Sensing trouble out at the network edge
The device could tell you that an earthquake had occurred, and it could provide some rather vague information about where it might have happened. It wasn’t much, but it was a start. Sensors and their accompanying recording devices have been growing steadily more sophisticated ever since.
In the modern world sensors are used in ways that Zhang Heng could never have imagined, recording everything from exhaust gas temperatures to the electrical activity in a human heart. And they don’t just collect data anymore. They’re becoming intelligent. Embedded microcontrollers and built-in software are turning them into thinking, communicating, and active nodes of intelligent networks, systems that demand localized intelligence and real time, sensor-driven analytics. You don’t just connect them anymore; they communicate with you.
The network periphery
Earlier generations of sensors reported their data via basic analog or digital connections. These days they’re becoming Ethernet-aware, with all of the accompanying advantages. Their use of universal Ethernet communications protocols enables them to work with off-the-shelf technology, and their ability to communicate via Ethernet means that they can be placed just about anywhere.
But Ethernet technology has its roots in safe, climate-controlled office environments. The IT world thought in terms of structured cabling systems, mature protocol and transport standards, and hardware vendors who would provide standardized products with near-seamless interoperability. Who knew, back then, that networking would outgrow its tame, office-based beginnings and move out into the real world?
Nowadays networks must function reliably in increasingly harsh terrain – on factory floors, on gas and oil pipelines, in industries ranging from mining to transportation. And sensors tend to live way out on the network periphery, where conditions are the very worst. Design engineers must know what to do about that.
Getting to the Ethernet
Connecting sensors to Ethernet can be problematic in real world scenarios. The first issue is distance. Copper wire-based Ethernet has a practical range limitation of 100 meters. That’s adequate for a network in an office or a small building, but it won’t do the job when you need to monitor the turbines on a wind farm or the chlorine levels at a water treatment plant. To be useful outside the office, Ethernet must function at far greater ranges.
One answer is a device called an Ethernet extender – see figure 1. Ethernet extenders use DSL technology to create a long-distance Ethernet bridge over virtually any available copper pair. There’s a drop in bandwidth as the range increases, but you can reliably extend Ethernet over thousands of feet while maintaining a quite serviceable connection rate of several Mbps. Better yet, Ethernet extenders give the system designer the freedom to use existing wiring infrastructure, like any telephone cabling or legacy coaxial cable that may be present. As the labor and materials involved in cable installation are often the most expensive element in setting up a network, the flexibility provided by Ethernet extenders can represent an enormous savings.
Fiber optics
A second option is fiber optic cable. There are two kinds. The cheaper option, multi-mode fiber optic cable, uses LED light and can carry data several miles. Its long-range cousin, single-mode fiber optic cable, transmits with a laser rather than an LED. It’s more expensive than multi-mode cable, as are the associated transceivers and receivers. But it has the ability to transmit data over great distances. Telephone and cable companies use it for their long haul applications. Both kinds of fiber optic cable provide far greater bandwidth than copper wire. And they’re impervious to EMI interference, as the data is carried on a beam of light rather than copper. This is invaluable in industrial applications, for example, where the electric motors on the machinery can generate powerful magnetic fields.
Given that labor costs typically represent the largest expense in a cabling installation, it often makes sense to go straight to single mode fiber, even if its capabilities exceed current requirements. Considering the rapid growth of vision-based sensor systems, and their ever-increasing need for bandwidth, starting with a fiber optic installation at the very beginning may soon turn out to be a very wise investment.
What happens when cabling isn’t a practical solution?
One solution to the range issue is 802.11, which has evolved far beyond the coffee shop “hot spot.” Today’s WiFi can provide robust IP-based connectivity out to thousands of meters. Using a pair of 802.11-based wireless Ethernet bridges, a local sensor network can easily be bridged back to a SCADA system or enterprise location. Some wireless bridges can even connect at distances of many kilometers while maintaining connectivity rates well into the tens of Mbps – enough to handle even high-bandwidth, vision-based sensor systems. In many cases sensor manufacturers are embedding WiFi directly at the sensor level. Rather than make the engineering and compliancy investment required to develop devices of their own, they often prefer to make use of pre-certified WiFi modules from reliable suppliers – see figure 2.
What about physical connections?
Ethernet hardware isn’t normally designed for tough conditions. For example, when installed in difficult environments, the popular RJ45 connector needs to be strengthened and protected by an over-connector. The standard switches, hubs, bridges, wireless transmitters and other physical devices that help make up a network suffer from similar weaknesses. They won’t stand up to the conditions they’ll encounter as networking continues to expand into new applications and new territory.
Designers will need to specify equipment built for the job, and it won’t be found at the local office supply store. Look for industry-hardened devices built by technology companies that specialize in ruggedized equipment.
Ground loops and power surges
Ethernet connections help sensors communicate, but they can also leave sensors vulnerable to damage from power surges and spikes. The greater the distance between two connected devices, the more likely it is that they will have different ground potentials and the associated risk of damaging ground loops. Lightning strikes and other power surges can also travel on copper Ethernet cable to burn out integrated circuits and connections. Sensors can be protected by Ethernet isolators, which allow data to pass unimpeded, but control electrical flows – see figure 3.
Figure 3: Sensors can be protected by Ethernet isolators.
Sensors may be out on the network edge, and they may have to function in some pretty rough environments. They’ll still be expected to communicate with reliability, accuracy and durability. And a sensor with poor Ethernet connections isn’t much of a sensor at all. No matter how well designed it is, and no matter how capable, if it can’t communicate it isn’t going to be much more useful than Zhang Heng’s old one-trick seismograph.
The author, Mike Fahrion, is Director of Product Management at B&B Electronics – www.bb-elec.com.
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
