hardware problem

[Ted Mittelstaedt]

> -----Original Message-----
> From: owner-freebsd-questions at freebsd.org
> [mailto:owner-freebsd-questions at freebsd.org]On Behalf Of D G Teed
> Sent: Wednesday, February 27, 2008 3:54 AM
> To: Ted Mittelstaedt
> Cc: DAve; FreeBSD Questions
> Subject: Re: hardware problem
> 
> 
> On Wed, Feb 27, 2008 at 1:58 AM, Ted Mittelstaedt 
> <tedm at toybox.placo.com> wrote:
> >
> >
> >  > -----Original Message-----
> >  > From: owner-freebsd-questions at freebsd.org
> >  > [mailto:owner-freebsd-questions at freebsd.org]On Behalf Of D G Teed
> >  > Sent: Monday, February 25, 2008 7:22 AM
> >  > To: DAve
> >  > Cc: FreeBSD Questions
> >  > Subject: Re: hardware problem
> >  >
> >  >
> >  > Every system I've seen with his description of the problem, where
> >  > the power supply can't even run it's own fan, is having a 
> power supply
> >  > problem.  Power supplies are very often low quality these 
> days and can't
> >  > handle the stresses of typical electrical grid fluctuations.
> >
> >  My experience has not been that the power supplies can't handle the
> >  electrical grid.
> >
> >  What I've mostly seen is that the power supply FANS get dust in them,
> >  the fans slow down or stop, airflow through the supply drops, and
> >  then the supply overheats.  Once it overheats, the supply will never
> >  be reliable again and must be thrown out.
> 
> I've been able to routinely clean out the dust with canned air, and
> they still die more frequently than say motherboards.  Even quality
> brands like Antec.  I often replace the fan if it is showing signs
> of noise from bearing getting burned out.  I'm speaking mainly
> of home and small office PCs.  This is something that won't
> happen as much in a server room since the air is cleaner, but
> I'd guess the O.P. wasn't in that environment since he is wasting
> 3 days before trying another power supply.
> 
> Power supplies do have a limit of life related to the quality
> of your electricity 

Not the good ones.  Seriously.

I run a NOC that has a 50kva natural gas fired generator.  Every Tue.
the generator is tested for 1/2 hour (basically we put the entire
NOC on generator power for 1/2 hour)  There is an automatic
transfer switch that switches the entire NOC, under load, including
the HVAC unit, onto generator power for 1/2 hour then switches it
back to mains power.  There is NO feedback circuit that syncs
the sinewave from the generator with mains power.  As you can
imagine the switch is tremendously disruptive.  All of the UPSs
in the place squawk and switch into UPS power for a couple minutes.
All of the UPSes in the place are cut-in types.

So far we have only had 1 system lose power supplies on a
regular basis, and this was a brand new, very expensive, HP
server.  (on UPSes of course) HP's replaced at least 8 power
supplies in it under warranty.

None of the others, including some of the most motley customer-owned
clone equipment you might imagine, have suffered power supply failure.

The HVAC unit of course heavily filters the air so there is
no dust so to speak.  I can pull the cover off 3 year old
servers and the interior is as pristine as when they are new.
And we keep the temp around 68 degrees.

Please keep in mind most computer power supples nowadays
are auto-switching and will run on anything from 110-220v.

It is NOT dirty power that does them in.  It is dust.  And
heat, as you said.  Overloading a supply will kill it also
- very few (retail) power supplies on the market will run
close to their rated power output for any length of time.

Today, the biggest problem I see is people demanding these
minitower systems, getting these tiny small cases and
stuffing them full of hard and optical drives.  There's 
dead air spaces throughout the layout, and small, low-volume
"quiet" fans.

Hard drives also suffer as a result of this.  A disk with
good cooling can last many years.  But few computers other
than server gear provide it to the drive bays.

>I can recall the bad
> electrolyte scandle with several motherboard brands 5 years ago.
> The explanation of the shortened capacitor lifespan due to the
> electrolyte missing an ingredient was a bit of an education
> into what capacitors do.  They do have a limited lifespan
> related to heat and the number of hours they are exposed to a high
> ripple current.
> 
> Here is an excellent wikipedia entry on capacitor plague
> which will explain it in layman's terms.
> 
> http://en.wikipedia.org/wiki/Capacitor_plague
> 
> If you have not read about this before, it may be an eye opener.
> 

I know all about that.  I also own several TV sets that date from
late, late 60's early 70's and still work.  Electrolytic capacitors
have been around a long, long time.  They had them during the tube
days, and tube gear ran very hot.  Like anything, they have a
lifespan, but it is in the multiple decades, and little dependent
on ripple current or heat.  The issue with the self-corroding
capacitors was corrected and while the equipment (like for
example, Apple eMacs of 1Ghz CPU) that had those suffered, it
isn't indicative of normal electrolytic capacitor lifespan.

The most heat-sensitive parts are the semiconductors, the
IC's and transistors.  There is a thing called ion migration
that happens to all semiconductors and causes them to go off
tolerance.  It's a chemical process that is hastened by heat.

Ted