Frequency variation
Since the transponder is designed for communication between 865 and 868 MHz, the highest signal strength of the transponder was expected at this frequency range when the composite is cured. Further, it was expected that the liquid resin would affect the transponder in such a way that the peak of the RSSI would shift as well as the RSSI would be increase during the curing process. This assumption is based on the physical relationships described in Eqs. 1 and 2. Accordingly, the usable energy by the transponder depends, among others, on the electric permittivity of the transmission medium and the wavelength. Due the changing electric permittivity of the resin, the peak of the RSSI should shift. Following Eq. 2, it was expected that the change of the electric permittivity of the resin would also affect the signal transmission and, thus, the RSSI would increase during the curing process. Figure 4 illustrates this expectation. The red arrow marks the expected shift and increase of the RSSI peak.
Figure 5 presents the result of the frequency variation. For reasons of clarity, the measurement values are connected with lines and only frequency variations at intervals of 120 min are displayed. The following findings are obtained from the measurement:
-
The RSSI of the integrated RFID transponder is highest at 800 MHz, both in cured and uncured state.
-
Another peak of the RSSI is around the UHF standard of 865–868 MHz, both in cured and uncured state.
-
The RSSI increases during the curing process.
Surprisingly, the highest RSSI was measured at 800 MHz. However, this is consistent with the result of Bernhard et al. They measured the highest reading range of an in GFRP integrated RFID transponder at 800 MHz. In contrast to Bernhard et al., a second peak at approx. 868 MHz was measured. The expected shift of this peak, as observed by Groth et al. in the GHz range, is not detectable in this frequency range with the used RFID transponder. A reason could be the auto-tune function of the used transponder. This function ensures a high performance of the transponder at 868 MHz in different environments.
So far, it can be concluded that no peak shift occurs with a Protag 3 mini between 800 and 1000 MHz. However, possibly this measurement method can work with another RFID transponder without the auto-tune function in this frequency range. Furthermore, a peak shift may also occur with the used transponder in the range below 800 MHz or above 1000 MHz.
The question remains where exactly the peak is below 800 MHz and whether this peak shifts. Due to technical restrictions, this question remains open.
RSSI measurement
Considering a specific frequency in Fig. 5, for instance 865 MHz, it can be determined that the RSSI increases over the curing time. Furthermore, the graph shows that this increase is over the first 240 min higher than during the time afterwards. Figure 6 illustrates the RSSI measurement at 865 MHz over the curing time (black asterisk) and compares this measurement with the ion viscosity measured by dielectric analysis at 1 Hz (red circle). Furthermore, the blue line displays the core temperature of the material under test. This measurement visualizes the self-heating of the laminate while the curing is starting. The RSSI and the ion viscosity measurements start with a significant slope, which decreases after approx. 300 min and finally levels out at an almost constant level. In this case, the slope of the ion viscosity is almost zero. However, by remembering the typical curve of ion viscosity illustrated in Fig. 1, it becomes clear that the ion viscosity can also level out with a low slope like the RFID measurement in this case. As the laminate heats up during curing and cools down accordingly at this late stage of curing, the slight increase of the RSSI may indicate that the RSSI measurement is influenced by temperature. As this is important for increasing the precision of the RSSI measurement (especially for curing in an oven or an autoclave), the degree to which the RSSI measurement is temperature depended should be investigated in further experiments.
In order to quantify the correlation between the measurement series of the RSSI and the ion viscosity, both measurement series was normalized to values between 0 and 1 and the correlation coefficient was calculated according to Eq. 3. The correlation coefficient of the two measurement series is 0.98. Consequently, the two measurement series possess a very high linear correlation. Hence, the correlation coefficient confirms the optical impression that there is a very high correlation between the two measurements.
A curve fitting was performed and the functions of the two measurement series were determined to examine them in detail. The curve fitting was executed with a Gauss function, a rational function, a polynomial function, and an exponential function. The Gauss function matches the measured values of both functions with the lowest sum of squared errors (SSE). Table 1 presents the general functions considered as well as the SSE and Table 2 details the functions with their variables.
Table 1 Result of curve fitting Table 2 Detailed functions Figure 7 visualizes the measured values, the determined gauss functions, and their first derivation. Thus, Fig. 7 displays:
-
Both measurement series start after the viscosity minimum tCP(2) occurs.
-
Both measurement series initially increase. The slope of increase of the RFID signal is higher than the increase of the DEA measurement.
-
The RFID measurement series shows a turning point after approx. 50 min. The turning point of the DEA measurement series follows at approx. 110 min. The turning point indicates the inflection point tCP(3).
-
The RFID measurement series levels out after approx. 400 min, but still shows a slight slope afterward. The DEA measurement series levels out after approx. 550 min. Subsequently, the slope is almost zero. The levelling out of the slope indicates a cessation of the polymerisation reaction and, thus, the end of curing tCP(4). In both cases, the user can define a specific slope, which can be considered as the end of curing.
A detailed comparison of the two measurement series illustrates their similarity. Both measurement series follow the typical course of the DEA (displayed in Fig. 1). Nevertheless, there are differences in detail. The turning points tCP(3) are about 60 min apart. The decrease in slope and the associated end of curing tCP(4) can also be interpreted differently.
The differences may be due to the different positions of the sensors. Since the material under test was produced in a hand laminate process, the amount of resin remaining at the sensor positions may vary. In addition, self-heating of the resin during curing can cause the resin in the center of the laminate (position of the RFID transponder) to cure slightly faster than at the edges (position of the DEA sensor).
The result confirms the previous results of Veigt et al. [26] and demonstrates the feasibility of the UHF RFID technology for cure monitoring. The low-interference environment inside the measuring chamber leads to an almost ideal measurement curve. This result emphasizes the suitability of the UHF RFID transponder for cure monitoring. The difference of the RSSI measurement to the DEA measurement can be explained with the different position of the DEA sensor and the RFID transponder. Experiments with two DEA sensors prove that these measurement curves can also be shifted slightly to each other [26]. Questions which remain are: To which degree is the RSSI measurement temperature depended, to which degree does the DEA sensor cause interferences and how interferences can be filtered out. The filtering is essential to increase the precision and the reliable of the RSSI measurement in an environment with interference as it was the case in the preliminary work by Veigt et al.