Archives

- Beowulf
- Beowulf Announce
- Scyld-users
- Beowulf on Debian

# [Beowulf] Re: UPS & power supply instability

Many of your questions may have already been answered in earlier discussions or in the FAQ. The search results page will indicate current discussions as well as past list serves, articles, and papers.

Search

Robert G. Brown rgb at phy.duke.edu
Fri Sep 30 08:38:36 PDT 2005

```David Kewley writes:

> One ground path travels with the computer power lines.  A second is
> grounding the raised floor metal components.  And a third ground
> travels with the power to the HVAC systems.  To complicate that third
> path, the HVAC ground path splits between the HVAC main breaker box and
> the HVACs, one path going through a devoted UPS, and one directly
> connected to the HVACs (the two branches power different subcomponents
> of the HVAC).  Very likely the HVAC and PDU chasses are tied to the
> floor grid, thereby connecting all 2-3 paths back to the earth
> electrode.
>
> If there's no leakage between neutral and ground anywhere, then the only
> other mechanism I can think of that would cause these currents is
> induction.  Can induction typically produce currents this large on a
> ~1000 ft^3 ground loop?  Am I missing something?

The HVAC is a nasty load.  Motors draw a lot of current, have lots of
harmonic content, usually have a phase angle (are highly inductive).
Remember, 3 A on a circuit loop with a resistance of order an ohm is
only a volt.  So the issue isn't can the induction produce a 3 A
current, it is can it produce a 1V potential around a current loop.

If the ground loop involved topologically is (say) 30'x30' \approx 1000
ft^2 (not cubed - surface not volume) then there is a lot of area to
subtend time varying flux, but there should be no such loop in series
with supply voltages or local grounds.  If you prefer, the potential
loops should be shorted out -- power supply lines next to returning
neutrals so that there is very LITTLE area subtended.  There should be
no way of forming the real loop that involves the power supplies -- the
neutrals should definitely NOT be tied to your chassis grounds anywhere
but at the One True Ground for the room (which isn't really ground, of
course, for anything BUT the room).  Your racks should be grounded with
the same ground wire that grounds the computers plugged into the rack.
In that case even if the raised floor supports support little eddy
currents, those currents cannot couple to the computer wiring -- they
are invisible ghosts responsible for absorbing the fairly miniscule
amount of energy radiated by the HVAC etc.

I'd expect most of the difficulty from HVAC to come from "noise", or
induction at startup, not from steady state operation.  I don't think it
or anything else is likely to make a volt on a loop

rgb

>
> David
> _______________________________________________
> Beowulf mailing list, Beowulf at beowulf.org
> To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
Url : http://www.scyld.com/pipermail/beowulf/attachments/20050930/b4a1008b/attachment.bin
```