Why ASUS south bridge don't last...

rkauer

Amiga fanboy
Joined
Dec 17, 2007
Posts
10,337
Country
Brazil
Region
São Leopoldo, RS
My desk pc was acting up: unexplainable freezes, drives disappearing with no explanation...

A plain IDE DVD burner (inherit from old pc) and the SATA HD...:|

Finally the mobo died two weeks ago: no drives recognized at all, no matter what (tested with old IDE PATA drive only, with old XP installation, you name it).

Finally I opened (again) the tower and sacked the mobo out for a replacement.

Checked for salvageable parts (I love to destroy pc boards for Amiga bits: heatsinks, battery holders, ATX connectors and the like).:nod:

When I sacked the south bridge heatsink... Shock! Horror!:blink:

ASUS, as cheap as they can be, simply put half of the thermal paste they must put on the heatsink, causing odd thermal dissipation and finally killing the board.:Doh:

NOTICE TO ALL: when buying a new pc board, take your time and do a complete thermal compound replacement on all heatsinks. This will save the pc mobo for a couple more years.
 

Attachments

  • 100_0630.JPG
    100_0630.JPG
    58 KB · Views: 1
wow.... cheapo!
will have your advice in mind rk!

(also I also butcher parts from old/semi-dead pc mobos LOL)
 
I've got an old Asus mobo that was playing up the last time I used it. It would sometimes not detect the SATA drives. If you left it off for a while, it would work for a bit, which really does suggest cooling issues.

I might try whacking a bunch of arctic silver between the southbridge and heatsink, see if that can save the thing. Thanks for the heads-up, rk!
 
Actually, thermal conduction doesnt require a full transmission layer from the heat producing device to the heat dissapation device in order for good efficiency.

So although only having half the cpu transmitting heat to the heatsink via the thermal compound is not as efficient as having the full surface area coated, it does not make a HUGE difference.

Any material can only conduct away so much heat before thermal resistance lowers the level being shared out, so it really means that it takes longer for the heatsink to heat to that "dead zone" than it would if there was a lrger surface area to saturate the heatsink with.

In other words, it is not good practice, but I would be surprised if this is what actually killed your MB.

---------- Post added at 16:29 ---------- Previous post was at 16:29 ----------

@ Andy - you are betting man? Ill bet it makes no difference. Are we on? :)
 
Heh, I'm probably clutching at straws anyway. It's one of those "I guess it's worth a try" moments. It's possible the bridge is already screwed and the damage is done. Assuming that was ever the problem.

Still, as I say, worth a go and if it works, hooray. Otherwise nowt ventured, nowt gained.
 
Definitely. What have you got to lose.

It IS possible the problem is heat related.... the heatsink under perfect conditions may be JUST able to keep the chip cool enough to be stable/reliable, and the lack of sufficient thermal surface area is dropping the cooling down by say 10%... which may be enough to cause probs.

Let us know if it made any difference.
 
Mate, the real problem is ASUS simply put the compound in one half of the HS, not covering the centre of the piece.

This way the centre (the part where the heat is building up) didn't have enough thermal transfer, killing the chip after some months.
 
@abraXXious

I am sorry but I really feel I must post the following, although its likely to take 20 minutes with my internet to update!

abbraxxious said:
Actually, thermal conduction doesnt require a full transmission layer from the heat producing device to the heat dissapation device in order for good efficiency.

So although only having half the cpu transmitting heat to the heatsink via the thermal compound is not as efficient as having the full surface area coated, it does not make a HUGE difference.

Any material can only conduct away so much heat before thermal resistance lowers the level being shared out, so it really means that it takes longer for the heatsink to heat to that "dead zone" than it would if there was a lrger surface area to saturate the heatsink with.

This is full of so much wrong I don't know where to begin! it is such incredibly poor advice if my exceptionally poor internet would allow me to moderate it I would.

