What Is Signal Conditioning?
Signal conditioning is the part of a data acquisition system that interfaces to a sensor. Most sensors require some type of accommodation to turn their input into a linear voltage that can be measured by a data acquisition computer. Often, the signal is too “small” and must be changed to a form that can be read by the computer, or the sensor needs power. The process is called signal conditioning. Signal conditioning includes the amplification, filtering, converting, excitation, and other techniques required to make sensor output compatible with data acquisition hardware. The proper design of the signal-conditioning module is critical for maximizing the accuracy of the DAQ system.
Common techniques to condition the input signal include:
- Simultaneous Sampling
- Cold-Junction Compensation
If the signal is smaller than a few Volts, it might need to be amplified. If it is larger than the maximum range of your analog input hardware (typically ±10 V), you will have to divide the signal down using a resistor network.
For example, a thermocouple generates a voltage that changes with temperature. This is called the thermoelectric effect or the Seebeck effect, and is caused by placing dissimilar metals in contact to form a junction. However, the voltage is very small. The most common thermocouples, types E, J, K, N and T change about 30 to 60 microvolts per degree C change in temperature. A data acquisition device that has a 10 V full scale range and 16-bit resolution can therefore only detect a change of 3 to 5 degrees C or larger. Therefore amplification is required.
The linearity must be corrected for almost any practical use of a thermocouple. There are two methods to perform the correction: (1) find the temperature in a look-up table of voltage vs temperature, or (2) calculate the temperature using an equation. Both of these can be highly accurate. The linearity correction is normally done in software on a PC where the complexity of the equation is not a problem. For devices not connected to a PC, such as stand-alone meters, temperature controllers, or many dataloggers; the linearity correction is done by a microprocessor in the device. A look-up table may be the easiest. A few extremely simple devices without a microprocessor do not have linearization.
A signal might contain unwanted frequency components and require filtering. Filtering removes undesirable noise that could lead to errors in erroneous measurements. Slowly varying signals such as temperature may require a low-pass filter to remove high frequency noise that can reduce the accuracy of your measurement. Rapidly varying signals such as vibration often require an anti-aliasing filter.
When an electrical signal contains a high-voltage component that could damage the computer, the sensor signals should be electrically isolated from the computer for protection from serious overvoltage. Electrical isolation makes sure that the measurements from the data acquisition system are not affected by differences in ground potentials.
For example, if a thermocouple is mounted on the windings of a motor, it may come in electrical contact with the high voltage wires. To protect the electronics from damage, and to make an accurate measurement, the amplifier can be electrically isolated from the thermocouple. In other cases, when the hardware device and the sensor signal are each referenced to ground, problems occur if there is a potential difference between the grounds of each device. This difference creates noise and might lead to erroneous measurements. Using electrically isolated signals eliminates the difference between the ground potentials and ensures that the signals are accurately measured.
A common technique for measuring rapidly changing signals is simultaneous sampling. For rapidly changing signals such as vibration, it is often important to measure multiple channels at the same time, so they can be compared. Simultaneous sampling does this by having separate amplifiers and “sample and hold” circuits. In systems without simultaneous sampling the A/D converter samples one channel, switches to the next channel and samples it, switches to the next channel, and continues this cycle. The effective sampling rate of each individual channel is inversely proportional to the number of channels sampled because the same A/D converter is sampling many channels. Care must be taken when using multiplexers so that the switched signal has sufficient time to settle.
Some sensors require an excitation source (power) to operate. For example, strain gauges, and thermistors require external voltage or current to operate. Signal conditioning modules can provide these signals. Resistive temperature measurements made by thermistors need to have a current pass through it to produce a measurable voltage output.
Cold junction compensation
The thermocouple has a unique additional correction called cold junction compensation. It is unique among temperature sensors, and it is often misunderstood. The reason for the compensation is that the thermocouple only reports the relative temperature difference between the sensing end and the terminals where it is connected. The absolute temperature of the connection must be known in order to calculate the true temperature the thermocouple is measuring. The absolute temperature of the connection is called a cold junction.
A cold junction sensor is typically a semiconductor sensor or a thermistor located at the connection terminals. The temperature measured by the cold junction sensor is added to the temperature of the thermocouple. This process is simple, with subtle ramifications.
Because of the cold junction compensation, the connection of thermocouples is more complex than other sensors. The wire type and location of the connections is important. The voltage generated by a thermocouple results from two dissimilar metals being placed in contact, which generates a voltage due to the thermocouple effect. The voltage changes with temperature.
Each lead of the thermocouple must have the same type of wire from the sensing end to the terminals where the cold junction sensor is located. If a copper wire is connected to the non-copper thermocouple wire, another thermocouple junction is created, and the voltage across the new junction changes with temperature. The voltage measured would be the result of the two junctions, not what was expected. The cold junction sensor must be located where the thermocouple wire changes to copper, usually inside the measuring device.
To avoid problems always use thermocouple wire from the thermocouple to the connection at the cold junction sensor. Extension grade thermocouple wire is sold for this purpose. It is less expensive than ordinary thermocouple wire.
This paper has demystified the signal conditioning part of data acquisition systems. Most sensors require some type of conditioning to turn their input into a linear voltage that can be measured by a data acquisition computer. Signal conditioning is one of the most important components of a data acquisition system and the proper design is critical for maximizing the accuracy of the DAQ system.