OT: UPS for FreeBSD

[Erich Dollansky]

Hi,

On Sat, 29 Nov 2014 00:49:07 -0800
perryh at pluto.rain.com (Perry Hutchison) wrote:

> Erich Dollansky <erichsfreebsdlist at alogt.com> wrote:
> > As switching voltage regulators use a few hundred volts,
> > it is a bit difficult to match this with a battery.
> 
> Depends on how many cells one is willing to string together
> in series.  IIUC the battery in a Toyota Prius supplies a
> few hundred volts.

are you sure?
> 
> However, there is no intrinsic reason why a switching regulator
> _has to_ take a few hundred volts as input.  Those under discussion
> here are set up that way because that is what is easily and cheaply
> available:
> 
The reason is very simple: cost.

> 120VAC ==> full-wave rectifier ==> ~300 VDC ==> switching regulator
>                                                        |   |
>                                               12VDC <--+   |
>                                                            |
>                                                5VDC <------+
> 
> (That is typical for North America; in Europe the AC would be
> ~240V and the DC therefore ~600V.)
> 
> To integrate a UPS with the PSU, one would instead build something
> along the lines of:
> 
>           multitapped
> 120VAC ==> step-down ==> 6VAC ==> full-wave ==> ~15VDC ==> battery1
>           transformer             rectifier
>                   |
>                   +---> 3VAC ==> full-wave ==> ~7.5VDC ==> battery2
>                                  rectifier
> 
> battery1 ==> regulator ==> 12VDC
> 
> battery2 ==> regulator ==> 5VDC
> 
How much energy is then wasted?

On the other side, there are concepts used in which the 'power supply'
converts the incoming 230V AC to something like 48V which is then
brought to the individual sections of a large computer and there
onverted into the required voltage.

> IOW the computer runs off the batteries, which are charged by the
> stepped-down but unregulated (or minimally-regulated) line voltage.
> 
> There is nothing the least bit new about this -- the telephone
> industry was doing it in the mid 20th century.  They installed
> enough battery capacity for their system to run on battery power
> alone for however long it would take to get their standby generators
> started up (or for at least a few days in smaller offices that did
> not have standby generators).

The real big data centres should still use the same concept. The last
big data centre used 100% the same concept the telecommunication
industry uses for their own network. The data centre used even to have
in the basement a 'swimming pool' which kept below the freezing point
to have even cooling available without the need to power it with the
generators.

Erich