Re: Building computer - power supplies
On Sat, 28 Sep 2013 16:18:30 -0400
Doug <dmcgarrett@optonline.net> wrote:
> On 09/28/2013 03:23 PM, Stan Hoeppner wrote:
> > On 9/27/2013 6:37 PM, Joel Rees wrote:
> >> On Fri, Sep 27, 2013 at 9:36 PM, Stan Hoeppner
> >> <stan@hardwarefreak.com> wrote:
> >>> A point I forgot to make. This is something everyone should know.
> >>>
> >>> Subject: The marketing myth of multiple +12V rails
> >>> [...]
> >>
> >> What I want to know is why Intel CPUs still need the +12V.
> >
> > They don't now and they never did. The 8088 through 80486 and the
> > first
>
> /snip/
>
> >
> > Installing two voltage regulators next to the CPU socket and using
> > standard ~22 gauge copper wires from the 12V rail of the PSU solves
> > the problem cheaply. The 12V rail was chosen because 3.6x less
> > current is required vs using the 3.3V rail as was used previously,
> > which means much smaller wires are needed.
> /snip/
>
> Do I understand correctly that there are two switching power supply
> chips at the input to the CPU to produce 3.3Volts? Obviously, a linera
> regulator cannot produce and gain in current. Iin = Iout for linear
> regulators.
>
Yes, though they are for different voltages, typically 3.3V and 2.5V or
1.8V, see the voltage monitoring in the BIOS or MB monitor for the
values for a particular board.
You'll find two inductors, probably toroids, and a group of tall
aluminium electrolytic capacitors in the area. These electrolytics
carry a very high ripple current, and tend to run warm and to be the
weakest links in the system, as they will dry out with age, even if
they are run within their temperature and ripple specifications. But
linear regulators could not even be considered for this task.
--
Joe
Reply to: