Sensor calibration techniques

Sensors are electronic devices. They are sensitive to the changes in their working
environment. Undesirable and sudden changes in the working environments of the sensors give
undesired output values. Thus, the expected output differs from the measured output. This
comparison between the Expected output and measured output is called Sensor Calibration.
Sensor calibration plays a crucial role in increasing the performance of the sensor. It is used to
measure the Structural errors caused by sensors. The difference between the expected value
and the measured value of the sensor is known as the Structural Error.
Working Principle
Sensor calibration helps in improving the performance and accuracy of the sensors. There
are two well-known processes in which sensor calibration is done by industries. In the first
method companies add an In-house calibration process to their manufacturing unit to perform
individual calibration of the sensors. Here the company also adds necessary hardware to their
design for sensor output correction. By this process, the sensor calibration can be changed to
match the application-specific requirements. But this process increases the time to market.
The alternative of this In-house calibration process, several manufacturing companies
provides sensor packages with a high-quality automotive-grade MEMS sensor along with
complete system-level calibration. In this process, the companies include an on-board digital
circuitry and software to help designers to improve the functionality and performance of the
sensors. To reduce the product design time and component count, digital circuitry such as
voltage regulation and Analog signal filtering techniques are included. To improve the overall
performance and functionality, the on-board processor is provided with sophisticated sensor
fusion algorithms. Some of the sophisticated on-board signal processing algorithms also help in
reducing the manufacturing time enabling the faster time to market.
Standard Reference Method
Here the sensor output is compared with a standard physical reference to know the error
in some sensors. Examples of sensor calibration are rulers and meter sticks, For temperature
sensors- Boiling water at 100C, Triple point of water, For Accelerometers- ”gravity is constant
1G on the surface of the earth”.
Calibration Methods
There are three standard calibration methods used for sensors. They are-
 One point calibration.
 Two-point calibration.
 Multi-Point Curve Fitting.
Before knowing these methods we have to know the concept of Characteristic curve.
Every sensor has a characteristic curve that shows the response of the senor to the given input
value. In the calibration process, this characteristic curve of the sensor is compared with its
ideal linear response.
Some of the terms used with the characteristic curve are-
 Offset – This value tells us whether the sensor output is higher or lower than the ideal
linear response.
  Sensitivity or Slope – This gives the rate of change of sensor output. A difference in
slope shows that the sensor output changes at a different rate than the ideal response.
 Linearity – Not all sensors have a linear characteristic curve over the given
measurement range.
One point calibration is used to correct the sensor offset errors when accurate
measurement of only a single level is required and the sensor is linear. Temperature sensors
are usually one point calibrated.
One-Point-Calibration
Two-point calibration is used to correct both slope and off-set errors. This calibration is
used in the cases when the sensor we know that the sensor output is reasonably linear over a
measurement range. Here two reference values are needed- reference high, reference Low.

Two-Point-Calibration
Multi-point Curve fitting is used for sensors that are not linear over the measurement
range and require some curve-fitting to get the accurate measurements. Multi-point curve fitting
is usually done for thermocouples when used in extremely hot or extremely cold conditions.
For all the above calibration process, the characteristic curves of the sensors are drawn and
compared with the linear response and error is known.

Applications of Sensor Calibration

Sensor Calibration in simple terms can be defined as the comparison between the desired
output and the measured output. These errors can be caused by various reasons. Some of the
errors seen in sensors are errors due to improper zero-reference, errors due to shifts in sensor
range, error due to mechanical damage, etc…Calibration is not similar to adjustment. The
calibration process includes placing the DUT-‘Device under Test’ into configurations whose
inertial input stimuli for the sensor are known, which helps us to determine the actual errors in
the measurements.
The calibration process helps us to determine the following results-
 No error noted on the DUT.
 An error is noted and no adjustment is made.
 An adjustment is made to remove the error and the error is corrected to the desired
level. For sensor calibration sensor models are used. Sensor calibration is applied in
Control systems to monitor and adjust the control processes. Automatic systems also
apply the sensor calibration to get error-free results.
Use of Sensor Calibration
The calibration process is used to increase the performance and functionality of the
system. It helps in reducing errors in the system. A calibrated sensor provides accurate results
and can be used as a reference reading for comparison.
With the increase in the embedded technology and low size of sensors, many sensors are
integrated over a single chip. Undetected errors in one sensor can cause the whole system to
degrade. It is important to calibrate the sensor to get the accurate performance of the
automated systems.

