[mythtv-users] Power outages and UPSs

f-myth-users at media.mit.edu f-myth-users at media.mit.edu
Mon Jul 30 22:25:14 UTC 2007


    Date: Mon, 30 Jul 2007 17:31:29 -0400
    From: "Jay R. Ashworth" <jra at baylink.com>

    > And this is an excellent selling point for continuous UPSs, too; the
    > charger is a sacrificial lamb for incoming spikes, and if you have a
    > backup charger module offline, you can plug it in after.

Er, uh, maybe; it depends on your fault model.

If the spikes are -just- sufficient to destroy the charger, then sure,
you swap in another charger.  But if you take an actual nearby strike
(or any really severe spike, strike-related or not) that manages to
propagate through the charger to its outputs, then everything past
that point is very poorly spike-isolated and will be fried.

OTOH, if you have an inverter after those batteries and a CPU PSU
(e.g., you're using either the "ghetto" UPS w/an inverter, or an
ordinary UPS which just has surge clamps but isn't running the
inverter in normal operation) then the PSU -itself- has lots of
spike-suppression and some HV isolation (the latter just to meet UL
approvals, etc) and hence the inverter & PSU also act as fuses of a
sort---expensive ones, but probably cheaper than the guts of the CPU
itself.

So it all depends on what sorts of failures you think are likely.

If you want -maximal- isolation, you hook a motor to a generator
and you can get better isolation than even an isolation transformer
will give you... :)  Especially if the shaft between them is long and
nonconductive.  Then your machine will be totally destroyed by the
lightning spike hitting your broadband connection and propagating
through all your Ethernet cables, so you'd better switch to some sort
of satellite-based IP provider and live with the latency... :)


More information about the mythtv-users mailing list