4.3        New OLCR set-ups

4.3.3       Balanced detection scheme

Fig. 4-9 OLCR set-up with balanced detection

The interference signals (AC) are very small compared to the constant part (DC) and amplitudes between -50 to -120 dB are expected for FBGs. Electronic filtering of the DC signal is not optimal as it adds a lot of noise that reduces the S/N. A balanced detection scheme is then preferred where the DC signal is differentially cancelled. Moreover, the balanced detection scheme within our set-ups (Fig. 4-9) not only suppresses the DC part but doubles the AC part at the same time. In order to discuss such point, Fig. 4-9 presents the principal parts of an OLCR set-up where the reference system has been omitted. A circulator is placed in the source arm and the returning signal in this arm is then redirected to the detector. An attenuator is used on the other arm to compensate the insertion loss of the circulator.

This balanced detection scheme is based on the properties of the coupler. For an input signal of amplitude E, the outgoing symmetric E_s and anti-symmetric E_a signals (Fig. 4-10a) exhibit a phase difference of p/2 for the same OPLD [4-10].

Fig. 4-10 Fiber coupler principle (a) and electric fields pertinent for balanced detection (b)

Considering the reference signal E_r and test signals E_t (Fig. 4-10b), the total signals in the source arm E_s and in the detection arm E_d are given by

where j_1,2 is a phase dependent factor that contains the propagation in the source and detection arm respectively. The corresponding intensities are

We observe that the DC intensity is identical but that the AC intensity part has a p phase factor difference. This means that when I_s shows a constructive interference, I_d shows a destructive interference. The intensity difference is obtained by using the relation e^ip = -1

For a given mirror position, the difference intensity I_diff is the real part of the AC interference signal that would be obtained directly with E_r and E_t for a mirror l/4 away from its current position. Experimentally, this means that the effective mirror position has a constant l/4 offset.

As it concerns this detection, another point must be discuss. The S/N is strongly related to the detector noise that depends on the total optical power of the incoming light [4-11]. In our set-ups, the total light power is reduced in the reference arm by the in- and out-coupling (-50 dB at least) and in the test arm by the FBG itself, allowing by this lower detector noise. Moreover, the chosen balanced detection scheme improves the S/N [4-12]. The observed noise limit in our measurements is about -120 dB for FBGs. This level corresponds to the Rayleigh backscattering in telecom fibers [4-13]. Experiments conducted on cleaved fibers as sample have shown lower S/N several centimeters after the fiber end position around -140 dB.