[time-nuts] Symmetricom chip scale atomic clock

Magnus Danielson magnus at rubidium.dyndns.org
Fri Apr 25 02:52:17 EDT 2014

On 04/25/2014 08:14 AM, Attila Kinali wrote:
> On Fri, 25 Apr 2014 08:33:06 +0300
> MailLists <lists at medesign.ro> wrote:
>> The recently acquired cash cow isn't working exactly as
>> expected/advertised. We still don't have a clue when/if the fundamental
>> (as in physics laws) design (we can't officially blame the cheap Chinese
>> manufacturer) flaw will be fixed (manufacturer replaced), but as our
>> main customer, which is used to (literally) blow up tons of (others')
>> money, isn't very concerned (for now), and the profit margin is (still)
>> high enough to replace (no questions asked, for the time being) the
>> failed units of the other (civilian/commercial) customer(s).
> Sorry, but this is was not necessary.
> Not every company is evil and not every company just works for the
> short term bottom line.
> It is very normal that problems show up in series production which
> were not visible before in the prototypes or pre-series production.
> It's part of the very nature how volume production work. And no,
> you cannot always attriubute it to less care taken in the volume
> production than in the pre-series. Some flaws are only visible if
> you get enough produced and then it's still one in a couple hundred
> if not a one in a couple hundred thousand.
> Every product i was ever involved with had some flaws uncovered during
> series production. Even if the gratest care was taken. IMHO it does not
> matter whether the product has a shows suddenly a flaw or not, but how
> the manufacturer reacts to it. And as it seems Microsemi is replacing
> the failing units without causing trouble.

You will find that every major product has had some child-hood issues 
which took time to trim in the design och production. Most companies 
want to move away from that hassle, as they really care about how they 
are perceived and that involved future sales. They are reluctant to say 
there is quality issues, but if they are honest, there is always some 
crud in there. It's actually when you do not see any errors you get 
afraid that it's a big one... that will hit you late in the series.

So I agree, it was not fair to give those comments.

> Also, please be aware that fixing a flaw that only very few units show
> is not an easy thing. You cannot just build a prototype and be sure
> that the bug is gone. You have to first produce enough to have a statistical
> significant sample size. This all takes time, weeks, months, or even years.
> So, please refrain from spreading false rumors that anyone is ignoring the
> issue when aparently the contrary is the case.



More information about the time-nuts mailing list