Most IoT devices initially fail EMI testing
According to IEEE, the basic definition of an IoT device boils down to any object that can be assigned an IP address and has the ability to transfer data over a network. Thus, an IoT device can be just about anything. Indeed, IoT devices range from low-cost consumer gadgets like $5 Bluetooth-enabled key finders to complex highly advanced sensor grids for medical, manufacturing, transportation, and utilities. Most (but not all) IoT products involve some sort of wireless, and for many designers, the RF component can be a stumbling block, with some engineers not taking wireless and EMI testing seriously.
Ultimately, there is no one-size-fits-all definition for IoT and what it means to be an IoT designer. Even the IEEE in the white paper mentioned above says the definition is “fuzzy.” An engineer working on the Bluetooth key fob faces very different challenges from someone designing industrial-grade sensors or life-saving medical devices. Similarly, they face very different test and measurement challenges.
One of the questions we often hear in the context of IoT revolves around whether RF testing is even necessary. For many lower-end projects, testing is often viewed as a low priority because designers use pre-certified wireless modules and often face tight deadlines to meet narrow time-to-market windows. The cost of test equipment capable of doing the job can also be a concern. Startup or boot-strap ventures often lack the capital to acquire vector signal analyzers or spectrum analyzers.
Another obstacle is the lack of expertise to handle RF testing, leading to some interesting workarounds. I recently had an opportunity to talk to a development team integrating WLAN to their airborne particle counter. They used a pre-certified WLAN module, and the project turned out to be mostly a software project. As the software neared completion, the testing involved looking at Wi-Fi speeds as an engineer moved their device various distances from a store-bought Wi-Fi router. It’s possible you could learn something from this, but it’s a far cry from true characterization and optimization.
As the above example illustrates, you can bolt on wireless modules to a new or existing design and get to a passable level of functionality. For cheap consumer-grade products, this approach makes some sense — even more so if you lack RF expertise and test equipment. This approach assumes that everything will work as intended and exactly the way your suppliers claim. That’s how things always play out in the real world of electronics engineering, right?
On the other hand, a casual approach to testing makes no sense for industrial-grade or medical devices, particularly ones that are mission-/life-critical or need to be in remote or hard-to-access places. Here, thorough testing isn’t something you should overlook or dismiss as unnecessary.
There’s a strong correlation between test and improved reliability and performance, even when using pre-certified components, because the RF environment can change significantly once they are dropped into the final product. Using pre-certified modules with custom antennas likely won’t provide the best possible power transfer and will benefit greatly from optimization using a vector network analyzer. And measuring and optimizing DC power efficiency is critical for wearables and other battery powered devices.
Testing can also help you track down problems or even get basic functions to work properly. One company I spoke with recently had issues with Bluetooth-pairing and tried different racks of premium test equipment to test the transceiver itself. Finally, they were prompted by a friend to capture traffic over the air using a USB spectrum analyzer. With analysis software to demodulate the packets being transmitted, they quickly tracked down the root cause of the problem.
The use of pre-certified RF modules won't remove requirements for EMI/EMC regulatory compliance testing. Many IoT design teams assume that EMI testing is a mere formality. Unfortunately, this is rarely a safe assumption: EMI test fees and repeated board turns can add up quickly. We recently heard from a test house in China that 90% of the products that they test fail the first time through. Given that EMI compliance test fees can range from $5,000 to $50,000, it's not unreasonable to suggest that pre-compliance testing would be a worthwhile endeavor.
The rise of IoT devices is changing RF test. Testing is often performed quickly or not at all. Unfortunately, products that don't work as well as they could may struggle to meet regulatory requirements. For us in the test and measurement industry, our response needs to center on delivering more capable, lower-cost instruments that can tackle projects such as evaluating wireless modules, basic debugging, performance optimization, and EMI pre-compliance testing. We also need to continue to improve usability so that even RF novices can get to valuable information. EMI pre-compliance testing and RF standards compliance need to be automated and well-documented.
RF and EMI testing of IoT devices, even with relatively inexpensive test equipment, can deliver significant design improvements. In future posts, my colleagues and I will look at such topics as improving IoT battery life, surviving EMI compliance testing for your IoT device, and system-level debugging.
—Dorine Gurney has over 14 years of experience as a product planner at Tektronix. She has supported product lines including high-performance oscilloscopes, signal sources, and spectrum analyzers.
Originally published on Embedded's sister site, EE Times: "IoT devices: most initially fail EMI testing."