Abstract

Examination of the sensitivity of a three-wavelength Raman fibre laser to variations in its output-coupler reflectivities shows configuring its power is not significantly more complicated than controlling multiplexed individual lasers, and the source can output relevant power/wavelength combinations.

Introduction

Raman amplification is a useful tool in extending the span lengths and the capacity of a transmission system[1]. An advantage of Raman scattering is that a broad flat gain can be achieved by using pumps at several different wavelengths and optical powers[2,3]. This eliminates the need for expensive and loss inducing gain flattening filters. Multiple-wavelength Raman fibre lasers (MWRFL) are potential Raman pump sources[4-7]. These lasers simultaneously emit depolarised radiation at several wavelengths from a single cavity, as compared to multiplexing together several semiconductor lasers or Raman fibre lasers. Recently a practical three-wavelength MWRFL (3?RFL) was demonstrated that exhibits a 38% 1,100nm to 14xx nm conversion efficiency[6,7]. This device varies the reflectivity of the cavity output coupler (OC) grating by an applied voltage to control the amount of power emitted at a given wavelength. Relative intensity noise (RIN) measurements demonstrated good stability[7]. However, the interdependence of the optical powers at different wavelengths in the single cavity laser suggests that proper power partitioning is a potentially complex problem.

This work shows that for a 3?RFL the required voltage precision for OC reflectivity control is easily obtainable and that power partitioning meets the requirements of different transmission fibres and span lengths. Section "Experimental set-up" describes the experimental arrangement and Section "Operating point stability" demonstrates the voltage control of the source around an operating point (OP). Section "Accessible optical power space" explores accessible power distributions. The results are summarised in the conclusions

Experimental set-up

A 3?RFL is shown schematically in Figure 1. It is pumped by a Yb-doped cladding-pumped fibre laser[8]. The RFL consists of a spool of enhanced Raman gain single mode fibre. Light is shifted from 1,100nm to 1,347nm in the Raman fibre with the efficiency aided by four nested pairs of high reflectors (HR) with fixed reflectivities of approximately 99%. The output grating set also contains a HR for the 1,100nm pump.

Figure 1: Three-wavelength Raman fibre laser

Resonant cavities at 1,428; 1,445 and 1,466nm (?1, ?2, ?3) are created by a set of adjustable-reflectivity voltage-controlled OCs[6,7] and a matching set of broadband HR gratings. The 3?RFL is placed in the experimental set-up shown in Figure 2. Eight C-band equally spaced signal lasers [1,538.2 - 1,560.6nm] are launched down a 100km span of Lucent TrueWave™RS (TWRS) optical fibre. Pump radiation from the 3?RFL is launched in the counter-propagating direction with the aid of the wavelength division multiplexer (WDM). A 2% tap provides radiation for pump power monitoring by the optical spectrum analyser (OSA-2). The amplified signal lasers pass through the WDM to OSA-1 for measurement of the gain and the gain ripple ?G. The OSAs and all voltage controls to the OCs are interfaced to a controlling computer for collection and storage of the gain flattening and power partitioning data.

Figure 2: Experimental arrangement for transmission span gain ripple and power partition measurements

Operating point stability:
Gain ripple and power partition voltage sensitivity

The total optical power and power partitioning were adjusted to achieve a gain ripple minimum at transparency. A ?G of 1.4dB at transparency was achieved at an initial launch pump power OP (P10 , P20 , P30 ) of (283, 194, 144) mW and a voltage OP (V10 , V20 , V30 ) of (2.30, 1.30, 1.05) volts. The stability of this OP was interrogated by fixing V1 and varying V2 and V3 in a range of ±0.5 in increments of 25mV. Figure 3 is a contour plot of the gain ripple at constant V1. The rectangle indicates a region of ±50mV about the initial OP corresponding to 1.1dB < ?G < 1.8dB yielding a voltage sensitivity of ~0.01dB/mV. The gain ripple minimum was 1.13dB.

