Desktop computer experiences complete lock-ups after power outage - Hardware & Software

Greetings Facepunch, Today, my community's power went out for about an hour, and it did so while I was using my desktop. I don't recall doing anything very disk-taxing at the moment the power went out, so I'm not so sure if there's any kind of corruption in the disk. Anyhow, after booting up and using my computer after the power came back on, it will completely lock up (no keyboard or mouse response) after about 5 to 10 minutes of being on. Whether I'm letting it idle or playing games, it will lock up. I'm running a checkdisk on the system drive right now, so far it has reported 2 EA (?) records processed, and 92 reparse (?) Records as processed. My system runs on an MSI P6NGM-FD motherboard with an Intel Core 2 Duo processor and two one-GB sticks of memory. This system has two hard drives; the primary one a 250GB SATA drive, and the other a 1TB of the same interface. The operating system installed is Windows Vista Ultimate 64-bit with SP 2 installed. I don't really think any more specs are necessary, other than a 585W power supply, and an ECS 9600 GT. If anyone can help me in diagnosing this lockup issue so I can isolate the cause, that would be great. I'd just like to save reinstalling my operating system for last.

What model of PSU is it.

[QUOTE=bohb;31147891]What model of PSU is it.[/QUOTE] It's an HEC HP585D.

Extraordinarily piece of shit PSU, it's most likely your problem. Get a name brand that doesn't cut corners.

[QUOTE=bohb;31150303]Extraordinarily piece of shit PSU, it's most likely your problem. Get a name brand that doesn't cut corners.[/QUOTE] I was actually reading the reviews two years after I had gotten it, I was surprised at how many people were having issues with it. It has worked without failure for me since I bought it in January of '08, and judging by the reviews, I think a lot worse should be happening if that PSU in particular was failing. Also, I would like to avoid having to spend any kind of money until we're absolutely sure it's the PSU. I wish I had a similar one in the house, but I'm afraid to say that I don't.

New development: This time, when the computer froze, the system speaker continuously played a high-pitched tone until I rebooted the computer. I also happened to be running DebugView when that happened, if running DebugView has anything to do with that.

[QUOTE=tgp1994;31153403]I was actually reading the reviews two years after I had gotten it, I was surprised at how many people were having issues with it. It has worked without failure for me since I bought it in January of '08, and judging by the reviews, I think a lot worse should be happening if that PSU in particular was failing. Also, I would like to avoid having to spend any kind of money until we're absolutely sure it's the PSU. I wish I had a similar one in the house, but I'm afraid to say that I don't.[/QUOTE] I can tell you right now it's a low quality piece of shit that can damage your components, even if it was working properly. Power supplies require several stages to output clean DC power with low ripple current, which shoddy manufacturers (like HEC) omit to save money. The missing parts that are supposed to be there cost pennies on the dollar to add, but they choose not to so they make like 1000% margins by ripping people off. [img]http://img62.imageshack.us/img62/2663/hec5852.jpg[/img] Here are the internals of your PSU, starting from the mains plug, I'll tell you what's wrong with it and what the missing parts do. On the mains plug, there is supposed to be a transient filtering stage (which consists of one double ferrite coil per hot and neutral line, two ceramic capacitors to ground, and one metalized polyester capacitor across the hot and neutral) These are completely missing. Some PSUs implement this stage on the main PCB, but more built up PSUs don't have the room so they put a small PCB attached to the mains plug and implement it there. Your PSU has it implemented on the board (Look at the top right of the board, the places are labeled C1, R1, and LF1.) At least the PSU has R1 (which is a bleeder resistor) so it discharges on power down. The next step is the second part of the transient filtering stage, which is also completely missing. This usually comprises of two more metalized polyester capacitors, three Y capacitors and another double ferrite coil (C2, C32, CY1, CY2, CY3 (under the big line capacitor on the bottom) and LF2) Since both stages are completely missing, this means the switching transistors can backfeed high frequency signals back into the line, AND has zero filtering for incoming interference or power spikes. [B]Now we move onto the most important part:[/B] The AC-> DC bridge. Proper PSUs use a bridge rectifier to rectify AC into DC, which this PSU clearly does not have (DQ 1-4). Instead it has four discrete diodes, which are rated at 3A each. This is a problem because [B]this means this is not a 585W PSU.[/B] Basically these four diodes can pull 3 amps from the power grid before they burn. Doing the math, 120V x 3A = 360W, but you can't continuously draw this, so you'll want to drop it down to a maximum duty cycle of 80%, which further degrades the output to 288W. The next steps are the primary and secondary switching transistors (which are attached to the aluminum heatsinks.) You can't see them here, but they are shown in another picture. Both sets of transistors are underrated for the load specified they can take, and will produce dirty more dirty DC as you get close to the 288W limit. I won't go too into detail with these as it gets complicated depending of the type of transistor. Finally going into the output phase (to the far left with the big bunch of colored wires) There is only one ring ferrite multi-coil instead of two (+5v usually has one dedicated to itself) but you can see the wire jumper going to the smaller ferrite cylinder, which is woefully inadequate for filtering ripple current. +12v seems to have two coils, but due to the shoddy construction of the rest of the PSU, it's useless. It also seems to be using "Teapo" capacitors, first of which are overloaded and underrated (not enough capacitors for the load and the farad rating is far too low), and Teapo capactiors are known to prematurely fail by venting or exploding due to faulty electrolyte. Get a better PSU, one that costs more than $15. Preferably one from OCZ, Corsair, Kingwin or some other brand name. Here's an example of a trash PSU I "upgraded" with the proper filtering stages, higher rated caps and coils: [URL=http://imageshack.us/photo/my-images/708/img0081kv.jpg/][IMG]http://img708.imageshack.us/img708/7795/img0081kv.th.jpg[/IMG][/URL] The parts I added won't fix the garbage transistors, but it will make the PSU have better filtered DC.

