M
McGrandpa
Fish said:Thanks for the info, McGrandpa - very useful!
I will check the voltages given in the BIOS and report what I find.
I have a question regarding what you said about regulator ICs (just to
make sure I've understood it correctly):
When you say, "If the regulator IC flakes out, so does your video.
There is no thermal sensor in these...", do you mean that a rise in
temperature is causing the regulator ICs to fail (causing the crash,
and they're somehow restored on reboot)? or that they've already
failed (for whatever reason and can never be restored) and this is
causing my garbled voltages? I take it you mean the latter?
I do mean that something would have failed inside the 3 pin Regulator
IC, and the device might be semi operable when load and temperature is
low, then the heat causes expansion and seperation of some element
inside, and it is then non operating. Yes, the latter It can go
from current condition to worse. It won't get better on its own. More
than one appear to be affected.
For whatever reason; a component type was selected for specific values
in its output range. In order to have more current available at a
particular voltage, two or more are placed in parallel on that power
circuit. The mobo mfgrs engineers determine how much load is going to
be placed on the bus they're providing power for, and they decide how
many of which per voltage to use. This is 'common' garden variety
stuff, most mobos are made this way. Gigabyte, ABit, Asus, Soyo, Intel
etc. do this. The component is actually the same common one you can buy
at Radio Shack for about a buck Their problem is providing a
somewhat variable amount of current on demand while maintaining voltage
level.
Some of the buses simply cannot be loaded beyond specification. Many
newer AGP cards DO load the AGP bus beyond what they consider the
industrial standard. Gigabyte has had a number of returns from
'blowout' on first startup with newer AGP cards lately. Then, as with
what happened to mine, the blowout occured over time in which the bus is
loaded at full spec. This was about 10 months for mine. The IC's that
died simply cooked while being loaded at (I'm guessing) 105% rated load.
I still have that mobo, there is no doubt the 5v +/- regulator IC's
burned. If there are 3 on the circuit, and all 3 work fine, all is
well. If one flakes out, that increases the load on the other 2, and
they don't last long either.
If your are a GOOD electronics tech, you can isolate and replace the bad
regulators. Not too many people I know are that good \
Here are the results of my PSU swapping and my intermediate
conclusions...
When I placed a known, working PSU (350 W) in my PC and used
Motherboard Monitor to measure the voltages, they were all incorrect
just as they were for my QTEC. This means that I have a Motherboard
problem. (For this test I removed some components from my PC so it
demanded a similar load to the PSU's normal PC, i.e. GF4, one HD, one
CD-RW, and no case fans.)
I moved my QTEC PSU to a friend's PC (again) but this time I also
moved the 9800 Pro as well. The voltages obtained were stable - the
same as the last time I moved it (with +12 V measured at +11.6 V). I
played a game without problems for over 2 hours (Jedi Academy for what
it's worth). However, due a limitation of the Motherboard I could use
only AGP 4x, and I haven't had that crash on my system (although I've
tried it only once).
So far then, I reckon my PSU is fine. Due to the large discrepancies
in the times before my PC freezes, I reckon that overheating is not
the problem either. This leaves the 9800 Pro and the Motherboard.
Given that the games work for a while, i.e. the 9800 Pro is
functioning, and the info given by McGrandpa, I reckon my Motherboard
is the problem. However, if it is the Motherboard, then why didn't I
see problems with my GF4? I suppose my Motherboard could have started
having problems after I installed the 9800 Pro, and I haven't used the
GF4 since then. Or maybe the 9800 Pro is just more sensitive to
voltage fluctuations than the GF4? Maybe I should try the GF4 again?
Good thinking and good testing! Look up the detailed specs on both
video cards and your motherboard. You're looking for the 'engineering
requirements', of voltages/current and variables for the AGP slot.
Your DX9 9800Pro will draw quite a bit more power while IN 3D modes than
the GF4. Which GF4 is it, btw?
My New 9800 Pro 128 seems to draw a lot LESS power than my 1 year old
FX5900 does. It's a smaller card, smaller heat sink and fan, and it
doesn't get hot, even when playing Far Cry with everything maxed out.
This is a happy surprise for me
Thanks to everyone who has replied to my post - your help is much
appreciated.
I will report my progress here, and any other useful stuff I find out
along the way.
Cheers.
I just hope you can get it all sorted out without a lot of undue expense
and time spent without a stable system.
I've found that motherboard problems are often hard to troubleshoot, and
isolating whether it's video card or mobo or CPU or PSU can really be a
mess without other systems to try proving a part good/bad. I had to
rely on a local vendor for that, and he was very accomodating. So I of
course bought my new parts from him. And he blew a mobo the very first
time he tested my FX5900 in a new one. He tried my vid card and
processors in 3 diff mobos and combinations. Both card and cpu fine.
PSU fine. Ram was PC800 Rambus and I ended up with PC400 DDR non dual
channel. This ram tests to be slightly less bandwidth than the old
rambus. This ram will go faster when 800fsb is used
Another upgrade comin this summer then!!!
McG.