We recently interviewed Justin Bessette, Manager, Wireless Systems and Software Engineering at LORD Corporation, Micro-Strain® Sensing Systems, about the nuts and bolts of a wireless sensor system.

Figure 1.

Justin Bessette: A Sensing system is comprised of four main elements:

A transducer senses a physical medium, such as pressure or temperature, and converts it to an electrical medium, in the form of an analog signal.

A node takes that electrical analog and converts it into a digital signal that can then be transmitted wirelessly. It includes signal conditioning, an analog to digital converter (ADC), and a transmitter.

A gateway takes the digital data and outputs it in a form that can be used by other software-controlled devices.

Software determines how the data is used. It could typically be used for a graphical user interface (GUI) or for machine-to-machine communication. The software could also direct the data to the “cloud” where it could be shared with others, anywhere.

We call these four main elements a wireless system.

Tech Briefs: Could you elaborate on gateways?

Bessette: Gateways convert a generic digital input signal from a node to a standards-based wired transmission. Our gateways are capable, of out-putting data onto a TCP type of connection over Ethernet; to a USB device; or RS-232 or RS-485 serial buses.

Figure 2. Typical transducer+node — a battery powered wireless 3-axis accelerometer.

Tech Briefs: How can the software be used?

Bessette: The software takes the raw data — data stored as a numerical value on a disk and it makes it presentable to a human to view and use the data or to another machine. It allows you to move data around. The software also allows you to change configuration options, for example: I'm changing a filter out on a node, or I'm changing the channel that's on, or I'm changing a scale factor — a calibration value. So, the software allows you to receive and manipulate the sensor data you're trying to measure. And it allows you to also configure and control the node or the data acquisition devices out on the remote side.

Tech Briefs: Could you describe machine-to-machine interaction within a wireless system.

Bessette: An applications program interface (API) is software that allows you to talk to hardware programmatically. Machines speak programming languages — protocols — so that having an API to your hardware allows you to get data to and from your devices to wherever the control system, or the remote system is. It's a structured piece of software that allows you to communicate with another system.

Tech Briefs: Suppose I have a factory with different, functioning sections like different assembly lines and machining functions, etc. How would I go about coordinating and using the data from all of the different sensors? How would I program the whole thing to work as an integrated system?

Bessette: In the early days of computer-controlled networking, you would use a programming environment such as C or C++. You'd look at a protocol document and you'd find out what string of bytes to send and parse to fit into a document. But there are many more ways of doing it today. With an API, it's done at a higher language level where you can make simple commands like, ‘Turn node X on and turn node Y off,’ instead of saying: ‘Send this byte string or look for that byte string.’ It gives you a higher-level way of communicating with devices.

There are also graphical systems like LabVIEW and scripting systems like Python or MATLAB. In the case of our products, we have an API that can work on all of those platforms, so that whatever platform the customer is using they're able to use our products within their system.

Figure 3. Vibration sensor for predictive maintenance.

Tech Briefs: So, the APIs kind of go between the sensors and the controlling computer?

Bessette: Yes, the controlling computer will have some program on it that speaks the API language and does what it's supposed to do.

Tech Briefs: How are the different parts of the system synchronized?

Bessette: In a traditional wired data acquisition system you have a single clock. Since everything is wired together, everything is inherently time synchronized.

In our wireless system, a master clock resides in the gateway and the nodes synchronize their clocks to the master. We use time division multiple access (TDMA), which is similar to the way cellular phone systems share bandwidth. With a really high precision clock and a distributed clock system, everybody can share a piece of space. With that sharing you also get time synchronization, which is really important when you're trying to have a sample system with a distributed clock. If you're measuring strain or acceleration, for example, at any number of locations, all at the same point in time, it's vital that the data is coordinated to reflect that. One more note: high timing accuracy is important. If you're trying to sample something at say, 10 kHz, a drift of a couple or even a few seconds a month is enough so that you would have a misalignment in time.

If you're trying to do, let's say, a nodal analysis — perhaps you're trying to look at an event that propagates down the wing of an aircraft, like a physical wave of strain or something like that. If you're trying to see how that propagation goes, you need samples to all be taken at a very similar time, if not exactly the same time. So, if clocks drift, you become less and less able to do an analysis of multiple distributed sensing points.

Tech Briefs: How do you get this high accuracy synchronous clock? What kind of clock do you use?

