Guide to Data Acquisition: Signal conditioning (part1)

C-1 Introduction to Signal Conditioning

PC based data acquisition (DAQ) systems and plug-in boards are used in a wide range of applications. Typically, general-purpose DAQ plug-in boards are used for measuring analog and digital input and output voltages. As we have seen, many transducers’ signals must be conditioned in some way before a DAQ board or measuring system can accurately acquire the desired signal. Signal conditioning is the term generally used to describe the front end pre-processing required to convert the electrical signals received from transducers into signals which DAQ plug-in boards or other forms of data acquisition hardware can accept.

• Amplification
• Isolation
• Filtering
• Excitation
• Linearization

The type of signal conditioning equipment required, and the manner in which this is interfaced within the DAQ system, is largely dependent on the number and type of transducers, their excitation and earthing requirements, and no less importantly, how far the transducers are located from the personal computer, which must acquire, analyze and store the transducer signal data.

This section discusses several of the main hardware configurations used when integrating signal conditioning products into a DAQ system, as well as the general signal conditioning functions that must be performed.

C-2 Types of Signal Conditioning

C-2.1 Amplification

Amplification is one of the primary tasks carried out by signal conditioning equipment. It performs two important functions:

• Increases the resolution of the signal measurement.
• Increases the signal-to-noise ratio (SNR).

Amplification is primarily used to increase the resolution of the signal measurement. Consider a low-level signal of the order of a fraction of an mV, fed directly to a 12-bit A/D converter with full-scale voltage of 10 V. There will be a resultant loss of precision because the A/D converter has a resolution of only 2.44 mV (15.2 µV for 16 bit resolution). The highest possible resolution can be achieved by amplifying the input signal so that the maximum input voltage swing equals the maximum input range of the ADC. Another important function of amplification is to increase the SNR. Where transducers are located a long way from the data acquisition board and the signal measurements are transmitted through an electrically noisy environment, then low-level voltage signals can be greatly affected by noise. Where the low-level signals are amplified at the data acquisition board after they have been transmitted through the noisy environment, then any noise superimposed on the signal will also be amplified by the same amount as the signal. If the noise is of the same order of magnitude as the signal itself (i.e. the SNR is low), then the signal measurement may be lost in noise, leading to inaccurate and meaningless measurements. Amplifying the low-level signals before they are transmitted through the noisy environment increases the level of the required signal before they are affected by noise, thereby increasing the SNR of the signal for the same level of noise. Consider e.g., a J-type thermocouple, which outputs a very low-level voltage signal that varies by about 50 µV/ºC. If the thermocouple leads were to travel through a noisy electrical environment for say 10 m, then it's possible that the amount of noise coupled onto the thermocouple leads could be of the order of 200 µV. This noise-induced error corresponds to 4ºC at the measuring device. Amplifying the signal with an amplifier gain of 500, close to the thermocouple, produces a thermocouple signal that varies by approximately 25 mV/ºC. At this higher signal level, the 200 µV of induced noise coupled onto the 10m cable would result in a much smaller error, adding only a fraction of a degree Celsius of noise to the measured temperature.

C-2.2 Isolation

An isolated signal conditioner passes a signal from its source to the measurement device without a galvanic or physical connection. The most common methods of circuit isolation include opto-isolation, magnetic or capacitive isolation. Opto-isolation is primarily used for digital signals. Magnetic and capacitive isolations are used for analog signals, modulating the signal to convert it from a voltage to a frequency and transmitting the frequency signal across a transformer or capacitor without a direct physical connection before being converted back to a voltage.

Isolation performs several important functions. Firstly, isolation provides an important safety function by protecting expensive computer equipment and DAQ boards, as well as the equipment operators, from high voltage transients that could be caused by electrostatic discharge, lightning, or high voltage equipment failure. While isolated signal conditioning equipment provides an effective physical barrier and transient voltage protection for the computer and DAQ equipment, typically up to 1500 V, separate overvoltage protection is usually provided at the input(s) of the signal conditioning equipment to prevent internal damage to the signal conditioning equipment itself. In medical applications, isolation prevents the possibility of potentially fatal voltage or current signals from reaching sensors or transducers attached to or implanted in the human body. Another important function of isolation is to ensure that ground loops or common-mode voltages don't affect the accuracy of measured signals. Ground loops, caused by a potential difference between the source ground and the ground reference of the measuring device, may cause inaccuracies in the measured signal, or if too large, may damage DAQ equipment. Using isolated signal conditioning modules will eliminate the ground loop, and ensure that the signals are accurately measured. We shall see later that common-mode voltage signals are those that appear equally on each input of a measurement system. They can be caused by potential differences in the ground references of the source and the measurement system (i.e. ground loops) or are a necessary part of the measurement process (e.g. measuring the temperature of a device that is many volts above ground potential).

