The applications for industrial process-control systems are diverse, ranging from simple traffic control to complex electrical power grids, from environmental control systems to oil-refinery process control. The intelligence of these automated systems lies in their measurement and control units. The two most common computer-based systems to control machines and processes, dealing with the various analogue and digital inputs and outputs, are programmable logic controllers (PLCs) and distributed control systems (DCS).These systems comprise power supplies, CPUs and a variety of analogue-input, analogueoutput, digital-input and digitaloutput modules.
The standard communications protocols have existed for many years; the ranges of analogue variables are dominated by 4 mA to 20 mA, 0 V to 5 V, 0 V to 10 V, ±5 V and ±10 V. There has been much discussion about wireless solutions for next generation systems, but designers still claim that 4 mA to 20 mA communications and control loops will continue to be used for many years. The criteria for the next generation of these systems will include higher performance, smaller size, better system diagnostics, higher levels of protection and lower cost-factors. This will help manufacturers differentiate their equipment from their competitors’. In this article we will cover the key performance requirements of process control systems and the analogue input/output modules they contain. This case study will also introduce an industrial processcontrol evaluation system and will look at the challenges of designing a robust system that will withstand the electrical fast transients (EFTs), electrostatic discharges (ESDs) and voltage surges found in industrial environments.
Overview of PLC
Figure 1 shows a basic processcontrol system building block. A process variable such as flow rate or gas concentration is monitored via the input module.The information is processed by the central control unit and appropriate action is taken by the output module, which, for example, drives an actuator.
Figure 2 shows a typical industrial subsystem of this type. Here a CO2 gas sensor determines the concentration of gas accumulated in a protected area and transmits the information to a central control point. The control unit consists of an analogue input module that conditions the 4 mA to 20 mA signal from the sensor, a CPU and an analogue output module that controls the required system variable. The current loop can handle large capacitive loads that are often found on hundreds-ofmetres long communication paths experienced in some industrial systems. The output of the sensor element, representing gas concentration levels, is transformed into a standard 4 mA to 20 mA signal, which is transmitted over the current loop.This simplified example shows a single 4 mA to 20 mA sensor output connected to a single channel input module and a single 0 V to 10 V output. In practice, most modules have multiple channels and configurable ranges. The resolution of input/output modules typically ranges from 12 to 16 bits, with 0.1 percent accuracy over the industrial temperature range. Input ranges can be as small as ±10 mV for bridge transducers and as large as ±10 V for actuator controllers or 4 mA to 20 mA currents in processcontrol systems. The analogue output voltage and current ranges typically include ±5 V, ±10 V, 0 V to 5 V, 0 V to 10 V, 4 mA to 20 mA, and 0 mA to 20 mA. Settling-time requirements for digital-toanalogue converters (DACs) vary from 10 microseconds to 10 milliseconds, depending on the application and the circuit load.
The 4 mA to 20 mA range is mapped to represent the normal gas detection range. The current values outside this range can be used to provide fault-diagnostic information as shown in Table 1.
PLC evaluation system
The PLC evaluation system integrates all the stages needed to generate a complete input/output design. It contains four fully isolated analogue-to-digital converter (ADC) channels, an ARM7 microprocessor with RS-232 interface and four fully isolated DAC output channels. The board is powered by a DC supply. Hardware-configurable input ranges include 0 V to 5 V, 0 V to 10 V, ±5 V, ±10 V, 4 mA to 20 mA, 0 mA to 20 mA, ±20 mA as well as thermocouple and resistance temperature device (RTD). Softwareprogrammable output ranges include 0 V to 5 V, 0 V to 10 V, ±5 V, ±10 V, 4 mA to 20 mA, 0 mA to 20 mA, and 0 mA to 24 mA.
Output module
Table 2 highlights some key specifications of PLC output modules. Since true system accuracy lies within the measurement channel (ADC), the control mechanism (DAC) requires only enough resolution to tune the output. For high-end systems, 16-bit resolution is required. This requirement is actually quite easy to satisfy using standard digital-toanalogue architectures. Accuracy is not crucial and 12-bit integral nonlinearity (INL) is generally adequate for high-end systems. Calibrated accuracy of 0.05 percent at 25°C is easily achievable by over ranging the output and trimming to achieve the desired value. Today’s 16-bit DACs, such as the AD5066, offer 0.05 mV typical offset error and 0.01 percent typical gain error at 25°C, eliminating the need for calibration in many cases. The total accuracy error of 0.15 percent sounds manageable, but is actually quite aggressive when specified over temperature. A 30 ppm/°C output drift can add a 0.18-percent error over the industrial temperature range.
Output modules may have current outputs, voltage outputs or a combination of the two. A classical solution uses discrete components to implement a 4 mA to 20 mA loop. The AD5660 16-bit nanoDAC converter provides a 0 V to 5 V output that sets the current through sense resistor, RS, and therefore, through R1. This current is mirrored through R2.
