Re: BAD Problems with Asus Mobos

I have the same board Paul...the P4C800-E Deluxe and have noticed some wierd
anomalies from time to time, especially when booting which makes me think
that this problems extends to the 875 northbridge as well. I wonder if
these problems have been fixed in the newer 915/925/945/955 chipsets?

That's very discouraging to hear about these problems and I think that Intel
owes it to it's customers for a fix, but I think you're right in that they
probably won't do anything.

I was going to install a separate PCI USB 2 card to get around the problem
but are you saying that doing that may not work? I will have to try it as
it's the only way to save my face since they use a portable notebook hard
drive on the front USB to dump/retrieve data. That is probably what has
been killing the boards...the increased current requirement for a portable
drive.

Thanks again Paul....I'm going to try and contact Intel about this and see
what they say. I don't want to start a class action lawsuit, but I will if
I have to.

-Steve



"Paul" <nospam@xxxxxxxxxx> wrote in message
news:nospam-1907051438140001@xxxxxxxxxxxxxxxx
In article <tvadnbdbEpjUmUDfRVn-rA@xxxxxxxxxxx>, "Steve"
<spfouche@xxxxxxxxxxx> wrote:

> Thanks Paul, this was a very informative post. I know that on one
occasion
> that the entire USB controller did seem to die, but the mobo worked
> otherwise. I had to RMA the board for the USB controller problem, but the
> board failed a month later completely...again. I wonder if this
southbridge
> problem could kill the entire board? From the one picture it appears
> possible, but I've checked the southbridge controllers on every RMA
because
> I remember reading where they could get extremely hot on occasion but I
> wasn't sure why. I didn't find any sign of burning or discoloring
however.
> The northbridge seems to run hot on these boards as well, but I haven't
> checked the temp on the southbridge. One thing that your post made me
> wonder was that in each of these configurations I was using a RAID 1 setup
> using the ICH5R chip with dual Seagate drives. I wonder if the failures
are
> due to the RAID 1 possibly combined with the USB problem?..or due to the
> RAID 1.
>
> Thanks again for your post, plenty of airflow in these systems so whatever
> is happening isn't normal.
>
> I wasn't even aware of the USB problem until your post.
>
> -Steve
>

I think one poster mentioned losing USB functionality, without the
chip being burned. If the bond wire (or gold button, if this is flip
chip) fails first, it is possible the USB I/O will just lose power,
without the rest of the chip being damaged. But there seem to be more
reports of board-killing meltdown of the Southbridge, then of just
the loss of USB functionality. (And, even if you use a PCI USB card
in each computer, to avoid the risk of ESD killing the motherboard
USB ports, I'm not even certain that will be enough to prevent all
failures.)

I own a P4C800-E Deluxe, with the ICH5R, so I cannot say I'm at all
happy about the situation. Asus does seem to be replacing the
Southbridge on the boards returned for RMA (and with the right
techniques and materials, a replacement like this can be done up
to three times, before the board is too beat up to be repaired).
But, I expect they are using the same mask revision chips from
Intel, to make the repair, and so another failure is also likely
to happen. The way the chip business works, is you don't spin
new versions of chips, unless there is volume to pay for the
respin, and the ICH5 family is at the end of its production life.
At $1 million max per respin, it takes a lot of chips to pay for
the NRE. The big question right now, is who is paying for the
replacement ICH5(R) chips.

The next question is, what will happen after the three year
warranty is up ? Will Asus refuse to repair them, or will we have to
go after them with a class action suit, to extend the warranty ?
Class action suits can be successful:

http://www.abitsettlement.com

And in the end, will playing "whack-a-mole" with class action
suits, improve the quality of the products offered ? Quality,
after all, is determined by industry competition - if all
competitors make crap, the manufacturers don't have to work
too hard.

As for chip power levels, a portion of chip power depends on the
amount of high speed I/O. Some of the low amplitude technologies
like PCI Express, SATA, USB2, and so on, are quite good on I/O
power. But the amount of high speed logic backing up those interfaces
inside the chip can also be an issue, and contribute more power
dissipation than the I/O.

