Data Drift of Six Axis Force Sensors

When selling six axis force sensors, customers often ask about sensor drift. Some customers believe that the sensor should have an absolute zero value as soon as it is powered on, regardless of how long it is turned on; Some customers can understand that sensors are not idealized devices that can have a nozero value when powered on, but they also have a misconception that they only need to perform a zero reset operation immediately after power on, and then require the sensor’s output to remain zero for a long time during startup.

We fully understand the customer’s needs. Such an idealized product can reduce the difficulty of data processing. But actual sensors always have zero drift and temperature drift. The reason for these drifts is due to the sensor’s self sensing principle. The six axis force sensor products in the industry are all strain gauge sensors. Specifically, when a sensor is subjected to a force, a small deformation corresponding to the force will occur in a certain part of itself. Then, this small deformation is induced into a change in resistance or voltage through a sensitive element. Then the internal circuit of the sensor measures this change and calculates its own force magnitude. So it can be said that a six axis force sensor is a device that senses small deformations. The more sensitive the sensor is to the perception of small deformations, the higher the sensitivity of the sensor, and we are more likely to make the sensor have higher accuracy.

But this actually brings two problems. Firstly, the core technical difficulty of the six axis force sensor is the decoupling solution of the six dimensional force signal. The decoupling solution itself is nonlinear, and this nonlinearity is further exacerbated by the induction nonlinearity of the sensing element itself and the nonlinearity of the temperature response. Therefore, considering that the analytical model of the temperature response sensor is a superposition state of higher-order nonlinearity, this greatly increases the difficulty of understanding. Secondly, temperature can also cause the sensor itself to experience a thermal expansion and contraction effect, which is a small deformation. The sensor cannot be sensitive only to the deformation caused by force, but not to the deformation caused by thermal expansion and contraction.

Since our company was founded, we have attached great importance to the temperature response of sensors. Currently, we will comprehensively utilize hardware temperature compensation including components and circuits, as well as software temperature compensation based on high-order response models, to achieve industry-leading zero drift and temperature drift performance of sensors. Specifically, we can achieve that the main drift of the sensor only occurs in the first half hour after power on, and the overall drift is less than 0.5% of the sensor’s full scale. Therefore, we also recommend that customers power on and preheat for a period of time before using the sensor.

In addition, due to the quasi-static drift characteristic of the sensor, the drift change of the force value is always very slow. So in fact, the vast majority of customer application scenarios can collect sensor data immediately after the sensor is powered on. The difference between the sensor before and after being subjected to force is always accurate after being powered on, and this difference will not cause any drift.

Therefore, we also recommend that customers regularly reset. Such as when the robot moves to the zero position or when there is no force on the equipment during operation, which can eliminate the impact of sensor drift and cumulative errors in equipment operation, which is very beneficial for force measurement.


Subscribe to our newsletter