Cradle to the grave
Two events happened recently that set me thinking about the lifetime of the data we use in the semiconductor industry and how that lifetime is extending in much the same way that human lifespans are extending. The gamut of the data we use in all areas of our lives is not just constantly growing but changing shape.
One Saturday morning, awoken from my planned late lie-in by the knock at the door of a parcel delivery courier, I somewhat bleary eyed signed for the delivery of my new ultra-wide monitor. A very useful aspect ratio for all those Excel spreadsheets where you just want one more column visible without reaching for the scroll bar. Ok so that is a battle, much like how much RAM do I need, that will perhaps never be won. There will always be one more column just off the screen. What caught my attention as I opened the box though was a single piece of A4 paper with three charts on it. Those charts turned out to be a set of individually measured and plotted gamma, color temperature and color difference plots for the particular individual monitor in the box. Now I’ve seen enough calibration reports in equipment boxes over the years, nothing new as such there but this was no high-end monitor but a low-cost consumer grade item purchased from an online retailer. My guess is that these reports will get read about as often as user manuals and software licence agreements.
This set me thinking however about another event a couple of weeks earlier. Having upgraded my broadband to a fibre optic service my provider sent out a new connection hub. There was nothing at all wrong with the old one it just wasn’t compatible with the new service. So dutifully I placed the old equipment in the postage paid return bag for recycling. It was of no further use to me anyway and it saved me a trip to the local recycling facility.
So how are these two events connected? Simply this, manufacturers are forced these days by various regulations in different territories to provide handling facilities for goods that are end of life. In our modern high-reliability consumer world, increasingly those end of life goods will still be functional. Because modern electronic items are usually uniquely identified with serial numbers we now have traceability of any item not just up to the point it leaves the factory but also when it returns years later to be recycled. In some cases this traceability could go all the way back to x-y probe data for individual parts if the x-y location is recorded electrically on the device itself during manufacture. So how about sample testing those end of life items and correlating to the original test data. Manufacturers, product designers and process engineers could gain useful insights into how parametric values shift over extended periods of real world use. How might such data compare to the results from accelerated ageing methods that are in use; could we learn a thing or two about the methods we accept as being a fair approximation to device ageing?
There is a trend of some very large players in the industry adopting automotive standards for high volume consumer items. A little while ago I heard an engineer presenting at a trade event say that he expected automotive standards to become the norm. At the time he was somewhat scoffed by others in the room but it looks like that guy was right; we see it happening daily. The drive for higher reliability at lower cost is clear and present and as we push that envelope we need to find smarter ways to learn from all that data that is right under our noses.