[mythtv-users] Power outages and UPSs

f-myth-users at media.mit.edu f-myth-users at media.mit.edu
Sat Jul 28 09:06:32 UTC 2007


    > Date: Sat, 28 Jul 2007 02:48:02 -0400
    > From: "Michael T. Dean" <mtdean at thirdcontact.com>

    > I could make an UPS from a good battery (I've been looking at the
    > Concorde Lifeline GPL-31T :) and a good-quality, high-efficiency
    > benchtop power supply.

I'd use a purpose-built battery charger, not a benchtop supply.
Getting this wrong will kill the lifetime of your battery fast.
Getting it -really- wrong can lead to messy venting (or bangs).  Even
(good) lead-acid chargers are surprisingly complex; other chemistries
(NiCd, NiMH, Li in various incarnations) are increasingly hairy.

A purpose-built charger will give you much faster recharging and
better charge termination, as well as being able to compensate for
cell temperature.  Ignoring float will sulfate your battery and
undercharge it as well.  Attempting to recharge only via float can
take days, and you need to set the float voltage typically to within
10mV/cell---and you've also got to make sure your charger (or your
benchtop supply) is matched to the cell chemistry, of which there are
half a dozen even in ordinary lead-acid varieties.  (Gel cells, for
example, typically want somewhat unusual float voltages; AGMs have
more typical voltages but will accept huge currents during fast
charge; etc.)  A benchtop supply is likely to produce insufficient
current for fast charge, be unable to differentiate between the
various charging regimes and thus will either charge very slowly,
undercharge, or incorrectly float, and otherwise do a poor job of
keeping your battery charged and conditioned---and likely cost more
than a purpose-built charger as well.  Especially after you factor in
replacing the battery every few months because you're mischarging it.

Note that the net is chock-full of various homebrew designs for all
kinds of lead-acid chargers, of varying complexity and efficiency,
and also chock-full of vendors willing to sell you commercial gear.

[More details would be far off-topic and inappropriate here.  You
could try searching for ``charging lead-acid'' and poke through the
1+ million hits therein.]

P.S.  It's completely unclear to me why people are so hung-up on the
"inefficiency" of the inverter; all consumer UPSes only run it when
you're on-battery, which by definition is rarely.  And you can only
get rid of the always-on "inefficiency" of a typical switching power
supply (already about as efficient as any supply you'll see) if you
can supply exactly the required voltage from your batteries, which you
can't because their cell voltages droop during use and are much higher
than nominal while charging---there's no way the +5/+3.3V rails in
your CPU are going to be directly connected to any battery.  (That
"low-efficiency" computer PSU can approach very high efficiencies if
correctly designed, and most are; if you think you're doing better as
an amateur, you're deluding yourself.  And yet, you -have- to have
exactly the same sort of power supply to keep your battery charged!
So whether that PSU is running the computer from main power, or is
keeping your battery charged and -that- is running the computer,
you've got one.  In fact, the "keep the battery charged and run from
that" solution requires -two- DC/DC converters, most likely---the one
that charges your battery, and the one that converts your battery to
-exactly- what the computer wants.  One of them may be hidden from
sight, but it's there.  And since "your" DC/DC converter must be sized
big enough to -run- the entire machine, plus charge a discharged
battery (as opposed to just to recharging a battery at a lower
amperage than might be required to run the machine 'cause the machine
is running directly off the mains), you'll have correspondingly higher
absolute losses in it even if it's as efficient as the computer's PSU,
-especially- since you'll have oversized it for the run+charge state
but will almost always be using it in the run+float state.  The only
way to win this game is to use a CPU that requires less power; you'll
still have the same percentage inefficiencies, but the absolute amount
of wasted power is lower.)


More information about the mythtv-users mailing list