Sherlock Ohms Discovers the Backward ICs

In the 1980's I was hired in the debug phase of a scientific supercomputer design. This used four large backplanes with 16 inch x 24 inch boards full of 100K ECL in a large cabinet. I was assigned to debug the memory subsystem. We did much of the debug on a test fixture driven by a DG Nova minicomputer.

One of the boards had failing memory locations. I asked the technician to pull the suspect chip and replace it with one of our meager stock of good ones. These were 4K ECL-compatible CMOS chips in a ceramic package. The ceramic package consisted of a ceramic slab on top and bottom, sandwiching the lead frame and die, all held together with a glass frit.

Upon examination of the failed IC, I noticed a pin-1 notch on both ends of the package, one on the top ceramic slab, and one at the other end on the bottom ceramic slab. When it was installed, they went by the top notch, which of course was at the wrong end. A close examination of the memory boards showed several more ICs installed backwards the same way, each corresponding to a failing memory location.

The computer was eventually brought into production, together with a lower cost and lower speed variant, achieved by a half-speed clock. They changed some signals on one board to make it non-interchangeable with the full-speed version. We privately called this the Half-Fast Array Processor. If we had actually sold any of these units, it would probably have bankrupted us.

Kevin Clark has been interested in electronics since the 1960's, and has worked in a variety of fields, including telecommunications; computer; consumer; and industrial electronics industries, in design, debug, programming, and support roles.

Tell us your experience in solving a knotty engineering problem. Send stories to Jennifer Campbell for Sherlock Ohms.

Related posts: