On Sat, 18 Dec 2010, Rod Smallwood wrote: It was very common with tubes, too. You can find many German tubes with the marking "G.Abl." (=Garantieabl?sung), meaning that the tube manufacturer (e.g. Valvo) would not guarantee anything as they were sold untested (and cheaper). They were mainly used by industrial and testing equipment manufacturers because they tested these parts anyways. Christian

On 19 Dec 2010 at 13:43, William Donzelli wrote: There *was* sort of a retread business going with some small vendors testing old tubes, repainting them (bases in the case of a glass tube; the whole thing in the case of a metal one). and a new logo attached. You could pick them out by using a bit of lacquer thinner on the base of glass tubes--the paint would come right off. Come to think of it, I've even got a couple of 70's-era 7400-series ICs that have been painted and relabeled. I suspect that was more of a "house numbering" scheme, however. --Chuck

In house testing was reasonably common back in the 1970s simply because integrated circuit technology just was not anywhere near what it is today. For the past ten or twenty years, a manufacturer could pretty much assume that every part received from an IC vendor is basically perfect - a 100 percent yield out of the chip tubes. Any failed parts that get stuffed on the boards are faulty due to improper handling. This super quality can be traced to the constant fine tuning of the mass production and QC lines over many, many years. Rewind to the 1970s, and the story is different. There were many more vendors, due to the nature of glue logic - every 7474 is supposed to be the same over the 20-some vendors around. The problem was that not all the glue was the same, with very subtle differences in the parts. Perhaps one vendor might have very strict quality controls, yet another might let a few things slide, especially towards the end of the year. Likewise, many of these vendors were pretty new to integrated circuit manufacturing, like Sylvania, Transitron, Spague, and Stewart-Warner, and still had things to learn. Even the established vendors had only really 10 or so years of "real" chip production. Bad parts routinely slipped out of the factory. In those early days, it was quite hard for some of these "newcomers" to reach even 3N reliability (99.9 percent yield - in contrast, today 6N is reasonable). Well, one bad IC stuffed on a complex board can really throw a wrench in the works of a production line, so the expense of a IC test station could be financially justified. And as for DEC - remember that they sometimes used ICs in somewhat unorthodox ways that may not be tested for back at the vendor. It seem to recall that one of their favorite tricks was to use some of the internal "diodeness" of the input and output pins of line transmitters and receivers for clamping. -- Will