Wow... that's amazing. I never knew there was so much wrong with my power supply. Too bad it ended up costing me ~$30 on Newegg :\ From the extremely in-depth post you gave me here (and thank you very much by the way, I have a feeling the manufacturer doesn't even know the power supply to such a depth as you seem to), it seems like the slightest mishap of current could cause this power supply to fail. However, from what I can tell, it is still working well. I can boot my system into safemode, and do whatever safemode allows me to do without so much as seeing a hesitation from my system. Call me tight-pocketed, but I still want to hold off on spending money to replace a part that I'm still not entirely convinced as being broken. I still have yet to run a system on this machine since the problem arose that can more completely utilize the hardware (I'm thinking of software like KNOPPIX, which I will run in just a moment) to see if this really is a hardware issue. Am I still barking up the wrong tree here? Could this power supply actually be causing my system to lock up? (If there's a possibility that there was a surge after the power came back on, that's no concern here, I had the power strip unplugged from the wall before the power came back on.) [b]EDIT:[/b] Just froze in KNOPPIX, too. Blast, this is looking like a hardware issue.

Yes, a shoddy PSU [B]can[/B] cause spontaneous lockups, reboots, crashes and other nasty problems. Even if there appears to be nothing wrong with it. You [I]probably[/I] don't have an oscilloscope, so you can't test the ripple current. The maximum allowed is 120 mV for +12 and -12V. +5v and +3.3v have a maximum of 50 mV. if these are out of range, it can cause crashing, lockups, reboots, etc. If you have a voltmeter, you can test all of the main rails (+12v is yellow wire, +5v is red wire, +3.3v is orange and blue is -12v) These shoddy PSUs are often based on an ancient PSU design from over a decade ago, which also includes -5v. This voltage was removed from the ATX standard ages ago and can be dangerous to modern boards because they often aren't designed to support it. You should test the rails when the PSU is under load as the voltage idle can be different. They all should be within 5% of their set values, except -12v and -5v which can drift up to 10%. having high or low voltages on any of these rails can cause crashes, lockups, reboots, etc. [B]and[/B] potentially hardware damage or accelerated wear on the motherboard/CPU.

[QUOTE=bohb;31167208]Yes, a shoddy PSU [B]can[/B] cause spontaneous lockups, reboots, crashes and other nasty problems. Even if there appears to be nothing wrong with it. You [I]probably[/I] don't have an oscilloscope, so you can't test the ripple current. The maximum allowed is 120 mV for +12 and -12V. +5v and +3.3v have a maximum of 50 mV. if these are out of range, it can cause crashing, lockups, reboots, etc. If you have a voltmeter, you can test all of the main rails (+12v is yellow wire, +5v is red wire, +3.3v is orange and blue is -12v) These shoddy PSUs are often based on an ancient PSU design from over a decade ago, which also includes -5v. This voltage was removed from the ATX standard ages ago and can be dangerous to modern boards because they often aren't designed to support it. You should test the rails when the PSU is under load as the voltage idle can be different. They all should be within 5% of their set values, except -12v and -5v which can drift up to 10%. having high or low voltages on any of these rails can cause crashes, lockups, reboots, etc. [B]and[/B] potentially hardware damage or accelerated wear on the motherboard/CPU.[/QUOTE] Certainly looks like I should stop using the desktop at this point! :P I'll see if I can borrow a voltmeter, I'd really like to try testing out my power supply with one. In the mean time, I do happen to have a toy voltmeter from a circuit set, do you think it would work? I have no idea what units it's actually measuring in, I'd like to avoid breaking more things than I have to. :P [URL=http://www.megapix.com/?p=YB88FUP6][IMG]http://i1003.megapix.com/6e080c29f4eef3d13e2fd76def8b8a04c30ade6aa69cb197.jpg[/IMG][/URL]

You need a digital voltmeter that can reliably detect voltages at least two decimal places to the right, an analog voltmeter isn't going to work. That voltmeter definitely won't work because first, it tops out at 10v and second, its not nearly accurate enough.

