No Post with more than one stick of RAM

[jbearlake]

Gaming PC, Elite Computer Group
MB - NFORCE6M-A
CPU - AMD Quad Core 9750
RAM - 8G DDR2 800
HD - 500GB Seagate
GPU - GeForce 8600GT 512MB
PSU - Logisys ATX 20/24 serial ATA 480W (3.3V, 5V, 12V, -5V, -12V)




Symptoms and step to date.

Rig shut down without warning, no power.
Restart without successful post.
Disconnected ATX12V, power stays on, but no post.
Cleared BIOS, nothing.
Stripped to the bones, MB, PSU, (1) 2GB RAM of (4) 2GB RAM (all identical) installed, cleared BIOS- successful post!
Tested each RAM and slot, one at a time - all good. One RAM in any slot will post. Addition of any other RAM leads to no post, even with clearing BIOS, no matter what combination of RAM / slot is used.

This is on an ESD mat with only the following right now:
MB, PSU, GPU, HD, (1) 2GB RAM, CPU fan, USB mouse & keyboard, LCD monitor.

I added everything else to make sure than my power supply wasn't failing. I thought it may be fluctuating, getting ready to fail, causing RAM voltage problems. So I added everything thinking that if it wouldn't post if that was the case, given the increase in power demand. Not the case, so I've ruled out the PSU.

Processor must be good, since I'm using it to post this right now! No glitches, other than running slowly on only 2 GB RAM, rather than 8GB (I was only using 3.5GB anyways with 32 bit Vista Ultimate installed, I intend to upgrade.)

Is there something I can do to check if the motherboard is going? I don't have another for bench testing.

I don't think it's software, it's been running fine in this configuration for over a year, and always updated, cleaned, lots of ventilation with low operating temps and a high temp automatic shutoff that has never been needed. Plus a UPS. There was a squealing noise that happened on power up over a month ago. Made me worry because it happened two power-ups lasting for about 3 seconds each time, but I cleaned it out thoroughly and it didn't happen again. (Noise was from PSU, I think)

Any ideas people?

 

[raybay]

Are both memory modules identical?
It should have been able to use 4 GB or RAM on the VISTA setup.
Does this make sense: "..Disconnected ATX12V, power stays on, but no post..."
Did you test the memory with MemTest86? If not, how did you determine they were both good?
The symptoms your describe are typical of a failed power supply... and nothing you have reported rules that out... the only test that works is with a different power supply, or a high quality power supply tester
The squealing noise is likely significant. You need to track it down, in order to rule it out.
You have either a filed power supply, or CPU Fan, or cooling fans,
You likely have one of your power supply rails failing.
CPU cooling may have too much thermal paste between the CPU and the cooling device.

 

[jbearlake]

All memory modules are identical, no tweaking performed.

I have 32 bit, so I can only use part of the memory anyways, but that isn't the problem, it's hardware.

The computer would stay on with the ATX12V line disconnected, but wouldn't post. It will only post when all RAM but one is removed

Each RAM will lead to post, no matter what slot it's in, so it's unlikely that it's the RAM or sockets.
I'm a 2M tech, so I tend to test hardware first before software testing.

I believe that the squealing was coming from the power supply (it had fine hairs wrapped around the spindle, which I think caused the noise....just an educated guess though, from experience with other systems (non-computer). It may have also been a failed front case fan, or possibly even the HD, since the noise came from the location they are both in. I haven't replaced the failed fan yet, but that's inconsequential. Cleaning the PSU seemed to do the trick though, with no problems till now.

I agree about the power supply rail though, since I've had to replace the power supply before. I replaced it with the exact same model (naively) before. I just find it odd that I'm using the computer right now to type this with just the one RAM and additional modules cause failure.

If it were a failing rail, how can I test? The ATX12V supplies steady 12v, no fluctuation over 3%, I don't think that one's bad. If one of those two rails is the problem though, I'm concerned...wouldn't it be an indication of a failing MB if that rail is steady until placed under load? Possibly a cold solder point within the board?

I'm capable of fixing this board or supply under a microscope if I have to, but a replacement for either is cheap enough to forget that aggravation. I'm just trying to narrow it down, or determine is there is some interruption of control of the RAM.

 

[jbearlake]

By the way, Memtest clears the RAM, 100% for all four.

 

[raybay]

We see this one differently.
There are a lot of power supplies that shut down if one rail is disconnected or disabled.

It is important to test your power supply under load, perhaps with a quality power supply tester.

We certainly had different points of view on this one.

One way to test would be with a different power supply, but if you use the same assumptions, you may get the same result when many power supplies are perfectly good.
It is under load that the fluctuations occur in a problematic power supply... Each rails must be tested separately with both connected or under load.

 

[jbearlake]

By the way, Memtest clears the RAM, 100% for all four.

 

[jbearlake]

Sorry, there's a little time lag on my posts....I was editting the first one when you replied, and writing the second when you made your second reply.....opps!

Anyways, I tested both ATX12V rails under load, both show 12.26V, with the greatest deviation being a < 1 second drop to 12.25V, well within tolerance. Any other ideas where to look next?

Sorry to have sounded curt, I was just jumping from one task to another, plus posting in another forum. I have no problem with different view-points at all. I'm use to fixing circuitry, mainly for radars and UHF radios, but I only deal with a few computers, most of which have fairly simple solutions. This one's just not as clear cut, yet.