Classification of sensors
There are several classifications of sensors made by different authors and experts. Some
are very simple and some are very complex. The following classification of sensors may already
be used by an expert in the subject but this is a very simple classification of sensors.

In the first classification of the sensors, they are divided in to Active and Passive. Active
Sensors are those which require an external excitation signal or a power signal. Passive Sensors,
on the other hand, do not require any external power signal and directly generates output
response.

The other type of classification is based on the means of detection used in the sensor.
Some of the means of detection are Electric, Biological, Chemical, Radioactive etc.

The next classification is based on conversion phenomenon i.e., the input and the output.
Some of the common conversion phenomena are Photoelectric, Thermoelectric, Electrochemical,
Electromagnetic, Thermo-optic, etc.

The final classification of the sensors are Analog and Digital Sensors. Analog Sensors
produce an analog output i.e., a continuous output signal (usually voltage but sometimes other
quantities like Resistance etc.) with respect to the quantity being measured. Digital Sensors, in
contrast to Analog Sensors, work with discrete or digital data. The data in digital sensors, which
is used for conversion and transmission, is digital in nature. Classification of sensors shown in
figure 6.
PRIMARY CLASSIFICATION OF TRANSDUCERS
 Sensors Output Types

The output signal of a sensor usually goes to an indicator, to a recorder or a controller.

There are a variety of sensors according to the variable to be measured, e.g. pressure,
temperature, level, flow, position (proximity), speed, weight, voltage, current, frequency,
viscosity, resistivity, radiation, pH, electrical conductivity, humidity and others.

When the sensors have digital outputs (only two possible values) are called switches and
when they have analog outputs (more than two possible values) are called transmitters, in this
sense they have pressure switches and pressure transmitters. The sensors have digital and
analog outputs, however sensors also may have serial outputs, which can send digital, analog
signal or both.
1. Digital output
An important aspect to consider is the type of output and voltage levels. In the sensors with
discrete output levels the more common voltages are 24Vdc, 120Vac and 220Vac.
The types of outputs are:
 Relay: The relay outputs handle AC and DC voltages up to 240 volts, the output current
is in the order of amps, 5 A is a typical value. They are used in general purpose
 applications, they are inexpensive. They have a long response time (15ms to open
contacts, 25ms to close the contacts). The outputs of multiple sensors can be wired in
series or parallel. Its main disadvantage is the wear suffered by mechanical contacts and
sizzle.

 Transistor: handle voltages from 0-30 V DC, the output current is in the order of mA
(100 mA typical). It has a low leakage current. Response times are short, 1ms or less.
They are used for general purpose operating DC.

 TRIAC: the output voltage is 120 or 240 V AC, output current usually is about 0.75A
maximum, has a long response time (8.3ms to enable or disable). It has a leakage
current about 1 mA. They are used in general purpose applications in AC operation.
Suitable for inductive loads.

 FET: The output voltage is 0-120 V AC or 10-200 V DC, output current is on the order of
tens of mA, has short response time (1ms or less), the leakage current is in the order of
uA. The outputs of multiple sensors can be wired in series and / or parallel.

 MOSFET: the output voltage is 0-120 V AC or 10-200 V DC, output current is on the
order of hundreds of mA. The leakage current is moderately high. The outputs from
multiple sensors can be wired in parallel.

2. Analog Outputs
The most widely used standards are current analog 4 to 20 mA and voltage of 0-10 VDC.
There are others like 1-5 VDC, 0-20 mA and 20mA -20mA.