C-2.3 Filtering

Filtering removes unwanted noise from signal measurements before they are amplified and presented to the A/D converter. In intelligent signal conditioning modules, integrating A/D converters go a long way to averaging (filtering) out any cyclical noise appearing at the input. Alternatively, software averaging may also be used to digitally filter out periodic noise signals such as mains hum. This technique involves taking many more measurements than is necessary to acquire the wanted signal, then averaging them to produce a single measurement. If the samples are averaged over the period of the cyclical noise signal then this signal will be averaged to zero. Where there is no other form of filtering, an analog hardware filter provides the cheapest option. There are two types of analog filter, namely passive filters that use only passive components (such as capacitors and resistors), and active filters that utilize operational amplifiers. Ideally, filters should eliminate all data at frequencies outside the specified frequency range, providing a very sharp transition between the frequencies that are passed and those that are filtered out. Most practical filters are not ideal and don't usually eliminate all the undesirable amplitude components outside a specified frequency range. Attributes common to filters are:

Cut-off frequency

This is the transition frequency at which the filter takes effect. It may be the high-pass cut-off or the low-pass cut-off frequency and is usually defined as the frequency at which the normalized gain drops 3 dB below unity.

Roll-off

This is the slope of the amplitude versus the frequency graph at the region of the cut-off frequency. This characteristic distinguishes an ideal filter from a practical (non-ideal) filter. The roll-off is usually measured on a logarithmic scale in units of decibels (dB).

Quality factor ‘Q’

This variable is an adjustable characteristic of a tuned filter and determines the gain of the filter at its resonant frequency, as well as the roll-off of the transfer characteristic, on either side of the resonant frequency. Active filters are more frequently used since they provide a sharper roll-off and better stability. Such filters are described below.

Low pass filter

Low pass filters pass low frequency components of the signal and filter out high frequency components above a specific high frequency. An active low pass filter is shown in Figr C-1.

Figr C-1 Active low pass filter

The transfer characteristic of an ideal low pass filter is shown in Figr C-2.

Figr C-2 Ideal low pass filter transfer characteristics

The transfer characteristics of a practical filter for minimum ‘Q’ and maximum ‘Q’ are shown in Figr C-3.

Figr C-3 Practical active low pass filter transfer characteristics

High pass filter

High pass filters pass high frequencies and filter out low frequencies beginning at a specific low frequency. An active high pass filter is shown in Figr C-4.

Figr C-4 Active high pass filter

The transfer characteristic of an ideal high pass filter is shown in Figr C-5.

Figr C-5 Ideal high pass filter transfer characteristics

The transfer characteristics of a practical filter for minimum 'Q' and maximum Q are shown in Figr C-6.

Figr C-6 Practical active high pass filter transfer characteristics

Band pass (selective) filter

Band pass filters pass only those frequencies within a certain range specified by a low and high cut-off frequency. This is also known as a selective filter and combines a low pass and high pass filter in series, each tuned to the low and high cut-off frequencies respectively. The ideal transfer characteristic of an active band pass filter is shown in Figr C-7.

Figr C-7 Ideal band pass filter transfer characteristics

The transfer characteristics of a practical filter for minimum ‘Q’ and maximum ‘Q’ are shown in Figr C-8.

Figr C-8 Practical active band pass filter transfer characteristics

Band stop (notch) filters

Notch filters filter out a certain range of frequencies specified by a start and stop frequency, and pass all others. These filters combine a high pass and a low pass in parallel, each tuned to the low and high cut-off frequencies respectively. The ideal transfer characteristic of an active band stop filter is shown in Figr C-9.

Figr C-9 Ideal band stop filter transfer characteristics

The transfer characteristics of a practical filter for minimum ‘Q’ and maximum ‘Q’ are shown in Figr C-10.

