Film-Tech Forum ARCHIVE: Help with new PC switching itself off

This is topic Help with new PC switching itself off in forum Film-Yak at Film-Tech Forum ARCHIVE.

To visit this topic, use this URL:
https://ft-forum.com/ft/cgi-bin/ubb/ultimatebb.cgi?ubb=get_topic;f=8;t=004514

Posted by Andy Muirhead (Member # 712) on 02-22-2006, 06:55 PM:

Hey all, need some feedback from the PC savvy crowd!

Just bought a new base unit which arrived monday, and all was cool until today when it just powered off completely and restarted. It got progressively worse until it would only be on for a few minutes before it powered off.

basic system spec is;

AMD Athlon 64 3500+ processor
2x Aries 512Mb PC3200 DDR 400 DIMM RAM
Maxtor 200GB 8Mb cache 7200RPM SATA HDD
K8 RS482-M Skt939 DDR PCI-E 16x 128Mb ATi R (motherboard)
ATi Radeon X550 256Mb PCIE graphics
Generic PSU (300W)

The system worked well for the first two days, although it powered down once yesterday, then today it just got worse.

I booted into the bios and found the processor was running at 56 degrees celsius at idle, which seems pretty high. However shutdown on high temperature in the bios is disabled. I tried removing a case side and pointing a fan at the MB which got the system and processor temp down a little, however it still shut down.

Anyone have any ideas whats up??

I don't think it's anything i have installed, ie software or virus issue as it occurs when the machine boots into the bios and is sitting idle. Also it did happen yesterday when i hadn't loaded much software on.

I think it may be hardware related, but if this is the case, why did it take perhaps 10 hours of runtime to occur? I'm also pretty sure these machines are run for 24 hours pre test before being dispatched. By the way, this is custom built Aries PC from Watford Electronics in the UK.

I'm also thinking perhaps the PSU is not up to the job (300W), but the BIOS reports voltage levels within tolerances, and again why would it take so long for the system to fail so dramatically?

Any ideas?? I would try things like re-seating the memory cards, etc, but since it's a new PC and chances are I'm going to have to send it back, I wouldn't like to mess anything.

Thanks for any suggestions!!

Posted by Matthew Bailey (Member # 601) on 02-22-2006, 07:18 PM:

If the supply is not putting out clean DC,it can cause the PC to behave like you described. A supply that does not have a clean DC output should be replaced.

Posted by Scott Norwood (Member # 30) on 02-22-2006, 07:44 PM:

Bad power switch? (or loose connection)

Posted by Monte L Fullmer (Member # 2797) on 02-22-2006, 07:44 PM:

AMD 64's should be powered with a minimum of a 500+w PS. With that much horsepower on the momboard, that AMD 64 is choking and draining the power from that 300w PS.

Plug in a 500w + PS for that BIG processor's need. This is why the temperature is quite high on that processor: not enough "oomph" from that 300w PS causing the resistance and temperature to climb and eventuall processor failure will result - let alone the north/southbridge chips as well can fail.

Also, that 300w PS has a safety inside when it's being demanded for too much power - opens up and easily can go "poof".

This also pertains to those STRONG switchers that a bunch of us have: Too much draw causes them to go as well and that is a big "poof" to handle. Computer PS's are actually "mini switchers" that is in our projectors - and need surge protection from the input power.

Quite suprised, if you homebuilt your system, that your supplier didn't recommend a correct PS for all that horsepower and accessories that is plugged on that momboard.

-Monte

Posted by Bobby Henderson (Member # 840) on 02-22-2006, 07:52 PM:

I agree with others about the power supply being too small.

Not only do you have a processor that demands a lot of power, but the real culprit could be the graphics board. Any ATI or nVidia board with 256MB of DDR SDRAM and a fast GPU is going to require hook up to at least a 450 watt power supply or better.

I'm surprised anyone even sells 300w power supplys for desktop machines anymore. The next desktop machine I buy will have at least double that amount of wattage.

Posted by Monte L Fullmer (Member # 2797) on 02-22-2006, 07:56 PM:

..or build you a "monster" tower with dbl Xeon processors, dual RAID capability, 6 SATA 500gb HDD'S, loaded to the guts with RAM "up the whoosegow" dbl gForce DVI video cards so you can do your 3-D work with...(lol)

-Monte

Posted by Anslem Rayburn (Member # 1267) on 02-22-2006, 08:01 PM:

I had this same issue when I used a cheap power supply that came with a case. I almost guarantee it's the power supply. Try a decent brand, like Antec, Thermaltake, or if you've got the deep pockets : PC Power & cooling or Zalman. I swapped in an Antec TruePower and everything worked perfectly.

I personally use Antec on all the computers I build now, and have never had a problem with any of them. I would use a 400 watt or higher rated supply. Just remember a 400 watt quality supply is far better than a 500 watt cheap-o supply.

Posted by Adam Wilbert (Member # 1184) on 02-22-2006, 08:12 PM:

something else you can check easily (and free!) would be the memory. Download memtest86 and burn the image to a cd. Boot to the cd, and let it run through the tests.

