Introduction to Data Acquisition and Signal Conditioning

SIGNALS, SENSORS, and SIGNAL CONDITIONING

All industrial processing systems, factories, machinery, test facilities, and vehicles consist of hardware components and computer software whose behavior follow the laws of physics as we understand them. These systems contain thousands of mechanical and electrical phenomena that are not steady state; rather, they’re continuously changing. The measurable quantities that represent the characteristics of all systems are called variables. The proper functioning of a particular system depends on certain events occurring in time and the parameters of these variables. Frequently, one is interested in the location, magnitude, and speed of the variables, and one uses instruments to measure them. One may assign the variables units-of-measure: e.g., volts, grams, and meters per second, etc.

Most variables must be measured with a device that converts the phenomena into a form that a human can perceive, such as a visual display, a transducer for sound, or vibrations to stimulate physical sensations. The conversion devices are called transducers or sensors, and they translate the physical phenomena to electrical signals (or vice versa) to be measured with electronic instruments. These instruments have traditionally been ammeters, voltmeters, and various other gages, and the variables can be observed in real time. However, an increasing need to record and store (or log) these phenomena -- and, often, analyze them at a later time -- forced engineers to develop data recorders and data acquisition (DAQ) systems.

Variables may be classified in several ways; usually, most experts prefer two classifications:

• by characteristic
• by type of measurement signal

Variables classified by characteristic include:

• thermal
• radiation
• force
• rate
• quantity
• time
• geometric
• physical properties
• chemical composition
• and electrical

Those classified by measurement signal include motion, force, electrical, and time-modulated. Measurement signals for variables often are hard to differentiate from the measuring system. Four factors require close consideration for measurement signals and systems:

(1) the types of transducers available for converting variables to measurement signals

(2) transmission characteristics

(3) data acquisition system input matching

(4) transducers available to convert from one type of measurement signal to another measurement signal

DATA ACQUISITION SYSTEMS (DAQs)

Data acquisition systems (DAQs) have evolved over time from electromechanical recorders containing normally from one to four channels to all-electronic systems capable of measuring hundreds of variables simultaneously. Early systems used paper charts and rolls or magnetic tape to permanently record the signals; however, since the advent of computers, particularly personal computers, the amount of data and the speed with which they could be collected increased dramatically. But, many of the classical (sometimes called legacy) data-collection systems still exist and are used regularly.

PERSONAL-COMPUTER- (PC-) BASED DATA-ACQUISITION EQUIPMENT

Before the 1960s, costly mainframe computers were extensively used for gathering multiple channels of data, mostly in large industrial or scientific applications. They were rarely used in small projects because of their relatively high cost. What changes all this was the introduction of small rack-mounted minicomputers that developed in the 1960’s and later desktop personal-type computers (PCs) that had microprocessors and explodes in use in the 1970’s justified their use for smaller projects. Not long after this, data-acquisition plug-in cards (as well as hundreds of other types of plug-in cards) for these small computers were a common means to collect and record data of all types.

Plug-in cards for computers didn’t always do what the users expected, however. Internal noise from rotating devices such as drives and electromagnetic and electrostatic noise from the computer’s internal busses frequently interfered with the measured variable, notably in data-acquisition cards. Isolation and shielding have really helped to solve the problem in most cases; nevertheless, many data acquisition manufacturers also provide signal-conditioning and signal-processing circuits in small, stand-alone, shielded enclosures. The separate box provides isolation by distance, expansion for hundreds of channels, and portability with laptop computers that desktop personal computers with plug-in cards do not possess.

All PC-based DAQ systems will record very accurate, repeatable, reliable, and error-free data -- but only if they are connected and operated according to the manufacturer’s recommended practices. These practices include:

• selecting the correct sensors for the application, the proper wire and shielded cable;
• capturing the signals in proper magnitude, range, and frequency;
• and paying close attention to grounding and shielding – notably eliminating ground loops.

Other items include choosing the right impedance and using doubled-ended (differential) inputs instead of single-ended where possible. The environment should also be considered, especially for extremes of ambient temperature, shock, and vibration. And here lies the major goal of this discussion -- to inform and guide users of the top recommended practices based upon a fundamental knowledge of the internal workings of DAQ system instrumentation.

After the information about a physical process -- such as temperature or pressure -- is measured by a sensor or transducer, the analog signal heads toward the data-acquisition system (DAQ). Analog signals can take many forms, but may be roughly categorized into type types:

• physical measurements -- pressure, temperature, flow, force, vibration, mass and density
• compositional measurements -- conductivity, pH, chemical analysis

Often, the signal is too "noisy" or "quiet" and must, therefore, be changed to a form suitable for analysis. The process of doing so is called signal conditioning. Signal conditioning includes amplification, filtering, converting, and other processes required to make sensor output suitable for reading by computer boards. It is primarily used for DAQ, systems in which sensor signals must be normalized and filtered to levels suitable for analog-to-digital conversion so they can be read by computers. In most cases, this is done by changing the sensor's output to a voltage (if it isn't already), modifying the sensor's dynamic range to maximize the accuracy of the DAQ system, removing unwanted signals, and limiting the sensor's spectrum. One may also think of signal conditioning as a deliberate attempt to increase signal fidelity or signal-to-noise ratio.

