My Bachelor's Thesis about developing a system capable of detecting counterfeit DS18B20 temperature sensors through a combination of electrical and optical analysis methods.
The goal of this research was to develop a system capable of detecting counterfeit DS18B20 temperature sensors through a combination of electrical and optical analysis methods. The investigation involved testing genuine and counterfeit sensors across multiple parameters, including temperature accuracy, conversion time, active and standby currents and ROM pattern. The analysis revealed that while temperature measurements and conversion times did not distinguish between authentic and counterfeit components, standby current was the most reliable indicator for identifying counterfeit sensors. Additionally, optical inspection, although promising for detecting visual markers, was limited by inconsistent text recognition results. The study concludes that a combined approach using ROM pattern analysis, electrical measurements and refined optical inspection techniques holds potential for reliable counterfeit detection. Automating these methods makes detection faster and easier to use, helping to identify counterfeit DS18B20 sensors.
Standby current was indirectly measured by placing a resistor in series with the Vdd pin and
measuring the voltage drop across it. Since the maximum current according to the data-sheet is
very low (1 μA), a large resistor value is required to be able to measure the voltage using the
ADC.
The suspected counterfeit components show a very low standby current with a resistor value of
100 kΩ-250 kΩ, which is around 40 nA. But it suddenly starts drawing a lot of current,
from two times to six times the maximum value, when under a resistance of around 250 kΩ to 1
MΩ.
One of the primary methods used to identify counterfeit DS18B20 sensors was analyzing the
ROM pattern. The ROM pattern for genuine sensors consistently followed the expected
structure of 28-xx-xx-xx-00-00-xx, while counterfeit units deviated from this pattern. There are a
few exceptions. One is when a sensor has the pattern 28-xx-xx-xx-00-00-00. In this case, the
presence of a “C4” marked on the sensor will show that the component is a counterfeit.
A camera system was utilized to take images of the sensors, with the aim of recognizing the "C4" marking via optical inspection. Using OpenCV and Tesseract for text recognition, a series of morphological operations such as top-hat transformation, closing, and thresholding were applied to enhance text contrast.
To create an automated system, integrating the standby current measurements and the ROM reading, a custom printed circuit board (PCB) was created, which can be connected to the Raspberry Pi and also holds the camera in place with a 3D printed bracket.
The current measurements, ROM readings and optical inspection data were collected using Python scripts and displayed in real- time through a GUI built with Tkinter.
