and operating a laser with what we refer to as a C-cavity that, just like the theta-cavity, allows for tuning without changing the repetition rate. Our laser has a very simple layout and only consists of three different optical components, namely a semiconductor optical amplifier (SOA), working as both gain material and modulator, an output coupler and a C-FBG.

2. Cavity design and operating principle

 

Fig. 1. (a) Schematic cavity design where SOA, OC and C-FBG denote semiconductor optical amplifier, output coupler and chirped fiber Bragg grating respectively. (b) Electrical pulse sequence driving the SOA.

Download Full Size | PDF

A schematic representation of the presented laser is depicted in Fig. 1(a): The cavity includes a 2 m long C-FBG (DCM-CB ± 650 ps/nm 1535-1565 nm, Proximion) for spectral control, an 80/20 fused fiber splitter used as an output coupler (OC) and a SOA (SOA1117S, Thorlabs) serving as both gain medium and modulator. The SOA is driven by a commercially available circuit board (T165-14 Laser Pulser, Highland Technology), which in turn is driven by an arbitrary waveform generator (Keysight 33600A). Due to the circuit board’s limitations in terms of repetition rate and trigger duration, we added spools of 40 m and 1 km standard single mode fiber to increase the propagation time on each side of the SOA in the cavity. The electrical gating sequence consists of a repeating pulse pair that drives the SOA, as shown in Fig. 1(b). The time intervals T₁ and T₂, respectively highlighted in blue and red, correspond to the time it takes the light to propagate from the SOA to a target section of the C-FBG and back on each side, as indicated by the double arrows in Fig. 1(a). This gating sequence prevents light passing through the C-FBG (i.e. circular propagation) to experience gain. Instead, laser oscillation is established for a pulse that returns to the SOA along the same path it exited. Thus, the light inside the cavity effectively oscillates as it would in a linear cavity, although the components visually seem to form a loop. We therefore refer to this design as a C-cavity, since the letter C resembles our design of a “bent” linear cavity. In this configuration, wavelength tuning is achieved by simply adjusting the values of T₁ and T₂ to different fractions of the total roundtrip time while maintaining a constant repetition rate.

3. Results and discussion

A wide temporal scan of the output and gate dynamics recorded on an oscilloscope (Tektronix TDS3034) is shown in Fig. 2. Note that these traces have been normalized and offset relative to each other for easier comparison. The electrical gate signal was measured from an electrical probe port on the circuit board that drives the SOA while the optical signal was recorded at output 1 with a photodiode (DET01CFC, Thorlabs). As can be seen, only one optical pulse is emitted for each gate pulse pair, which confirms that the operating principle discussed above was achieved.

 

Fig. 2. Wide scan traces of the electrical gate pulses driving the SOA (blue) and the optical pulses recorded at output 1 (red). The time delays from Fig. 1 are also indicated.

Download Full Size | PDF

The resulting tuning performance, measured with an optical spectrum analyzer (ANDO AQ6317B), when running the laser with ${\sim} $2.3 ns gate pulses and an operating current of 480 mA is shown Fig. 3(a), where it is seen that the entire reflection band of the C-FBG was covered. Using a dual-channel pulse generator and phase modulation to adjust the value of T₁, it was possible to scan the central wavelength from 1535-1570 nm. As expected from a linearly C-FBG, the delay versus wavelength shows a linear trend as highlighted by the peak wavelengths, indicated by the red crosses, and the green dashed linear fit. The spectral width was around 3 nm over the tuning range except for at the edges of the C-FBG’s reflection curve. Typical temporal profiles, measured with a 1 GHz photodiode (D400FC, Thorlabs) and a 1 GHz oscilloscope (Tektronix TDS784A), of the pulses from the two different output ports, normalized and overlapped in time for easier comparison, are shown in Fig. 3(b). It is seen that the pulse width from output 2 is slightly longer than from output 1. Both pulses are in the range ∼2 ns, consistent with a physical length ∼20 cm on the grating, which corresponds to ∼3 nm bandwidth. We attribute the slight difference in pulse durations to a slight detuning of the correct delay times T₁ and T₂, which should be controlled with sub-nanosecond precision and was not possible with our current electronics. For correctly set values and negligible ASE contribution, we expect an essentially constant pulse duration across the entire cavity. The output power from output 1 gradually increased from 85 nW to 603 nW when the laser was tuned to longer wavelengths, the corresponding increase from output 2 was 56-331 nW. With a repetition rate of 90 kHz, these output powers correspond to pulse energies between 0.95-6.70 pJ and 0.61-3.68 pJ from output 1 and output 2 respectively. The pulse energies increase with the SOA’s operating current, however this comes at the expense of broader spectra.

 

Fig. 3. Spectral profiles at different delays between the gating pulses are displayed in (a).The delay relative to an initial separation is indicated by the baselines on the y-axis. For visual aid, the red crosses mark the peak wavelength and the green dashed line is a linear fit. Typical temporal profiles of the optical pulses from output 1 and 2 are shown in (b).