Contact surface is critical for heat transmission, otherwise there would be no need for LARGE heatsink contact surfaces and large air heat dissapation surfaces. Simply put Alu430 (which most anodized aluminum sinks are made from) only has a finite heat-transfer co-efficient per cm2 of surface contact.

Thermal pastes provide a more effcient heat transfer between the source and the sink by reducing pits and ridges of contacting surfaces, if this is applied incorrectly it can introduce air vacuoles whereby there is no contact surface, as such this will heat up a lot more than the surrounding fully contacted area - leading to hot-spots on the heat source as there is only so much heat (watts) an Aluminum surface can dissipate.

To prove your point I suggest that you your coat half a cpu in thermal paste and apply the sink, you will notice in the best case a serious increase in temperature, and in the worst case a dead processor - your argument to the contrary of not having as much contact surface as possible for heatsinks and processors (Northbridge / CPU's) is very poor advice indeed.
Mozilla/5.0 (Windows NT 5.1; rv:2.0.1) Gecko/20100101 Firefox/4.0.1
 
@ Zetr0. I understand the point you ae making Zetr0, but in I think you may have misunderstood mine.

What I was trying to say was a full surface transition of heat is not necessary for good heat conduction. Of course, the more squares surface area interfacing between the heat device and the heat conductor the better, but it is a case of deminishing returns.

For example. If a heat producing device, lets say a microchip, has a surface area of 1600 square mm (40x40) and with no heatsink etc it would typically heat to a critical thermal load of 90 degrees C.

Then you put on a heatsink with the same surface area (1600 square mm) using a thermal transition layer. The transition layer is never a perfect heat conductor (which is impossible, governed by the laws of thermal dynamics) and so has some thermal resistance, as does the heatsink above it. So, lets say this setup will conduct away 30 joules of thermal energy per second when the air surrounding the heatsink is say 25 degrees C, 70% humidity (now we are getting technical :)). Lets also assume this drops the Micro chips surface temperature down to 50 degrees C.

Now, if you had the EXACT same setup, but only HALF the microchips surface area was interfaced to the heatsink via the thermal transition layer, you would probably still be conducting away 26 joules of energy, and the microchip surface area would probably be 60 degrees C instead of 55.

In other words, having a bottleneck to the heat conduction is not good, but as the thermal conducting device still gets saturated by thermal energy and still has to dump this to the surrounding area it is not a one for one reduction in efficiency.

I went through all of this when I did some testing back in the 486 days.... remember the 100 mhz overdrive chip? :)
 
NOTICE TO ALL: when buying a new pc board, take your time and do a complete thermal compound replacement on all heatsinks. This will save the pc mobo for a couple more years.

I always do this with any board i buy... I apply Arctic MX-4 compound which i personally find to be the best :)

What Asus board do you have? Mine is an Asus M4A88TD-M EVO/USB3 and the thermal paste seemed to be applied ok (but i changed it anyway as all mobo manufacturers use cheap paste!).... Also i just held my finger on the southbridge (AMD SB850) heatsink and it's only slightly warm after being used for most of the day. Is your board using a volcanic Nvidia chipset or an older Intel/AMD chipset by any chance?

---------- Post added at 21:58 ---------- Previous post was at 21:54 ----------

I had one bad asus board back in 2000, never bought their crap again.

I think their sister company 'Asrock' are pumping out some quality boards these days whilst keeping a low price tag... I don't think they like to be known as the lower end of Asus anymore :D

Oh and i've only ever owned one Asus board (which i'm using now) and it's been perfect so far... The board i bought before this had to be returned as it was faulty (a Gigabyte board which is also the first board i've had from them).
 
Powerpie is spot on with his comments. Asus used to be a quality brand, but on the overclocking and 'bleeding edge' hardware forums, Asus is becoming a name to avoid. They have some reliability issues that seemed to creep in around the Pentium 4 to Pentium D era and it seems to have gone downhill from there.

