Mesa Production 1958 Qualify Customer Specification 2n696

# Qualify the 2N696/2N697 to a Customer Specification: Passing This Line's Test Is Not the Same as Passing Theirs

## 1. Why a Device That Satisfies Fairchild Can Still Fail IBM

This step takes a batch of sealed, screened devices and runs them against a customer's own written specification — electrical parameters at specific bias conditions, temperature range, mechanical dimensions, lead pull strength — a document this line did not write and whose acceptance limits may be narrower, in specific places, than anything this production process has measured itself against before. Every prior step in this series validated the device against this line's own internal standards: the furnace's own tolerance, the coating's own defect rate, the burn-in's own stress criteria. A customer specification is a different document entirely, written by the customer's own engineers for their own application, and a device can clear every one of this line's internal gates while still falling outside a customer's narrower acceptance band on some parameter this line never had reason to tighten:

where $\rho(x)$ is this line's own measured distribution of a given parameter, $[L_{\text{internal}}, U_{\text{internal}}]$ the band this process has been controlling to since Step 4's statistical process control, and $[L_c, U_c]$ the customer's acceptance window, which may sit strictly inside the internal band. Qualification is the act of discovering exactly how much of this line's own distribution actually survives being measured against someone else's narrower standard, and the fraction that does is not something this line can simply assume — it has to measure it, batch by batch, for every customer whose requirements differ from the last.

## 2. Real Diagram: A Specification Is a Document Before It Is a Number

Qualification is not a single electrical test. It works through a written document clause by clause — electrical limits at specified bias and temperature, mechanical outline dimensions, lead pull and solderability, storage and operating temperature range — each clause potentially demanding its own measurement, its own fixture, and its own pass/fail criterion that this line's own internal standards never required anyone to check in exactly that form.

## 3. Why No Earlier Series Needed to Satisfy Someone Else's Paperwork

Every device this project has documented before this point — the 1947 point-contact transistor, the 1954 diffused-base series, the 1956 drift transistor, and the first ten steps of this own series — was validated against this project's own documented criteria, implicitly treating "correct according to our process" as the final word. The 2N696 and 2N697 are being qualified here against a document written outside this project's own process entirely, by engineers at a customer who will put this device into equipment this line will never see built. That is a genuinely different kind of validation: not "does this device do what we designed it to do," but "does this device do what someone else, with their own requirements, needs it to do" — and the gap between those two questions is exactly what this step exists to close.

Step 11 does not change anything about the device; it decides whether this line's idea of a good transistor and the customer's idea of a good transistor are actually the same idea, clause by clause, before a single unit ships under that customer's part number.