Designing an ORT

I received a question about setting up an ORT the other day. Below is my response.

—

There is not a hard and fast rule for how much life to take out of a product during ORT and still be able to sell the unit.

There are two different reasons to run ORT and each may take a different approach.

1st. There is a possibility of a fraction of units that contain a defect that can be revealed when exposed to some stress. The balance is how much stress provides an effective screen and finds the defective units, and does not adversely impact the warranty or expected life of the product. This is generally done on all units or on some form of lot sampling to catch lots with bad units. When done on 100% sample, it’s done to screen for latent defects.

2nd. The ORT is done to detect shifts in the production line, supply chain or design that may adversely shorten the expected lifetime of the unit (or increase warranty exposure). These units are generally run for a lifetime or longer (often under accelerated stress conditions) in order to estimate the lifetime of the current production. This is usually a small sample of units and not sold as the testing is generally considered destructive.

In each case, being very specific on the reasons for the testing and the specific failure mechanisms under investigation permit a rational test plan.

—

To this I would add – do provide some effort to not running any ORT or HASS and rather fix and improve the design and/or process.

I’ve seen some very poor ORT’s often done only because that was what we always did. One was set up and run on every product costing $100,000s per quarter and nobody looked at the results for did anything based on the results. Total waste of time and money.

I’m sure you’ve seen some pretty ‘good/bad’ examples. Share them here.

cheers,

Fred