Figr C-10 Practical active notch transfer characteristics

Butterworth filter

Butterworth filters provide a higher level of low pass filtering, containing two or more low pass filter stages. The number of stages ‘n’ of the filter determines how sharp the roll-off is at the cut-off frequency. A two-stage filter of this type is known as a second order Butterworth filter as shown in Figr C-11. A fourth order Butterworth filter would have two of the filter sections shown in Figr C-11 cascaded together.

Figr C-11 Two-stage Butterworth filter

C-2.4 Linearization

As we have seen, the output signals from transducers such as thermocouples exhibit a non-linear relationship to the phenomena being measured over a given input range. The data acquisition software typically performs linearization of these signals. However, where the non-linear relationship is predictable and repeatable this task can be performed by intelligent signal conditioning hardware. This typically requires the signal conditioning equipment to be programmed for a particular type of transducer, but once completed, the measurements returned to the host PC or stored as part of the measurement process are directly related to the phenomena (e.g. temperature) being measured.

C-3 Classes of Signal Cond.

Signal conditioning products, available from many different equipment manufacturers, are provided in many different forms covering a range of price, performance, modularity and ease of use. The type of signal conditioning hardware should be matched to the specific application. The main forms are discussed below.

C-3-1 Plug-in board signal conditioning

This range of signal conditioning hardware typically covers specialty plug-in data acquisition boards where the signal conditioning hardware is contained on the board itself. This is shown in Figr C-12.

Figr C-12 Plug-in DAQ board signal conditioning

Each board specializes in one type of transducer; thermocouple boards for interfacing to thermocouples, strain gauge boards for strain gauges, etc. These boards are typically used for small, specialized data acquisition systems that have a limited number of transducers located near the host computer.

C-3-2 Direct connect modular – two-wire transmitters

Two-wire transmitters are two-port modular signal conditioning modules that input an unconditioned signal on the input port and output a conditioned signal on the output port. A single module is required for each type of transducer (or actuator). These signal conditioning modules are not intelligent devices and don't perform on-board A/D conversion. Instead, the conditioned analog signal is transmitted over two lines to the data acquisition board in the host PC, either as a voltage, or converted into a standard current loop signal (4–20 mA) to the data acquisition board, hence the name two-wire transmitters. The simplified functional block diagram of a typical two-wire transmitter signal conditioning module is shown in Figr C-13.

Figr C-13 Functional block diagram of a two-wire transmitter signal conditioning module

Voltage outputs ( ±10 V or 0–10 V), compatible with the single ended inputs of most data acquisition boards allow easy interfacing to the latest data acquisition board technology. However, due to voltage drops, which may occur on the signal lines and the effects of noise that is proportional to the length of the transmission lines, voltage outputs should only be used for short transmission lines. Current signals have much greater immunity to noise and can be transmitted over hundreds of meters (up to 1000 meters), to a receiver that converts the currents back into a voltage, for A/D conversion at the PC. The receiver is principally a resistor, nominally in the order of 500 for full-scale deviation of 10 V (500 X 20 mA). A separate pair of wires is used for the current loop of each individual sensor, resulting in many cable pairs to the PC. A power supply (between 15–40 V), capable of driving as many current loops as there are modules, is required.

As individual signal conditioning modules require external power, they are typically designed to plug into a mounting board with on-board power supply as shown in Figr C-14.

Figr C-14 Board mounted modular signal conditioning

A single connector on the mounting board is used for easy cable connection between the mounting board and the I/O of the plug-in data acquisition board. Cables are typically a multi-core twisted-pair. This allows many different types of transducers to be interfaced to the latest plug-in data acquisition boards, but does not facilitate distributed I/O.