Setting RS = 15 kiloohm, R1 = 3 kiloohm, R2 = 50 kiloohm and using a 5-V DAC will result in IR2 = 20 mA max.
This discrete design suffers from many drawbacks. Its high component count engenders significant system complexity, board size and cost. Calculating total error is difficult, with multiple components adding varying degrees of error with coefficients that can be of differing polarities.The design does not provide shortcircuit detection/protection or any level of fault diagnostics. It does not include a voltage output, which is required in many industrial control modules. Adding any of these features would increase the design complexity and the number of components. A better solution would be to integrate all of the above onto a single IC, such as the AD5412/AD5422 low-cost, highprecision, 12-/16-bit digital-toanalogue converters.
The output current range is programmable to 4 mA to 20 mA, 0 mA to 20 mA or 0 mA to 24 mA over-range function. A voltage output, available on a separate pin, can be configured to provide 0 V to 5 V, 0 V to 10 V, ±5 V, or ±10 V ranges, with a 10 percent over range available on all ranges. Analogue outputs have a short circuit protection, which is a critical feature in the event of miswired outputs. For example, when the user connects the output to the ground instead of the load.
The AD5422 also has an open circuit detection feature that monitors the current output channel to ensure that there is no fault between the output and the load. In the event of an open circuit, the fault pin becomes active, alerting the system controller. The AD5750 programmable current/voltage output driver features both short-circuit detection and protection.
While earlier systems typically needed 500 V to 1 kV of isolation, today >2 kV is generally required. The ADuM1401 digital isolator uses iCoupler technology to provide the necessary isolation between the MCU and remote loads or between the input/output module and the backplane. Three channels of the ADuM1401 communicate in one direction, while the fourth channel communicates in the opposite direction, providing isolated data readback from the converters.For newer industrial designs, the ADuM3401 and other members of its family of digital isolators provide enhanced systemlevel ESD protection. The AD5422 generates its own logic supply (DVCC), which can be directly connected to the field side of the ADuM1401, eliminating the need to bring a logical supply across the isolation barrier.
The AD5422 includes an internal sense resistor, but an external resistor (R1) can be used when a lower drift is required. Since the sense resistor controls the output current, any drift of its resistance will affect the output. The typical temperature coefficient of the internal sense resistor is 15 ppm/°C to 20 ppm/°C, which could add a 0.12 percent error over a 60°C temperature range. In highperformance system applications, an external 2-ppm/°C sense resistor could be used to keep drift to less than 0.016 percent.
The AD5422 has an internal 10-ppm/°C max voltage reference that can be enabled on all four output channels in the PLC evaluation system. Alternatively, the ADR445 ultra-low-noise XFET voltage reference, with its 0.04 percent initial accuracy and 3 ppm/°C, can be used on two output channels, allowing performance comparison and a choice of internal versus external reference, depending on the total required system performance.
Input module
The input module design specifications are similar to those of the output module. In industrial applications, a differential input is required when measuring low-level signals from thermocouples, strain gages and bridge-type pressure sensors to reject common mode interference from motors, AC power lines or other noise sources that inject noise into the analogue inputs of the ADCs.
Sigma-delta ADCs are the most popular choice for input modules as they provide high accuracy and resolution. In addition, internal programmable-gain amplifiers (PGAs) allow small input signals to be measured accurately. The AD7793 3-channel, 24-bit sigmadelta ADC is configured to accommodate a large range of input signals, such as 4 mA to 20 mA, ±10 V as well as small signal inputs directly from sensors.
Care was taken to allow this universal input design to be easily adapted for RTD or thermocouple modules.Usually, two input terminal blocks are provided per input channel. One input allows for a direct connection to the AD7793. The user can programme the internal PGA to provide analogue gains up to 128. The second input allows the signal to be conditioned through the AD8220 JFET-input instrumentation amplifier. In this case, the input signal is attenuated, amplified and the level shifted to provide a single-ended input to the ADC.
The low-power, high-performance AD7793 consumes <500 microampere, and the AD8220 consumes <750 microampere. This channel is designed to accept 4 mA to 20 mA, 0 V to 5 V and 0 V to 10 V analogue inputs. Other channels in the input module have been designed for bipolar operation to accept ±5 V and ±10 V input signals.
To measure a 4 mA to 20 mA input signal, a low-drift precision resistor can be switched (S4) into the circuit. In this design, its resistance is 250 ohm, but any value can be used as long as the generated voltage is within the input range of the AD8220. S4 is left open when measuring a voltage. Isolation is required for most input-module designs. The ADuM5401 4-channel digital isolator uses isoPower6 technology to provide 2.5-kV rms signal and power isolation.
