Want to move to all 208V for server racks

Leo Bicknell bicknell at ufp.org
Fri Dec 3 17:42:10 UTC 2010


In a message written on Fri, Dec 03, 2010 at 04:57:03PM +0000, Gary Buhrmaster wrote:
> limits so that ones life has increased protection.  A protective trip
> is better than
> the alternative.

Not always.

I worked in a data center with something I thought was very, very cool.

http://www.hilkar.com/highresistance.htm

The concept, at a high level, is rather than tie the (service, not
signal) ground back to grounding rods directly you run it through a
large resistor.  Now when a phase is "grounded" it runs through the
resistor, allowing a small but safe current to flow.

Why is this cool?  Well, say you have a power strip running at 10A
with a bunch of servers on it.  If you took a paperclip and inserted it
in an empty plug connecting hot to ground with a normal system
(simulating a faulty bit of gear) the breaker would trip, all your
servers would go off.

If you did this with a high resistance setup the paperclip would conduct
about 0.5A, maybe less.  An alarm, dectecting current, at the resistor
would go off to say there was a fault.  Your circuit would draw 10.5
amps and everything would stay up and running.  That faulty bit of gear
didn't take down your entire power strip.

This totally eliminates arc faults, and there isn't enough current to
ground to arc.  I think GFCI's are also unnecessary, as the fault can't
conduct enough current to be harmful.

Not there are a TON of other details to building such a system.  Cost is
a factor why more folks don't do it, and it takes a lot of pencil
scratching by your EE types.  Still, one of the coolest things I've ever
seen, and I wish more data centers were built this way.

-- 
       Leo Bicknell - bicknell at ufp.org - CCIE 3440
        PGP keys at http://www.ufp.org/~bicknell/
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 826 bytes
Desc: not available
URL: <http://mailman.nanog.org/pipermail/nanog/attachments/20101203/2a5f12de/attachment.sig>


More information about the NANOG mailing list