C-3.3 Distributed I/O – digital transmitters

Often sensors must be remotely located from the personal computer in which the processing and A/D conversion of the analog data takes place. This is especially true in industrial environments where sensors such as thermocouples and strain gauges are located in hostile environments over a wide area, possibly hundreds of meters away. In noisy environments, it's very difficult for the very small signals received from sensors, such as thermocouples and strain gauges (in the order of mV), to survive transmission over such long distances, especially in their raw form, without the quality of the sensor data being compromised. An alternative to running long (and possibly expensive) wires from the transducers directly, or from two-wire transmitter modules, is the use of distributed I/O. Distributed I/O is available in the form of signal conditioning modules that are remotely located from the host PC, near the sensors to which they are interfaced. One module is required for each sensor used, allowing for high levels of modularity (single point up to hundreds per location). While this can add a reasonable expense to systems with large point counts, the benefits in terms of signal quality and accuracy may be worth it. One of the most commonly implemented forms of distributed I/O is the digital transmitter. These intelligent devices perform all the functions of simple signal conditioning modules (two-wire transmitters) but also contain a micro-controller and A/D converter to perform the digital conversion of the signal within the module itself. Converted data is transmitted to the computer via an RS-232 or RS-485 communications interface. The simplified functional block diagram of a typical digital transmitter is shown in Figr C-15.

Figr C-15 Functional block diagram of a digital transmitter signal conditioning module

The use of RS-485 multi-drop networks greatly reduces the amount of cabling required because each signal conditioning module shares the same cable pair. It does however require an RS-232 to RS-485 converter to allow communications between the computer and the remote signal conditioning modules. Digital transmitters are available that provide two alternatives for configuring the distributed I/O system. In the first system configuration, shown in Figr C-16, the digital transmitter modules are designed to plug into a mounting board with facilities to accept an external power supply.

Figr C-16 Distributed I/O signal conditioning network using board mounted digital transmitter modules

The second distributed I/O system configuration, shown in Figr C-17 makes use of individual digital transmitter modules. Individual modules can be easily stacked together where many transducers are located in close proximity or can be positioned individually where they are required.

Figr C-17 Distributed I/O signal conditioning network using individual digital transmitter modules

Like other signal conditioning modules, these devices require an external power supply. The power supply should be located to supply as many signal-conditioning modules as its rating will allow.

C-4 Field wiring and signal measurement

When measuring analog input signals from transducers, it's unfortunately not just a simple matter of wiring the transducer leads to the signal conditioning equipment or data acquisition board, or connecting the signal conditioning equipment to the data acquisition board itself. Signal conditioning equipment and data acquisition boards typically provide a variety of methods for taking measurements of input signals. When determining the wiring connections and analog input configuration that will produce accurate and noise free measurements, careful consideration must be given not only to the type of signal produced by the transducer but also to the nature of the signal source. The most common electrical signal output by transducers or signal conditioning equipment is in the form of voltage. In certain situations, where the output signal from signal conditioning equipment must be transmitted over long distances or is particularly susceptible to noise, it may be converted to a current or frequency signal. In most cases however, the signal is converted back to a voltage signal before a measurement is taken. it's therefore necessary to understand the voltage signal source and the various methods of taking measurements of voltage signals. Two categories of voltage signal source are defined:

• Grounded signal source
• Floating (ungrounded) signal source

Three types of measurement are available on most signal conditioning equipment and data acquisition boards:

• Single-ended
• Differential
• Pseudo-differential

Since an understanding of the types of signal sources and measurement systems is necessary to determine the best methods of taking analog signal measurements, these topics are discussed in the following sections.

C-4.1 Grounded signal sources

By definition, voltage is a measurement of the potential difference between two points. Grounded signal sources have one of their signal leads connected to the system ground as shown in Figr C-18. This is theoretically shown as earth potential, although the system ground is not necessarily at earth potential. The voltage output from the signal source is the potential difference between the system ground and the positive signal lead of the signal source.

Figr C-18 Grounded signal source

A common example of a grounded signal source is an instrument that is earthed via its AC plug to the building ground.

C-4.2 Floating signal sources

Floating or ungrounded signal sources, as shown in Figr C-19, don't have either of the signal source leads connected to the system ground. This means that the signal source is not referenced to any absolute reference. The potential difference, that each of the signal lines may have, with respect to the system ground or earth potential between the signal lines, is not indicated in anyway by the voltage potential.

Figr C-19 Ungrounded signal source

Examples of floating signal sources are transformers, isolation amplifiers, batteries, and battery powered instruments.

C-4.3 Single-ended measurement

A ground-referenced measurement system, as shown in Figr C-20, is one in which the voltage measurement is taken with respect to ‘ground’. it's known as a single-ended measurement because only one signal line is required to determine the signal voltage, provided it's ground-referenced.

