Data Acquisition system basics and working

Overview of Data Acquisition system: Data acquisition is a process of automatically obtaining data from one or more sensors or transducers directly into the computer system. A sensor is a device that responds to a physical change and outputs an electrical signal and a transducer is a device that converts energy from one form to another. E.g. a thermocouple generates an emf due to two dissimilar metal joined at the it’s junction. The emf generated is a low level voltage. When this voltage is sent to a wire it is called a voltage signal. A transducer can be used to convert the low level voltage signal to a high level voltage signal. I have already posted a article on how to get data serially on labview which is a example of Data Acquisition system.

A simple system consists of

• Transducer
• Data Acquisition Hardware (DAQ)
• Computer

Once the sensor outputs a signal that can be transmitted to a wire there are still cases when there could be a mismatch in terms of type and range of signal and type and range of data acquisition hardware. e.g.

• Some transducer might be designed to generate a current signal as output but our data acquisition hardware could have been designed to take a voltage signal as input.
• A transducer might be designed to give output voltage in the range of 50-100mV but our data acquisition hardware accepts the input voltage signal in the range of 0-10V.

Signal conditioning may be required to adjust the signal type and range of the output signal to align with the requirements of the data acquisition system. Once mismatches between transducers and data acquisition hardware are removed, data can be collected.

The computer needs to know

• Which signal to measure
• How often to take readings
• How many readings to collect or how long to continue reading the signal.

This is known as configuring the data acquisition process and is done before collecting the data.

The ability of LabVIEW to take in data from an external source, process it and send processed data back to control external devices is what distinguishes LabVIEW from other softwares. The collected data is called waveform in LabVIEW. This data can be displayed, analyzed, modified or stored.

Sensor and Transducer

The bottom line when it comes to measuring a variable of interest is identifying some physical phenomenon that changes when the variable of interest changes. For example, the volume of a fluid expands as its temperature increases, and that physical phenomenon has been utilized to create thermometers. While lots of physical phenomena have been used for measurements, phenomena that can be related to an electrical property are more useful for automated data acquisition systems. A sensor is a device that responds to a physical change and outputs an electrical signal and a transducer is a device that converts energy from one form to another.

Signal Conditioning:
The process of modifying the output of a sensor is called signal conditioning. Signal condition is required for following reasons.

• Dealing with noisy signal (filtering)

If signal is noisy, the gauge needle bounces around while taking a reading. Signal noise is very common. The sources of signal noise includes the system itself and sometimes noise is also created by the electronic sensor itself. We can deal signal noise in following ways:

• Modifying the system
• Using better quality sensor
• Ignoring the noise
• Using a bunch of readings so that result can be averaged
• Filtering the system before it gets into the data acquisition system
• Filtering the signal after the data acquisition system

• Aligning sensor output with DAQ (Amplification and offset)

When the range of a sensor’s signal is different than the input range of the data acquisition system, signal conditioning can be used to rescale the signal from the sensor. Common changes include amplification and offset. Amplification causes the span to increase whereas adjusting the offset causes all values in the span to be increased or decreased by the same amount.

Sensor Calibration:
Sometimes there is an equation relating a sensor’s output to the measured variable. For example, the temperature at the junction of a thermocouple can be calculated if the voltage is known. If you use signal conditioning to adjust the range of the sensor output signal, you must account for the offset and amplification values to determine the measured value from the sensor output. This is extremely important when the sensor output signal must be related to an external standard.

Data Acquisition Hardware:

Data acquisition devices are a bridge between sensor and computer system. When considering which data acquisition system to use, there are several things to consider:

1. What types of signals will the data acquisition system need to handle?

• How many AI channels (analog inputs) are required
• How many AO channels (analog outputs) are required
• How many DI channels (digital inputs) are required
• How many DO channels (digital outputs) are required

2. Will your AI channels be wired as differential inputs or single-ended?
3. What level of precision is required in the analog-to-digital converter?
4. How fast will you need to take samples?

Differential Inputs:

When AIs are wired as differential inputs, the two signal wires are connected to two AI channels that are configured to work together to measure the voltage difference between the two wires.

Single-Ended Inputs:

Single-ended inputs only require one AI channel per measurement and can work under the right conditions.

Analog to Digital Converters:

Once the signal passes through the AI port, it must be converted to a digital value corresponding to the signal voltage. An ADC performs this action. One of the characteristics of an ADC is the number of bits used to describe the voltage level. The higher the bit count, the greater the resolution of the ADC. Common ADCs are 12–22 bit devices.

Sample Rate:
Another consideration when selecting a data acquisition system is the required sampling rate. The analog-to-digital conversion process takes a finite amount of time. If you want to take one sample a second, any data acquisition system will work fine. If you need to sample each of 32 AI channels at 1000 samples a second (total of 32,000 samples/second), that is a tougher challenge but there are data acquisition systems that can handle it.