[Tfug] UPS sizing

[Bexley Hall]

Hi, John,

--- On Fri, 1/2/09, John Gruenenfelder <johng at as.arizona.edu> wrote:

> >I'd be surprised that you had good enough *service*
> >to run much in terms of electronics, etc.  Many of those
> >homes only had 60A services.  And, if they did NOT have
> >a gas stove, you can be sure the oven ate up the biggest
> >part of that!  :-/
> 
> Given the age of the house, there had been a few additions
> to the wiring over the years.

Ah, OK.

> In my apartment I had access to three separate circuits.  The
> newest was high load on two outlets and used by the oven,
> fridge, and microwave.
> 
> The second oldest was relatively new (70s I think?) and
> this was what I had my equipment plugged into, though I 
> needed an extension cord to plug the UPS into

I *think* you are better off plugging the UPS into an
extension cord than plugging the extension cord into
the UPS.  But, I don't have any quantitative data to
support that.

> it.  It was three prong.  The landlord, who had only lived
> there six months more than I, said he didn't think the third 
> prong was actually grounded, but my UPS never complained.

I suspect it *was* "grounded" -- but, not necessarily a
"safety ground" (i.e., ground is ground is ground -- except
when it isn't!  :> )

> The oldest circuit was the original ancient low load stuff
> which I only used for a lamp and a clock.  My neighbor, however,
> kept stupidly plugging a hot plate into it which would instantly 
> trip the upstairs breaker.  But... even with the modifications, 
> it still all wound its way to an aging rats nest of
> wires in the basement.  I'm surprised it didn't all
> trip more often.

One "giveaway" re: age is whether or not there are *individual*
conductors (e.g., knob and tube), *pairs* of conductors (with a
'drain" wire) in "armored" cable (e.g., BX) or two/three conductors
in a vinyl sheath (e.g., ROMEX).

You can also look at the *tape* used on connections (e.g., K&T
used asphalt soaked cloth tape -- predating the vinyl adhesive
tape used nowadays... "electrical tape")

> My biggest power problem was still linked to the UPS.  I
> think I posted to TFUG about it a couple of years ago.  About
> once every 1-2 months, the UPS would shut off power to all 
> devices, rapidly blink the red light on the front,
> and emit a constant siren.  This error "code" did
> not correspond to anything in the manual and the closest match
> I could find on the web suggested that the
> devices were drawing too much power from the UPS and so
> caused the shutdown.

Possible.  I've seen APC units "click on" (i.e., go into
backup mode) spontaneously.  I've always assumed it was the
result of a brown-out -- though have never been able to
correlate this with "dimmed lights", etc.  <shrug>  As long
as it didn't happen too often, I just shrugged it off (I've
got enough *other* stuff to do so why go looking for issues?)

> This was a major pain for a number of reasons.  It could
> only be reset via manual intervention which made remote
> action impossible.

That sounds like an overload, then.  Note that most UPS's
don't sequence power to their loads (outlets).  Since most
of those loads, nowadays, are reactive, I am surprised that
the UPS's don't "trip" on power up!  This could be why
many of the cheaper UPS's will not power on *unless* AC
mains are available (i.e., let the mains deal with that
start-up surge; thereafter, the UPS can "carry" the load)

> Once I had to have my landlord reset it for me while I was
> in Tucson.  Also, I bought the UPS specifically to insulate
> myself from anything bizarre on the incoming line.  I
> expected that if anything bad happened - too much or too
> little power, etc. - the UPS would switch to battery until 
> the problem passed.

<frown>  I think you are being overly optimistic, there! :>

There are (more or less) two classes of UPS's:  on-line and
standby.  The standby UPS is basically an extension cord
99% of the time; it just brings the AC mains to a set of
outlets for your loads.  As such, anything that happens on
the AC line is seen by your loads[1].  When the UPS detects
an anomaly on the line, it switches (usually via a genuine
relay) your loads to the output of a DC-AC inverter within
the UPS.  I.e., now you are running off battery (though the
battery may still be "charging" from the AC mains, if they
are still capable of running the charger).

The switchover is not instantaneous.  It is typically on the
order of 1 cycle (~15ms) -- the "transfer time".  So, during
this time (while the UPS decides if the AC line is "OK"),
your loads are still plugged into the AC line.  And, "see" what's
happening thereon.[2]

An "on-line" UPS avoids this transfer issue entirely.  Here,
your loads are *always* powered by the UPS's inverter.  At
the same time, the battery is being charged from the mains
*and* DC power supplied by the mains to run the inverter
(i.e., AC mains provides DC to inverter -- and charger -- which
then converts it back to AC for your loads).

In this topology, your loads are isolated (largely) from the
AC line at all times.  It's as if you were running off the
UPS's battery *and* that battery seemed to last forever!
(which it does -- as long as there is AC power available!  :> )

Here, there is no transfer time.  No "relay" in series with 
your loads.

> This problem stopped after I moved back to Tucson which
> seems to indicate that it was indeed some issue with the 
> incoming power.

If you *really* are concerned about power, you can buy
tools that will watch the AC line and log anomalies
(brownouts, blackouts, dropped cycles, spikes, etc.).
But, $$$$

> But, if so, why did the UPS not do its job?  I could find 
> no errata or firmware update on APC's website either.

See above.

> >Ah, all of your stuff is bunched together?  (I have stuff
> >spread around the house, hence the need for more/smaller
> >devices)
> 
> One of the few benefits of living in a small apartment.  I
> have a closet near the TV where I keep everything but the 
> desktop PC (so as to cut down on the noise) and I have a
> heavy gauge 20' extension cord so I can run my desktop
> machine off the same UPS.

Note that the output of many UPS's isn't very "clean".
If you get neighbors complaining about RF interference,
this is one quick fix to try (eliminating the long cord).

[1]  This ignores the effects of "surge supressors" in the UPS
Personally, I think these "supressors" are a complete waste!
They'll only handle trivial problems -- problems that the
power supply in most electronic devices will already handle!

[2]  This transfer time means that you can't use a "standby" UPS
in applications where your load can't hold itself up for the
duration of that transfer.