Volume 40 August 2006
Precision Signal-Processing and Data-Conversion ICs for PLCs Now Have More Performance at Less Power, Size, and Cost
By Albert O'Grady, (firstname.lastname@example.org)
A programmable-logic controller (PLC) is a compact computer-based electronic system that uses digital or analog input/output modules to control machines, processes, and other control modules. A PLC is able to receive (input) and transmit (output) various types of electrical and electronic signals and use them to control and monitor practically any kind of mechanical and/or electrical system. PLCs are classified by the number of I/O functions provided. For example, a nano PLC incorporates fewer than 32 I/Os, a micro PLC has between 32 and 128 I/Os, a small PLC has between 128 and 256 I/Os, and so on. A typical PLC system is outlined in Figure 1.
Figure 1. PLC system architecture, showing various I/O-module functions.
PLC systems comprise input modules, output modules, and input/output modules. Since many of the inputs and outputs involve real-world analog variables—while the controller is digital—PLC system hardware-design tasks focus on the requirements for digital-to-analog converters (DACs) and analog-to-digital converters (ADCs), input- and output signal-conditioning, and isolation of the electrical wiring of the input- and output modules from the controller and each other.
Resolutions of I/O modules typically range from 12 bits to 16 bits, with 0.1% accuracy over the industrial temperature range. Analog output voltage and current ranges include ±5 V, ±10 V, or 0 V to 5 V, 0 V to 10 V, and 4 to 20 mA or 0 to 20 mA. Settling-time requirements for DACs vary from 10 µs to 100 ms, depending on the application. Analog input ranges can be as small as ±10 mV, from bridge transducers, as large as ±10 V, from actuator controllers, or 4-to-20 mA currents, in industrial process-control systems. Conversion times, depending on the required accuracy and choice of ADC architecture, vary from 10 samples per second to hundreds of kilosamples per second.
Digital isolators-—optocouplers or electromagnetic isolators—are used to isolate the ADCs, DACs, and signal-conditioning circuitry on the field side of the system from the controller on the digital side. If the system must also be fully isolated on the analog side, a converter would be necessary on each channel of the input or output to maximize the isolation between channels—and isolated power would be necessary, via transformers or Analog Devices isoPower™ technology.
The iCMOS™ Process
This powerful combination allows a single
chip design to mix-and-match 5-V CMOS circuits with higher-voltage
Thin-film resistors, with their 12-bit initial matching, 16-bit trimmed matching, and temperature- and voltage coefficients up to 20 times better than conventional polysilicon resistors, are ideal for high-precision, high-accuracy digital-to-analog converters. On-chip thin-film fuses allow digital techniques to be used for calibration of integral nonlinearity, offset, and gain in high-precision converters.
PLC Output Module
Two architectures are commonly used in PLC-system analog output modules: DAC per channel and sample-hold per channel. The first uses a dedicated DAC in each channel to generate its analog control voltage or current. A variety of available multiple DACs provide economy of space at low per-channel cost—but channels requiring interchannel isolation usually employ single-channel DACs. Figure 2 shows a typical DAC-per-channel configuration. The simplest DACs are low-voltage single-supply types, operating from a 2.5-V to 5.5-V supply with an output range of 0 V to VREF. Their output signal may be conditioned to generate any required voltage- or current range. Bipolar-output converters, which require dual supplies, are available for use in output modules that must supply bipolar output voltage ranges.
Figure 2. DAC-per-channel architecture.
Table 1 shows a selection of 16-bit multichannel D/A converters suitable for implementing PLC output modules. These products offer either bipolar or unipolar output ranges, with settling times of 10 µs. Other members of these families, available with 12-bit and 14-bit resolution, are pin-compatible with the 16-bit versions—allowing a direct upgrade path from 12 bits to 16 bits with no hardware changes and minimal software changes. Most of these devices include an on-chip reference, thus offering a fully integrated output solution.
Quad D/A converters are ideally suited to nonisolated multichannel output designs where up to four different output configurations can be implemented using external signal-conditioning circuitry. For example, Figure 3 shows how the AD5664R, a 16-bit quad voltage-output DAC, can deliver its specified 0-V to 5-V output range—or be connected for a variety of standard output voltage ranges, or for current-sink output, using an external quad op amp. In the bipolar-output configurations, the external output of its internal reference provides the necessary tracking offset voltage. The AD5664R operates from a single 5-V power supply, includes an internal 2.5-V, 5 ppm/°C reference, and is available in a tiny 3 mm × 3 mm LFCSP package.
Figure 3. Implementing ±5-V, ±10-V, 0-V-to-10-V, 0-V-to-5-V, and current-sink outputs using a multichannel D/A converter.
