E wrote:
>
> Hello
>
> I have PC where an AGP video card stopped sending a signal to a flat
> screen monitor in the middle of playing the Sims 2. The video card and
> flat screen monitor where linked via a standard VGA cable. The user says
> that the screen froze, or went black in the middle of a game. This
> happened only an hour after installing the game. The machine now has no
> display, during POST, or booted into full graphics mode. Before I
> troubleshooted anything, I knew it was properly booting into Windows
> because I could hear it through the speakers.
>
> This is an Asus P4P800SE motherboard, with an e-GeForce 7600GS, 256MB
> DDR2 AGP video card.
>
> To troubleshoot, I connected the PC to a known working CRT monitor, and
> also changed out the AGP video card. No signal in either attempt. I did
> however get a PCI S3 video card to work, and then checked out a few
> things in Windows. The machine did not 'crash' or blue screen per say,
> so it did not leave a minidump file. There is nothing wrong in device
> manager. Nothing of relevance in the system logs. It is interesting to
> note that if an AGP card, and the old PCI S3 video card are installed in
> the system together, there will still be no signal from the PCI S3. It
> must be by itself.
>
> My current thinking is that the AGP slot has failed. I didn't see any
> obvious blown capacitors around the AGP slot. I may try a third AGP card
> in the slot before I go to something else. If this is the case, I will
> have to change out the mainboard, maybe even consider building another
> system from scratch. The user is tight on funds now, so I may have to
> find a good working replacement mainboard.

I've seen the exact same symptoms on a system that had a wonky power
supply. Apparently the supply was not providing enough to properly
drive the AGP slot.

E wrote:
> Grinder wrote:
>> E wrote:
>>>
>
> snip
>
>>>
>>> My current thinking is that the AGP slot has failed. I didn't see any
>>> obvious blown capacitors around the AGP slot. I may try a third AGP
>>> card in the slot before I go to something else. If this is the case,
>>> I will have to change out the mainboard, maybe even consider building
>>> another system from scratch. The user is tight on funds now, so I may
>>> have to find a good working replacement mainboard.
>>
>> I've seen the exact same symptoms on a system that had a wonky power
>> supply. Apparently the supply was not providing enough to properly
>> drive the AGP slot.
>
> Thanks for the reply
>
> I haven't ruled this out, but at the end of 2007 I installed a new power
> supply in this machine. Its an Antec Basiq 500W.
>
> The nVidia card has its own female molex connector. I've switched to
> another line from the PSU to feed power to the AGP card and still got no
> signal to the monitor. The older Radeon 9600 Pro I used to troubleshoot
> will not send a signal either. The Radeon has no separate power input.
>
> Is it possible for an increase in power draw from playing The Sims2, to
> burn something out in the PSU?

There are AGP I/O power pins, in the AGP slot. On my motherboard, from
the same era as that board, a number of power requirements are met by
linear regulation of the 3.3V rail. With an ammeter, I get 3.3V at
14 amps, drawn by the motherboard while running something like 3DMark.
Part of the power is drawn by memory and Northbridge, but some of that
could go to power for AGP I/O signal drivers.

There should be something, near to the AGP slot, that takes in 3.3V
and puts out 1.5V for the AGP slot. 1.5V drives the AGP interface
power requirements, whether the actual signal level is 1.5V or 0.8V
amplitude.

Paul

E wrote:
> kony wrote:
>> On Thu, 29 May 2008 21:11:55 -0400, E
>> <(E-Mail Removed)> wrote:
>>
>>
>>> Maybe I should just get a pinout of the AGP slot and try to test
>>> voltage in there (if possible without shorting contacts together), or
>>> like you say, look for a voltage regulator that feeds the AGP slot
>>> and see if I can test it's output.
>>
>> Might not be necessary to probe the pins, locate the
>> regulator and take a voltage reading there, inspect the pins
>> for uniformity and no burn marks using a strong light, and
>> try the 3rd AGP card... I would also try clearing CMOS with
>> AC power off inbetween each card attempt.
>
> Are these voltage regulators?...
> http://members.localnet.com/~eddie18.../100_2300w.jpg
>
> I metered the one right under the AGP slot, without an AGP card
> installed, and got anywhere from 1.89V to 1.72V between the two outside
> pins. It read different every time I rebooted the machine.
>
> I pulled the old passively cooled Radeon VE out of my box and installed
> it in the problem PC, cleared the CMOS and still got no signal. I also
> disconnected the hard drive to reduce power consumption and got no signal.
>
>>
>> Even though the PSU seems of ample capacity and branded
>> reasonable quality, sometimes things just break anyway so it
>> would be good to also measure voltages from PSU and if all
>> else fails try another PSU. The old PCI card was certainly
>> using less power than either AGP card.
>
> I checked voltages on different leads coming from the PSU, and measured
> slightly over all the specified voltages, ~12.11V, ~5.03V, and ~3.3V. I
> have not yet tried another PSU, although I have few of questionable
> integrity laying around. A 500W, 400W, and 350W.
>
> I'm hoping to find a socket 478 ATX form factor board that will take a
> P4 2.8Ghz Northwood 800Mhz FSB. It would be nice to have a PCI-E board
> with the above specs and onboard video good enough to play the Sims2. I
> haven't really kept up with hardware over the last couple years, so I'll
> need to learn what my options are.
>
> Thanks
> Eddie
>

It took me a while, to find web.archive.org in a working state.

