DEMONSTRATING RELIABILITY

by Mr. Chet Haibel, M.S.E.E.

This two-day seminar will equip participant with both theory and best practices in planning and executing good Reliability Demonstration tests.  Such tests become necessary for one of two reasons.  The first reason is that Wear-Out (end-of-life) failure modes are often not easily discovered by overstress tests such as HALT.

The second situation is where several failure modes have been discovered, the design has been improved, but it is not clear whether the design is good enough yet, or whether it’s possibly “too good”; potentially weight, size, cost, or some other attribute can be traded off.  Reliability must be quantified to move forward.

HALT deals wonderfully with design margin issues in electrical and mechanical devices, and with some overstress fatigue issues in mechanical devices, “surfacing” these issues so they can be understood and remedied.  But what about cable flexing; connector insertion and extraction; battery recharging; switch cycling; belt, pulley, gear, and bearing wear; evaporation; corrosion; & many more “endurance” issues?

This seminar zooms in on those Wear-Out (end-of-life) failures that are caused by some “process” (failure mechanism) consuming some “reservoir” of material.

What about product and process designs that have been improved?  How good is good?  What is the shortest, least expensive test to substantiate improvement?  Can you place confidence in tests of small quantity samples with few failures or even in the absence of failures?

Starting with a review of the nature of failures and the mathematics that describe them, optimal test strategies are laid out to save you time and money.  Various methods of reducing test time, number of cycles, and / or the number of samples are shown for many genres of devices.

Pass / fail tests of single-shot devices; sudden death testing of sample groups; (constant hazard rate) MTBF demonstrations; Sequential Probability Ratio Tests (SPRT); cycle tests of electrical, mechanical, and electro-mechanical components and assemblies; censored and interval testing; and many more techniques are presented.

For higher reliability and / or shorter tests, acceleration must be applied.  This can be as simple as removing the dead time between cycles, but what are the risks of overlooking some failure modes or introducing foolish failures with this?

For higher acceleration, stresses must be applied.  The standard known acceleration relations between stress and life are presented:  Arrhenius (temperature acceleration) with examples of activation energies, Inverse Power Laws such as the Coffin-Manson relation for low cycle fatigue, Peck’s modification of the Arrhenius relation to include effects of humidity, etc.

The class culminates with how to use several accelerating stresses (that are not independent) with unknown stress-life relations to demonstrate reliability.

PAPERS

Why Should My Company Develop a Basic Understanding of Reliability?
by Chet Haibel, M.S., CRE, CQE
When a company wants to move to the next level in reliability, for example when the Reliability Engineers want to institute HALT, Cycle Testing, Failure Analysis, Design for Reliability Guidelines, HASS, and such, they need the support of New Product Development engineers, project managers, supplier managers, material engineers, etc. After all, these are the folk who fear their schedule and budget may be adversely impacted. They need to understand that properly applied reliability techniques shorten schedules and save money. Everyone must be rowing the boat in the same direction. This starts with a common understanding of terms: the three kinds of failures, how each is caused, and therefore how each is discovered, corrected, and prevented. The role of stress in reliability is another key concept everyone in the company should understand.
(click to download the complete paper...)

Weibull Analysis Points to the Problem Location
by Chet Haibel, M.S., CRE, CQE,
as demonstrated in the "Introduction to Reliability & Weibull Analysis" Seminar.
One of the companies for whom I consulted last year had a vexing field issue with front panel keypads. Customers were complaining that keypads were hard to press. 136 complaints had occurred in the last five years on a product that was currently shipping at about 850 units per month. Service technicians were typically able to make customers happy by swapping out the keypad. But since the keypad was only a flat panel with a flexible printed circuit board "pigtail" with no replaceable parts, the keypads were not returned to the service depot. Measurements on new keypad assemblies showed that every one tested good, requiring well less than the specified 2 pounds of pushing force to achieve less than the specified maximum resistance of the switches.

That resulted in speculation that keypads were wearing out and ought to be included in preventive maintenance and swapped out along with the rechargeable batteries every two years. However, a life test on new keypads showed that while their resistance varied quite a bit from unit-to-unit and rose somewhat with initial use, they still met specifications at a million cycles, well beyond what customers could experience.
(click to download the complete paper...)