Bessette: At its simplest form it's just a precision oscillator with low temperature drift. We can also align our system's time to GPS's, which are atomic clocks in the cloud. We don't need atomic clocks in every device because we can remotely synchronize them. This same kind of process is used in cell phones. The only way all cell phones can operate together is if they have accurate clocks and they know when they can be on and when they cannot be on. Wireless is a shared resource, so all devices in a single spectrum need to coordinate, otherwise it's just going to be chaos and noise, and nothing will work.

Tech Briefs: Not all sensors are created equal. Can you discuss some of the differences.

Bessette: It's something like a Ferrari vs. a Fiat. They both drive, so they both do a single task, but one can do it a lot faster and with a lot more precision. You have the same thing in a data acquisition system. You can make a voltage measurement — let's say you have a tenth of a volt precision — that might be good for some systems. But the typical users of our systems are looking at instrumentation grade sensing, so you want to have more decimal places, you want to have better filtering, you want to have a high-fidelity signal.

Tech Briefs: Could you give me an example of what could make a sensor perform better.

Bessette: Better signal conditioning.

Tech Briefs: Where is the signal conditioning done?

Bessette: It happens at the node, at the wireless data acquisition box. High quality signal conditioning allows you to amplify the signal while rejecting the noise. Done properly, you can have better insights into what you're trying to measure. It could mean earlier warning, it could mean seeing something you wouldn't otherwise see.

Tech Briefs: So, the level of signal quality is determined by the node?

Bessette: Yes, because this isn't just a data acquisition system, it's a distributed data acquisition system, and more specifically a wireless distributed data acquisition system. And the data acquisition system is done right out of the node. That's where we take that physical data, in the form of an analog signal and convert it into a digital form. At that point you've captured the maximum amount of precision.

Tech Briefs: Could you talk about how the quality of sensors can vary.

Bessette: In my world, we call it a transducer. A transducer acts as the conduit from the physical to the electrical. The most common example I could think of would be a microphone, which takes the sound pressure wave and converts it to a voltage by a change in resistance or capacitance in an element or by varying a magnetic field. These methods have different degrees of fidelity to the actual pressure waves.

Tech Briefs: What about the reliability of the wireless connections between the nodes and gateways, especially in a factory?

Bessette: Utilizing best practice techniques for transmission, using the latest transceiver technology from the semiconductor manufacturers, we can mitigate that. When you're transmitting the data and trying to receive it with the latest semiconductor technology, you can see down into the noise a little bit further — that gives you a little more distance of transmission. But you can only improve that so much. But there are other areas you can improve. A data acquisition system like ours, needs to send the data and then make sure that the receiver receives it. If it didn't receive it, we transmit it again and again until it gets it. Because we are dealing with this imperfect system: wireless, we have to employ tricks and techniques to make it perform as well as it should. That involves retransmitting data, the time synchronization we talked about, and the latest semiconductor technology to peer down into the noise floor and pull out the data stream that's there. These days we can go to a semiconductor fabricator and buy a product they have that's tuned really well for our purposes. This is actually what is done by all the consumer electronics device-makers, such as cell phones and computers that are wireless. They buy chips from major manufacturers that have invested tons of money on their transceivers.

Tech Briefs: Actually, I'm just realizing that I'm recording our conversation on my smartphone and it's blocking all of the ambient noise around me.

Bessette: That's exactly what I'm talking about. In this case, it's sound, which is vibration, which is a typical thing you measure on equipment.

Tech Briefs: What about the ease of getting a system up and running?

Bessette: One of the biggest benefits of a wireless system, is that you don't need as much time to install it. With a wired system, you have to run data lines in a factory, or on a vehicle that has rotating or articulating parts, which can be very difficult. It takes a lot of care as to how you wire that — the signals can be corrupted in that length of wire. In a wireless system, you simply need to install the transducer at the location, and you're done. So, for a manufacturer who's installing or retrofitting, this means a big savings of time and money.

Tech Briefs: For retrofitting, would you just cut short the signal wires from the transducer and connect them to a node?

Bessette: Yes, that's one way. Or if you have an old machine that never had a sensor on it, you could just slap one on and you'd have a vibration monitoring system that took five minutes to install.

Tech Briefs: What kinds of distances are achievable between a node and the gateway?

Bessette: That's a difficult question to answer. In open air, it's very easy — point-to-point, it's about a kilometer. But more typically, they're installed in factories or vehicles. It reduces the distance, in an office environment by about half, in a factory environment, maybe by another half — 250 meters.

Ultimately, to deploy a wireless system, you should do a site survey — put the system out there and see if it works. We provide a “network reception tool,” that allows you to look at the spectrum and see which channels would be best to use.

This article was written by Ed Brown, Editor of Sensor Technology. For more information, visit here.