Is it okay to plug a power-strip into a UPS?

[Bill Sconce]

Oo.  We're going to agree again.  And Ben caught me in a mistake.


On Fri, 17 Jun 2005 09:59:30 -0400 (EDT)
Benjamin Scott <dragonhawk at iname.com> wrote:

> Remember, we're talking alternating current.   In the US, AC 
> goes from zero to (IIRC) 180 volts (absolute) twice per cycle,
> or 120 times per second.  In other words, AC is *constantly 
> surging*.

Woops.  I mistakenly said 3 times; it should have been _2_ times.

For a sine wave the peak voltage is the square root of two times
the RMS value -- 165.46V for 117V mains.

You are right about 120 times per second.  (In the U.S.)  I wouldn't
say "constantly surging", though, just "alternating".  A "surge" is
an excursion beyond the normal alternating limits.

So a clamping value of 330V would be (only) twice the mains' peak
value.  Which means that your equipment would see 165 volts excess
before the "supressor" does anything.  No doubt most equipment is
designed to survive twice the nominal voltage.


> Now, I would expect it is possible to build a device that clamps
> if the  voltage rise is too fast or out-of-phase, but I'm really
> just guessing here.

Indeed it's possible.  You expect correctly.  It's possible to clamp
at just two volts more than 165, or at half a volt, or at ten, or if
the rise is too fast, or even if the phase is messed up.  Provided 
you're willing to pay for it.  (Hint: MOVs won't do the job.)

The usual first need, though, is to clamp the voltage at some maximum.
Clamping at 10V excess would be reasonable.  (Just for the sake of
illustration.  Half a volt, while also possible, would be getting
pretty expensive.)  And good UPSs can do things in addition to voltage
clamping, such as cleaning up the waveform (related to your "rise too 
fast".)

A good UPS is a good investment.  (Of course a UPS also supplies
power for a while in an outage, an unquestionable Good Thing.) 

It's MOVs which are junk.  (In power strips, our current subject.)


> Is a fluctuation during 0.000003% of the cycle really something
> to worry about?

Oh YES.  

Consider a lightning strike.

For a voltage-sensitive device the time-to-damage is really,
really short.  

A less-extreme example than lightning would be blowing a chip
by static electricity you picked up from the carpet.  The total
energy, risetime, and duration of a static discharge are all
really small, but the damage is fatal.  The % of a cycle doesn't
enter into it.  Same with a pulse or "surge" on the AC line. 
(Which, indeed, is what the MOV-power-strip manufacturers are 
trying to scare you with).

What they don't say is how often you GET pulses like that.  The
power from the utility company may be quite good.

(It just occurred to me that one thing we may not have mentioned
is our own influence on the AC lines - e.g., the other devices
we have around.  Something like an air conditioner motor cutting
on and off, especially on the same circuit, can create strong
voltage spikes.)

Generally, though, I'd say that the risks ARE overblown.  When
is the last time you lost an answering machine?  Clock radio?
Television set?  A computer's power supply is NOT that special.


>    I'm sure if you spend $200 on a really elaborate device
> from a quality manufacturer, you're getting something for
> the money. 

OK.


> The question is, does it really matter?  Are you paying for
> protection that simply does nothing, due to the fundamental
> nature of AC?

I agree that it may NOT really matter.  Not because of the
fundamental nature of AC -- good, clean AC is fine.  If you 
HAVE good, clean AC.(*)  And now we're back at the beginning.


>    As I said, surprisingly complicated.  :-)

Yep.    :)

-Bill


(*) My workstations are all behind UPSs, which are behind
series-mode surge suppressors.  But the second-hand Compaq
Deskpro 4000 which runs as firewall/CUPS server sits on a 
bare, unprotected AC outlet.

The poor thing crashes every time the power goes out.
(Which it does out here, maybe twice every three months.)

I "intend" to put it on a UPS someday.  In the meantime,
thank goodness for reiserfs - and for what must be at
least reasonably clean AC power in spite of everything.
That machine has  been running 24/7 (except for the outages
and two kernel upgrades) for four and a half years now, 
thunderstorms and all.