Birefringent Fiber Enhances Fiber Optic Strain Sensor

Breck Hitz

There are multiple solutions to the dilemma posed by a fiber sensor’s response to both temperature and strain (see Wrinkles Improve Fiber Optic Strain Sensor, page 27). For example, another recent experiment has shown that the contrasting responses of a long-period grating and a birefringent fiber can be used to discriminate between a change in temperature and a change in strain.

The scientists who performed the experiment were affiliated with the Portuguese institutions INESC Porto (Instituto de Engenharia de Sistemas e Computradores do Porto), ISEP (Instituto Superior de Engenharia do Porto) and Universidade do Porto, all in Porto; and with the Universidade da Madeira in Funchal.

Conceptually, the scientists designed a fiber sensor that had loss peaks at two wavelengths, one peak caused by a long-period grating and the other by a Mach-Zehnder-like interference in the fiber circuit (Figure 1). When the sensor experienced a temperature increase, one peak moved upward in wavelength, and the other moved downward by a smaller amount. Both peaks exhibited exactly the opposite behavior when the sensor experienced an increase in strain. Thus, by observing the wavelength shift of both loss peaks, the scientists could infer the amount of change in both temperature and strain.

Figure 1. For a strain measuring technique, the sensor head had two spectral loss peaks, one caused by a long-period grating (LPG) and one by a Mach-Zehnder-like interference (M-Z) in the fiber circuit. The two peaks responded differently to the changes in temperature and strain, allowing the scientists to discriminate between the two variables. Images reprinted with permission of IEEE Photonics Technology Letters.

Instead of adding notches to the long-period grating, the scientists spliced a highly birefringent fiber to it, thereby creating the head of their sensor. They placed the sensor head in a fiber-loop mirror together with a polarization controller and illuminated the circuit with a broadband source (Figure 2). Light from the source — centered at 1550 nm, with a 100-nm bandwidth — propagated in both directions around the ring, and was recombined at the 50:50 splitter and detected by an optical spectrum analyzer. Light resonant with the long-period grating was coupled out of the fiber core, resulting in one loss peak observed at the spectrum analyzer.

Figure 2. The fiber sensor head comprised a fiber long-period grating (LPG) and a highly birefringent fiber (HiBi Fiber) in series. The scientists placed the sensor head inside a tubular oven, stretching it between stationary and translatable mounts, so that its temperature and strain could both be independently varied.

The second peak resulted from the asymmetric placement of the fiber polarization controller, which functioned as a λ/2 plate, rotating the polarization of light passing through it by 90°. The counterclockwise-propagating light passed through the birefringent fiber before its polarization was flipped, while the clockwise light did not. Thus, when the two waves were recombined at the 50:50 splitter, they were out of phase by an amount determined by the total birefringence in the birefringent fiber. In essence, the two directions around the ring corresponded to the two arms of an unbalanced Mach-Zehnder interferometer.

Figure 3. The wavelength shift of the two loss peaks was opposite for a change in temperature (left) and a change in strain (right).

The scientists placed their sensor head in a tubular oven and stretched it between a fixed mount and a translatable one. They observed that an increase in temperature moved the long-period-grating loss peak to a shorter wavelength, while simultaneously moving the Mach-Zehnder destructive interference to a longer wavelength (Figure 3a); an increase in strain produced exactly the opposite effect (Figure 3b).

IEEE Photonics Technology Letters, Nov. 15, 2006, pp. 2407-2409.