[time-nuts] OCXO and fluctuations after EFC adjustment

[Bob kb8tq]

Hi

In most cases, the “new part” (and three or four other candidates) were each built 
into their own batch of a couple hundred oscillators along with batches using the 
currently qualified parts. The normal production process / production testing was
done on the batches. That was followed by a (possibly abbreviated) first article
test. 

The parts normally would go on long term aging for months / years depending on
the part involved. Routine frequency monitoring would detect gross failures. At the
end of the process, normal testing would be preformed to detect any more subtle 
failures ( = frequency test would not catch the output power dropping out of spec …).

None of this is all that different than the normal approach when buying eBay surplus
parts. Bring them in, check them out. Put them on power for a month or two. Then
check them out again. Compare parts in the group to others in the same group
(or other groups) to detect the weird ones …..

Bob

> On Apr 11, 2020, at 10:05 PM, Taka Kamiya <tkamiya9 at yahoo.com> wrote:
> 
> I'm STILL reading this, with interest.
> 
> I want to know the method of fault discovery, thought process that ensued, analysis conducted, testing process, and eventual root cause analysis.  We rely too much on automated processes and computers.  I want to know how engineers did more with less.  Before they are forever lost, we got to document it, or better yet, pass it on to newcomers.
> 
> I care more about how it was done before, not how it can be done better with this or that million dollar tool.
> 
> --------------------------------------- 
> (Mr.) Taka Kamiya
> KB4EMF / ex JF2DKG
> 
> 
> On Saturday, April 11, 2020, 8:59:32 PM EDT, Bob kb8tq <kb8tq at n1k.org> wrote:
> 
> 
> Hi
> 
> 
> > On Apr 11, 2020, at 8:10 PM, Richard (Rick) Karlquist <richard at karlquist.com <mailto:richard at karlquist.com>> wrote:
> > 
> > 
> > 
> > On 4/11/2020 2:25 PM, Bob kb8tq wrote:
> >> Hi
> >> Would you *really* want to read a book about how from August of 1986 to
> >> January of 1993 AVX NPO’s had some sort of issue ( not that the issue is
> >> clearly known, just that they are flakey) and that by 1994 the parts with
> >> values below 220 pf in 0805 seemed to be fixed?
> >> Again, the task was never to *fix* a component, simply to sort out the parts
> >> that worked from the parts you didn’t want to use. The only feedback to the
> >> manufacturer was via the (lack of) purchase orders.
> >> Somehow I doubt anybody would make it past the first page ….
> >> Bob
> > 
> > Back when before HP broke up into pieces, capacitor vendors considered
> > it a computer company and assumed that all capacitor orders were for
> > "computer grade" capacitors.  This envisioned huge motherboards with
> > thousands of bypass capacitors.  Like monitor specs where it is OK
> > for so many pixels to be bad, as long as 99% of the capacitors were
> > good ... ship them.  As long as the average leakage current met some
> > spec, it didn't matter if a few of them were very leaky.  The current
> > wouldn't be noticed.  Tempco and dissipation factor didn't matter.
> > 
> > We did actually give the manufacturer feedback, but it was not accepted
> > because we as an instrument division were not in their target market.
> > They didn't support repurposing.  It didn't matter than we were owned
> > by HP; we were using them for the wrong end use.  It's like those
> > disclaimers that say "We do not authorize for the life support
> > market" etc.
> > 
> > Some vendors flat out would not sell to our division although they
> > were fine with the computer divisions.
> > 
> > By now, few people besides Bob are still reading this. :-)
> > 
> > Rick N6RK
> 
> Indeed over the years, our experience was that feedback on components was
> at best unwelcome and at worst a major waste of everybody’s time. Lots of 
> “dialog” and very little benefit. Unless you are a precision crystal company, 
> oscillator companies are *not* a big customer for any component outfit ….
> 
> 
> Bob