DX4300 desktop randomly blanks out

[caeadas]

I have a Gateway DX4300-15e that has recently started randomly blanking out. Not sure how to describe it, the screen disconnects and all the external devices lose power, the hard drive will continue reading for a few seconds but the HDD light will not blink. The power button stays lit, but is unresponsive to either pressing or holding. With music playing, the sound dies connected to either front or rear jacks. The fans continue to run. I have to completely unplug the system to get it to shut off. System logs don't show anything that could be causing it.

I've tried system restoring it back to the earliest save i had, and then disabling auto updates but that didn't help. I tried running it in safe mode, also didn't help. I tried disconnecting all the external devices, still cuts off. I've checked the voltages on the PSU with a voltmeter and all are within spec. I pulled all the internal pci cards out of the case, disconnected the DVD drive, and moved the monitor from my graphics card ports to the motherboards and removed the card, still cuts off. I've ran memtest and it shows no memory failures.

I'm not sure what to do next, I'm afraid if i try to reinstall windows it may lock up midway and screw up everything. I don't know how to check anything else. Any advice would be appreciated.

 

[C12Friedman]

Welcome to Tom's Hardware Forums,
Sounds like you've been rather thorough in your troubleshooting so far.

Going to suggest looking at heat as a possible suspect here. Are you currently monitoring your temps? If not, a program like coretemp, speedfan, CPUID, or... should be used to monitor your CPU temperature.
Heat can also affect PSU's, make sure the PSU fan works and blow it out good with compressed air.

A good cleaning of the inside of the whole computer may help. And also the reapplication of thermal compound and reseating the HSF can be helpful if temps are high.
If while monitoring your temps, you find they are not bad, I'd be suspecting a serious hardware failure (mobo or CPU - possibly PSU still) - we will hope that is not the case.

 

[caeadas]

I've checked the temps, they are all within spec, and the computer gets cleaned regularly. It usually dies when i'm doing something on it, but I don't do anything strenuous. Mostly internet and music stuff. It has died within minutes of turning it on, but it will also stay on for hours running idle. (I disabled the screen power saver stuff so the screensaver runs constantly) I've also got a first-gen Asus g73 laptop that I game on, so i'm all too aware of heat. I usually go after both comps about once a month with compressed air.

 

[C12Friedman]

Although at this point you seemed to have ruled out the PSU, I am still highly suspect of it (only because of symptoms). I am thinking fluctuations on the 5V rail or... you can inspect you mobo for hairline cracks or possibly leaking capacitors.
If you have access to another PSU, I will suggest swapping it out - don't go buy one just yet though... gonna consult with a friend, he may prove helpful here...

 

[caeadas]

Yeah, I didn't check the PSU for any long term stuff, the voltages i was getting out of HWmonitor were way off spec, so I checked them with a volt meter and they were fine. Knowing the voltages aren't calibrated right, could i still trust HWmonitor to watch for fluctuations? Or what would be the best way to watch it long-term?

BTW, thanks a ton for your help.

 

[C12Friedman]

I'm back after "consulting" with a few friends only to decide nothing is specifically isolated at this point (one did insist it is the PSU - we'll ignore him for now). I'm going to direct you to Gateway's response to random shutdowns
http://support.gateway.com/s/Checklists/Q42006/ck2006110919.shtml
I found this article which I think may be too brief
http://www.computerhope.com/issues/ch000689.htm
but, I also found this article to be somewhat enlightening perhaps something here will help
http://www.techtnt.com/pc-turns-off-randomly.html
Happy reading

 

[caeadas]

So, more testing done...

I connected the tower to a different surge protector, on a different plug, and connected it to a different monitor, keyboard and mouse. still died. That's basically every single peripheral device and external influence eliminated. (other than my house current...)

Rechecked the PSU voltages again, this time making sure to watch them while the comp died. I found that the easiest way to kill it is to start programs up.

