K8N-E ATX PS Voltage problem?

  • Thread starter Thread starter Bryon van Prooyen
  • Start date Start date
B

Bryon van Prooyen

Hi Guys,

I have posted a couple of times now trying to find a fix for my system
freezing everytime I do MPEG2 to MPEG2 re-encoding, BTW thanks to those who
have given me things to look at so far, it all helps.

My system again...

Asus K8N-E Deluxe Mobo Bios 1.05
Athlon64 2300 (Not overclocked)
1024MB Kingmax DDR-466
Asus 9600XT Video
WD 120GB SATA (Boot)
WD 80GB ATA (Data)
480W Tagan PS (TG480-U01)
Windows XP SP2
VisionPlus DTV Digital TV tuner

I am still having problems where re-encoding MPEG2 streams from my
VisionPlus DTV card using pretty much any MPEG2 software I can find
(including TMpgEnc) eventually freezes my computer. It does not BSOD or
reboot - just freezes.

I am now convinced the system is otherwise stable. I have tried Memtest86+,
prime95, and just about all of the load tests on the "Ultimate Boot CD"
(which is great BTW) and all pan out flawlessly.

I have noticed one thing of concern, I have checked all of the PSU rails
going to the Mobo and all are in perfect spec. The supply that concerns me
is the 12v rail. With a multimeter at to 20pin and 4pin connectors I get
12.01V exactly - does not vary at all even with load change using Prime95 or
TMpgEnc loading the system down. However, the BIOS, Asus PC Probe and
HMonitor all report the 12v rail as varying between 11.56 and 11.72 volts,
never any higher.

Has anyone else with one of these boards noticed this? Is this normal, or
have I found a problem?

Tanks again in advance

Bryon
 
Bryon van Prooyen said:
Hi Guys,

I have posted a couple of times now trying to find a fix for my system
freezing everytime I do MPEG2 to MPEG2 re-encoding, BTW thanks to those who
have given me things to look at so far, it all helps.

My system again...

Asus K8N-E Deluxe Mobo Bios 1.05
Athlon64 2300 (Not overclocked)
1024MB Kingmax DDR-466
Asus 9600XT Video
WD 120GB SATA (Boot)
WD 80GB ATA (Data)
480W Tagan PS (TG480-U01)
Windows XP SP2
VisionPlus DTV Digital TV tuner

I am still having problems where re-encoding MPEG2 streams from my
VisionPlus DTV card using pretty much any MPEG2 software I can find
(including TMpgEnc) eventually freezes my computer. It does not BSOD or
reboot - just freezes.

I am now convinced the system is otherwise stable. I have tried Memtest86+,
prime95, and just about all of the load tests on the "Ultimate Boot CD"
(which is great BTW) and all pan out flawlessly.

I have noticed one thing of concern, I have checked all of the PSU rails
going to the Mobo and all are in perfect spec. The supply that concerns me
is the 12v rail. With a multimeter at to 20pin and 4pin connectors I get
12.01V exactly - does not vary at all even with load change using Prime95 or
TMpgEnc loading the system down. However, the BIOS, Asus PC Probe and
HMonitor all report the 12v rail as varying between 11.56 and 11.72 volts,
never any higher.

Has anyone else with one of these boards noticed this? Is this normal, or
have I found a problem?

Tanks again in advance

Bryon
Click to expand...

I've noticed the difference between the Power Monitor and a real
multimeter as well. My Antec power supply reads bang on, when read
with the multimeter, whereas the power monitor will give numbers that
are off by a bit. I would consider that to be normal, as you cannot
expect the ADC in the monitor chip to be that good (accurate).

Now that you've tried Prime95, and passed, that suggests to look
elsewhere. I don't know if you mentioned this "VisionPlus DTV" card
in your previous post or not.

Are you pulling data from the DTV card, processing the data, and
writing it out ? Or are you storing the data on disk first, then
the DTV card is ignored while the computing takes place ? The
reason I ask, is to determine if DMA transfer is happening when
the freezing happens.

It could be, if the computer dies while data is being transferred
from the DTV card, that there is a PCI bus problem. Some of the
old TV cards, for example, would move 25MB/sec from the TV card
to the frame buffer on the video card. Moving that much data was,
in effect, a stability test for the PCI bus. Some computers fared
poorly, and froze up in no time. (This was back in the days of
BT848/BT878 WinTV cards.)

The problem could be happening during the DMA bus transfer, or it
could be some kind of interrupt problem. You could try stuffing
the DTV card in PCI Slot 5 (furthest from the AGP slot), while at
the same time having no PCI card in Slot 1 (nearest the AGP slot).
That would give a unique, physically unshared, interrupt signal.

I expect though, that this is a data transfer problem. And the
only solution for that, is to either RMA the PCI card, or try
a different brand of card. You might try the DTV card in any
other PCs you own, to see if this is a problem with the chipset
on the K8N-E board. Either could be at fault.

You could report this to Asus, and maybe at some future point
they can make a BIOS change to improve the stability of PCI
cards that move a lot of data. These are the three interfaces
I've found so far - don't know which one would work best, so
maybe try all three :-)

(e-mail address removed)
http://Helpdesk.asus.com/
http://vip.asus.com/eservice/techserv.aspx

There is at least one kind of hardware problem that cannot be
fixed by a BIOS change. PCI chips have setup and hold time specs
to meet, and the PCI card designer has to adjust the length of
the clock track on the PCI card design, to make the timing right
for his card. It tooks some companies a while to figure this out.