Complete system
The ADuC7027 precision analogue microcontroller is the main system controller. Featuring the ARM7TDMI core, its 32-bit architecture allows easy interface to 24-bit ADCs. It also supports a 16-bit thumb mode, which allows for greater code density, if required. The ADuC7027 has 16 KB of on-board flash memory and allows interfacing of up to 512 KB external memory. The ADP3339 highaccuracy, low-dropout regulator (LDO) provides the regulated supply to the microcontroller.
Communication between the evaluation board and the PC is provided via the ADM3251E isolated RS-232 transceiver. The ADM3251E incorporates isoPower technology, making a separate isolated DC-to-DC converter unnecessary. It is ideally suited to operate in electrically harsh environments or where RS-232 cables are frequently plugged in or unplugged, as the RS-232 pins, Rx and Tx, are protected against electrostatic discharges of up to ±15 kV.
Evaluation system software and evaluation tools
The evaluation system is very versatile. Communication with the PC is achieved using LabView8. The firmware for the microcontroller (ADuC7027) is written in C, which controls the low-level commands to and from the ADC and DAC channels.
In the main screen interface, the pull-down menus on the left side allows the user to choose active ADC and DAC channels. Under each ADC and DAC menu, there is a pulldown range menu, which is used to select the desired input and output ranges to be measured and controlled. The following input and output ranges are available: 4 mA to 20 mA, 0 mA to 20 mA, 0 mA to 24 mA, 0 V to 5 V, 0 V to 10 V, ±5 V and ±10 V.
The ADC configure screen is used to set the ADC channel, update rate and PGA gain, to enable or disable excitation currents and for other general ADC settings. Each ADC channel is calibrated by connecting the corresponding DAC output channel to the ADC input terminal and adjusting each range. Therefore, when using this calibration method, the offset and gain errors of the AD5422 dictate the offset and gain of each channel. If these provide insufficient accuracy, ultra-high-precision current and voltage sources can be used for calibration.
After selecting the ADC’s input channel, input range and update rate, the ADC stats screen can now be used to display some measured data. On this screen, the user chooses the number of data points to record, the software generates a histogram of the selected channel, calculates the peak-to-peak and rms noise and displays the results. For example, the input is connected through the AD8220 to the AD7793: gain = 1, update rate = 16.7 Hz, number of samples = 512, input range = ±10 V, input voltage = 2.5 V. The peak-topeak resolution is 18.2 bits.
The input is connected directly to the AD7793, bypassing the AD8220.The on-chip 2.5-V reference is connected directly to the AIN+ and AIN- channels of the AD7793, providing a 0-V differential signal to the ADC. The peak-to-peak resolution is 20 bits. If the ADC conditions remain the same but the 2.5-V input is connected through the AD8220, the peak-to-peak resolution degrades to 18.9 bits.This happens because of two reasons: at low gains, the AD8220 contributes some noise to the system and the scaling resistors that provide the input attenuation result in some range loss to the ADC. The PLC evaluation system allows the user to change the scaling resistors to optimise the ADC's full-scale range, thereby improving the peak-topeak resolution.
Power supply input protection
The PLC evaluation system uses best practices for electromagnetic compatibility (EMC). A regulated DC supply (18 V to 36 V) is connected to the board through a 2- or 3-wire interface. This supply must be protected against faults and electromagnetic interference (EMI). The following precautions were taken in the board design to ensure that the PLC evaluation system will survive any interference that may be generated on the power ports.
A piezo resistor, R1, is connected to the ground adjacent to the power input ports. During normal operation, the resistance of R1 is very high (megohms), so the leakage current is very low (microamperes). When an electric current surge (caused by lightning, for example) is induced on the port, the piezo resistor breaks down, and tiny voltage changes produce rapid current changes. Within tens of nanoseconds, the resistance of the piezo resistor drops dramatically. This low-resistance path allows the unwanted energy surge to return to the input, thus protecting the integrated circuitry. Three optional piezo resistors (R2, R3, and R4) are also connected in the input path to provide protection in cases when the PLC board is powered using the 3-wire configuration.
A positive temperature coefficient resistor, PTC1, is connected in a series with the power input trace. The PTC1 resistance appears very low during normal operation, with no impact to the rest of the circuit. When the current exceeds the nominal, PTC1’s temperature and resistance rapidly increase. This high-resistance mode limits the current and protects the input circuit.
Y capacitors C2, C3 and C4 suppress the common-mode conductive EMI when the PLC board operates with a floating ground. These safety capacitors require low resistance and highvoltage endurance. Designers must use Y capacitors that have UL or CAS certification and comply with the regulatory standard for insulation strength.
Inductors L1 and L2 filter out the common-mode conducted interference coming in from the power ports. Diode D1 protects the system from reverse voltages. A general purpose silicon or Schottky diode specifying a low forward voltage at the working current can be used.
