Design and Construction of the FP-FBG

The FP-FBG is made by the Grupo Comunicaciones Opticas (GCO) at the Universidad Politecnica de Valencia (UPV) in Spain. The design target was to have a FBG with a FSR of 10 GHz and a contrast ratio between 3 and 10 dB. Two gratings separated by a certain cavity length are written in a 62.5µm core multimode fiber. A 62.5µm core fiber was used because that fiber was available in the GCO lab. The grating separation is chosen such, that the desired FSR of about 10 GHz is obtained. Before writing, the FBG was put in to water (hydrogen) for two weeks to improve the penetration of the ultraviolet light into the glass. The writing was done just before the water has diffused out of the fiber again.

The fiber was first characterized at 1500nm because measuring equipment was available at 1500nm only. After that the design parameters were converted to the 1300nm window.

Then the gratings were written ’blindly’. We have spliced 62.5µm core multimode fiber pigtails to the bare fiber ends of the FP-FBG. A drawing of the final FP-FBG is shown in Figure 4.7.

4.2. FABRY-PEROT FIBER BRAGG GRATING 41

Figure 4.7: Fabry-Perot Fiber Bragg Grating made in 62.5 µm core multimode fiber.

4.2.2 Measurement Results

The FP-FBG was measured with a temporary high-resolution OSA (Advantest Q8384 with resolution of 10pm and singlemode input). The measuring setup is shown in Figure 4.8. The SOA output, functioning as a broadband source (of which the spectrum from

Figure 4.8: Measuring setup for the Fabry-Perot Fiber Bragg Grating

1318 to 1320 nm is used during the measurements and Pout = 33.1 dBm), is connected by using a singlemode fiber, to the 62.5µm core FBG. The output of the FBG is connected to the OSA. The measurement results can be found in Figure 4.9. The measured FBG has a maximum contrast ratio of about 5.5 dB and a loss of about 3 dB. The contrast ratio of C = 5.5 dB corresponds to a effective mirror reflectance of about R = 30%, according to Equation (4.3) and (4.6). Secondly, the contrast ratio is not constant over the operating wavelength range of the FBG. Table 4.5 shows the wavelength of the transmission peaks and the wavelength separation between the peaks. The peak spacing is not constant over the FBG which can have consequences to the performance of the RoPOF system, but is quite near the design targets of 10 GHz.

λ ∆λ ∆f

[nm] [nm] [GHz]

1318.666 0.060 10.35 1318.726 0.060 10.35 1318.784 0.058 10.00 1318.846 0.062 10.69 1318.906 0.060 10.35 1318.968 0.062 10.69 1319.030 0.062 10.69

Table 4.5: Spacing of the transmission peaks of the FP-FBG

Figure 4.9: Measurement of the FP-FBG filter

We wanted to see the performance of the FBG in combination with the POF link as well, but that was not possible due to the large modal noise introduced at the interface of the 62.5µm core multimode fiber and the singlemode input of the OSA. The OSA needs to be extended with a MMF input, which will be obtained in the near future.

4.3 Conclusions

• The 400 µm thick InP wafer has a effective reflectivity of R=17% when measured with singlemode fiber.

• The flatness of the 3mm thick InP wafer is not sufficient. About λ/100 is needed.

• Due to the divergence of the nearly parallel beam, the pinhole finesse degrades as well as degrading of finesse due to ’walk-off’. This is partly compensated by placing the fiber further away from the collimating lens its focal point because the beam divergence is then decreased.

• The pinhole finesse is increased by tapering the POF fiber.

• The calculated pinhole finesse is better in our solid FP wafer because the high refractive index of the wafer will reduce the beam divergence in the FP itself due to refraction.

• The FSRs of the FP wafer filter are quite equally spaced.

4.3. CONCLUSIONS 43

• The FSRs of the FP-FBG are not equally spaced.

• The contrast ratio of the FP is constant over a broad wavelength range.

• The contrast ratio of the FP-FBG is not constant over the operating wavelength.

Recommendations

• The pinhole finesse and ’walk-off’ can be improved by using a collimating lens with a larger focal length (but that requires a larger FP aperture). Or the pinhole finesse can be improved by making the input fiber core diameter smaller (e.g. by tapering).

Or the refractive index of the medium between the FP mirrors (such in our solid wafer FP) can be increased to increase the refraction and thus the divergence of the light beam.

• Due to the need for reduction of pinhole finesse when a POF core is used as emitting fiber, the use of lenses with a large focal length is recommended. Fabricate the 3mm thick wafer with sufficient optical flatness and use in combination with a lens with a focal length of 7.5mm.

