[Server-devel] School networks and electrical equipment damage

[James Cameron]

On Thu, Jun 06, 2013 at 01:58:58PM -0600, Daniel Drake wrote:
> And the most surprising thing - we had not even turned on the network
> yet, pending some electrical work. Everything was connected up except
> one crucial link - the UPS was not plugged into mains power. So all of
> this damage happened without any of the devices having a connection to
> the mains.

Actually not plugged in?  The whole network was therefore either
isolated from building electrical earth ("ground") or had a series of
surreptitious connections, the critical one being the telephone line.

Hmm.  Lightning wasn't necessary.

Here's my theory:

Each long run of ethernet twisted pair becomes one side of a capacitor,
the other side being the building wiring, piping, or structure.

The long run of telephone wire picks up static charge from wind,
lightning ground currents, test currents from the telephone exchange
or line workers, or induced currents from other subscribers or power
network switching.  Even turning on many lights.

http://www.pololu.com/docs/0J16/all can give you an idea of what can
happen.  Any of these events will induce a starting pulse of DC in the
telephone wire, which is analogous to the length of DC power cable in
the Pololu explanation; the inductor.  The low equivalent series
resistance (ESR) capacitor can be thought of as the long cabling
against the building.

As a result of the capacitor and the inductor, the voltage is
amplified until it reaches the breakdown voltage of whatever is
connected.

Having the UPS plugged in might have prevented this voltage from
finding a route through something more precious.  Instead, it might
have found a route through a series of surge protection devices in the
UPS, and then the only damaged equipment would be the ADSL modem.

The convention is to build from the ground up.  Don't plug the cables
in until the ground is available, and then plug them in in strict
order.  People get away with not doing this because the damaging pulse
isn't constant.

(Reminds me of the time that I pulled a UPS input power plug out
instead of just turning it off.  A bad idea.  The last pin to separate
was active.  The connected equipment lost ground reference, the only
ground reference that remained was through a device, so it took a full
hit and died.  The resulting current passed into the building ground,
and triggered the earth leakage breaker on the circuit that the UPS
was originally connected to.  Other equipment connected to that
circuit powered down.)

(Reminds me also of working for a cable contractor in the 1980s,
looking after their cable management system on a VAX ... they were
putting millions of cables into a power station, and the build was
done from the ground up; that is cables were tracked as to whether
they had been terminated yet, and the list of unterminated cables was
a special report from the database that they always wanted to see.)

> I have seen that some UPSs (unfortunately not these ones) allow a
> phone line to be passed through them, supposedly offering some
> protection. Would such a system protect against a lightening bolt,
> assuming thats what happened here?

Yes, but only if the UPS was earthed.  It would also protect the ADSL
modem.  It would also protect from most other causes of a current
pulse arriving on the phone line.

-- 
James Cameron
http://quozl.linux.org.au/