Download Full Size | PDF

The fast recovery time of the SOA allowed us to operate the laser in a time-multiplexed regime, generating either a single wavelength or multiple wavelengths of choice per roundtrip. Figure 4 illustrates the laser output when sustaining four different wavelengths. The top inset shows that two gate pulses at $\textrm{T}_1^\textrm{a}$ and $\textrm{T}_2^\textrm{a}$ (blue) give rise to one laser pulse (red). This takes place four times (a, b, c, d) over one gating sequence, which is repeated. The amplitude variations in the top inset is attributed to an insufficient number of sampling points in the oscilloscope. It is worth emphasizing that the relative delay between these pulses can be preset and will remain fixed as they all share the same repetition rate. This operating regime could be interesting for pump-probe experiments as well as for measuring distances between gratings in FBG arrays and could also be used to boost the average output power of the laser.

 

Fig. 4. Running the laser with four pairs of gate pulses with different delays enables multi-wavelength operation. Dotted lines were added for visual aid in identifying the four individual spectra. The top inset shows the corresponding time traces, where it can be seen that four pairs of gate pulses result in the generation of four optical pulses.

Download Full Size | PDF

 

Fig. 5. Spectral FWHM for different optical pulse durations for the C-cavity (blue crosses) and a sigma cavity, using the same C-FBG, with short/long wavelengths reflected first (red/green circles). Schematic representations of the two cavity designs are also included in the vicinity of the respective data sets.

Download Full Size | PDF

The dependence of the spectral full-width at half maximum (FWHM) of the laser was studied as a function of the optical pulse duration, shown by the blue crosses in Fig. 5. For reference, we also investigated the spectral FWHM versus optical pulse duration when having the same C-FBG in a sigma cavity, using the same SOA and output coupler together with a circulator [3]. Schematic illustrations of the two cavities are shown with the corresponding data in Fig. 5. Red and green data points correspond to having short and long wavelengths reflected first, respectively. A ${\sim} $7 times difference in the linewidths associated with the two cavity configurations is seen in Fig. 5. We attribute this to the fact that wavelengths are spread in one direction in the sigma cavity, which allows for spectral narrowing over consecutive roundtrips, while the wavelengths are spread in opposite directions at every roundtrip in the C-cavity. The spectra from the C-cavity can in principle be made narrower by offsetting the gating time intervals T₁ and T₂ from their ideal values where they properly overlap with the optical pulses returning to the SOA after reflection from the C-FBG. This creates effectively shorter openings as the gate is closed for a part of the optical pulses reaching/leaving the SOA. However, this comes at a price of an increased level of ASE that quickly gives rise to a non-negligible tail in the spectra. Furthermore, the spectral shape also tends to vary to a greater extent when tuning the wavelength for a set offset in this operating regime. Another intuitive approach to narrow the spectra is to slightly detune the arrival time between consecutive gate pulse pairs, such that they effectively shorten the circulating pulses. However, we found that the effective narrowing was very marginal and that this way of operating the laser could lead to strong temporal pulse shape fluctuations. Stable operation of the C-cavity thus inherently supports broader spectra than a corresponding sigma cavity.

It is also observed in Fig. 5 that longer optical pulse durations are associated with broader spectra. In recent work with a SOA-based sigma cavity [3], it was noted that the spectra predominantly broadened towards longer wavelengths for increased pulse durations. We verified that this tendency is observed regardless of the C-FBG orientation. Besides, this is also present in the C-cavity, as illustrated by the spectra in Fig. 6. We attribute this behavior to the gain dynamics in the SOA in conjunction with the use of a C-FBG. Rapid thermalization in the SOA replenishes the population of the lasing level at the expense of short wavelength emission. In contrast, the energy levels corresponding to long wavelengths can experience gain. The C-FBG allows the emitted long wavelengths to make it back to the SOA and experience further gain and thus contribute to the output spectrum.

 

Fig. 6. Logarithmic spectra of the laser output for different pulse durations, the corresponding linear spectra are shown in the inset.

Download Full Size | PDF

In order to investigate the importance of the feedback provided by the C-FBG on the broad spectra in Fig. 6, we conducted experiments where the C-FBG was replaced by two discrete non-chirped FBGs. The pulse durations were varied from 3 to 39 ns but no spectral changes were observed. An example of the ability of maintaining narrow spectra for long gate pulses in this configuration is shown in Fig. 7. These FBGs were also used in a sigma cavity for comparison, and the greatest difference was a minor increase in linewidth of 0.04 nm in the C-cavity from the corresponding sigma cavity linewidth of 0.10 nm. From these results, it is clear that the spectral broadening in Fig. 6 depends on the feedback from the C-FBG.

 

Fig. 7. Output spectra, not taken simultaneously, of the laser operated with the two non-chirped FBGs, the corresponding temporal profiles are given in the inset.

Download Full Size | PDF

4. Conclusion

We have presented a simple C-cavity fiber laser which allows for wavelength tuning by adjusting the temporal separation between the gate pulses driving the SOA. We demonstrated a 35 nm tuning range, which was limited by the C-FBG’s reflection bandwidth, as well as time-multiplexed multi-wavelength lasing. The spectrum of the laser is unusually broad and smooth for a nanosecond laser pulse. This laser could find use in applications where speckle is to be avoided, for example in imaging through scattering media.