Boot up troubles, power related - system always stable once booted

Question

I've got a system that's been causing me a lot of trouble over the last few months. Below is a somewhat lengthy saga of my experience with it but it should hopefully provide some thorough problem history. I've provided a TL;DR at the end to provide some summary.

A while ago, the machine started having trouble booting up. I would hit the power switch in the morning, and come back a few minutes later to a blank screen (no input to the monitor). Rebooting didn't help, so I tried plugging directly into the VGA output of the motherboard instead of the DVI output of the GPU. While I didn't get any output from the motherboard, I eventually found that plugging something in to the VGA out would cause the system to boot up.

For a while, I thus ran the system by having a dummy vga cable (that led nowhere) plugged into the motherboard and the actual monitor input plugged into the GPU, which for some reason let the system boot up.

After a while, this dummy cable method stopped working, but I found that by fiddling around in the back a bit (plugging/unplugging the vga and dvi cables) I could eventually get the system to boot.

On several occasions I tried just "waiting it out" but this was never successful, the computer required some sort of interaction to boot and just letting it sit there for a long time did not work.

Another thing to note is that once the system gets past the boot, it is perfectly fine. Soft-resets (doing a "restart" through Windows) also seem to always work, but I haven't performed enough of them to say it always works.

Eventually, I decided that the machine wasn't worth the effort and ceased trying to boot it. Recently, however, I had a bit more time on my hands and wanted to see if I could diagnose and fix the problem.

I started by examining the computer during the "won't boot" phase, after I had pressed the power button but with still no input to the monitor. All the case fans/lights spin up to full blast, and when I get close to it a notice a distinct pulsing sound. The indicator light in the ethernet port also flashes in sync with this pulse. I tried to find out what the source of the pulse was (I suspected the hard drive) so I opened up the case and put my hand on the chassis near the hard drive to see if I could feel the pulse directly there.

Instead, the system suddenly booted up. I repeated this experiment a couple of times and it seemed to me like my direct contact with metal in the case was allowing the system to boot. I suspected that there was some sort of short which I was solving by grounding the machine.

For a while this technique seemed to work but it too decreased in effectiveness over time.

At this point, I decided to finally do some hardware swapping to see if I could isolate the problem. Based on the shorting issue, I figured it was probably related to the power supply. I wasn't too happy with this as it was a fairly new and costly PSU (Silverstone Strider Plus ST50F-P 500W). However, I swapped the PSU out entirely for a slightly older one I had on hand and the issue shortly returned.

I was also a little bit suspicious of the front panel since the issue seemed to be boot related, and the case is by far the cheapest component of the system. I tried unplugging all the front panel plugs and just using a spare power switch from an older computer, but the issue still persisted.

The issue seemed to be very graphic related so I then assumed it might be graphics card related. I tried two other graphics cards, though (one with auxiliary PCIe power and an older one with auxiliary molex) and with both cards the problem still persisted (I'm using the older card now because I'm less concerned if it gets fried).

I've also tried booting without the hard drives and other peripherals but to no avail. So far, I think the only time I had a degree of success was with the GPU removed entirely, but I need to test that a bit further.

More recently, on a couple of occasions the machine has simply shut down a few seconds after boot. After this occurs, the green light on the motherboard remains lit but I have to unplug the PSU entirely and wait for the light to turn off before plugging it in again or the machine will not respond at all to pressing the power switch. After a couple of tries with this, and some plugging/unplugging of the rear VGA and DVI cables, I am able to get the machine to boot.

I also tried resetting the CMOS and flashing the BIOS to the latest version - neither of which helped.

So at this point, I feel like I've narrowed down the problem to either the CPU, the Motherboard, or the RAM. I have yet to try booting with some of the RAM removed because the issue seems so separate, but to be thorough this test is next on my list (after more trying without a GPU). There are 4 sticks of RAM in the system, DDR2, 2x2GB and 2x1GB (all Corsair).

Here is a detailed spec breakdown:

Potential Problem Parts:

ASUS M4A785-M (AM2)

AMD Athlon II X4 635

6.00 GB Dual-Channel DDR2 @ 266MHz

GPUs Tried:

XFX GeForce 250 GTS

Radeon x1950 XT

XFX GeForce 680 XT

Power Supplies Tried:

Silvestone Strider Plus ST50F-P 500W

Ultra ULT-500P 500W

Peripherals: 195GB Western Digital WDC WD2000JS (SATA)

So, does anyone have any thoughts on this issue? I'd like to be able to narrow it down to the Motherboard so I can seek some sort of warranty repair or replacement (the board is only about 1.5 years old).