Figure 4 shows a single-channel converter in an isolated 4-to-20-mA current-loop control circuit. The AD5662, available in an SOT-23 package, is ideally suited to applications where full isolation between the analog output channels is required.
Figure 4. A 4-to-20-mA current-control circuit.
The AD5662 has a maximum output voltage span of 5 V, provided by the ADR02 voltage reference, which derives a precise, regulated supply from the variable loop voltage. The 5-V DAC-output span is converted to a 4-to-20-mA current output using an operational amplifier and transistor circuit. Since the noninverting input of the op amp (N1) is a virtual ground, the op amp regulates the current, IS, to maintain equal voltage drops across RS and R3, hence
Current summation at N2 provides the loop current,
Summing the currents at N1:
The 4-mA offset component of the loop current is provided by the reference:
The programmable 0-to-16-mA component of the loop current is provided by the DAC:
Figure 5. Single-DAC architecture.
Table 2 shows a selection of single-channel, 16-bit D/A converters, with full-scale settling times from 4 µs to 10 µs. Well-suited to sample-and-hold output architectures, they are provided in small-form-factor surface-mount packages.
Switches and Multiplexers
For other applications, where very low on resistance is necessary, the ADG1408 and ADG1409 ±15-V multiplexers offer a maximum of 9 V over the full signal range. Besides having low RON, their excellent on-resistance flatness (with voltage level) makes them an ideal solution in applications where low distortion is essential for reliable, predictable circuit performance.
Table 3 tabulates capacitance, charge injection, and RON for a selection of iCMOS switches and multiplexers. They are compared with a popular earlier type, the ADG508/ADG509.
Galvanic Isolation for Power and
The popular optocouplers contain light-emitting diodes (LEDs)—to convert electrical signals to corresponding light intensities—and photodetectors to convert the light back to electrical signals. In general, their LEDs are characterized by low conversion efficiencies—and the photodetectors by slow response; overall, optocouplers tend to have limited life, as well as excessive performance variation over temperature, speed, and power consumption. They are generally limited to one- or two-channel configurations and require external components to configure complete functions.
A new approach to isolation has been developed at Analog Devices, combining chip-scale transformer technology with integrated CMOS inputs and outputs. These iCoupler® devices are easy to use—with smaller size, lower cost, and lower power requirements than optocouplers. Available in a wide variety of channel configurations and performance levels, with standard CMOS interfaces, iCouplers require no external components—and provide high performance and stability over temperature, supply voltage, and life. A typical iCoupler isolation IC, the ADuM2400 quad isolator, with its interfaces and coupling transformers, is shown in Figure 6.
Figure 6. Block diagram of ADuM2400 quad isolator.
iCouplers have two- to four-times faster data rates and timing specifications than commonly used high-speed optocouplers—and they operate with as little as 1/50 the power of optocouplers, with correspondingly less heat dissipation, improved reliability, and reduced cost. Table 4 shows available channel configuration options.
Providing isolated power from the system side to the field side in fully isolated systems is another challenge that has an emerging solution. Traditional techniques employed in transferring power across an isolation barrier include either a separate, relatively large, expensive dc-to-dc converter, or a difficult-to-design-and-interface discrete assembly. A newer and better approach, currently available to supply up to 50 mW, is to use a complete, fully integrated isolation solution involving signal- and power transfer across an isolation barrier using microtransformers. The ADuM524x isoPower family of products provides isolation up to 5 kV, for both signal and power, within a single component—eliminating the need for a separate, isolated power supply and significantly reducing the total isolation system cost, board area requirements, and design time. A typical device is shown in Figure 7. All products have achieved UL, CSA, and VDE safety approval.
Figure 7. Block
dual-channel isolator with integrated dc-to-dc converter
PLC Input Module
Analog Input-Module Low-Level Signal Ranges
A variety of A/D converter types can be used for industrial and PLC applications—including successive-approximation (SAR), flash/parallel, integrating (including sigma-delta), and ramp/counting. The selection of an ADC for a particular application is primarily dictated by the input-signal range required by the input transducer—as well as the required accuracy, signal-frequency content, maximum signal level, and dynamic range. The most widely used architectures are successive-approximation and sigma-delta.
Successive-approximation ADCs provide resolutions from 12 bits to 18 bits with high throughput rates; they are ideally suited to multichannel multiplexed applications, where a number of input channels need to be monitored at reasonably high sample rates.