This picture shows a Vdimm mod. The LM358 8 pin DIP next to the
two MOSFETS, is an op amp used to implement a linear regulator.
The author of this picture, shows a place to measure Vdimm.
So those things with the three legs, are MOSFETS and not three
terminal regulators. The opamp varies their conductivity, and
provides a fine and automated control, giving a smooth DC
voltage. (A separate chip somewhere, would be used to create
the termination voltage for the memory bus.)

http://web.archive.org/web/200408220...s/p4pvdimm.jpg

There is another Vdimm ("Vmem") picture here, giving an idea where
to look for a place to measure.

http://www.xtremesystems.org/forums/...1&d=1082886312

The AGP slot could very well be using a similar approach. For
example, imagine you took that 2.5V level from the Vdimm circuit,
and dropped it further to 1.5V for the AGP slot. That could be
done with a MOSFET as well. There might have been a hint in
this picture, but I cannot find this picture now. (It is not in
the archive.) This picture would have been for a P4C800 board, but
could use the same ideas as the P4P800 family.

http://malvescorner.com/images/p4cvagp.jpg

I wonder what the AGP voltage value looks like, when you're using
a PCI video card, and there is nothing in the AGP slot. That
should give you an unloaded readout of the voltage.

Paul

kony wrote:

>
> I would guess one of the two on the left of the picture, but
> I could be wrong. Paul has the right idea about process of
> elimination since we also question whether the voltage is
> even right from the regulator, though I'm doubting his
> speculation that the 1.5V might be derived from the 2.5V,
> think it more likely taken from a higher value rail voltage.
>

On my motherboard, I have precious little 5V consumption, and
the 3.3V measures 14 amps, when using a dual channel memory
configuration. So as near as I can figure, with only switching
conversion for Vcore, and linear regulation for everything else,
the power starts at 3.3V.

This Intel schematic, gives an example of a linear regulator
to make 1.5V. It starts at 3.3V, makes intermediate voltage 2.2V,
and uses a second stage to regulate from there to 1.5V for what
they call the core voltage rail. Asus won't do it exactly that
way, because Asus has adjustable AGP voltage, and so the core
rail and AGP can't come from the same regulator. So PDF page 88
would be an example of the technique. The intermediate voltage
could be done for two reasons - to share power dissipation
across two MOSFETs, and to provide extra filtering by doing
two stages of regulation (since the linear would have poor
high frequency noise rejection).

http://download.intel.com/design/chi...s/25281202.pdf

It wouldn't make too much different to the circuit on PDF page
88, to make the intermediate voltage equal to 2.5V.

Paul

E wrote:
> kony wrote:
>
>>
>> You could always call, but if you were at their own website
>> instead of still at pricewatch and they didn't have have
>> accurate stock status, it might be a bad omen about their
>> proficiency as a seller.
>>
>>
>
> If I understand correctly, Pricewatch doesn't do any selling of hardware
> themselves. They just refer retailers, similar to Google Shopping. I
> agree about the possible lack of proficiency of a random retailer.
>
> I'm leaning back towards a board with an AGP slot, even if its mATX. My
> user doesn't need many PCI slots.
>
> Here is an Intel made i865G chipset board referred by pricewatch.com ...
> http://www.pcpartsohio.com/BookDetail.aspx?item_id=852
> This isn't the same board or site I wrote of in a previous message. This
> one seems to be in stock.
>
> Ever deal with this seller? Its an OEM board, but they aren't listing it
> as a system pull, so I'm assuming it is unused. As you say, I could
> call. I like the fact that it is Intel made.
>
> Like you I am a little wary of dealing with random retailers through
> Pricewatch, but the price is right and I wouldn't think I would have to
> reinstall the OS, since this board is about the same chipset.

Resellerratings.com can give you some info on these companies.
PCPartsohio maps to this.

http://www.resellerratings.com/store/PC_Parts_Unlimited

In looking through Pricewatch a while back, with the same
objective in mind (AGP8X/S478/DDR motherboard), virtually
everything I looked at, was listed as a "pull".

Also, I note the picture on the PCPartsohio site, is linked
to the Gateway site. So the board could have been inside
a Gateway computer, or used as a replacement for a Gateway
motherboard. More info here.

http://support.gateway.com/s/MOTHERB...521363nv.shtml

And if the purchase is "motherboard only", then you won't get
the I/O plate to snap into the back of the computer. Not a
big deal.

Paul

E wrote:

>
> Yes, they have a link to Gateway for drivers for this board as well.
>
> The ATX power pinout on a a Gateway board is standard, correct? Not with
> wires crossed like Dell?
>

I don't know a way to guarantee that. It looks like a pretty standard
looking microATX board. Perhaps the vendor could answer that question.
It doesn't look goofy, like some of the Dells that had extra power
input connectors.

Paul