Below is a question from Paul Paroff asking:
“how to deal with these electromechanical components that do not have published MTBF values but do have rated life values in an MTBF prediction?”
Call: (408) 654-0499, x204
Email: info@OPSaLaCarte.com
Engineering Success with Reliable Products.
Generally MTBF values apply to electronic components. However in the case of push-button switches that are rated for a life of something on the order of 1 to 5 millon cycles. This is not an MTBF number and if you were to take 1 million cycles and if the use model gives you 1 cycle per second (think of a 60 wpm typist) then you are talking about 278 hours to failure. If you have 100 switches then plug that number into an MTBF calculation the overall MTBF number plummits like a 2 inch diameter steel ball.
So my question is how to deal with these electromechanical components that do not have published MTBF values but do have rated life values in an MTBF prediction?
You can get part failure rate data from the NPRD95 database of mechanical part failure rates; it is available from the Reliability Information Analysis Center (RIAC). Their website is
http://theriac.org/
Testing is good, if that is an option.
Two general comments here.
First, many companies that perform demonstration tests do not test to failure and therefore, you wind up with situations where the component vendor believes they have done “enough” of a demonstration and stop too early. If you are using their device for a low duty cycle, that number may be fine, but if you are using it for a higher duty cycle like you describe, then their demonstration appears inadequate. That is why it is important to plan out the testing properly and have the proper goals set early. I am a big advocate of using HALT before starting any demonstration test so that you can determine product margins and know how much you can accelerate the test. In the case of this switch, acceleration make come in the form of additional force to the keys.
Second, many tests such as this are more of life tests rather than demonstration tests. If what you are seeing is life test results, then you cannot mix that with the MTBF results because what they are really describing is how many actuations until end of life. A good example is a fan. If you look up in Telcordia for the failure rate of a fan, it gives a number of 100 failures/billion hours. If you talk with the manufacturer, they are talking about an MTBF of 50,000 hours. So why the big difference? Becuase the fan vendor is giving you the L10 life test results and Telcordia is giving you the steady state failure rate. These are describing two different parts of the bathtub curve.
Your switch example sounds more like the former issue, but be careful with the latter issue as well because once you jump into the world of mechanical components, this also tends to come up.