ELECTRONICS LETTERS 10th October 1996 Vol. 32 No. 21 Dispersion variable fibre Bragg grating using a piezoelectric stack M.M. Ohn, A.T. Alavie, R. Maaskant, M.G. Xu, F. Bilodeau and K.O. H ill Index ing terms: Gratings in fibres, Piezoelectric transducers The authors describe novel dispersion variable fibre Bragg grating device using a segmented piezoelectric stack to combat chromatic dispersion. By applying appropriate voltages to the independently controlled constituent segments, it is shown that the dispersion can be varied from ~8770 to ~940ps/nm with an increase in bandwidth from ~0.03 to ~0.32nm. Introduction: It has been shown that properly apodised linearly chirped fibre Bragg gratings (FBGs) can be used to compensate for chromatic dispersion in long links with installed 1300 nm zero dis- persion standard telecommunications fibre [1]. It has also been dem- onstrated that the same FBG can be used to compensate for different amounts of link dispersion, by varying the degree of linear chirp [2 – 5]. The technique applied by Garthe et al. involved bond- ing the FBG to a cantilevered beam; by deflecting one end, an almost linear strain gradient was established along the grating length. In a variation of this technique, LeBlanc et al. tapered an otherwise uniform cross-section beam, thus producing an increased strain gradient for the same load. Although tunable dispersion was achieved in both cases, the somewhat cumbersome and mechanically intensive loading mechanism of these techniques means that com- pact packaging cannot be used and rapid tuning is not possible. In this Letter we propose a more versatile and compact mecha- nism for tunable grating dispersion. A segmented piezoelectric stack with individually controlled segments was used to control the local strain at 21 positions along the grating [6]. By stepwise monotonic increase or decrease of the strain along the grating, a group delay response close to that of a perfectly linearly chirped FBG was obtained. Tunable dispersion device design: The piezoelectric stack (shown in Fig. 1), consists of a series of cylindrical piezoceramic segments (PZT-5H). These segments, each of which has a pair of electrodes, are physically connected in series to form a 45mm long cylindrical stack. Each segment is 1.4mm in thickness with 0.56mm thick iso- lation segments placed between neighbouring segments. The pie- zoceramic segments are polled in the axial direction such that the application of a voltage bias across the electrodes causes expansion of the segment in the axial direction. This is transferred directly to the FBG which is bonded to the stack in the axial direction. The seg- ments are electrically isolated from each other, and are attached using a bonding material of appropriately chosen stiffness to pro- duce a smooth variation of the axial strain. This smooth variation occurs due to the radial strain coupling between segments across the bond line, and is necessary for producing the required linear group delay response in the FBG. The FBG used was produced using a 248nm UV beam scanned along a uniform pitch phase mask. A 4.5cm long FBG with a measured peak reflectivity of ~30% was pro- duced to match the length of the piezoelectric stack. Dispersion measurement system: Direct group delay measurements were performed using a system originally described in [7]. A tuna- ble Hewlett-Packard (HP) external cavity laser was used as the source, which was then intensity modulated with a Mach-Zehnder interferometer at ~1GHz. The light was then transferred to the tuna- ble FBG dispersion device via an optical circulator and the output measured directly with a fast pin diode. The group delay was meas- ured relative to the short wavelength side of the FBG by comparison of the modulated and detected signal with an HP RF network ana- lyser. R esults and discussion: Fig. 2 shows the grating reflection response in the initial state (A) and with three different stepwise voltage ramps applied to the stack (B, C and D). The spectra were measured with an HP optical spectrum analyser at a resolution of 0.1nm. The cor- responding stepwise segment voltages (A, B, C and D) are repre- sented as a percentage of a 250V full scale (in the inset of Fig. 2). The reflectivity responses were in reasonable agreement with the pre- dicted value of the strain, taking into account the inter-segmental strain coupling and a piezoelectric constant of ~590pm/V. Fig. 3 shows the respective group delay measurements of the chirped FBGs. The group delay data window corresponds to the –1dB reflec- tivity, relative to the centre wavelength reflectivity. As seen in curve A, the measurement for the FBG in the relaxed state shows a small positive chirp, giving a large dispersion value of ~8770ps/nm over a band of 0.03nm on the lower wavelength side. This delay measure- ment and the abnormal FWHM value of 0.175nm for the 4.5cm long FBG (Fig. 2(i)) indicates that a nonlinear pitch distribution was produced during fabrication. Despite this, dispersion tuning to a value as low as ~940ps/nm over a band of 0.32nm was demonstrated by the device on application of ramp D. This results in a wider reflec- tion band with appreciably less grating dispersion, since the FBG has a larger strain gradient imposed on it. Fig. 3 shows that ripples are present on the dispersion curve which have been caused by a deviation from perfect linearity in the applied strain, and the lack of suitable grating strength tapering (apodisation) at the ends of the FBG. Fig. 1 S chematic diagram of 21 segment piezoelectric stack Fig. 2 Grating reflection response (i) Reflection response (state A) without applied strain, (ii) step-wise ramps B, (iii) step-wise ramps C, (iv) step-wise ramps D Fig. 3 Group delay response ● 8768ps/nm – – – – 4480ps/nm —●— 1634ps/nm ——— 942 ps/nm