Sigma-delta architectures provide resolutions from 16 bits to 24 bits. They employ high oversampling rates and digital filtering to achieve their high resolution and accuracy—but at lower throughput rates than are achieved by SAR types. Sigma-delta architectures generally incorporate programmable-gain amplifiers (PGAs) on the front end; in converter-per-channel applications, this allows a direct interface—without signal conditioning—between the sensor and the ADC.
A key requirement when measuring low-level signals from thermocouples, strain gages, and bridge-type pressure sensors is the ability to perform a differential measurement to reject common-mode interference and provide a more stable reading in the presence of noise. In industrial applications, for example, differential inputs are used to cancel common-mode noise or interference from motors, ac power lines, or noise sources that inject noise into the analog inputs of the A/D converter.
Single-ended inputs, which are lower in cost, provide twice the number of input channels for the same number of input pins, since they require only one analog input per channel and are all referenced to the same ground point. They are mainly used in applications with high signal levels, low noise, and a stable common ground.
Figure 8 shows many of the elements that could be included in a discrete implementation of an isolated PLC input module—including excitation and input signal conditioning, fault-protected multiplexers to handle a number of input signals, programmable-gain amplifier, and A/D converter. Many of these functions, formerly assembled with discrete collections of ICs and passive elements, are now available—fully integrated and characterized—in A/D-converter and analog front-end ICs.
Figure 8. Functions embodied in a typical discrete PLC input module.
For example, the AD761x (16-bit) and AD763x (18-bit) families of iCMOS PulSAR® ADCs with internal references provide programmable input voltage ranges (0 V to 5 V, 0 V to 10 V, ±5 V, and ±10 V) that allow the designer to change inputs on-the-fly. For these devices, all switching is done via internal registers, eliminating data latency and providing improved channel switching speeds. Table 5 shows a selection of 16-bit/18-bit PulSAR ADCs ideally suited to PLC applications.
Another, more highly integrated example is the AD7792/ AD7793/AD7794/AD7795/AD7798/AD7799 family of sigma-delta ADCs. Besides their ultralow noise (40 nV) and low power requirement (400 µA), the family also provides features such as an on-chip PGA (gains from 1 to 128), voltage reference, sensor-excitation current sources, and clock—in a small TSSOP package. The combination of very low noise and low power makes these devices ideal for applications requiring high-accuracy measurements.
These ADCs can be connected directly to the sensor interface in many applications, including PLCs, temperature measurement, weigh scales, pressure- and flow measurement, and general measurement equipment. Their update rate is programmable from 4 Hz to 500 Hz, and they offer simultaneous rejection of both 50-Hz and 60-Hz signals at selected update rates. Table 6 shows features and functions offered by the AD779x family of converters.
Figure 9 shows a typical configuration using the AD7794/AD7795 to measure input signals from bridge transducers and resistance-based temperature sensors.
Figure 9. Low-level measurement using AD7794/AD7795.
When high-precision fault-protected analog measurement with voltage-input capability up to ±10 V is needed in PLCs and industrial I/O, and where high throughput for multiple channels is important, the AD7732 (two fully differential input channels), AD7734 (four single-ended input channels), and AD7738 (four fully differential or eight single-ended input channels) are ideal.
shows a typical configuration using the AD7734 to measure high
level signals typically seen in PLC- and process-control applications.
The analog front-end features four single-ended input channels
with unipolar or true-bipolar input ranges to
±10 V, while operating from a single
Figure 10. High-voltage signal acquisition using the AD7734.
Amplifiers (Instrumentation and
The AD8220 is an example of a highly versatile instrumentation amplifier that can be used in a variety of applications, such as signal conditioning between sensors—such as strain gages—and ADCs, in medical applications, programmable logic controllers, data-acquisition cards, and analog I/O cards. It offers resistor-programmable gains from 1 to 1000, with 80-dB CMR, 1-mV offset, and 10-µV/°C drift.
Operational amplifiers are the “workhorses” of analog circuitry, and Analog Devices offers one of the largest op-amp stables available in the marketplace today. Innovative design, combined with recent IC process developments, such as the high-voltage iCMOS and high-performance iPolar™, have made possible the introduction of devices for the industrial marketplace with dramatically improved performance and far greater functionality—in one-quarter the board space of older process geometries. Table 8 and Table 9, respectively, outline single- and multichannel amplifier products that are frequently used in PLC applications. They easily handle the high supply voltages generally required to support a ±10-V output range, and they feature low offset voltage and low supply current—and are housed in a small package.
APPENDIX: SELECTION TABLES (Updated versions of these tables can be found at http://www.analog.com.)
Table 1. 16-Bit Multichannel D/A Converters Targeted at DAC-Per-Channel Applications
Copyright 1995- Analog Devices, Inc. All rights reserved.