Does it seem likely/worthwhile to check each of the RAM sticks? Is there anything I should be looking out for, or anything that I should try? Does the issue sound like its related to any specific part?

TL;DR: Computer boot up issues, seems to be related to plugging a vga cable into the motherboard and some sort of shorting problem, but is power-supply and gpu independent.

UPDATE: I swapped the computer to another case to make sure there wasn't any shorting issue with the original case, and so I could get a better look at the motherboard. I didn't see any anomalies during the switch, but as soon as the system was in the other case I found myself unable to boot it at all. I tried swapping out all of the RAM sticks and tried booting sans GPU but I kept getting the same problem. I figured that the motherboard was completely gone and decided to move the CPU to another (older) machine.

However, with the CPU in the older machine, I started to get the exact same problem. I moved the older CPU into what I thought was the problem motherboard and now the newer system is working just fine. I did several boots and I think I can say with certainty that the issue is definitely the CPU.

So I guess the CPU is causing the shorting issue. Is there any chance of salvaging it? During the transfer I noticed that the thermal paste was pretty hardened and I don't have too much confidence in the heat sink - could this possibly be an over-heating issue? I never had any trouble with the system once it was booted, so I'm a bit perplexed by this outcome.

Answer 1

I thus ran the system by having a dummy vga cable (that led nowhere)

This is important...

this dummy cable method stopped working, but I found that by fiddling around in the back a bit

... the plot thickens...

I suspected that there was some sort of short which I was solving by grounding the machine.

... on the right track...

In my opinion, everything you've said points to the motherboard... and likely physical damage like a loose solder contact. Moving external cables around (especially the onboard VGA), providing a ground contact within the casing, all would point to this. It could also be damaged capacitors (solid, or liquid, even a single capacitor), but I put a low priority on that.

You eliminated the power supply by putting another in the unit... I'd still test them both with a tester because it is possible that a bad mobo damaged the PSUs... not trying to scare you, but it also sounded like a bad PSU until you swapped that out... until the next one started doing it too.

If you are planning on RMAing the board, I'd grab a magnifying glass, take it out of the case, and examine it everywhere, starting with the back. Start at the VGA connector and examine all the solder contacts and start moving outwards. Check the back, check the front... and check both your PSUs in another computer to make sure they are fine.

EDIT I was thinking last night... it's possible that this also could be caused by motherboard standoffs that are too short/too long. I'm not saying it's probable... just possible. Having standoffs that are the wrong height for your case (and believe me, it happens) could cause cards to not seat properly in their slots, which could cause intermittent connection problems, which could cause any minor adjustment to the motherboard (IE moving or even just attaching a cable to an integrated VGA port) to make that connection. Again... it's a stretch.

EDIT to EDITS Wow. I would have put the CPU as the issue rather low on the probability scale, since when the system would work, it worked fine. I guess it is possible that an improperly fastened heat sink could have been putting it's weight on the processor, and pulling it away from the socket just enough to disrupt contact... where jiggling wires/creating vibrations from plugging cables in was enough to reconnect contact. You said the thermal paste was quite old... did it appear to be even? I remember an eMachines that had it's heat sink put on at a slight angle, and when we got it into the shop and pulled the heat sink off, you could clearly see that the heat sink hadn't been flush on the processor by the impression in the paste (and the line of dust on the surface of the processor at the edge of the thermal paste UNDER the heat sink).

It is definitely possible there is a problem with the processor... likely a physical issue like a damaged pin, since you could make the processor work like it was supposed to with physical solutions. Examine the heat sink more closely and how it mounts into place. Examine the brackets that hold the heat sink in place. Try a different one.

Answer 2

If switching the PSU doesn't do a thing (it usually does, it can be performing poorly and giving you trouble when a peak is demanded in the startup) it could be a short, most certainly in the motherboard. Also some SMDs can become quite sensitive to static or even stop working properly outside a narrower range of temperatures (e.g. until they're heated). Try sticking some tape where you suspect it may be. Seems silly and it's definitely a poor solution but I learned it from some Korean guys who fixed my laptop when I send it back there on warranty (those bastards didn't put a new one) and it has worked for me several times since. Also look for any defective cable in your setup (data too), specially defective power cords (I've had too many of those) and try plugging in a different socket and power strip.