Figr C-20 Single-ended measurement

C-4.4 Differential measurement

A differential measurement system, as shown in Figr C-21, has neither of its inputs tied to a fixed reference, such as earth or system ground.

Figr C-21 Differential measurement

Differential measurement is beneficial because, noise induced equally into each of the signal lines appears as a common mode voltage at the input and is largely rejected (see Common mode voltages and CMRR, below).

C-4.5 Common mode voltages and CMRR

Common mode voltages

Ideally a differential measurement system measures only the potential difference between its positive and negative terminals. Where a signal source is measured using differential inputs, and there is a voltage measured with respect to the measurement ground that is present on both input lines, then this voltage is referred to as a common mode voltage. This is shown in Figr C-22

Figr C-22 Common mode voltages

The common mode voltage Vcm can be calculated from the following:

V cm = (Va +Vb ) / 2

Where:

Va = Voltage at the non-inverting terminal of the measurement system with respect to the instrumentation amplifier ground

Vb = Voltage at the inverting terminal of the measurement system with respect to the instrumentation amplifier ground.

An example of a common mode voltage is the output from a bridge circuit, in which the small differential signal is superimposed over a much larger common mode voltage introduced by the excitation of the bridge circuit.

Common mode rejection ratio (CMRR)

Ideally, a differential amplifier would completely reject any common mode voltages present on its input signal lines and only amplify the potential difference between them. Practically, however, these devices don't totally reject common mode voltages. The common mode rejection ratio (CMRR) measures the ability of a differential input amplifier to reject signals that are common to both signal inputs. The CMRR is defined as the ratio between the common mode signal present at the input to the amplifier and the signal produced by this voltage at the output of the amplifier, as defined by the following equation:

CMRR = 20 log 10(Vcm / Vout)

This ratio, normally expressed in dB, can be used to calculate the output voltage error, which would occur due to a common mode voltage appearing at the input. The higher the CMRR, the better the rejection of common mode signals, and the more accurate the output due to the differential signal being measured. Typically, a CMRR of 60 dB–80 dB could be expected for a well-designed system.

Common mode input voltage limits

Practically, measurement systems also have another limitation, and this is that there is a maximum and minimum common mode input voltage allowable on each input, with respect to the measurement system ground. Applying common mode voltages to either input beyond this input range will result in measurement errors, or, in the worst case, possible damage to the measurement circuitry.

C-4.6 Measuring grounded signal sources

Differential measurement of grounded signal source

A grounded signal source is best measured with a differential or pseudo-differential measurement system as shown in Figr C-23. In this configuration, any potential difference ( Δ Vg) between the ground references of the source and the measurement system appears as a common-mode voltage to the measurement system. The measured differential voltage is defined as:

Vm = (Vs+ ΔVg) – ΔVg = Vs

Figr C-23 Differential measurement of a grounded signal source

Single-ended measurement of a grounded signal source

When a single-ended measurement system is used to measure a grounded signal source, as shown in Figr C-24, measurement problems may occur. In this configuration, any potential difference (Δ Vg) between the signal source ground and the measuring system ground is added to the signal source voltage as part of the measurement. The measured voltage is defined as:

Vm = Vs+ Δ Vg

Figr C-24 Single-ended measurement of a grounded signal source

If the signal voltage levels are quite high compared to the reference ground potential difference, and the wiring between the source and the measurement system has low impedance, then the inaccuracies in the signal voltage measurement may be acceptable.

C-4.7 Ground loops

The classic ground-loop problem arises because true earth ground is not necessarily the same potential at different locations. Where the ends of a wire are earth grounded at different locations, the potential difference between them (which may vary from micro-volts to many volts) can cause significant currents, referred to as ground-loop currents, to flow through the wire. In addition, this potential difference is not necessarily a DC level. As well as introducing DC offset errors, ground-loop currents contain AC components, such as AC mains hum (50–60 Hz), and are a continual source of noise. This is especially true when multiple ground points in a system separated by large distances are connected to AC power ground, or when the magnitude of signal levels in analog circuits is low compared to the noise voltage levels. Where signal lines are used to connect grounds then ground currents will flow with unpredictable results. A possibly more serious result of ground loops is the undefined current loop area, which may couple magnetic fields and induce other unwanted noise voltages in the signal conductors.

