One good way to do well in thermal testing: Wide Temperature Components
A system’s operating temperature looks like a simple range of temperatures, but it is really a game of 3D chess. The components (their operating temperatures and thermal limits), the ambient temperature in the environment and the system utilization come together to give a system a rating that is the key to knowing if your computer can take the heat common in industrial & manufacturing environments. The wrong 30°C PC in a 40°C environment can spell disaster: unintended shutdown, data loss, WiFi failure and more can all happen thanks to temperature. So how do you insulate yourself from failures due to heat? Read on.
Selecting a PC: Its about the parts . . . sort of
Each major component in a PC has its own operating temperature. So to get a system capable of withstanding 60C heat, you need only select systems with parts with specs above 60C, right? Not exactly. Operating temperature represent the operating limits of the part and physical temperature at which the component will fail. Each component will have a range, like the EMPHASE wide temperature drive’s -40°C to 85°C. But components such as RAM, hard drive, WiFi, and processor don’t run in isolation, and they all generate heat. They warm the air around them, and pretty soon a room that is 30°C has parts operating in a 60°C environment. Anyone who has put their hand on the side of PC case will tell you they can get pretty warm.
The case makes the difference
Given that components work in an enclosed area, the most important item when it comes to helping determine the rating is the case. The design of a case determines how much air flow there is, how close components are mounted to each other, how heat distribution systems — like heat pipes, heat sinks or fans — are deployed and much more. As the components begin to heat up, the case can be the differentiator. A good case will pull heat away from the components. A bad one is an oven in its own right.
System Operating Temperature Determination, pt 1
To get to the computers operating temperature, first you need a baseline. To establish this, the thermal performance of a PC is assessed at room temperature (20-21°C). Every computer manufacturer worth its salt has a testing process they run machines through. They are be proprietary and confidential, but all tend to do the same thing: engage the system’s resources to various degrees, while recording the temperatures of the components.
Using this data, the laws of thermodynamics, and some complicated math, they can extrapolate temperatures above and below room temperature and estimate the temperature point where the various components will reach their fail points. This gives the system a preliminary operating temperature, and is often used to quickly assess a configuration.
Case Study – The ML300: The ML300 is a Fanless Ventless NUC from OnLogic. It was recently sent to Intel for Thermal Testing, which uses a 100% utilization. The thermal limits on components are:
- Processor was 105°C,
- WiFi was 80°C,
- Hard drive (a SSD) was 80°C
- Memory was 85°C
Intel tested it, and verified the system for 50°C, which was the highest the ambient temperature could be raised to before one of those parts ticked over its limit. Interestingly at 55°C the WiFi was going strong, but was reading over 80°C and the other parts were still below their limits.
This is both a good example of good case design, with well placed heat sinks (the ML300’s case is one big heat sink), but also a good example of the intricacies of testings. The ML300 still works at 55°C. If that usage is dialed back to normal, what does the operating temperature look like? Its hard to say because that’s not part of the test.
System Operating Temperature Determination, pt 2
The next level of thermal testing involves what is known as a Thermal Shock Chamber. These are basically one part oven, one part freezer. The PC is turned on, set to run at a set rate and is heated up to a specific temperature. Components are monitored as their operating temperatures tick up. For example, a WiFi card in a room thats 20°C (Room temp) might register as operating at 50°C. When the ambient temperature rises to 40°C, that WiFi will be correspondingly hotter. The process is then repeated with incrementally higher and higher temperatures.
The shock comes in when it is rapidly cooled, usually with liquid nitrogen, from a target temperature. As the temperature drops, it’s doing two things. It’s seeing if the rapid shift causes failure and it is testing how well the various parts function at, say, 0°C.
Then they begin cranking it back up again and the whole process is repeated several times. How and when parts fail is recorded and the result is that system’s operating temperature.
And all this means what?
One of the problems with operating temperature, and testing, is the lack of standardization. Some manufacturers use full utilization, which is to say push a computer to its maximum. Some labs use optimal utilization, an ill defined estimate of standard usage levels. Some utilize a Thermal Shock Chamber, some do not. The variations go on and on, and are not always meant to be confusing or misleading, they are just all proprietary methods and thus hard to judge against each other. A PC might reach 50°C in one test and 40°C in another, only because the one tester uses software to test the temperature, and the other uses sensors on the components.
You will want to look carefully at the system rating, the case design, the wide-temp component options, the enclosure (kiosk, cabinet, etc), the software application and its resource draw, and the ambient environment. If you have an application using relatively few system resources, a 50C rated computer for a factory that never gets hotter than 50C may be OK. But if there is a big system load or local thermal increases due to an enclosure or poor ventilation, you may need more wiggle room. Don’t be afraid to ask your vendor how they verified their numbers.
Stay tuned, we are going to kick off some blog posts on:
environmental stress screening (ESS)
IP – Ingress Protection Ratings