View Single Post
  #3  
Old June 13th 21, 01:34 PM posted to alt.comp.hardware.pc-homebuilt
Paul[_28_]
external usenet poster
 
Posts: 1,467
Default Something weird going on with DreamPC from 2006, bad smell comingfrom heatsink on north bridge chip ?

skybuck2000 wrote:
New hypothesis:

Short version:

"Flaw in the design of the NFORCE4 northbridge chip"
"It cannot handle DUAL PCI Express traffic and will OVERHEAT".

I haven't even checked any specifications about that the **** a north bridge chip does ! HAHA ! But this is my hypothesis for now people ! =D See below for mooorrreee detective work ! =D

1. I played submarine test of World of Warships from USB

3.0 PCI Express Controller on DreamPC 2006

while at the same time:

2. Also playing submarine test of world of warships from

GT 1030 PCI Express graphics card.

Perhaps the combination of both led to the burn through of

the north bridge chip.

The north bridge chip now has to handle TWO PCI devices

with super high bandwidth ?!?!?

Perhaps this became to much for the north bridge chip

?!?!?!?

Perhaps this also explains why the DUAL 7900 GTX Graphics

Card SLI setup burned THROUGH multiple motherboards !

Perhaps now after 15 years the thruth is coming out and

there was a DESIGN flag in the NFORCE4 chipset and the

north bridge simply cannot handle so much PCI Traffic and

it BURNS THROUGH !!!

Could this be the culprit !

Skybuck detective is on the case ! =D

Bye for now,
Skybuck ! =D


If you were going to remove the heatsink, then a
heatpipe replacement would have been a good idea.

This one is an example. There used to be another one
that was L-shaped, for allowing a video card to
pass over the Northbridge when plugged in. There's
no branding on this, so I don't know who makes it.

https://www.ebay.com/itm/393071031098

*******

When you change out hardware, you should always turn
off all power to the PC.

Unplug it, for example.

NVidia PCI Express bus are ESD sensitive. Some of their
chipsets seemed to have little in the way of ESD protection,
This means, for owners of these older systems, it pays
to wear an ESD strap when inserting new cards in any slot.

Some batches of NV boards were so bad, they were
arriving with one PCI Express slot already blown.
Even though the motherboards are functional-tested
at the factory, in the two minute long test of such.

One of the failure modes for silicon is called latchup.
Latchup makes a chip so hot, that it will actually
burn a hole in the top of the chip. And the "magic smoke"
escapes.

To recover from latchup, requires complete power removal.
The junctions have to drain below maybe 0.5V for the
latchup PNPN junction to stop conducting. I had a chip
at work which went into latchup, and the chip did not
"recover" until eight hours later with the bench power
turned off. What was neat about that chip, is the
chip did not get hot. IBM designed the chip, and they
have some clever clever people. None of my stock
of chips burned out.

There is another possible explanation for excess heat,
and that would be an onboard regulator failure that
supplies core voltages to the Northbridge. Some of these
chipset chips, have five power supplies feeding them.
3.3V, 2.5V, 1.5V, 1.2V and so on. The majority of logic
gates, run at the lowest voltage. Only a few I/O signals
run off 3.3V. If one of the tiny regulators making
those supplies rises to too high a voltage, that
can make the chip hot. But those regulators also have
current limiting, and the motherboard could simply
refuse to start, if the consumption is too high.

You can try unplugging for a day or two, but if it
still raises a stink when next started, it could be
a regulator failure. If you had an infrared camera,
you might spot which additional components are hot.

Paul