• To obtain a FP based receiver with a finesse of 4.5 (about 10 dB contrast ratio), if the pinhole finesse is the limiting factor (and thus equal to 4.5), together with the 120 µm POF, the focal length of the collimating lens should be 13.6 mm if the emitting fiber is in the focal point of the collimating lens (worst-case situation). To let the pinhole finesse be the limiting factor, the reflectivity finesse should be larger than 20, which means that a reflectivity of 98 % is needed. These figures are for an air-spaced FP.

• Investigate the requirement for the optical periodic filter’s optimum contrast ratio needed in the system with the system parameters as used in the laboratory.

• The FP-FBG is very promising but the FSRs should be made more equal, the con-trast ratio more constant over the operating range and the operation range could be increased.

Chapter 5

Photodetector and Microwave Bandpass Filter

After the light passes the optical periodic filter, the light is coupled into a photodetector (Figure 1.6). The photodetector converts the optical signal to an electrical signal. From the electrical spectrum, the desired harmonic component is selected by a bandpass filter.

In this chapter, the problems and solutions regarding coupling light to the photodetector are discussed as well as the choice and implementation of the microwave bandpass filter.

5.1 Photodetector

A photodetector converts optical power to electrical current. The conversion from optical power to electrical current is according to Equation (5.1).

I_electrical = η · P_optical (5.1)

in which η is the responsivity in [A/W]; a measure for the ability to convert optical power to electrical current.

If a resistor is added, a voltage appears due to the photodetector’s imposed current and an electrical power is generated. The optical power is converted to electrical voltage according to Equation (5.2).

V_electrical = Cgain· P_optical (5.2)

in which Cgain is the conversion gain in [V/W]; a measure for the ability to convert optical power to electrical voltage.

The conversion gain and the responsivity are related with the load resistor R by Equa-tion (5.3)[19].

Cgain= η · R

2 (5.3)

The factor 1/2 is introduced because in a practical situation, signals are measured by using equipment with an internal 50 Ω resistor, which is the same as the impedance of the photodetector, to prevent reflections (which do not occur if the source and load resistance are equal). Thus, basically we have two resistors in parallel.

45

Noise generated by the photodetector is determined by the noise-equivalent power (NEP) which is a figure of merit for the weakest optical signal that can be detected. The NEP is the root-mean-square (rms) of the optical power that will produce a signal-to-noise ratio (SNR) of 1 in a 1 Hz bandwidth [19]. The minimum optical power that can be detected depends therefore on the measurement bandwidth by Equation (5.4).

P_opt,min = NEP ·√

B [rms] (5.4)

in which B is the measuring bandwidth. The total received noise power can therefore be greatly reduced by using an electrical band-pass filter. The minimum optical noise power translates into an output voltage according to Equation (5.5)

V_min = P_opt,min· C_gain [rms] (5.5)

The peak-to-peak voltage is approximately six times higher than the root-mean-square voltage for Gaussian noise according to Equation (5.6)

Vmin,pp≈ 6 · V_min,rms (5.6)

The bandwidth of the photodetector should be at least equal to the frequency of highest desired microwave carrier.

A high-bandwidth photodetector requires a small active area to overcome the band-width limits due to the parasitic capacitances between, and resistances in the active area and the bulk material. A small active area makes coupling of light from a fiber to the active area very difficult, unless a fiber with a small core is used. Usually singlemode fibers are required in high-speed data communication links. Singlemode fibers have a small core which makes the coupling of light to the active area easy.

We have found in Chapter 2 and 3 that spatial filtering in a multimode link will in-troduce modal noise. These optical intensity fluctuations will inin-troduce strong electrical current fluctuations when the light is converted to an electrical signal by the photode-tector. So, we can conclude that, for our application, coupling the POF to a singlemode pigtailed photodetector (which is widely available in the market) will introduce unaccept-able electrical signal intensity fluctuations. It is thus necessary to obtain a photodetector with an optical input where all light can be coupled to a large active area in order to eliminate spatial filtering and thus modal noise.

For multimode applications, the need for high-bandwidth multimode photodetectors is not yet trivial because multimode fibers (large core fibers) are not used in high-speed transmission links. The availability of high-bandwidth multimode photodetectors will increase because mostly of the increasing data rates in LANs, in which MMF is extensively used as well as the upcoming radio-over-fiber techniques for in-building applications for which the use of multimode fiber is preferable (as described in Chapter 1).

The active area must be large because of the need for coupling most of the light from the fiber to the active area, but the active area must be large as well to obtain a large responsivity. A large active area however decreases the bandwidth. The final requirement for the photodetector is therefore quite contradicting: a high-bandwidth, large active area photodetector. However, after doing a market search (Appendix F) on high-bandwidth large active area photodetectors, surprisingly a suitable photodetector