Article tools sponsored by

Simple. Not Stupid.

Smart sensors are too complicated for many applications. Plug-and-play sensors, although simpler, offer some of the same advantages.

Gary Legg, Contributing Editor -- Design News, January 13, 2003

Once upon a time—about ten years ago, actually—visionary technologists dreamed of an easier way to apply sensors in industrial automation, process control, and test and measurement systems. Instead of numerous sensors connecting to a system's control hardware via individual cables, each sensor would simply tie to a network, and all the sensors would then communicate over a single cable. Inches-thick bundles of impossible-to-sort-out wires would be a thing of the past.

The visionary technologists also dreamed that all the networked sensors would be smart. Each sensor would have built-in intelligence and could therefore automatically convert what it senses (a voltage, say) to what a sensor user actually wants (temperature in degrees Celsius, perhaps). A smart sensor could also, by examining its own calibration data, perform minor miracles like compensating for its own deviations from sensor perfection. And, of course, all the sensors would have digital outputs-the better to communicate with the digital network. The network would even be a universal, standard network-no more bloody battles among dozens of incompatible fieldbuses.

Alas, not everyone lived happily ever after in this sensor fairy tale. Although the visionaries' creation—smart-sensor standard IEEE 1451—received official sanction, it was so complicated that virtually no one applied it to commercial applications. And yet, using large numbers of sensors in an application remains complicated and difficult, something that the IEEE standard, for all its own complexities, intended to alleviate. In structural test, for example, you can have hundreds of sensors connected to an airplane or an automobile chassis, and just keeping track of which sensor is plugged into which instrumentation channel is both time consuming and prone to costly error.

Try Again

And so, the visionary technologists are trying again with some proposed modifications to IEEE 1451, one of which—IEEE 1451.4—takes a backward step toward simplicity. Instead of attempting to define know-it-all, do-it-all smart sensors, the proposed IEEE 1451.4 defines so-called plug-and-play sensors that are simply easier to work with. When you add a plug-and-play sensor to a system, the sensor tells the system what kind of sensor it is and what its characteristics are. System software can then use that information to configure the setup automatically. Tedious, error-prone, manual configuration is unnecessary. And, best of all, some of these sensors are already available.

Dual Ouptuts Key - Plug-and-play sensors retain the traditional analog sensor output and add a digtal interface fro reading and writing information that describes the sensor.

Compared to the smart sensors defined by the original IEEE 1451, plug-and-play sensors are very simple. They're not networked; each one connects directly to the instrumentation system. They're not smart, either; they don't contain any processing ability, so they can't perform self diagnostics or use their own calibration data to compensate for inaccurate measurements. What they can do, though, is simply work properly when they're plugged in, with little or no manual intervention.

And plug-and-play operation by itself is a laudable goal, says David Potter, vice chair of the IEEE 1451.4 standards committee and a platform manager at National Instruments (Austin, TX). "It doesn't solve the networking problem," Potter says, "but it does start to add some intelligence to sensors." Also, Potter notes, IEEE 1451.4 establishes an analog sensor connection, rather than digital, as in the standard's first incarnation. Because the dominant sensor interface by far is analog, Potter says, the new proposed standard doesn't require, as the earlier standard did, a major shift in how people use sensors.

The heart of a plug-and-play sensor-and what makes plug-and-play operation possible-is a built-in "transducer electronics data sheet" (TEDS) that describes the sensor and its operating parameters. The TEDS resides in a sensor in a small, inexpensive electrically erasable programmable memory chip (EEPROM). It includes information such as manufacturer, model number, serial number, measurement range, electrical output range, sensitivity, and calibration data. TEDS-aware signal conditioners or instrumentation software can read this information and thereby know how to configure the sensor into the instrumentation system. The system then uses the information to convert sensor data into accurate measurement units.

Advantages

The advantages of plug-and-play are many, says Martin Armson, marketing director of sensor manufacturer Sensotec (Columbus, OH). For one thing, he says, "Plug-and-play eliminates the need for you to read and enter data from a paper calibration sheet, and you needn't worry that a calibration sheet will get misplaced or lost." Also, Armson says, because sensors vary in sensitivity, swapping one sensor for another in a setup always requires some system readjustment, which plug-and-play systems can handle automatically. "The signal conditioner reads the data from the new sensor's TEDS, adjusts its electronics, and you're testing again almost immediately." Armson notes that reading the TEDS to acquire information about a sensor also makes it possible for a sensor from one manufacturer to work with a signal conditioner from another.

These same advantages ostensibly were available in the original IEEE 1451, however, because that standard also defined a TEDS that made them possible. So why didn't 1451 take off? In short, says Potter, "It was too complicated." Smart sensors as defined by the original 1451 had digital outputs, Potter notes, so even if a sensor were an analog device at heart, it would have to digitize its measurements before presenting them to the measurement system that it operates in. That's an unnecessary step, Potter says, because "we have systems and infrastructure and instruments and sensors all working with analog interfaces."