Posted by Andy Muirhead (Member # 712) on 02-22-2006, 08:14 PM:

Monte, Bobby, and others, thanks for that info. I was a bit surprised to find out it was only a 300W PSU in the machine.

Monte, your remarks are about the processor and resistance etc were very informative, I didn't think about that. I'm glad I posted here about this. It wasn't a home build system, just a pre build that they (the supplier) had. I perhaps stupidly thought the company knew their stuff with regards to the PSU.

Just out of interest, if it is the PSU not being rated high enough, would the system have been ok for so long then failed as it did?

Adam, thanks, may look at that.

Posted by Adam Wilbert (Member # 1184) on 02-22-2006, 08:20 PM:

i had a memory issue a few weeks ago that displayed the same random shutdown symptoms. After installing new memory, the computer worked fine for a few days, then started shutting down occasionally, then got progressively worse. Memtest boots itself right after the bios, it is not run from windows, and wouldn't use the videocard, or hard drives or anything else that would tax the psu, if that is indeed the problem. I would just hate for you to go out and buy a new power supply if that isn't the problem! [Big Grin]

(course replacing 300w isn't a bad idea anyway)

[edit - this is in response to something that andy edited out of his post. hahaha]

Posted by Bruce Hansen (Member # 281) on 02-22-2006, 08:25 PM:

This sounds like a problem I had. The heat sink was glued to the processor, in addition to the normal "clip" (I guess so that it would not come off in shippment). There was a THICK coating of heat sink grease (there should be only a very thin coat, the grease is there to fill microscopic voids in the metal serfaces, it is not a very good conductor of heat. The more direct metal to metal contact, the better). Either the glue caused the processor and heat sink to seperate, or the thick coat of grease started to impead heat transfer. The computer would reboot now and then, usually when I was running programing the required the processor to work a lot. It started rebooting more often, untill it would not run more than about a minute before rebooting.

I figured it almost had to be something with the processor, so I tried to remove the heat sink, and found that I had to pry it off, due to the glue. I chipped the glue off, and cleaned off most of the grease. I then put it back together, and it has been working fine ever since.

Posted by Andy Muirhead (Member # 712) on 02-22-2006, 08:37 PM:

Adam, Sorry! I posted something then checked your link to see that it booted after the bios, you're too fast for me, haha! so thanks and may well be helpful.

Bruce, that's also interesting, and something that I have heard mentioned about this problem.

Posted by Monte L Fullmer (Member # 2797) on 02-22-2006, 11:04 PM:

Yea, use that "Arctic Silver" compound when mounting headsinks to these high-powered CPU's.

quote: Andy Muirhead
Just out of interest, if it is the PSU not being rated high enough, would the system have been ok for so long then failed as it did?

It might have, if there wasn't any high drawing video cards, all of the slots loaded with the RAM, added devices in the slots...et.al...

-Monte

Posted by Dustin Mitchell (Member # 372) on 02-23-2006, 12:43 AM:

Did you connect to the internet with this machine right away? Is it possible you have an un-updated version of Windows XP and you got hit by blasterworm? This sucker is still out there and if you are running an unupdated version of windows it doesn't take long to hit you.

Posted by Thomas King (Member # 2745) on 02-23-2006, 04:10 AM:

Just to add to the comments about power supplies:

Running with a cheap/underpowered supply is one of the worst things you can do to any system (not just computers!). I've measured cheap supplies that were providing voltages way over their (already wide) tolerances. Nothing compares, however, to what happened when someone I know upgraded their ancient PC to a sizzling new P4. They kept the unbranded 200W psu and case. Long story short; it overloaded and broke down, totally destroying every single attatched component. I use this inspirational tale when finance managers refuse to sign for decent kit on the grounds that "the cheaper stuff is just as good, right?".

Posted by Steve Scott (Member # 630) on 02-23-2006, 10:24 AM:

I just replaced a CPU fan on an Athlon 64 as well. Might wanna see if that's spinning inside too.

Posted by Brad Allen (Member # 470) on 02-23-2006, 03:13 PM:

Make sure the heatsink on the processor is installed correctly. I've had computers arrive with the clips that hold the heatsink on not clamped down correctly. Or broken off.

Posted by David Stambaugh (Member # 1102) on 02-23-2006, 04:24 PM:

Some of those cooling towers for AMD processors seem poorly designed as far as how they mechanically attach to the CPU and socket. I've seen 2 machines where the cooling tower came loose during shipping. One of them pulled the CPU partially out of the socket with it and bent the pins so badly it was ruined. I suspect the cooling tower was incorrectly installed during system assembly and the weight of that tall heavy tower really did a number on the CPU during shipping.

Posted by Andy Muirhead (Member # 712) on 02-23-2006, 05:56 PM:

Brief update

Couldn't get a hold of tech support today (surprise!), but tried the machine this morning and it ran for a half hour ok. I turned it off myself. Now back from work and have tried it again, and it's still running after 45 minutes. This is what EVEREST reports at the moment, after 30 minutes gaming with GTA, just for a test.

