Everyone knows the worst-case scenario: A critical component cannot be found anywhere in the supply chain because it is no longer being produced, and even the original manufacturer has no clue as to how a buyer could solve the problem.
It's a problem many in the electronics industry grapple with every day. Here's how it goes: You have just released a product to manufacturing from R&D; Marketing has made preliminary product announcements (brochures, press releases, and ad placements in the various media); the factory has reviewed all the build and test specs; and the ball has been passed to Purchasing to send out the Request for Quotation (RFQ) and assess the MRP status on the parts already in the Item Master showing materials on order and on hand.
Everything looks normal so far. But then you hit an unexpected snag. One-by-one, the RFQs from vendors and suppliers come back with the various costs and availabilities, but one part comes up as "No BID" from all distributors -- not even the original part-maker has any inventory in stock. Purchasing starts to call around to find out more details and discovers that the part went "end of life" (EOL) over a year ago, and the last-time-buys expired before the bill-of-materials (BOM) was released from R&D to production. I did say "worst case." But, as we all know, this really does happen, and all too frequently. What really happened?
The design engineer (DE) found a part number on the Item Master, retrieved it from the stockroom, and began to build the first breadboards or prototypes. The DE knows he may go through several iterations of the design and is planning for a minimum number of board spins. When he checks out the engineering stock quantity, he gets enough parts to stock the parts bins, and he is off to the races. He pulls the part specification either from Document Control or an attachment to the Part Master software tool, or he goes online and downloads the datasheet from a service like Datasheets.com, which claims over 180 million part documents in its archive. Any new-to-the-company parts are researched very carefully and qualified for use by the DE, Purchasing, and the component engineer. Again, the focus is on new parts.
As the development progresses, the factory has bigger fish to fry, and Engineering is working against a schedule agreed upon by management across many company departments. The message is: "Leave the engineers alone and let them do their work. They have a deadline to meet." The materials department is focused on existing product factory forecast and orders. As the Buyer becomes aware of the part unavailability, a red flag goes up and a decision must be made quickly. This usually involves two immediate possibilities: 1) Can the DE or CE find another substitute part quickly? or 2), Can Purchasing try to find any stock from any source, gray market included? (The gray market is becoming an increasingly dangerous alternative as part counterfeiters thrive in this venue.) (See: Counterfeiters Meet Their Match in DNA Tagging.)
This did happen to an engineer friend who works at a world-class company we would all quickly recognize. The offending part was a Xilinx Virtex 2 series FPGA, and the DE had not been told that Xilinx had released obsolescence notices with the last-time-buy information over a year previous to this fire drill. The company had been using the Virtex part on several of its board designs, and the core logic was "cut and pasted" from an earlier FPGA design to facilitate R&D development. Let me be clear. This new board had several Virtex parts using old cores lifted from previous designs, so the designs had already been proven.
Now, anyone working with Ball Grid Array (BGA) packaging knows that trying to find another manufacturer's part and package with the same pin count and pin-callouts is nearly impossible. When the engineer heard about the FPGA's unobtainability, he hit the roof. Nobody had told him not to use the part in new designs. The engineer became so upset with Xilinx that he switched to Altera and swore never to do business with Xilinx again. I asked him if he had told Xilinx about his discontinuing the use of its products and he said, "It wouldn't do any good. They don't care." I did not take the conversation any further, but we all know it was not Xilinx's fault at all.
If we analyze this at the systems level, we find that there were several omissions in procedures that could have spared this company a lot of money, time, and anguish. Here is a shortlist. You may think of others:
The company did not have a functional, organized part obsolescence management process.
The company did not receive an obsolescence (EOL) notice.
The company did not incorporate a "Materials Strategy Procedure" in its development cycle.
The part in the Item Master was not flagged as going EOL.
Purchasing had failed to notify engineering of its last "risk" purchase.
The CAD system had not been updated to eliminate the part number from new designs.
Materials did not participate in any New Product Introduction (NPI) or Concurrent Engineering meetings.
EOL notices were ignored.
Everyone assumed incorrectly that the Item Master had been maintained and kept current.
All of the above.
When you think about how much finger-pointing goes on in a situation like this, the cost to the company is not just measured in scrambled resources: There is the intangible economic hit brought about by an atmosphere of distrust in the company's existing processes and procedures. The design engineer most likely took the blame and is not the happiest camper in the park.
At the very least, a company should employ the use of a part alert service such as EEContent.com, Total Parts Plus, IHS, or a number of others that exist to prevent these kinds of wildfires. Don't let your company get burned. Be sure you have a service that pushes to your email any PCNs, EOLs, and other critical alerts that are matched up to your Item Master or BOM part listings. Failing to do so will put your product's continuing availability in jeopardy.
The next Best-Practices article will cover what is involved in a Materials Strategy Procedure.