Another problem with the original standard was that, for many applications, it was overkill. In test and measurement, for example, says Armson, networking is often unnecessary. "You don't have the distances that require networking," he says, "and you don't have the permanent installations, either." In an automotive test application, Armson notes, "You might put in an engine, put all the test equipment on, and then three days later take it all off again. So hardwiring would be much more the norm."

The original IEEE 1451 was sometimes underkill, too, and again in the test and measurement arena. For test, Armson says, as opposed to many other applications, "The data rates required are much higher generally, and often the network doesn't give you the real-time data rates you need."

Analog and Digital

To give test engineers something more akin to what they really need, IEEE 1451.4 keeps sensor signals in analog form and adds a simple digital interface to read the TEDS. Some plug-and-play sensors (notably accelerometers) connect with two wires, just as traditional sensors do, and an attached signal conditioner—by activating a simple circuit embedded in the sensor—can alternately place analog sensor data or digital TEDS data on this two-wire interface. Other sensors have four wires, two for analog sensor data and two additional ones for TEDS data. The TEDS itself is smaller than in the earlier versions of IEEE 1451—as small as 64 bits-because it doesn't have to include some information needed to describe the earlier, more complicated sensors.

In an interesting wrinkle, National Instruments has also introduced the concept of a virtual TEDS, which makes plug-and-play possible even for existing sensors that weren't manufactured with plug-and-play in mind. As Potter notes, "One of the main strengths of 1451.4 is the TEDS, a standardized definition of the critical parameters to describe a sensor and what a sensor does and what its electrical interface is." But that same information is useful to any sensor, Potter says, whether or not it actually has an EEPROM embedded in it. Consequently, National Instruments is creating a database of TEDS files for legacy sensors, which appropriately implemented instrumentation software can access much as it would an EEPROM-based TEDS.

The concept of virtual TEDS is also appealing for some harsh environments. For example, Potter says, "If you have a sensor or a thermocouple that you can't put a chip in, because it's going into a an environment at a thousand degrees, you can get the TEDS information from a file, and your software can use it in exactly the same way."

To increase the chances for IEEE 1451.4's formal adoption, National Instruments and twelve sensor manufacturers have recently established a Plug & Play Sensors Program. As part of the program, NI has introduced a TEDS library that enables use of plug-and-play sensors with its LabVIEW software. The twelve sensors companies are making, or plan to make, products that conform to IEEE 1451.4, even before the standard is ratified, which will probably occur sometime this year. Any changes in the final, adopted standard, they say, can easily be accommodated with software changes.

Real Sensors

Plug-and-play sensors don't yet exist in great abundance, obviously, but they do exist. Sensotec, for example, has made plug-and-play sensors for the past eight years. Its SIG CAL sensors, in fact, served as the prototype for IEEE 1451.4, which makes only minor changes to the SIG CAL's TEDS format. Sensotec has now switched production to 1451.4-compliant sensors and can easily retrofit SIG CAL sensors, via software, to make them compliant also. The company can also retrofit ordinary sensors, by adding both hardware and software, to make them 1451.4-compliant. Other sensor manufacturers active in plug-and-play sensor development include Endevco (San Juan Capistrano, CA), PCB Piezotronics (Depew, NY), and Kistler (Amherst, NY).

Will plug-and-play sensors achieve the success that smart sensors haven't? That's not a foregone conclusion, despite their obvious appeal. Armson notes that Sensotec, a plug-and-play pioneer, still sells far fewer plug-and-play sensors than traditional sensors, with many of them going into test rigs that are set up by engineers, but used by technicians. "This is pretty neat technology because it's simple," Armson says. "Our challenge now is to find people who really want to use it."

Contributing writer Gary Legg can be reached at garylegg@media16.com.

• related stories
• resource center
• by this author

 

 

 

• Slick Slidin'

• MEMS Sensors Rev Their Engines

• Liquid Assets

• Wireless Act

• In the Air, Everywhere

• Silicon Power Device Challenges Relays

• Phone home-or elsewhere

• Getting the signal

View more articles by Gary Legg

Advertisement

Sponsored Content

Technology Marketplace

Article tools sponsored by

Find a supplier on oemsuppliersearch.com

Talkback

We would love your feedback!

» View All Talkback Threads

» Submit talk back

Advertisement

DN's Resource Center Get Free Information, Made Easy

• Embedded System Design for Machine Control
  National Instruments | White Paper
  Request Item
• Mechatronics Technical Library
  National Instruments | White Paper
  Request Item

View More >>

View all DN resources >>