Figure 3: Gain ripple (in dB) contour map as a function of V2 and V3 about the OP with V1 constant. Minimum gain ripple is 1.1dB. (Box interior indicates ±50mV - Dot shows OP)

Contour plots for constant V2 and constant V3 show similar results with reduced voltage sensitivity. The power space (P1, P2, P3) was explored in the vicinity of the power OP. Deviations from the power OP were quantified by the fractional power partition deviation: ?? = ?(P1 - P10)2 + (P2 - P20)2 + (P3 - P30)2 / P0 where P0 is the total power. Figure 4 shows that voltage excursions of ±50mV generate <4% variations in the 3?RFL spectral power distribution indicating a ???voltage sensitivity of ~0.04%/mV. Data at constant V2 and constant V3 show comparable results with lower voltage sensitivity. Hence, ?G and/or ?? can serve as an error signal in a suitable power partitioning control algorithm.

Figure 4: Fractional power partition deviation ?? (%) as a function of V2 and V3 about the OP with V1 constant. Box interior indicates ±50mV. Dot shows OP

Accessible optical power space

The complete accessible power space was explored by varying the three OC voltages from 0-3V in 0.2V increments. This was done for total output powers of 600; 860; 1,120 and 1,330mW. Figure 5 shows that the power points corresponding to each output power lie on a plane in a Cartesian optical power space confirming that the total power is constant.

Figure 5: 3?RFL optical power planes in 3D optical power space for 600; 860; 1,120 and 1,330mW total powers

The three-dimensional power state can be conveniently represented in two dimensions by the simplex diagram shown in Figure 6. An equilateral triangle is constructed with vertices (x, y) equal to (0, ?3/2), (-1/2, 0) and (0, 1/2), representing the power states (P1, 0, 0), (0, P2, 0) and (0, 0, P3) respectively. This construction can be generalised to an arbitrary power state (P1, P2, P3) according to the following equations:

x = 0 • (P1 / P0) - (1/2) • (P2 / P0) + (1/2) • (P3 / P0)
and
y = (?3/2) • (P1 / P0) - (0) • (P2 / P0) + (0) • (P3 / P0)

The small dots in Figure 6 show the power states spanned by the 3?RFL at a total launch power of 620mW. Data was taken with voltage a resolution of (0.20, 0.20, 0.05) volts for (V1, V2, V3). The open symbols correspond to experimental results for 60, 100 and 140km span lengths of Lucent TWRS™ fibre and the adjacent solid circles are simulation results. Simulation results for similar lengths of Corning SMF-28™ optical fibre are also shown. The experimental data points and simulation data points fall within the laser reach.

Figure 6: Power simplex plot showing accessible power states, experimental OPs and simulated OPs. Simplex coverage is not strongly dependent upon total power

Conclusions

The performance of a 3?RFL with variable OC reflectivities was investigated in 60km, 100km and 140km transmission spans. The data indicates that this device is practical and can provide stable broadband optical amplification. A gain ripple minimum of 1.1dB in a 100km span, in agreement with simulations, is achieved with an OC voltage precision of 25mV. The gain ripple ?G and power partition deviation ?? voltage sensitivities were found to be ~0.01dB/mV and ~0.04%/mV, respectively. This low sensitivity means control of the output powers of a MWRFL is not considerably more complicated than controlling multiplexed individual lasers.

Authors

by M. D. Mermelstein, C. Horn, Z. Huang, M. Luvalle, J.-C. Bouteiller, C. Headley and B. J. Eggleton - OFS, USA
and P. Steinvurzel, K. Feder - Bell Laboratories - Lucent Technologies, USA

References

[1] P.B. Hansen et al., IEEE Photonics Technology letters, Vol. 9 (1997)
[2] K. Rotwitt et al. OFC PD6 (1998)
[3] Y. Emori et al. OFC PD19-1 (1999)
[4] Do Il Chang et al. OFC MA6 (2001).
[5] S. B. Papernyi et al. OFC WDD15 (2001).
[6] M. Mermelstein et al. OFC PD3-1 (2001).
[7] M. Mermelstein et al., IEEE Photonics Technology Letters to be published
[8] S. G. Grubb et al. OAA SA4 (1995).

OFS
2000 Northeast Expressway, B230
Norcross, GA 30071 - USA
Fax: Int'l +1 770 798 3872
E-mail: [email protected]
Website: www.ofsoptics.com

Lucent Technologies
Specialty Fiber Group
25 Schoolhouse Road
Somerset, NJ 08873 - USA
Fax: Int'l +1 732 748 7596
E-mail: [email protected]
Website: www.lucent.com

Bell Laboratories
600 Mountain Avenue
Murray Hill, NJ 07974 - USA
E-mail: [email protected]
Website: www.bell-labs.com