Smart Technology, (Almost) Everywhere
After hearing the initial buzz around the “Internet of Things” and how connected devices were the future, it was hard not to joke about refrigerators sending Tweets when you ran out of eggs. While it is amazing we can get real-time inventory updates from the icebox, that IoT has permeated nearly every industry as it makes data more accessible and actionable is incredible.
MarketWatch reports the global edge computing market will double to $21 billion (€18.7 billion) in 2023, meaning the growth of IoT will push the edge farther than ever before. As businesses realize what they can do and efficiencies gained using the cloud, the need for connected technology in more remote and challenging locations is also increasing.
But, the next frontier of edge computing is anything but friendly to a computer’s sensitive internal components and require considerable ruggedization to thrive in these harsh environments.
Take autonomous vehicles, for instance, which need to be able to manage, monitor, and respond to thousands of constantly changing variables and conditions in real-time. This requires tremendous amounts of computation locally and we’ve seen projects where dual Xeon with GPU systems were an absolute must.
Raw performance isn’t always enough, especially in these circumstances. For in-vehicle applications, the amount of shock, vibration, and extreme temperatures the computer will be subjected to must also be considered. And we haven’t even gotten into the intricacies of automotive power, which only specifically-designed systems can handle.
This is one of many examples, but it shows the growing need and complexity of putting more compute at the edge where a typical computer doesn’t belong. When pushing farther out to the edge, one of the most important things to consider is the environment your hardware will be operating in and what protection it needs to operate reliably.
Enter Rugged Edge Computers
There’s only so much a computer can do when faced against the calculable reality of physics. Extreme temperatures, dust, debris, shock, and vibration – all have been the untimely end of many a PC (may their silicon rest in peace).
Rugged edge computers were developed to survive these threats and these specialized systems can operate in environments ranging from -40C up to 70C and withstand up to 50Gs of shock force (with some extremely specialized systems rated even beyond that!).
This level of durability is achieved through a combination of mechanical engineering and board design, thermal engineering, and component selection. Simply mashing all that together isn’t enough, though. Everything from design and component selection needs to be tested thoroughly to ensure the overall reliability and performance as a whole and the testing is not easy on technology, as you’ll soon see.
Mechanical Engineering and Board Design
The board and everything on it is the core piece of any computer and consequently requires the most protection. Motherboards for rugged edge computers are electrically designed to handle wider power inputs and fluctuations. The board also needs to be able to handle electrostatic discharge, extreme temperatures, and vibration which is achieved by a combination of material selection, vibration damping, and soldered components.
It’s also at the board level where feature sets are established. Because rugged edge computers are often deployed in industrial environments, having I/O that can interface with legacy equipment, receive power from different sources (such as automotive power or power from a UPS backup) also need to be accounted for. It’s not uncommon for these systems to be populated with CAN bus, DIO, and serial connectors for this reason.
As you can see, rugged edge computer motherboards have significant built-in protections, but they’re still a sensitive piece of technology.
The mechanical design of the chassis is the main defense for the motherboard and its design is primarily driven by the board size, I/O population, and thermal solutions. Not to mention, the chassis itself must also be designed in a way that can transfer or tolerate kinetic energy including impacts (drops or falling objects) and vibration (in a car or train).
One of the main failure points for computers is the fan. Removing this failure point and sealing the system from dust and debris extends system longevity, but the heat it generates must then be ejected in other ways.
Cooling for fanless systems must be done passively through the heatsink and chassis of the system, which you can learn more about in more detail here. But generally speaking, the more heat a system generates, the larger the heatsink required to keep the system operating normally. Because there’s a limit on the size of the heatsink that sits directly on the processor, the chassis must be able to accommodate the additional heat.
While a surprising amount of heat can be ejected from the CPU this way, processors will still throttle down their performance at higher temperatures (especially in environments that are hot to begin with). This built-in protection keeps the processor from cooking itself but can cause significant performance and reliability degradation.
We’ll talk more about this later, but it’s worth noting that if a system is pushed to its max constantly and is throttling as a result (as we’ve seen happen in some projects), it generally means that the hardware, as configured, isn’t up to the task. For systems under constant load, it’s far more preferable to be operating at 75-85% capacity, with room to boost when necessary, but then coming back down again. It can be easy to over or under spec (both can be costly), which is why testing and working with a hardware specialist are so important.
Adding to this growing list of sensitive technology in need of protection (while creating more heat to deal with) are the various components that populate the board; SSD and HD drives, RAM, co-processors (like Movidius VPUs), wireless and cellular cards, and even GPUs. While many of these components have robust rugged versions, they still have their own individual performance traits and temperature ranges that must be considered as the system works as a whole. And sometimes, like in the case of GPUs, only commercial options are available (which require their own unique design approach to protect and cool).
The Importance of Testing and Verification
While the combination of ruggedized components, design, and thermal engineering help create incredibly tough computers, it’s the testing of these features during the design process that helps inform how they’ll perform out in the real world.
Every manufacturer is responsible for its own testing, which can be done in-house (if testing equipment is available), at a lab (which has specialized testing equipment), or a combination of the two.
For our own rugged edge computer designs, we run a battery of tests that push the system to its limits, keep it there, and monitor its performance over time. During this testing process, we also populate the system with as much I/O and internal connectors as possible to simulate real-world usage and subject it to conditions that are worse than what the system would typically be exposed to.
Recalling our earlier comment before about throttling, there’s also a nuance between reliability testing, which refers to whether or not the system can remain on in the conditions it’s being subjected to, and performance testing, which describes how well the system can perform in those conditions. Unfortunately, there’s a bit of disparity that currently exists around this.
As we began testing our Karbon 300 rugged edge computer, for example, we had specific performance benchmarks we wanted to reach. At the design level, we wanted the system to perform to its full potential throughout the extreme temperature ranges without significant throttling, which is not an uncommon occurrence at these end ranges.
During this process, we tested several other systems to get an understanding of how they perform comparatively. What we noticed is that there is a significant performance tradeoff that occurs as a system reaches the far end of its temperature range.
What’s important to note here is that while these systems can technically operate at these extremes, how well they perform at these ranges is largely left undocumented. But this level of insight is important when you’re counting on a specific level of performance for your application, especially when speccing out systems to deploy.
Finding the Right Fit
With so many factors and data points to consider, it can be tough to find the right solution when speccing out hardware. Especially when the performance you’re expecting doesn’t align with the specifications that are attributed to the system.
Since 2003 we’ve been working with innovators all over the world integrate reliable computer hardware into their application. We’ve seen what works well and we’ve seen the challenges that hardware integrators have faced when deploying computers in places that are not conducive to their sensitive technology.
Armed with this information, we also encourage you to talk to your hardware supplier about the testing that’s been done on the systems you’re looking to spec. Has it been tested at all extremes, fully populated, and what are the tradeoffs? Ensuring that what is on paper actually matches your application requirements sometimes requires more information than what’s generally provided.
By building more flexible options with ruggedized features, we’ve been able to develop a portfolio of rugged edge computers that fit a wide range of applications and functionality.
Are you looking for a rugged edge computer for your application? Click here to browse our full line of rugged edge PCs or click here to get in touch with a system expert who can help you find the solution that’s right for you.