How to calibrate a sensor: methods of calibration and how often do sensors need calibration

Sensor calibration is an adjustment of the sensor to perform as accurately and consistently as possible. When a sensor needs to provide readings in standard units – there needs to be a Standard Reference to calibrate the sensor against. Let us explore this topic deeper.

Calibration is performed on a measurement instrument to confirm its accuracy and precision, in other words, to verify the dependability of the instrument. The calibration of measurement tools – sensors is the most important precondition for the reliability of the values it provides, thus the cornerstone of quality control.

There are several types, methods or general truths to know of sensor calibration, depending on the type of sensor being used. Some common ones are:

1. Linearity calibration: This type of calibration is used to measure the linearity of a sensor over its full range of measurement.

2. Span calibration: used to determine the full range of measurement of a sensor.

3. Zero calibration: used to determine the zero point or offset of a sensor.

4. Sensitivity calibration: used to determine the sensitivity of a sensor.

5. Temperature calibration: measure the effect of temperature on a sensor's perfor­mance.

6. Hysteresis calibration: This type of calibration is used to measure the hysteresis of a sensor, which is the difference in output at a given input when the input is approached from different directions.

7. Non-linearity calibration: This type of calibration is used to measure the non-linearity of a sensor over its full range of measurement.

8. Repeatability calibration: This type of calibration is used to measure the repeatability of a sensor, which is the degree to which the sensor produces the same output for the same input over time.

9. Stray field calibration: This type of calibration is used to measure the effect of magnetic stray fields on sensor's perfor­mance.

It is important to note that the type of sensor and its application will determine which type of calibration is required.

The calibration standards, required time, and investments:

To put it simply – in order to calibrate a sensor – one needs a reference standard, which is usually another calibrated tool (a sensor, measurement machine, etc.) which will be used to make reference readings of comparison. The already-calibrated sensor needs to be accurate (specifically, more accurate). For this reason there are various standards to be considered such as the standards of National Institute of Standards and Technology (NIST) or ones by the International Organization for Standardization (ISO), which is widely used by many laboratories around the world that have calibration tools.

These, come with the specific references of calibration and the correction factors that may be necessary, and are each custom aligned by the type of the sensor such as the ISO9001:2000 Standard for Quality Management Systems, or the commonly known standard ISO17025:2005 Standard entitled, “General Requirements for the Competence of Calibration and Testing Laboratories”.

Standard physical references are the reasonably accurate physical standards for some types of sensors. For Rangefinders those are the Rulers, Meter sticks; for Temperature Sensors: Boiling Water – 100 °C at sea-level and the triple point of pure water is at 0.01 °C (used to calibrate thermometers); and for Accelerometers standard physical references are Gravity as it is a constant 1G on the surface of the earth.

Characteristic Curve is typical for each sensor, and it showcases the sensor’s response to an input. During the calibration it the response of the sensor is compared to the available “ideal” response.

• Offset is the difference in the output gathered from a sensor compared to the one from an ideal output (meaning the available best output): it can be lower or higher. Single point calibration is considered the easier way to calibrate an offset
• Sensitivity/Slo­pe – The difference in sensor output slope indicates the output change in comparison with the ideal rate. This can be corrected with two point calibration.
• Linearity – In general there are only a few sensors with completely linear characteristic curves. For some, it is no issue, however some require more complex calculations to make the output linear.

One Point Calibration requires a single point for calibration, that can be applied the rest of the way once offset is adjusted. Good examples may be the temperature sensors, control systems that need to keep the same temperature for extended periods of time. These sensors are linear, and within certain measurement ranges have the correct slope.

Two Point Calibration is a bit more complex as it re-scales a sensor output against two points instead of one. A simple example is calibrating the temperature sensor through an ice water bath (0.01°C) and boiling water (100°C at sea-level). Two Point calibration helps to correct offset as well as slope, and can be used when the output of the sensor is known to be linear (Reference value is reasonably linear in difference with the actual response, meaning on the places where it should be higher, it is lawyer and vice versa).

Multi-Point calibration is the method that usually requires the most time and gives the best results. Occasionally, transducers will have inconsistency in linearity throughout the range. This can cause errors in a variety of points through the range. From three to eleven reference points could be used. To achieve the currently available best accuracy, in some cases curve-fitting is performed.

Recalibration: Why is it necessary?

Even the most widely known and commercially available accurate sensors with the most sensitive measurement showed some deterioration through wear, environmental physical damages, etc. Very few sensors come with proper calibration instructions, even though some standards clearly state the need for them. For example the DIN EN ISO 9001 specifies the need for calibration of the measurement equipment, and a certain process for the monitoring of such equipment should be set in place. A calibration in an accredited laboratory means that the calibration took place according to the DIN EN ISO/IEC 17025, and provides clarifications on measurement uncertainties if such shall occur. These recordings are also necessary for traceability of the calibration.

How frequently a sensor needs calibration depends on the type of the sensor, sometimes even the certain use case (nature of the application, required accuracy, the environmental details around the system, etc.). This topic is important to have under the attention as even some sensors of the same manufacturer, of the same type could have different stability of measurements over the time.

What should a calibration certificate include?

• Designation of the test specimen;
• clarification of responsible-entities’ competence;
• reference standards, traceability and documentation stating these;
• ambient conditions;
• measurement result which includes the uncertainty of the measurements;
• manuals (documentation) of the conducted procedure;
• dates to make sure the proper calibration intervals are documented.

The Client’s Point of View

The above-mentioned has mostly covered the industry basics for the topic, to ensure the safety regulations connected to the calibration, as well as to increase the general awareness of the time and effort required to keep a sensor calibrated, and in a good working condition. At the end of the day, the end user and a major beneficiary of the sensor is the company that integrates the sensor into a product. Be it the integration of the sensor into an everyday IoT device or a traditional sensor in an industrial, manufacturing environment – the developer wants the most reliable and accurate sensor possible, that matches the industry standards, and surely is as economical as can be. Usually the R&D departments are responsible for finding these sensors or the companies that can custom-develop those for them. However, which is the most accurate sensor, and how is it calibrated?

As we clarified above – a calibration is conducted with reference readings from a sensor, a measurement device or generally known standards that is at least one order more accurate. As the industry standards are aligned with these devices, it is close to impossible to prove that a sensor is more accurate or even as accurate as the most accurate known measurement system.

An example of this is one that RVmagnetics has faced over the years of experience in the sensor manufacturing, R&D and measurement system prototyping industries.

RVmagnetics representatives mentioned: “When a client requires proof of accuracy of 0.01 °C, as for example PT100 sensors, there is a necessity of another, one order more accurate sensor than our own MicroWires to be able to prove the 0.01 °C accuracy of a MicroWire, which currently does not exist”.

RVmagnetics manufactures exceptionally accurate sensors and custom-develops measurement solutions based on their own, smallest passive sensor in the world, for temperature, pressure, position, vibrations and other physical quantities. A standard calibration process is quite a fixed solution, and the MicroWire sensor has proven to be more accurate, and sensitive than lots of conventionally available systems and just as reliable as the standardized certifications can currently request.