Microprocessor-based Magnetic Angle Displacement Sensor Temperature Compensation System Design Studies

In modern high precision instrument, the sensor temperature error has become one of the obstacle in enhanceing the sensor performance. The Temperature compensation technology plays an important part in the sensor technology. Magnetosensitive sensor is an important branch of the sensor family. Magnetosensitive sensor can be used to detect the non-electronical characteristics through the magnetic, without contact. This paper studies the temperature characteristic of the magnetosensitive angle sensor that uses spin-valve GMR as the sensitive unit. Design a temperature compensation system.Firstly, the paper introduces the magnetosensitive angle sensor which base on GMR. The introduce include the GMR chip structure; the sensor structure A the the principle of work and the signal processing circuit of the sensor. And then analyzes the temperature characteristic of the magnetosensitive angle sensor, including the GMR chip temperature characteristic, the electric bridge primitive center point output voltage temperature characteristic, the sensitivity temperature characteristic. Obtained the sensor overall temperature characteristic. And establishes the sensor temperature drifting mathematical model base on the microprocessor.Obtains several important conclusions; (1) Under the magnetic field environment, sensor zero output assumes the non-linearity, in certain area it's random distribution. It's difficult to establish one mathematical model to describe the sensor temperature drift. (2)The sensor sensitivity temperature drifts seriously. The temperature drift assumes the linearity, elevates along with the temperature reduces. (3) It's capable to establish the precise mathematical model to the single sensor, but it just can establish the approximate model regarding many sensors.In view of the sensor temperature characteristic, compares two compensate methods that are hardware compensate and software compensate. Then determines the sensor compensation project. The key point of this paper is proposed one kind of software and the hardware unify compensate method. The method used the Fuzzy control theory and the neural network. Microprocessor control the digital potentiometer to compensate the zero and the sensitivty. After the compensation project complementate, carry on the system compensate test. The result is that, after compensates, sensor zero drift 14mV compare with 138mV that before compensates. The zero standard error reduces to 18‰, cmopare with 21% that before compensates. And the sensitivty drift 0.9mv compare with 12.3mV that before compensates. The sensitivity standard error reduces to 13.7‰, compare with 20.5% that before compensates. The compensates effect to be extremely obvious. What is more, the process of the compensates does not need too many manual intervention. When the system trained completely, the sensor could word with compensate automaticly.