Signal drift of oxygen optical sensors. Part II: “Smart” drift correction algorithm and its experimental check with a light intensity detection based sensor

The drift nature associated to oxygen optical sensors was used to mathematically setup a “smart” iterative algorithm allowing the accurate determination of oxygen percentage present in gas mixtures even with the light intensity detection based sensor usually considered unsuitable for that measurement owing to intrinsic weakness of the detection system. Oxidative, photochemical and thermal degradations, acting on both luminophore and polymeric membrane were considered as drift sources. The only requirement for the algorithm correct functioning was the reading of the signal coming from a reference oxygen mixture (air). It allowed the accurate quantification of the nominal %O2 concentration in all 9 days experiment demonstrating to be drift free. On the other hand, the “classical” Stern–Volmer approach, in the same experimental conditions, was useless as it always produced a large positive systematic error increasing with time and with %O2. The proposed algorithm gave sensor accuracy as good as the most expensive phase shift based commercial sensors even in the chosen test conditions producing, on purpose, heavy membrane degradation. Even in those drastic conditions, the mean accuracy was %O2 = 0.11, %RSD always lower than 5% and close to 1% for higher %O2.