14 March, 2017
Designing heat out of electronics systems has been a challenge over many years and, inevitably, one that has built up its own folklore about how to go about it. Tom Gregory from 6SigmaET explains why we’ve come a long way with smart thermal simulation software.
Tom Gregory, Product Manager at Future Facilities:
I spend a lot of time talking to our customers’ designers and engineers about solving electronics cooling problems – something they’re having to do with increasing frequency as electronics becomes ever more powerful. Yet despite these power increases, we’re trying to deploy electronics in packages that need to be small, lightweight and consume little power – this is known as the ‘SWaP’ challenge to improve the ‘Size, Weight and Power’ performance.
Nowadays, it’s becoming less and less the case that we’re able to use fans for cooling. They’re too bulky, they consume power, and - being mechanical - they’re potentially a high source of unreliability. But previously, it wasn’t uncommon for someone to determine the size of fan you’d need to cool something like a PCB, for example. You’d build it, but you wouldn’t get close to the airflow or cooling that you needed. That’s because a 200CFM fan will only deliver 200CFM through a totally unobstructed airway – a very different proposition to a PCB on a bench.
The lesson is a simple one. Cooling design is a function: not of the components, but of the components in the context of the system.
The 20°C rule
Something else that made me smile was when an engineer – who had a good deal of experience – reminisced fondly about the 20°C rule. “The what?” I asked him. Previously, it wasn’t uncommon for the specification to call for the maximum rise in air temperature from inlet to outlet to not exceed 20°C, for example. If this was achieved, it was believed that the system was being appropriately cooled. It meant nothing of the sort, as it took no account of the individual components within the system. It did, however, have the virtue of being easy to measure.
System-level cooling is important, but not at the expense of its constituent parts.
I’ve also heard some huge misapprehensions about heat sinks: what they do, how they work, what the different kinds are, how to best make use of the various types of thermal paste/grease. But grease alone is not a substitute for a properly implemented heat sink.
Another myth I’ve heard is related to the 20°C rule - it’s the 10°C rule. However, this one is all about the impact that heat/cooling has on the reliability and lifetime of an electronic component. It used to be believed that for every 10°C you could lower the temperature of an electronic device, you’d double its life. I’m not sure anyone believes or uses that any more, but it was once held to be true. Yes, cooling has an impact on reliability/longevity – but it’s somewhat more scientific than an arbitrary rule of thumb.
There’s another belief that - just like you can never be too rich or too thin - your electronic devices can never be too cool. In fact, they can. Electronics that are too cold simply don’t operate well – and sometimes don’t operate at all. We focus on managing heat out in what we expect to be the highest ambient temperature a product will operate in. But what if it’s destined for outdoor use, or in an environment that’s already very cold? You can design too much heat out of a system.
One of my favourite design goals states that in order to achieve the required reliability, any component used may not exceed 75% of its manufacturer-specified operating temperature limit. Superficially, like so many of these random golden rules, that may make sense. However, it assumes that an operating temperature is an entity with no connection to any other factor - like performance, for example. Most of us know that one way of managing a component’s heat is to de-rate it to perform at less than its maximum. The question is: do the performance goals of the system allow that to be done?
Art or science?
Sadly, there are still too many instances in the development of electronics products where cooling is regarded as an arcane art rather than a science.
Fortunately, help is at hand. It has long been recognised that computational fluid dynamics (CFD) has plenty to offer when it comes to predicting the heat characteristics of a system. CFD is at the heart of 6SigmaET; one of the most sophisticated and flexible - yet easy to use - simulation software tools available for electronics design. With substantial automation and inbuilt intelligence, it is designed to help manage heat at the outset of the design, rather than trying to do so after the fact. The result is a significant reduction in cycle time, risk, cost, and time to market – which is often a key element in competitive advantage.
Yes, experience and judgement - not to mention creativity - are still vital attributes for those designing electronics systems. And ‘rules of thumb’ still have their place – if not the ones I’ve mentioned above. Nevertheless, cooling electronics is best seen as both an art and a science, with 6SigmaET providing science and support for electronics engineers.