"Digital Design Obsolescence"
Many devices still in use today for military specification and high reliability applications were designed and fabricated in the 1970s and early 1980s. Integrated Circuits (ICs) engineered over thirty years ago continue to be specified in new applications due to their reliability in harsh environments and their high performance. Unfortunately, the same functionality has been unachievable or compromised by the advanced fine geometry processes used in most of today's semiconductor fabrication facilities. Older integrated circuits are no longer profitable to manufacture and are rapidly becoming obsolete.
Thus the problems this paper will address are the dwindling supply of older (obsolete) devices and the concomitant demand for new designs of these parts. These parts must meet the original device specifications while also being fabricated in modern high volume semiconductor facilities.
Several factors contribute to the reduced desire of modern semiconductor fabrication facilities to address this need directly:
Low Volumes – a fundamental “disconnect”
The necessarily low wafer volumes needed to supply the obsolete parts market are totally unattractive to semiconductor fabs/foundries. These facilities are accustomed to thinking in terms of 200-300 wafers per month per part number for the “small jobs” as compared to the hundreds of die per year mentality of the military equipment manufacturer (especially since 1000 die is about equal to 1 wafer)
Lack of Process Knowledge – a real gap.
Modern IC designers place blocks-- similar to the system designers of thirty years ago. That’s both good and bad. Without the “systems” approach to IC design we would never have been able to design the microprocessors, DSPs, and other complex circuits we have today. But the designers enjoy no real process knowledge. The obsolete device designer needs to have process knowledge such that he can tailor the design to take advantage of process anomalies.
Economic Imperatives – there’s just no room for “Specials”
Today wafer throughput and yield, that is large scale semiconductor manufacturing economics, drive the fab. The end result is a process flow optimized for the “few” as compared to the past when numerous flows were available to provide optimal product performance. The lack of economic motivation by the fabs requires that the obsolete parts maker think in new ways to accomplish the goal of making older parts.
Scarce Expertise – “Well, back in my day…”
Finally, the “graying” of IC designers impacts the ability of companies to make the older specification parts. There are fewer and fewer process engineers around who understand the architecture of designs invented in the 70s.
This paper addresses each of the above points in addition to describing a successfully implemented solution that expands on both the older technique of making several “options” on a given die and the “multi-project wafer”. This “Multi-Project Die” effectively makes multiple devices on the same die. It is done in such a fashion that it duplicates the performance of the obsolete devices. Examples of this solution will be shown.
2005 MAPLD International Conference Home Page