Distributed Data Acquisition
Moving Acquisition to the Point of Measurement
Communication, the cognizant transfer of information from sender A to receiver B, is the foundation of what we humans are and do. Telegraph, telephone, television, LAN, WAN, Internet, Web, e-mail, FAX, Ethernet, cellular, wireless, satellite…technology of the ever-increasing facilitation of communication. Whether you are a scientist at the bench recording data from a pH probe suspended in a beaker or an engineer in the control room monitoring thousands of inputs from the production floor, it’s all about — you guessed it — communication.
The goal of data acquisition is to communicate signal between the system under study and the human — from anywhere. The transducer handles communication between the physical system and the electronics; the analog-to-digital (A/D) converter between the electronics and the computer; the network between the nodes and the central station; the software between the raw data and useful information; and the operator between information and knowledge. While the technology of each component continues to evolve, advances in network speed and architecture are dramatically increasing data acquisition capability and applications.
The simplest data acquisition system is a direct copper-wire connection between a single transducer and the input channel on an A/D board plugged into the data bus of a computer. The signal propagates between the transducer and the A/D board near the speed of light. This configuration, however, accommodates a very limited number of channels. One method of increasing the channel density is to incorporate a signal multiplexer that rotates the connections between each transducer and the single A/D channel at the cost of data acquisition speed. As the distance between the transducers and the A/D board increases, so does the signal’s susceptibility to noise and the cost in materials and time required to add channels to the data acquisition system.
The strategy used to overcome these limitations is to perform the A/D conversion in close proximity to the transducer and communicate the value rather than the signal to the controlling computer. This requires the employment of a digital communication network.
For the network to be useful, the communication protocol must be standardized in order to connect the wide variety of transducers available. The simplest standard protocol is maintained by the Electronic Industries Alliance (EIA) and is known as RS-232, with the RS meaning “recommended standard.” The specification allows for serial data transmission from one transmitter to one receiver at data rates up to 20 KBits/s over distances up to 50 ft. A more capable protocol is RS-485, which can accommodate 32 transmitters and 32 receivers at data rates up to 10 Mbits/s over distances up to 4000 ft.
The Institute of Electrical and Electronics Engineers (IEEE) maintains the specification for the IEEE488.2 protocol currently known as the General Purpose Interface Bus (GPIB). The protocol was developed by Hewlett-Packard in 1965 and originally designated the Hewlett-Packard Interface Bus (HP-IB). This network is based on an 8-bit parallel connection that supports data transfer speeds up to 8 Mbytes/s. The susceptibility to cross-talk between the eight data lines limits the total number of devices to 15 and the separation between any two devices to 4 m with a total cable length of 20 m in the network. Repeaters, isolators, and extenders can increase the total number of devices and extend the total distance to kilometers.
The IEEE802.3 protocol standardizes the incorporation of serial Ethernet networks as the backbone for distributed data acquisition. This protocol uses the same Ethernet equipment and cabling used by computer communication networks commonly used to build local area networks, namely, 10BaseT and 100BaseTX. These networks support 10 Mbits/s and 100 Mbits/s, respectively, and a maximum separation of 100 m between Ethernet segments. Concomitant with the increase in speed, the maximum number of nodes (channels) per network segment increases dramatically.
In recent years, the speed and capacity of distributed data networks has been exploding. In 1995 the PCI Industrial Computer Manufacturers Group (PICMG) defined the CompactPCI specification, which is based on the omnipresent PCI (Peripheral Component Interconnect) local bus architecture. In 1997, National Instruments extended the CompactPCI specification for data acquisition and automation systems creating an open specification called PXI (PCI eXtensions for Instrumentation). Using PXI A/D modules in conjunction with the communication protocol termed MXI-3 (for Multi-system Extension Interface, Generation III — also developed by National Instruments) yields a serial network running at 1.5 Gbits/s. This results in a network with 132 Mbytes/s burst and 80 Mbytes/s sustained data transfer rates at distances up to 200 m between repeaters.
With the arrival of such high data transfer rates and nearly unlimited number of nodes available to the modular network, data from temperature transducers, strain gauges, current transducers, voltage transducers, flow monitors, pressure transducers, position indicators, and (gasp!) digital cameras…now the limiting reagent in distributed data acquisition is not the network. No, it is creating a rational and workable approach to handling all of the information. Communication is great, but we all need a little peace and quiet now and then.
This material originally appeared as a Contributed Editorial in Scientific Computing and Instrumentation 17:5 April 2000, pg. 14.
William L. Weaver is an Associate Professor in the Department of Integrated Science, Business, and Technology at La Salle University in Philadelphia, PA USA. He holds a B.S. Degree with Double Majors in Chemistry and Physics and earned his Ph.D. in Analytical Chemistry with expertise in Ultrafast LASER Spectroscopy. He teaches, writes, and speaks on the application of Systems Thinking to the development of New Products and Innovation.