In a classic two chip Northbridge/Southbridge pair, the number of
I/Os and their speed is such, that the Southbridge generally has
much lower power dissipation than the Northbridge. In the Intel
doc 25267301, the TDP for the ICH5 is listed as 2.4 watts. That
should be pretty easy to cool. For Northbridges, the 865G is likely
to be the worst, and the Intel 25251902 document lists

Intel 82865G GMCH (Integrated Graphics @ 266 MHz)
Dual-channel / 4 DIMMs / 400 MHz DDR / 800 MHz FSB 12.9 W
Dual-channel / 2 DIMMs / 400 MHz DDR / 800 MHz FSB 11.8 W
Single-channel / 2 DIMMs / 400 MHz DDR / 800 MHz FSB 10.0 W

That is getting warm enough, that a tall heatsink with some
airflow over it should be used. (And, that is one reason to
be careful in the selection of third party HSF for the CPU.
Asus may assume the "air spill pattern" of the Intel retail
cooling solution, and it is a good idea to check that a
third party solution blows down onto the motherboard strong
enough, to keep MOSFETs and other warmer components cool. If
the Northbridge has a fan on it, that is one less thing to
worry about.)

One trend you will notice on motherboards, is the power level
of processors is rising, but the number of phases used in the
Vcore circuit remain the same. At one time, a single phase of
a regulator only provided about 15 amps. That number has climbed
steadily. On a P5WD2, there is a four phase Vcore circuit, and
the 840XE dual core draws 130W. The datasheet for the 840
lists 125 amps maximum current, which is 30 amps per phase on
the four phase regulator. I bet even Intel doesn't use more
than four phases on their Vcore regulator circuit, so the
stress level is the same.

http://processorfinder.intel.com/scripts/details.asp?sSpec=SL8FK
ftp://download.intel.com/design/PentiumXE/datashts/30683101.pdf

(Picture of Intel D955XBK - it has four phases on Vcore, and more
substantial looking heatsinks than the Asus equivalent)
http://developer.intel.com/design/motherbd/bk/bk_photo.htm

The potentially most stressful part of a motherboard design,
is how the powering is done. On the P4P/P4C series, Asus made
extensive use of linear regulation (MOSFETs and quad op amps),
and I'm not too impressed by that. On my board, two transistors
run detectably warm (40-45C measured with a digital thermometer),
while the rest are OK. I notice on the P5WD2, there are now a
couple of switching regulators used, where we might have found
linear regulators. That is a good sign, as a switcher is more
efficient (more expensive), and better able to do the job. Mobo
manufacturers are very reluctant to use them, due to the cost.

I still see reasonably good engineering principles being used,
with corners being cut on the same old things they've always cut
corners on (power regulation). Only someone who distributes the
boards, and keeps records of return rates or warranty claims,
will have a good picture of whether there is an epidemic out
there. Considering the small number of postings made to this
group, I'd say there are a significant number of P4P/P4C failures
out there. That is because most people seem to suffer in silence.

Paul


.

Relevant Pages

  • Re: BAD Problems with Asus Mobos
    ... I had to RMA the board for the USB controller problem, ... > using the ICH5R chip with dual Seagate drives. ... As for chip power levels, a portion of chip power depends on the ... a regulator only provided about 15 amps. ...
    (alt.comp.periphs.mainboard.asus)
  • Re: PPC/RISC STILL ON!!
    ... When Intel announced the CSI memory/cpu ... If it are first to market, ... not using the chip that is ahead. ... I am not sure what differences exist between Power and PowerPC, ...
    (comp.sys.mac.system)
  • Re: Computer
    ... Apple went for the Intel chip because of the power consumption vs. ...
    (rec.photo.digital)
  • Former Alpha Designers at Work
    ... DEC veterans prepare chip challenge for Intel, AMD, IBM and Sun ... PA Semi is this start-up hoping to turn a concept many might see as hopeless ... systems run fast while consuming less power than in the past. ...
    (comp.os.vms)
  • Re: P4P800 Deluxe Hangs...
    ... There was even one report, of this latchup-type failure, ... where the USB pads get their I/O power. ... throws a short across USB internal power. ... If the chip power wiring sustains ...
    (alt.comp.periphs.mainboard.asus)