In article <FMoxd.548948$%k.478661@pd7tw2no>, Robert Hancock
<(E-Mail Removed)> wrote:
> captnorm wrote:
> > Recently bought Asus A8V deluxe, AMD 3500+ socket 939 cou, Corsair
> > ram... System won't POST, no video at all, no beeps on speaker.
> >
> > Vendor thinks cpu or A8V or both got fried as I forgot to connect the
> > 4-pin 12-volt power connector.
> >
> > In re-reading the Asus manual, it says:
> > ... this motherboard requires that you connect the 4-pin to provide
> > sufficient power to the cpu.
> >
> > ... otherwise the system doesn't boot up.
> >
> > NOTHING about the consequences of not connecting it, like frying
> > anything, just that "otherwise the system doesn't boot up."
> >
> > I've read the threads about bios version issues with this combination
> > & forwarded them to my vendor. I would just like to get a concensus if
> > anyone thinks I did NOT fry the cpu or MB, or if that is very possible
> > by providing no power to the cpu.
> >
> > BTW, my email address has a 101 before the @, not 102
> >
> > Norm Perron
>
> I can't say for certain that it's impossible, without a detailed circuit
> schematic of the board, but I think it's very unlikely that not
> connecting the ATX12V connector would damage the board. It likely just
> won't boot.
The power sequencing requirement is section 7.8.3 "Power Supply
Relationships" on page 76:
http://www.amd.com/us-en/assets/cont...docs/31411.pdf
All that this info adds to the puzzle, is that it is possible.
(VDDA and VDD come from a +12V powered circuit. The kicker is,
where does VLDT come from, as VLDT should not be allowed to
rise, if VDDA/VDD are not running yet.)
But I would think unlikely, as there _should_ be provision in the
circuit to meet the requirements under normal circuit operation.
When a circuit has a sequence requirement, there should be stuff
in place to meet it, such as diodes or MOSFET switches. In my
experience with circuit design, there are extensive discussions
during the design phase, about exactly which of these requirements
really need to be met, as cutting costs and meeting sequence
requirements are competing objectives. In some cases, it takes
3 months to get answers from the factory, on whether all these
requirements are necessary or not. In extreme cases, the circuit
schematic may be reviewed by the chip manufacturer, for
correctness.
I suppose Asus could think along the lines "well, only the
processor will get burned if we do this", but imagine how much
future support Asus would get from AMD, if Asus causes a few
million Athlon64s to get replaced under warranty. And if the
user community became aware of it, it could have quite an
impact on future business. There is no upside to cheating on
the circuit requirements, which are clearly stated in the
datasheet.
HTH,
Paul