Archived from groups: alt.comp.hardware.homebuilt (
More info?)
On Sat, 30 Oct 2004 18:03:43 GMT, "jimjim"
<netuser@blueyonder.co.uk> wrote:
>>I loaded that page on the box with the a7n8x-dlx in it, no
>>noise, but not an XP2800.
>yeah, my friend has the a7n8x-dlx, XP2800 combo and has no problems at
>all.This is one of the reasons that I chose the same cobination of
>components, but apparently there are some differences btw the a7n8x-dlx and
>the a7n8x-x
The VRM circuit is almost identical, but as I mentioned in a
prior post the a7n8x-x may have lower-spec capacitors, like
those seen in this pic,
http://www.cdromland.nl/hardware/moederbordamd/asus-a7n8xx-big.jpg
Not that they're "bad" caps, not at all they are good but
not AS good when it comes to ESR. It's not like you can
easily fix that though, Rubycon's MBZ used on the -DLX are
about as low as possible for an electrolytic (ignoring their
later MCZ, not even sure if MCZ come in suitable range for
this application). Frank Zabkar made a good point in the
previous thread, that ideally the two inductors should
match. I think it'd work with a "near" match but if you had
a pair of identical inductors then it should be easy enough
to just swap them in... but frankly I'd just put epoxy on
them if the board works ok otherwise, since disassembling
system, getting RMA/ship/buy/receive/reinstall, THEN
reinstalling OS if board is different, is a whole lot of
time and effort when a blob of epoxy will do the job,
assuming it is ONLY that inductor making noise, which
should be easy enough to tell by firmly grasping it with
your fingers while it's making the noise.
>
>> If you were to change the CPU's default voltage or over (or
>> under) clock it, that might be enough to reduce occurance of
>> noise if not eliminate it.
>yeah, this will be a good idea, as I will be able to verify whether it is
>the power regulation circuit that causes this problem. Suppose I try to
>under-clock it (assuming that I will still have a stable system), what
>voltage should I set?
If it's that inductor #1 in the pic, it is definitely the
CPU power regulation circuit, no question about it.
If you tried underclocking you wouln't "necessarily" even
need to change the voltage. Actually a better first attempt
might be leaving the CPU speed the same and trying slightly
lower or higher voltage. I'd sooner overclock a little
rather than underclock but if you dont' want to then yes,
decreasing the voltage can be expected to work with enough
of a CPU clock speed decrease... how much depending on how
much the CPU speed is reduced. I don't even remember the
default voltage for an XP2800 but possibly 1.65V?
It may work at 1.6V at stock Xp2800 speed, try Prime95's
torture test "In-place Large FFTs" to check stability, often
a voltage too low will cause error within a minute or two in
that test. Maybe even lower voltage will work, it can vary
per CPU. If you were to raise voltage to 1.7V it'd likely
run fine at least past XP3100 speed.