Many of your applications such as in the industrial, military and aerospace area are in exceptionally demanding environments with extreme levels of vibration, high humidity, dust, severe voltage variations and of course, that good old chestnut – extended temperature variations.
When manufacturers of computing hardware (particularly single board computers) provide their products for integration in these environments, they will often quote ‘extended temperature’ ranges. This should give you a warm and fuzzy feeling that all is well if you are operating within these extended temperature ranges in the field.
However, sadly, this is not always the case…..
Watch out for Engineering Untruths
The definition of extended temperature ranges can vary by vendor with typical ranges as follows:
-40 to +85 degrees Celsius
-20 to +70 degrees Celsius
-40 to +75 degrees Celsius
However, the picture isn’t as rosy as it would first seem.
As most of you hard headed veterans of engineering procurement know from bitter experience - an important rule always applies to purchases: ‘There is no such thing as a free lunch’ or ‘If it is too good to be true, then it is too good to be true’. A cheaper system doesn’t always mean you are saving money. Many manufacturers will unashamedly use lower specification commercial grade components in extended temperature products which will fail if used repeatedly near their stated temperature limits.
And as you know – replacement of a component on a circuit board for a critical application can run into thousands of dollars.
Always Assume the Worst
First of all, many users are optimistic about their temperature ranges and tend to underestimate the extremes and thus select -20 to +70 degrees Celsius range when they should be going for a far greater swing.
For example, an embedded system operated in a vehicle may see the temperature of the car reach over 90 degrees Celsius within 30 minutes on a hot day (yes – Western Australia!). At the other end of the spectrum, temperatures in cities such as Chicago (and remote mine sites high up in the Andes) see temperatures frequently falling below -20 degrees Celsius.
What Goes Wrong With Your Components
Thermal runaway is a major problem for an electronic component (e.g. CPU) operating outside its temperature range. Power dissipation increases as a function of the temperature which raises the temperature of the device further until it fails (or explodes).
Crystal oscillators and capacitors do not like temperatures below their stated limits and tend to cause timing errors and unpredictable change in parameters before component failure.
Testing Short Cuts
Some manufacturers take short cuts when designing and testing. Before buying any product, verify exactly how they do their testing. Some vendors will provide extended temperature ranges for their systems but only test them intermittently at these extended temperatures (or sometimes once-off).
As far as I am aware, there is no industry standard for designing, manufacturing and testing extended temperature embedded computing products. So this means that vendors can define their own standards and approaches. You have to thus be vigilant about what you use and investigate what extended temperature ranges mean for your application.
I would like to acknowledge an interesting non-sales whitepaper entitled Embedded Computing Mythbuster by Versalogic Corporation.
Aldous Huxley remarks apply to engineering professionals on a daily basis: Experience is not what happens to a man; it is what a man does with what happens to him.
Yours in engineering learning