A novel low-cost turbidity sensor for in-situ extraction in TCM using spectral components of transmitted and scattered light

Abstract

Abstract Compared with light intensity measurements of turbid liquids based on direct or lock-in amplification methods, the square excitation method (SEM) can be used to obtained abundant information from both the time- and frequency-domains. Voltage-based turbidimetry in the time domain (The voltage readout here refers to the signal obtained directly by the photodetector followed by an amplifier and is a function of the light intensity attenuated by scattering and absorption effects) is subject to measurement error because of the heterogeneous size distribution of suspended particles, fluctuations of the light source, and measurement noise. In this work, a novel low-cost turbidity sensor was proposed by combining spectral component extraction (SCE) with the transmitted (It) and scattered (Is) light signals and their ratio (It/Is). A near-infrared Light Emitting Diode (NIR-LED) at 860 nm was employed as the light source to reduce the system cost. Two independent preprocessing channels composed of a photodiode, a transimpedance amplifier, and a band-pass filter converted the modulated light to signals of a suitable scale. For conversion of the conditioned analog signals, a 16-bit analog-to-digital converter was used and the signal was further processed using a high-performance microcontroller (STM32F405). The SCE algorithm and turbidity correlation were implemented via the integrated digital-signal-processer combined with Fast Fourier Transform digital algorithm. Averaging of multiple measurements further improved the measurement precision when using the first, third, and fifth spectral components that were related to a known turbidity. The calibration measurements for the sensor revealed that the correlation coefficients from the linear fitting were reliable and satisfactory when using It, Is, and (It/Is). Additionally, validation measurements further demonstrated that the proposed sensor had low measurement error. The maximum measured error was 6% for a turbidity of 10 NTU while the lowest relative error was 0.4% for the 280 NTU solution. The results demonstrate that the sensor has potential for automatic in-situ turbidity measurement applications, particularly for in-situ extraction processes used in Traditional Chinese Medicine.