3.3v rail: 3.39-3.41v. Flickered twice down into the 2 volt range but thats when I was holding the probe on the connector, so I can't be sure it wasn't me moving it. Once i got the probe to stay on its own it didn't flicker again. Stayed steady when comp died.

5v rail: 5.03-5.06v. Didn't flicker, stayed steady when died.

12v rail: 11.87-12.02v. It seems like the 12v rail is a lot more bouncy than the other two. At startup, it was around 11.95 to 12.00, and would jump several hundredths each tick of the voltmeter within about .10v of average, where the other two would only shift a single hundredth, and not shift every tick. It eventually started moving down into the 11.90-11.95 range, but was still very bouncy. I started running everything I had on the desktop to load it the best I could, once I started minecraft the bounciness stopped, it went steady at 11.95 and it refused to die.

I also noticed while playing minecraft with HWmonitor running, the fan speeds started acting odd. the first reading was normal, couple thousand rpms constantly. the second one would randomly go to 0 rpm (even though i could clearly see them both with the case open and neither slowed down). Then, a third fan reading magically appeared, with rpms ranging from 0 to 600000 (yeah, six hundred thousand RPM.) Theres only two fans in the case, not counting the PSU.

Watching the temps, with a youtube video running at 720p and minecraft running it never went over 55 degrees Celsius on any of the sensors. Running my screensaver (Electric Sheep) its the same, all four cores at 40 degrees, and all the other sensors less than 52 degrees. I really don't think its a heat issue, since the computer sometimes dies within moments of booting up, and then sometimes stays running for hours if i don't mess with it.

I'm going to keep an eye out for an inexpensive PSU this weekend and see if that helps. On a side note, I've been sitting here now for an hour on screensaver, and the damn thing still hasn't cut off. I think it's taunting me....

 

[C12Friedman]

I think your computer is taunting me also.
Wondering here, have you had to reset system date/time at all?

 

[caeadas]

Not yet, I haven't noticed it being wrong. Before this, the computer has been flawless since i bought it.

 

[caeadas]

Well i think we have solved it. Found a Thermaltake 430watt PSU from best buy for $55 and tossed it in the tower, and it has been running for several hours of normal use. Been web surfing, playing games, and 0 lockups. Only issues i have found are the 2 sata power connectors are on the same line, but unable to reach from the DVD drive to the HDD at the same time, The old PSU has inline sata power splice connectors that connect to the molex wires, I was thinking about reusing one for the HDD power, the clearance is not gonna allow me to put the new PSU's power connector in the HDD and close the case up anyway.

There was also a wire bundle inside the PSU that was hitting the fan, pinched between the fan housing and a heat sink. I pushed it out of the way best as i could from the outside, but its still kinda touching the heatsink, which I don't care for.

The only other thing I don't like is going from a 300w to a 430w supply. Granted, the new one is probably higher quality than the old one, but I know the wattage should be close to what the computer needs. Will it be that big of a deal?

 

[C12Friedman]

Actually, I was going to advise on your PSU but...
As far as PSU wattage, more is actually better (within reason) since PSU's run their best at about 40-75% load (my personal numbers - close enough) as that is where they are the most stable in voltage and amperage output.
The wire bundle you are talking about in the PSU fan area - are you talking new or old PSU?

 

[caeadas]

It looked like some wires inside the new PSU covered in heatshrink that had been kinked, but after having a closer look at it, it appears to be a bead of black silicone they put between the fan and heatsink to keep them separate. It was kinda hitting the fan and making a ticking sound but once i pushed it away, it was fine. I used the sata inline splice connector to tap into one of the unused molex harnesses and checked continuity with my voltmeter to make sure there was no resistance and had a good connection. Stuffed all the extra harnesses into a drive bay, zip-tied everything in place, installed all the cards, gave the whole thing a good dusting just for the hell of it, and closed the damn thing back up.

I am typing this reply on the fixed computer (which is like a half-hour long test, since I type horribly slow). Everything seems to be in order. Thanks a ton for your help. I've google-searched my through plenty of fixes, but this one really stumped me.

 

[caeadas]

Best answer selected by caeadas.