I have been involved in this type of testing for over 20 years. The Mil-810 fatigue relationship using an exponent of 3 to 4 works well for electronics if the expected failure modes are in Lb-SN solder or copper component leads. We also avoid raising the level too much to reduce test time / cost. One approach is to assume the specified long duration level is at the endurance limit for some material and then limit the increased level to avoid yield stress for the chosen material. This is intended to avoid precipitating failure modes that would never occur at the specified level. When testing electro-mechanical devices use this with caution! The fatigue relationship does not predict failure from wear out, which may occur with increased levels!
In your case, going from a 20 hour test to a 94 hour duration I would reduce the PSD level to about 70% of the 20 hr test level.
One additional note..... some vibration controllers don't do well with very long duration random vibration testing. The environment does not remain truly random in nature, but then the actual use environment is also not ideal random vibration!
|