Maybe this site will help ? This could be a software problem too.
http://forums.dvbowners.com/index.php?showtopic=1311&hl=freeze

HTH,
Paul
 
Thanks for your input again Paul.

I suspected this as well, I originally had the PCI card in slot 2, then
moved it to 5, then totally removed it. Problem still there.

BTW, I am capturing the stream using the software provided with the TV card
(which is just a basic MPEG2 stream) then a couple of days later going back
and editing/re-encoding the streams to better suit DVD players.

I have just today tried running TMpgEnc under XP 64bit edition and it still
crashes (this is without the TV card installed). I guess this rules out SP2
being the problem.

I think I may have found a fix for this now, but there may be a couple of
reasons for it. I tried disabling all the MMX, 3DNow, SSE, SSE2, etc options
in TMpgEnc and it did not crash - just took 5 times longer to process (About
1hr for 7mins of video).

It may be just the fact that IDE and SATA transfers were slowed down that
avoided the problem, or there are problems with these boards (or my board)
executing these enhanced command sets.

The very last thing I can think of is temperature of the nForce3 chip - it
must be running 50 or 60degC as it is too hot to touch for more than a
second or two. I am considering buying a fan-forced VGA cooler and
installing it in place of the pathetic one factory installed. I have added a
lot of cooling to the case and this has helped reduce all the temps, but
this chip is still VERY hot.

Bryon
 
Bryon said:
Thanks for your input again Paul.

I suspected this as well, I originally had the PCI card in slot 2, then
moved it to 5, then totally removed it. Problem still there.

BTW, I am capturing the stream using the software provided with the TV card
(which is just a basic MPEG2 stream) then a couple of days later going back
and editing/re-encoding the streams to better suit DVD players.

I have just today tried running TMpgEnc under XP 64bit edition and it still
crashes (this is without the TV card installed). I guess this rules out SP2
being the problem.

I think I may have found a fix for this now, but there may be a couple of
reasons for it. I tried disabling all the MMX, 3DNow, SSE, SSE2, etc options
in TMpgEnc and it did not crash - just took 5 times longer to process (About
1hr for 7mins of video).

It may be just the fact that IDE and SATA transfers were slowed down that
avoided the problem, or there are problems with these boards (or my board)
executing these enhanced command sets.

The very last thing I can think of is temperature of the nForce3 chip - it
must be running 50 or 60degC as it is too hot to touch for more than a
second or two. I am considering buying a fan-forced VGA cooler and
installing it in place of the pathetic one factory installed. I have added a
lot of cooling to the case and this has helped reduce all the temps, but
this chip is still VERY hot.

Bryon
Click to expand...
<<snip>>

Regarding the Northbridge heat, maybe you could try adjusting things
that might influence its temperature. The HT multiplier, for example,
if reduced from 4 to 3, might make a difference. (The option isn't
shown in the manual, but appears in a picture on a review site. It
might only appear in the BIOS when Cool N' Quiet is enabled.)
Also, the AGP voltage setting would make a tiny difference to power
dissipation.

http://www.pcper.com/article.php?aid=78&type=expert&pid=5

Since memory is controlled by the processor, I don't think a hot
Northbridge would affect an actual computation. If you are using
Prime95, that shouldn't need to go near the chipset. But, if
TmpgEnc is doing disk accesses, that would go from Northbridge to
Southbridge, and maybe with the hot Northbridge, that is when
the problem happens. Could you set up a smallish ramdisk and place
the source video material on it, then run the encoder ? If that
works without problems, then it could be the storage end of things
that is letting you down.

Another possibility is the chipset voltage is wrong on your
board. The only way to check for that, is to see if other
users are complaining about the hot Northbridge.

In this thread, K8N-E owners change the heatsink for a different
one, but I don't see them using a fan. A fan can make a tremendous
difference to temperature. Maybe you could purchase a 40mm fan
and use metal screws jammed between the fins to hold it in place.

http://www.nforcershq.com/forum/viewtopic.php?t=51241&highlight=k8ne

Just another one of those mysteries I guess.

Paul
 
Once again a very helpful response Paul :)

I will try and install one of my spare 40mm fans onto the chipset heatsink
and let you know if it helps. I will also try lowering the HT bus speed, be
interesting to see what performance is lost doing this, but if it makes it
stable it will be worthwhile.

As for people using huge heatsinks and no fans, great for making a nice
quiet PC but mine is already pretty loud - counted 7 fans in it before so I
can put up with another.

As for your link regarding HT and AGP settings
http://www.pcper.com/article.php?aid=78&type=expert&pid=5 I noticed a screen
showing power supply voltages in BIOS, consistent with mone. I am guessing
the 12v rail is measured at a different point on these boards, possibly
after another voltage regulator.

I will let you know how I progress.

Bryon
 
I think I have stumbled across the problem!!!!

I dismantled my PC to install the Zalman NB47 heatsink onto the nForce3
chip, reassembled and started up again. Tried video encoding and about 30
minutes in it froze again.

Frustrated, I rebooted and read some emails. Froze just doing that! Rebooted
again, this time into Memtest86+ and found a bad byte in one of my DDR-466
sticks. Thought it may have been a fluke, so rebooted and tried again - same
address and same fault. Pulled the sticks out and swapped them, faulty
memory address moved accordingly.

Looks like I had a memory stick that was intermittently faulty - as I have
run Memtest many times on this system with no faults reported whatsoever.
Maybe disturbing the PC by pulling it apart has helped uncover the fault?
Beats the hell out of me...
 
Back
Top