He's also right about Asrock; I have always liked Asrock as they 'buck the trend' and come up with some really left-field ideas; their 775-Dual VSTA and 775-Quad VSTA boards were absolute genius for those without big pockets, who wanted a sensible upgrade path. I mean, how many other motherboards can you name, that crammed support for DDR and DDR2, SATA and IDE, along with AGP and PCI-E all on one board and still had room left for four PCI slots? These motherboards were the 'Swiss Army Knife' of Socket LGA 775 boards -Brilliant!!

I still have a 775-Dual VSTA and was able to overclock the Celeron D352 that is still in it up to 4Ghz on the stock AMD cooler; for a board that is considered 'budget', that's not to be looked down on. I guess that makes me a bit of an Asrocker fanboi.....:cool:

Edit: I found the original CPU-Z report from my overclock - here it is; this was originally posted on OldskoolUK. I had managed to get the FSB up from 133 to 166 by using Corsair XMS2 memory rated at 675Mhz. Keeping all of the clock settings within multiples of 33 I ended up with this:-

Overclock2.jpg


The machine was (As)rock stable and ran at 45C maximum under load.

Asrock FTW!! (y) Asus could learn a lot from them.....although I *think* that Asus and Asrock are separate companies these days.
 
ASUS are alright as a mainboard brand - Now Gigabyte on the other hand are a complete pile of excrement IMO
 
TBH, Abit went off the boil in terms of quality for quite a while as well. I remember that the BH-6 and BM-6 Celeron boards were ace, but then I got an ST-6 Raid board and I took that back to Aria four times before I gave up. The non-raid ST-6 was equally bad and I settled for a Gigabyte board as a replacement.

The Aria technicians tested my RAM, CPU and everything checked out but they wouldn't work on those ST-6 boards.

It took quite a while before Abit came back to form and they got hardcore with the Fatal1ty branded boards.
 
@PP5k: mine was an ASUS P5gc-mx socket 775, Intel 945 chipset.

A tad old, but still a runner if not for the crapped south bridge.(n)
 
My last Abit board was the IX38 QuadGT and it had my good old Q6600 running stable at 3ghz with no voltage increase needed (something my MSI P45 Platinum couldn't manage for some reason!). I had a Foxconn BloodRage board in my old i7 940 rig but that PC was extremely power hungry and would heat up the whole room!! It also had dual Radeon 5850's in crossfire, but i decided to sell it while it was still worth something as i spend most of my gaming time on the X360 or PS3. I now just stick with a humble quad core AMD build with a Radeon 6870 (which still handles anything i throw at it) It was cheap to build and is very energy efficient compared to my old i7 rig :)

I'm not too sure what the best board brands are these days... So many quality brands such as Abit, EPoX, Chaintech and Soyo are no longer with us. One of my favourite socket 7 boards was made by a company called 'Commate' and it was brilliant!

Anyway, most mobo manufacturers seem to concentrate too much on overclocking and core unlocking functions these days... Something i'm not too fussed about anymore (anyone remember the Celeron 300A glory days with an easily acheived 50% overclock?). I may look into getting an Asrock board for my next build (built around either the AMD Bulldozer or Intel Ivy Bridge).
 
I had one bad asus board back in 2000, never bought their crap again.

I think their sister company 'Asrock' are pumping out some quality boards these days whilst keeping a low price tag... I don't think they like to be known as the lower end of Asus anymore :D

Agreed, my old reliable board was an ASROCK.

Now Im running a quality MSI board, :).
 
@PP5k: mine was an ASUS P5gc-mx socket 775, Intel 945 chipset.

A tad old, but still a runner if not for the crapped south bridge.(n)

Yeah those old Intel chipsets (and especially the CPU's) ran pretty hot compared to todays standards... Now would be a good time to buy something better, even if it's another SKT 775 board :)
 
Bought a replacement 775 board with same memory and CPU tolerance, as I keep my other parts (2x 2Gb DDR2, PentiumD 3.0 until my Dual Core 3.6 arrives).

I'm not a PC gamer, anyway...

I just play a session of Portal from time to time and Amiga emulation (mostly to save my Amiga partitions and file transferring).
 
Back
Top Bottom