[time-nuts] TAPR TICC boxed (input protection)
kb8tq at n1k.org
Tue Apr 11 08:21:55 EDT 2017
Testing can mean a lot of different things. Did they test every single part they shipped for every parameter?
Did they just do a sample of parts and decide the lot was good? Did they test a sample of parts for a sub-set
of the specs and decide they were good? Did they test them after packaging or at the wafer level? Did they test
a completely different (but much easier to test) part at the wafer level and decide the whole wafer was good? Did
they test one wafer out of the batch and decide the rest of the day’s production was good?
The further down that list you go, the cheaper the part gets. I rarely go looking for the most expensive part when
I’m doing a sort on the distributor site. I do toss out a few outfits I don’t trust, but that’s about it. I doubt I’m the only
one who shops this way. That drives the whole process to ever lower cost approaches.
If you *really* need a specific parameter, test it yourself. Depending on a supplier to 100% test this or that is *not*
a good idea. Unless you have an agreement with them to do the testing and get the data from the tests, there is
no certainty that your idea of “tested” and their idea are the same thing.
Semiconductors are by no means unique in this regard. Your wrist watch, wall clock, or Cesium standard has
the same dynamics driving it’s production. They all are impacted. That’s not always a bad thing. We get stuff
for less money. Other approaches to QA now drive the quality of the product where 100% testing once ruled.
> On Apr 11, 2017, at 7:44 AM, Charles Steinmetz <csteinmetz at yandex.com> wrote:
> David wrote:
>> If the 10pA specification is guaranteed by design, then wouldn't they
>> have to be testing the 1pA "A" parts?
> That assumes the parts are produced by exactly the same process, which is very often not a safe assumption. One of them may undergo extra process steps, for example, or one or more process steps may be modified. That's not at all uncommon, BTW -- "A" versions are often the product of process tweaks, not selected "non-A" devices.
> Best regards,
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.
More information about the time-nuts