[time-nuts] Symmetricom chip scale atomic clock
jimlux at earthlink.net
Fri Apr 25 07:57:19 EDT 2014
On 4/24/14, 11:43 PM, Magnus Danielson wrote:
> On 04/25/2014 05:32 AM, Jim Lux wrote:
>> On 4/24/14, 6:26 PM, Said Jackson wrote:
>>> Hi Magnus, Bob,
>>> Thanks much for your kind words.
>>> The failure rate is thankfully so low that we are not greatly alarmed,
>>> and Microsemi has been a champ in resolving any failures with/for us
>>> when they did show up. We are awaiting the results of the full
>>> re-qualify that Microsemi is doing on the CSAC and were going to
>>> announce the issue at that time..
>> Sometimes it is the low failure rates which make it so troublesome.
>> Everyone gets excited, but the vast majority don't have the problem, but
>> then, every little anomaly or unexpected event prompts a "is it the
> When I started to see failures on a particular OCXO, I was alarmed. It
> ten turned out to be about 4% of all we bought, and I have checked them
> all myself. Seems that those that had problems have been detected or
> died out since we have not had any more reports.
> Single units is freak cases, but when several have issues within a short
> time, then we react to try to see the systematics. Let me tell you that
> the OCXO case was benign compared to fans that died much earlier than
> advertised. Guess if that vendor is now blacklisted.
> Another vendor got blacklisted as they did not take responsibility for
> the failures their poor design chip caused. The OCXO vendor handled the
> situation nicely, by replacing the broken units when requested.
> Failures happens. Vendors taking responsibility when failure happens are
> keepers, but you don't really feel motivated to deal with those that don't.
> That's why I'm happy to hear that Said have that support from
> Jim's boxes is another story.
Yes.. I fully agree with you. I think the most frustrating case is
where you have one failure in a thousand, and you sell a few hundred a
month, and it's something that people are buying in small quantities, so
you have lots of different customers. It makes it difficult to figure
out what's going on: Is it a usage corner case? Or is it just a low
probability defect that is random? Or is it a harbinger of worse to come.
In space stuff, especially for things that are in flight, something
going wrong triggers a whole lot of speculation, complicated by the fact
that you can't actually see the failed box to test it. All you have is
some last gasp telemetry or something; that may not even be valid. We
in the space radio business grimly joke that all major spacecraft
failures begin with a "loss of comm" anomaly. The spacecraft exploded
and the first question people ask is "why did the radio stop working?".
All those people standing around waiting in vain for Mars Polar Lander
to respond (Maybe it landed ok, and the radio isn't working, but the
spacecraft is really ok? Please?)
More information about the time-nuts