A pointer to a recent article I just saw,
As interesting as the prospect of fake parts showing up was, I thought, the note on the fallout rate and that the parts are fully labelled before they are finished testing.
I don't usually notes on what the actual device fallout rate is.
Robert
Whatever happened to "the quality goes in before the name goes on". ;-)
Best regards, Spehro Pefhany
Probably the parts were stolen by a frustrated customer that could not get the parts any other way... ..and that fallout rate shows the high quality of their fab areas....
It's owned by Lucky Goldstar. :-(
Now its, "Quality? Get the hell out of here!"
Not really. It shows that the package is a small part of the cost.
Fairchild, National Semiconductor, Signetics and other reputable IC makers do their best to create a reliable design and high *wafer* yield. Extensive testing and correlation is done so that data sheet parameters that cannot be (directly) tested at wafer sort is reasonably guaranteed by some test(s) that can be done at wafer sort. In the case of "jelly bean" parts, wafer yield has usually been rather close to theoretical maximum (ie: take into account any secant cuts, incomplete die at edge, and poor masking at edge due to curvature). Only parts that pass wafer sort are then packaged and tested; no flim-flam about sending them to another planet before final test. Failure analysis of the rare final test rejects almost always show assembly problems. so, like i said, 30 percent final test rejection rate shows the high quality of their fab areas [NOT].
Not really. Packaging and Tesing is e substantial part of the costs...
Post-packaging testing is often cheaper (and better) than pre-package testing. If the cost of the package is low enough, it's often better to test after since a final test is needed anyway (sometimes after burn-in). Sometimes both are needed to find the lowest overall cost. A 30% gross fallout (wafer to final test) isn't unheard of. I know some people who would be quite happy with 70% final yield. ;-)
Funny they that didn't include the list of part numbers and date codes in the news release, if they were marked..
RL
When I was in the business you sure didn't want to waste money packaging bad die and you symbolised after packaged testing. I can just about buy logistical reasons for symbolising packages before testing.
The 30% only shows something if you believe it. Sounds more like hype to scare potential customers away from the stolen product. Just like all counterfeit software is claimed to contain viri.
...
...
Maxim's webpage
Have something to add? Share your thoughts — no account required.
Ask the community — no account required