Analogue input protection
The PLC board can accommodate both the voltage and current inputs. Load resistor R5 is switched on for current mode. Resistors R6 and R7 attenuate the input. Resistor R8 sets the gain of the AD8220.
These analogue input ports can be subjected to electric surge or electrostatic discharge on the external terminal connections. Transient voltage suppressors (TVS) provide highly effective protection against such discharges. When a high-energy transient appears on the analogue input, the TVS goes from high impedance to low impedance within a few nanoseconds. It can absorb thousands of watts of surge power and clamp the analogue input to a preset voltage, thus protecting precision components from being damaged by the surge. Its advantages include fast response time, high-transient power absorption, low leakage current, low breakdown voltage error and a small package size. Instrumentation amplifiers are often used to process the analogue input signal. These precision, low-noise components are sensitive to interference, so the current flowing into the analogue input should be limited to less than a few milli-amperes. External Schottky diodes generally protect the instrumentation amplifier. Even when internal ESD protection diodes are provided, the use of external diodes allows smaller limiting resistors and lower noise and offset errors. Dual series Schottky barrier diodes D4-A and D4-B divert the overcurrent to the power supply or the ground.
When connecting external sensors, such as thermocouples (TC) or resistance temperature devices (RTD), directly to the ADC, similar protection is needed.
Two-quad TVS networks, D5-C and D5-D, are put in after the J2 input pins to suppress transients
coming from the port.
C7, C8, C9, R9 and R10 form the RF attenuation filter ahead of the ADC. The filter has three functions: to remove as much RF energy from the input lines as possible, to preserve the AC signal balance between each line and ground, and to maintain a high enough input impedance over the measurement bandwidth to avoid loading the signal source. The -3-dB differentialmode and common mode bandwidth of this filter are 7.9 kHz and 1.6 MHz, respectively. The RTD input channel to AIN2+ and AIN2- is protected in the same manner.
Analogue output protection
The PLC evaluation system can be software configured to output analogue voltages or currents in various ranges. The output is provided by the AD5422 precision, low-cost, fully integrated, 16-bit digital-to-analogue converter, which offers a programmable current source and programmable voltage output. The AD5422 voltage and current outputs may be directly connected to the external loads, so they are susceptible to voltage surges and EFT pulses.
A TVS (D11) is used to filter and suppress any transients coming from port J5.
A nonconductive ceramic ferrite bead (L3) is connected in a series with the output path to add isolation and decoupling from highfrequency transient noises. At low frequencies (<100 kHz), ferrites are inductive; thus, they are useful in low-pass LC filters. Above 100 kHz, ferrites become resistive, an important characteristic in highfrequency filter designs. The ferrite bead provides three functions: localising the noise in the system, preventing external high frequency noise from reaching the AD5422, and keeping internally generated noise from propagating to the rest of the system. When ferrites saturate, they become nonlinear, losing their filtering properties. Thus, the DC saturation current of the ferrites must not cross the limit.
Dual series Schottky barrier diodes D9-A and D9-B divert any overcurrent to the positive or the negative power supply. C22 provides the voltage output buffer and the phase compensation when the AD5422 drives capacitive loads up to 1 microfarad.
The protection circuitry on the current output channel is quite similar to that on the voltage output channel except that a 10-ohm resistor (R17) replaces the ferrite bead. The current output from the AD5422 is boosted by the external discrete NPN transistor Q1. The addition of the external boost transistor will reduce the power dissipated in the AD5422 by reducing the current flowing in the on-chip output transistor. The breakdown voltage BVCEO of Q1 should be greater than 60 V. The external boost capability is useful in applications where the AD5422 is used at the extremes of the supply voltage, load current and temperature range.
A 15-kiloohm, precision, low-drift current-setting resistor (R15) is connected to RSET to improve the stability of the current output over temperature.
The PLC demo system can be configured to provide a voltage output higher than 15 V when the AD5422 is powered by an external voltage. A TVS is used to protect the power input port. Diodes D6 and D7 provide protection from reverse biasing.All the supplies are decoupled by a 10-microfarad solid tantalum electrolytic and 0.1-microfarad ceramic capacitors.
IEC tests and results
The results show the deviations of the DAC output that occurred during the testing. The output recovered to the original values after the tests were completed. This is generally referred to as Class B. Class A means that the deviation was within the allowed system accuracy during the test. Typical industrial control system accuracies are approximately 0.05 percent.
About the authors:
Colm Slattery is an Applications Engineer in the Precision Converters group in Limerick, Ireland. He can be contacted at colm.slattery@analog.com.
Derrick Hartmann is an Applications Engineer in the DAC group at Analog Devices in Limerick, Ireland. He can be contacted at derrick.hartmann@analog.com.
Li Ke is as an Applications Engineer with the Precision Converters product line, located in Shanghai, China. He can be contacted at li.ke@analog.com. |
