1248 Views

The True Cost of IC Test

December 14, 2017, anysilicon

The other day I was having lunch with a friend I had not seen in some time who had recently come back from a ten-day trip overseas. This was his sixth such trip in the past twelve months made to check on production test activities at the OSAT his company had picked to run their production test.

 

My friend looked tired and I asked him why he had to make so many trips to manage what should have been a pretty smooth activity. He explained what so many of our customers have told me over the years: “things don’t always work out as planned.”

 

Cost of test has long been a significant factor in the production cycle of ICs.  The goal is to have adequate test coverage in test so as to not ship problematic units to customers, but also to minimize test times and their associated costs.   There has long been the belief that the quicker you can get your product production test needs shifted to low-cost factories (often in locals such as southeast Asia), the lower cost you will incur—and as a result, there has been an imperative to do so.

 

But does this plan always lower overall cost? And is it the best plan for everyone?

 

I won’t argue that in large volumes even fractions of pennies make a huge difference to the bottom line. But is that where your company is today?  Over time, it may certainly make sense to take mature, stable, simple production activity in high volumes to factories that are set up to process exactly that, but there are numerous reasons, as my friend can validate, that keep products from getting there and why he needed to make six trips to “check” on things.

 

#1: Complex Test Flow

Large-scale IC production test is a factory; factories work well with predictable, organized flows. Needing to step outside of that is painful and that pain often translates into errors, and inability to execute or price increases/adders to pay for the customization and delays.

 

#2: Engineering Changes

If you think releasing a test program to production is a onetime event for a given product, guess again.  Test program changes do get made, and often engineers need to step in and validate, alongside their production partner, what effect those changes have. Again, change is not a friend of the high-volume factory and a place where problems and cost can be introduced. Further complication is brought about when language and/or geographic constant come into play.  Engineering iterations can really slow down and mistakes can be made.  Some of the trips my friend made were out of frustration of not being able to interact effectively remotely.  Also consider that the trips themselves are expensive–and while that expense does not hit the gross margin of the product, it certainly does hit the bottom line of your company and takes a toll on the constrained engineering resources.

 

#3: Low Volumes

Perhaps I should have said lower-than-expected volumes.  Of course, we all aim for success and for many that means hockey stick forecasts, with production volumes soaring.  That’s a great plan; and when you reach the hockey stick and the volumes are there, go get the benefit of high-volume cost reductions. But until you are, there is often little to no cost savings on lower volumes. And given fixed costs and some of the other inflexibilities of the high-volume factories, consider slowing down your rush to move product there.

 

When my friend and I examined the test flow they were employing, the differing flows for each of the few customers they had and the volumes of units against the all-in pricing he was getting, it turned out I could have offered him very competitive pricing to test the units on our test floor in the Silicon Valley (not an area known to be low-cost).   Intuitively this doesn’t make sense, but when you look more closely, the fact is that EAG’s test floor is geared to NPI (New Production Introduction) efforts, which means we have a flexible engineering focus and in the early days of production ramp supporting dynamic changes and complexities can cut time and effort to quickly arrive at solutions and continue to support them in production. Flexibility is not only a feature of our business, but an essential part of it. This means that while we have robust processes, we can adapt to serve customers’ needs without customers bending to fit the needs of our ‘factory.’

 

So, as you consider your production test plans, I ask you to take a pause and give some serious thought to what your needs really are.  Ask yourself what the true costs are.  Of course, there is the test cost, but make sure you consider the cost of time delay, the cost of troubleshooting and debugging, and decide what wear and tear and other headaches cost your team your organization in dollars… and please reach out to EAG if you think we can help.

 

On the other hand, if you really enjoy long plane rides, traveling last minute over the holidays to deal with production issues at quarter end, or managing through similar unanticipated adventures, you can talk to my friend…. I think they are hiring.

 

 

This is a guest post by: Aram Sarkissian, General Manager at EAG Laboratories