Last night I finished up Reliability Control for Electronic Systems (again for work). I was thinking there was even less people would be interested in from this book than the burn-in book, but there is one thing that is sorta interesting.
If you've ever seen a company advertise a brand new product and then guarantee that it will last 5 years and wonder how the hell they know that, well basically the answer is they got it really hot and then ran it for a much shorter period of time. If you've never wondered about this then stop reading here and save your patients for future posts.
Actually there are several ways to accelerate life testing.
The most straight forward is to increase the duty cycle. Ex: if you assume someone will use a light bulb 4 hours a day then by running the bulb 24 hours a day you can accelerate the life test by a factor of 6 so a 5 year life requires less than 1 year of test. With some things like cell phones which are actually communicating a tiny portion of the time this method can be a huge benefit.
Another method is if turning a device on and off puts more stress on it than just leaving it on, then for life testing it can quickly be turned on and off to simulate a much longer life.
The first two methods depend on the usage pattern of the device - the method that can be used for anything is to increase the stress on the device. For an electrical product the supplies can be run at a higher voltage than normal. For mechanical the device can be stretched or vibrated or... But the main one is to just increase the temperature since this accelerates basically all failure mechanisms. The amount it accelerates each mechanism varies significantly. But if the primary failure mechanism is known then the acceleration factor can be calculated. A very common method in electronics is to run the device at 125 C (257 F) because it is hot enough to get decent acceleration, but not so hot that it generates failures just because of the temperature (such as melting the packaging).
The methods are all a bit questionable, but it gives a decent chance of getting the life right (or at least making people feel comfortable that most devices will last at least as long as the guarantee (or at least something to show the judge to prove you didn't just pull a number out of the air)).
Friday, October 05, 2007
Subscribe to:
Post Comments (Atom)
1 comment:
for some reason, every time you say "stop reading now", I feel compelled to keep on reading.
That last post was just a listing of things I learned while studying for GREs (for French or Philosophy of science), and then it spiraled into just things I learned/observed in general that day. Now that I look back at it, it does seem kind of confusing and random.
Post a Comment