C-4.8 Signal circuit isolation

Where a signal conductor is required to be earthed at both ends and additional noise immunity is required, the ground loop should be broken by isolating the signal source from the measuring equipment. Isolation by the use of transformers, opto-couplers and common mode chokes, is shown in Figures C-25, C-26 and C-27 respectively.

Figr C-25 Transformer isolation of ground loop

When a transformer is used to isolate the signal source from the measurement system the common mode voltage appears between the windings of the transformer and not at the input to the measurement circuit. Noise coupling between the circuits is very small and dependent on any stray capacitance between the transformer windings. Disadvantages with using transformers are that they are quite large and costly, especially where several signal circuits have to be isolated. In addition, transformers have limited frequency response and provide no DC continuity from the signal source to the measurement system. The opto-isolated circuit, shown in Figr C-26, is more typically used for digital signals because of the non-linearity of the opto-coupler to analog signals.

When a transformer is connected as a common mode choke, as shown in Figr C-27, DC and differential analog signals are transmitted while common mode AC signals are rejected. The common mode noise voltage appears across the windings of the choke. One big advantage with this type of isolation circuit is that multiple signal circuits can be wound on a common core without coupling.

Figr C-26 Opto-coupler isolation of ground loop ; Figr C-27 Common mode choke isolation of ground loop

C-4.9 Measuring ungrounded signal sources

Ungrounded or floating signal sources can be measured using the single-ended, pseudo-differential or differential measurement methods.

Differential measurement of ungrounded signal sources

When using the differential measurement system to measure the voltage signal from an ungrounded source, care should be taken to ensure that the common mode voltage level of the signal with respect to the measurement ground does not exceed the common mode input voltage limits of the measurement device. In addition, where there is no return path to the measurement system earth for the instrumentation amplifier input bias currents, then the flow of these currents through the source impedance, as well as charging stray capacitances, can cause the voltage level of the source to float beyond the valid range of the input stage of the measurement system. This is especially true where the source impedance is high. The degree to which the source voltage will float depends on the magnitude of the input bias currents and the system imbalance. A balanced measurement system meets the following criteria:

• The input impedances to ground of each terminal of the instrumentation amplifier are equal.
• The impedances of each signal cable to ground are equal.
• The impedances to ground of each terminal of the source are equal.

Increased noise immunity is also achieved using a balanced system, since induced noise voltages appearing on the signal wires, are equal and should be cancelled out by the differential amplifier measurement. Bias resistors, connected between each input lead and the ground reference of the measurement system, as shown in Figr C-28, provide a DC return path for bias currents from the inputs of the instrumentation amplifier to the reference ground.

Figr C-28 Differential measurement of an ungrounded signal source

Where the signal contains both AC and DC components (i.e. DC coupled) and the signal source has low impedance, only one bias resistor is required to be connected between the negative input and the ground reference. If the source impedance is relatively high compared to the input impedance of the instrumentation amplifier, then the imbalance caused by using a single bias resistor could lead to erroneous results. Therefore, for high source impedances both input bias resistors should be used. For input signals, which contain no DC component (i.e. AC coupled), both bias resistors are required. The bias resistors should be large enough to allow the source to float with respect to the measurement system ground and not to load the signal source (i.e. much greater than the source impedance), but small enough to keep the voltage at each input terminal within the input stage common mode voltage range of the measurement system. Bias resistors between 10 k and 100 k are typically used for low impedance sources such as thermocouples or when connecting the outputs of signal conditioning modules to data acquisition boards.

C-4.10 System isolation

To allow the measurement of signals that contain large common mode voltages, special hardware and measurement techniques are used. This typically involves isolating the measurement system from the ground reference so that signal lines, such as amplifiers, commonly used as a measurement reference, become a floating reference point.

System isolation can be carried out in the following ways:

• Using isolation transformers to reject the common mode voltage appearing on the signal lines.
• Using isolation amplifiers to isolate the input signals from the measurement system ground reference.
• Permanently isolating the measurement system ground using isolation transformers.
• Temporarily isolating the measurement system ground reference with a digital switch whilst an input signal measurement is taken.

NEXT: Signal conditioning (part 2)

PREV: Analog and digital signals

All related articles   Top of Page   Home