----

Temperatures
Motherboard 41 °C (106 °F)
CPU 55 °C (131 °F)
Aux 40 °C (104 °F)
Maxtor 6L200M0 40 °C (104 °F)

Cooling Fans
CPU 3245 RPM
Power Supply 2411 RPM

Voltage Values
CPU Core 1.42 V
+2.5 V 1.18 V
+3.3 V 2.58 V
+5 V 5.59 V
+12 V 12.22 V
+5 V Standby 5.13 V
VBAT Battery 3.22 V

---

Will now try to test the memory using Adams idea, if the system can stay up!

Posted by Bruce Hansen (Member # 281) on 02-23-2006, 06:55 PM:

quote:
+2.5 V 1.18 V
+3.3 V 2.58 V

IF you are saying that there is 1.18 V where there should be 2.5 V, and 2.58 V where there should be 3.3 V, that looks like there are power supply problems. In the case of the 3.3 V supply, 2.58 V would be about 22% low.

Posted by David Stambaugh (Member # 1102) on 02-23-2006, 06:56 PM:

quote:
+2.5 V 1.18 V
+3.3 V 2.58 V
+5 V 5.59 V

If each pair of numbers is the nominal Voltage and the measured [reported] Voltage, something isn't right.

Posted by Andy Muirhead (Member # 712) on 02-23-2006, 08:12 PM:

Yes, I agree, It just doesn't look right, however I'm no expert on PC voltages and tolerances.

System uptime as of now is almost 3 hours and no problems, this is very bizarre. Considering yesterday 2 minutes was about the most it would stay on.

Adam's suggestion for memtest was a success. No problems there.

I'm loath to send the unit back, as it would probably be months before i saw it again, if ever, a google search on the company revealed they are not the best at customer support.

Posted by Anslem Rayburn (Member # 1267) on 02-23-2006, 08:23 PM:

Link

According to this site, the Athlon 64 3500 draws 89 watts and has a max temp rating of 70°C. Heat would not appear to be an issue, but the rails you report for the power supply are terrible. I am confident it's a power supply problem, and would upgrade to a decent quality 400 watt supply. There is honestly no reason to spend a ton of money on a monster power supply when you don't need it. Just get a decent brand.

Posted by Monte L Fullmer (Member # 2797) on 02-24-2006, 02:50 AM:

..I would try a 500w PS and then do the reading, again - just to "condemn" each of the reading down the line.

See, if you can 'borrow' a PS to try out for the heck of it for a couple of days.

Kinda like what we do when processor have their days - start swapping out cards from one to another and it's amazing where one can find the culprit with this procedure.

-Monte

Posted by Andy Muirhead (Member # 712) on 02-24-2006, 07:37 PM:

Thanks for all your help. Since this was a brand new base unit, I'm not going to try swapping parts. The customer support from the seller (which is actually one of the biggest in the UK) is non existant. My emails have been ignored, and phoning is no better, having been kept on hold for 20 minutes each time before I give up. If I had been able to speak to a support technician I'm sure we could have figured out the problem. As it is, since I couldn't speak to anyone at all, I've decided to return the unit as faulty. Which is easier said than done. I will still have to contact the company to obtain a return code. How to do that when they don't answer phones or emails will be a challenge.

I am so pissed off at this companys' (lack of) customer support. If only I could have spoken to someone, I'm sure the problem could have been resolved.

Posted by David Stambaugh (Member # 1102) on 02-24-2006, 07:42 PM:

Can you buy a larger power supply from a store that would let you return it if it doesn't resolve the problem? You might spend perhaps $75 (US) on a good power supply. The cost of returning the system, with shipping & insurance, might approach the cost of just buying a power supply.

Posted by Andy Muirhead (Member # 712) on 02-24-2006, 07:52 PM:

David, I'm not sure I want to go down that road. At least if I give up and return the unit, the seller is resonsible for all costs (including return shipping) and I will get a full refund. Trying a new PSU will at least leave me out of pocket whether it works or not. If it works, then that would be great, If it doesn't then the seller would be under no obligation to refund/replace the unit since I would have 'fucked' about with it, if you see what I mean!

Posted by Jim Ziegler (Member # 86) on 02-24-2006, 09:33 PM:

I was having a ton of crashes and such recently. I talked to a tech support friend of mine and he said chage the PS.. Changed it, and the system is completely stable now.

Posted by Monte L Fullmer (Member # 2797) on 02-25-2006, 02:23 AM:

..course, why I mention a 500w PS, is to have that 'buffer power' in case of any added cards/devices that might be installed later on. Doesn't hurt to be a bit over sized

..what I did with my Athlon 64 setup - I could easily got away with a 400, but opted for the 500 to have that buffer. And I'm glad that I did. System runs cool and smooth without any glitches.

Intels demand quite a bit more...

-Monte