The proper design of the signal-conditioning system is critical in mapping the sensor output to the data-acquisition input. Poor choices can affect the way the DAQ system reacts to input signals. Therefore, it is important to note the changes in the properties of the sensor signal caused by the conditioning circuitry.

In a nutshell, the following signal-conditioning technologies are used to improve the performance and accuracy of data-acquisition systems:

• Amplification
• Attenuation
• Calibration
• Cold-Junction Compensation
• Excitation
• Filtering
• Isolation
• Simultaneous Sampling

The following table lists the type of signal-conditioning technologies required or recommended for data-acquisition input components:

Important requirements for Analog-to-Digital Conversion (ADC)

As stated above, the main reason for analog signal conditioning is to change the sensor output into a form that can be optimally converted to a digital data stream by the DAQ system. Important input requirements of most DAQ systems include:

• Input signal being in voltage waveform: The process of converting the sensor output to a voltage can also be used to reduce undesirable signal noise.
  
  
• Input signal dynamic range: This should be at or near the dynamic range of the data acquisition system (usually equal to the voltage reference level, V_ref, or 2V_ref,). This is important in maximizing the resolution of the analog to digital converter (ADC).
  
  
• Input signal source impedance, or R_s: should be low enough so that changes in the input impedance, R_in, of the DAQ system do not affect the input signal.
  
  
• Input signal bandwidth: must be limited to less than half of the sampling rate of the analog to digital conversion. More on this topic here.

Other Reasons for Signal Conditioning:

Signal isolation: In certain applications it may be necessary to isolate the sensor from the power supply of the computer. This may be done in one of two ways: magnetic isolation or optical isolation. Magnetic isolation is primarily used for coupling power from the computer or the wall outlet to the sensor. This is done by using a transformer. Optical isolation is used for coupling the sensor signal to the data acquisition input. This is usually done through the use of a light-emitting diode and a photodetector. Optical isolation eliminates ground loops, reduces noise, and blocks transient signals. left: A Multi-Channel, Universal DC I/O, from Acromag, that uses optical isolation. This unit receives DC current or voltage input signals, provide isolation, and output proportional DC current or voltage signals.

Signal pre-processing: Often, it is desirable to perform pre-processing on the sensor signal before data acquisition. Depending on the application, this can help lower the required computer processing time, lower the necessary system sampling rate, or even perform functions that will enable the use of a much simpler data acquisition system entirely. For example, while a transducer system can output a voltage proportional to amount of change in the physical phenomena, it may be desired to only tell the computer when the change of the phenomena is greater than a certain amount. This can be done using analog signal-conditioning circuitry. Therefore, the DAQ system is reduced to only having a single binary input (no need for an ADC (analog-to-digital conversion)).

Removal of undesired signals: Many sensors output signals that have many different components to them. It may be desirable or even necessary to remove such components before the signal is digitized. Other added signals may actually corrupt the sensor output. This "noise" can also be removed using analog filter circuitry. For example, 60 Hz interference can distort the output of low-output sensors. Signal-conditioning circuitry can remove this before it is amplified and digitized.

When Shopping for Signal-Conditioning Devices, look for:

• Signal Conditioning device type: this includes signal filters, instrument amplifiers, sample-and-hold amplifiers, isolation amplifiers, signal isolators, multiplexers, bridge conditioners, analog-to-digital converters, digital-to-analog converters, frequency converters or translators, voltage converters or inverters, frequency-to-voltage converters, voltage-to-frequency converters, current-to-voltage converters, current loop converters, and charge converters.
  
  
• Form Factors: IC or board mount, circuit board, panel or chassis mount, modular bay or slot system, rack mount, DIN rail, and stand-alone.
  
  
• Device specifications:
  • differential analog input channels: Differential channels use the difference between two signals as an input; common mode is filtered out. In some cases, differential inputs are combinations of two single-ended inputs; therefore, twice the number of differential channels are available as single-ended inputs.
  • digital I/O channels: used for low-level on-off signals. Applications include: communication, user interface, or control
  • accuracy: relies on the signal conditioning linearity, hysteresis, temperature considerations, etc. It is represented as percent full-scale of measurement range.
    
    
• Signal inputs accepted by signal conditioners: include DC voltage and current, AC voltage and current, frequency and charge
  
  
• Sensor inputs: accelerometer, thermocouple, thermistor, RTD, strain gauge or bridge, and LVDT or RVDT
  
  
• Specialized inputs: encoder, counter or tachometer, timer or clock, relay or switch, etc.
  
  
• Transducers and Excitation
  
  
• Outputs: voltage, current, frequency, timer or counter, relay, resistance or potentiometer, etc.
  
  
• User Interface

Also look for:

• application
• software
• memory and storage
• network specifications
• filter specifications (e.g. bandwidth, frequency range)
• amplifier specifications
• environmental parameters

Further reading : Signal Conditioning: Buy vs. Do-it-Yourself