FBG Thermal Test Results

To prove the functional performance of the designed interrogator, two strings with eight FBGs are connected to the two channels of the interrogator. The FBGs were drawn-tower gratings with an average reflectivity of 30% and an FWHM of approximately 300 pm [71]. The setup for this test is illustrated in figure 6.2.

Figure 6.2: Thermal test setup for measuring two FBG strings with the designed inter-rogator module.

The temperature profile has a minimum temperature of -20^◦C increasing in steps of 5^◦C to a maximum temperature of +65^◦C. The measurement period was set to one minute at which the complete reflection spectrum of both strings was queried. After the test, the measurement data-set was reassigned to the single FBG wavelengths. The measure-ment result of the FBG no. 1 of channel 1 with a central wavelength of 1528.159 nm is illustrated in figure 6.3. Three different types of fit-algorithms were applied to the

data, namely the coupled-mode fit (CM), the Gaussian fit (GS) and the Trip-Hop (TH) algorithm to find the FBG’s central wavelength. Taking into account the sensitivity of the Bragg wavelength shift, estimated in section 6.1.1 to 11.2 pm/^◦C, a conclusion to the sensor’s temperature can be made. In the inlet on top right of figure 6.3 the temperature computed with the three different algorithms is shown. It can be seen that the Trip-Hop algorithm shows a higher deviation for this data set around a testing time of 9 hours and 30 minutes for a setpoint temperature of 20^◦C. For the temperature setpoints of 25^◦C and 15^◦C all three algorithms came to the same results within an error boundary of 0.2^◦C.

5 1 0 1 5 2 0 2 5 3 0 3 5 4 0

Figure 6.3: Temperature measured with the FBG sensor and the designed interrogator during the thermal test. Three different fits are applied to the spectral data. Inlet: Zoom into timeframe between 9h:00s and 9h:30s, the Trip-Hop shows deviations.

For further investigation, the error of the Gaussian fit and the Trip-Hot algorithm with respect to the coupled-mode fit is plotted in figure 6.4 at the bottom. The red line illustrated the error of the trip-hop algorithm whereas the black line shows the error for the Gaussian fit. The data of the temperature profile correlates well with the larger error of the trip-hop algorithm, reaching nearly±1^◦C. For the Gaussian fit no excessive errors were observed, the maximum error stayed within a boundary of ±0.4^◦C.

To investigate the origin of the higher deviation for the trip-hop algorithm, a more detailed look into the spectral data was taken. The reflection spectrum for a temperature of 25^◦C at a runtime of 9 hours is depicted in figure 6.5 on the left, whereas the spectrum for 20^◦C and a runtime of 9 hours and 30 minutes is shown on the right. Here a small dip within the right slope of the spectrum can be seen. As described in figure 3.48 the trip-hop algorithm uses the data on the left and right slope of the spectrum, so corrupt data will led to a non-ideal central wavelength estimation. The spectral impurity comes from a discrepancy between the laser set-point wavelength and the real output wavelength. Note,

5 1 0 1 5 2 0 2 5 3 0 3 5 4 0

Figure 6.4: Temperature error with respect to the coupled-mode fit for the trip-hop algo-rithm (red line) and Gaussian fit (black line) respectively.

up to this test no wavelength referencing system was introduced within the measurement system. Laser instabilities (a deviation ±10pm is enough) at a certain wavelength leads to a result like shown in figure 6.5 on the right.

1 5 2 7 , 5 1 5 2 8 , 0 1 5 2 8 , 5 1 5 2 9 , 0

Figure 6.5: Left: Spectral data for stable wavelength of the laser. Right: Laser instability for lower control voltages generates impurities in the spectrum.

Having a further look into the used LUT for the wavelengths around 1528 nm, the control parameters can be found to be 0.98 V/0.17 V/2.48 V for right-, left- reflector and phase control values. It has been proved that for lower control voltages, as it is the case for the left reflector with 0.17 V, the current noise of the used voltage-to-current converter circuit influences the stability of the laser’s wavelength dramatically. To overcome this problem, a detailed investigation addressed to current noise was carried out leading to an improved design of the converter circuit. The previous current controller was replaced by

a low noise circuit design as illustrated in section 5.1.1. Afterwards a new thermal test of an FBG sensor was carried out, obtaining the results given by figure 6.6. The estimated Bragg wavelength is plotted in blue color, following the temperature profile measured by a PT1000 temperature sensor. The small overshoots have their origin in the non-optimum settings of the used temperature controller during the test. A zoomed view into three cycles in figure 6.6 is given in figure 6.7. It can be clearly seen that the estimated Bragg wavelength even follows the overshoots in temperature.

15:15:32 15:57:12 16:38:52 17:20:32 18:02:12 18:43:52 19:25:32 20:07:12 20:48:52 21:30:32 22:12:12 22:53:52 23:35:32 00:17:12

-20 0 20 40 60

T e m p e r a t u r e F B G W a v e l e n g t h

T i m e

Measured Temperature [°C]

1531.0 1531.5 1532.0 1532.5 1533.0 1533.5

FBG Wavelength [nm]

Figure 6.6: Thermal cycle of a single FBG sensor illustrating the temperature (red) and the estimated FBG peak wavelength (blue).

Zooming into the estimated Bragg wavelength for one temperature step, a peak to peak wavelength error of 6.5 pm can be obtained. This corresponds to a temperature error of approximately 0.58 ^◦C for the above given sensitivity of 11.2 pm/^◦C. This is illustrated in figure 6.8 in more detail.

For the measurements here, the polarization effects, as discussed in section 3.4.4, are not taken into account. This was mainly because the working mitigation strategy as discussed in section 4.3.2 was implemented later in the interrogator design. For the used UV written gratings [71] the Bragg wavelength error was estimated to be in the range of 10 pm and it was assumed that the test of the functional performance will not be limited by this effect. As long as the full setup is static, meaning that the fibers are not moved or additional fiber lengths between sensors are introduced, as long the error introduced by birefringence stays constant.

17:20:32 17:28:52 17:37:12 17:45:32 17:53:52 18:02:12 18:10:32 18:18:52 18:27:12 18:35:32 18:43:52 30

35 40 45 50 55 60

T e m p e r a t u r e F B G W a v e l e n g t h

T i m e

Measured Temperature [°C]

1532.25 1532.30 1532.35 1532.40 1532.45 1532.50 1532.55 1532.60 1532.65 1532.70 1532.75 1532.80 1532.85 1532.90 1532.95 1533.00

FBG Wavelength [nm]

Figure 6.7: Zoom into three temperature steps from the diagram in figure 6.6.

Figure 6.8: Zoom into estimated Bragg wavelength for one temperature step showing a peak to peak wavelength deviation of 6.5pm.