[QUOTE=bohb;31167789]You need a digital voltmeter that can reliably detect voltages at least two decimal places to the right, an analog voltmeter isn't going to work. That voltmeter definitely won't work because first, it tops out at 10v and second, its not nearly accurate enough.[/QUOTE] Had a feeling it wouldn't quite cut it, I'm glad I asked. I guess I'll just be in a waiting period now.

I'm really trying to milk every last bit of this power supply purchase from HEC, I'm expecting a reply from them on Monday to see if they still honor any kind of warranty before I crack this thing open. Meanwhile, I have my multimeter, and I'm ready to measure stuff. I'm not so sure if what I did was as accurate as you suggested, or if I actually did do what you suggested, but I tested the metal contacts (soldering points?) Directly under the 24-pin ATX socket on my motherboard. All of the voltages seemed to show up in perfectly normal ranges as I tested each one in reference to my motherboard's manual. What concerned me, though, was that as I was testing and rebooting my computer, it would seem to seize up earlier on in each reboot. It got to the point where it just halted at my BIOS logo screen with many artifacts on it, so I just shut down my computer for the night. Have I been causing the situation to progressively become worse? Is it possible that I have just now incurred some kind of permenant damage on other hardware? What should my plan of action from this point on be? Try to find another power supply and continue testing?

Please don't use the replacement if you manage to get an RMA..Buy something of quality

[U][B]STOP TURNING ON THE MACHINE WITH THAT PSU.[/B][/U] And don't waste your time trying to RMA that piece of shit. HEC will only send you another PSU of the same terrible quality with the same missing parts. Buy a quality PSU like Thermaltake, Corsair, Cooler Master or some other brand name. Here are three I'd recommend: [url]http://www.newegg.com/Product/Product.aspx?Item=N82E16817171060[/url] [url]http://www.newegg.com/Product/Product.aspx?Item=N82E16817153023[/url] [url]http://www.newegg.com/Product/Product.aspx?Item=N82E16817139026[/url]

[QUOTE=bohb;31187031][U][B]STOP TURNING ON THE MACHINE WITH THAT PSU.[/B][/U][/quote] Agree'd. [QUOTE=bohb;31187031] And don't waste your time trying to RMA that piece of shit. HEC will only send you another PSU of the same terrible quality with the same missing parts. Buy a quality PSU like Thermaltake, Corsair, Cooler Master or some other brand name. Here are three I'd recommend: [url]http://www.newegg.com/Product/Product.aspx?Item=N82E16817171060[/url] [url]http://www.newegg.com/Product/Product.aspx?Item=N82E16817153023[/url] [url]http://www.newegg.com/Product/Product.aspx?Item=N82E16817139026[/url][/QUOTE] True, definitely not worth it anymore. Those look like awesome power supplies. But do they offer enough? As a starter for adding up the usage, my motherboard's manual recommend 350W. Then I have the two 1-gb sticks, a Geforce 9600 GT, the two SATA drivers (one's a beast by the way, the motors probably make it the loudest thing in my computer), two IDE drives (not necessarily used simultaneously), as well as two 120mm sleeve fans. If I were to go for a 430W, would it suffice and still give me some wriggle room if I were planning on a moderate upgrade relatively soon? [b]EDIT:[/b] Hmm... if you say that, theoretically, my maximum power draw from this should have been in the ~300's of watts, maybe these 400 watt PSs would do just fine. I just opened the power supply now, two of the resistors (or transistors, I don't really know) seemed to have some sort of brownish, hardened material ontop of them. I have this circled in this picture. It looks like my powersupply has a few subtle differences compared to the one you showed me, although, that obviously didn't help it from dieing :P [img]http://www.megapix.com/?p=TBAJYD1U.jpg[/img]

[QUOTE=tgp1994;31190243]two of the resistors (or transistors, I don't really know) seemed to have some sort of brownish, hardened material ontop of them. I have this circled in this picture. It looks like my powersupply has a few subtle differences compared to the one you showed me, although, that obviously didn't help it from dieing :P [img]http://www.megapix.com/?p=TBAJYD1U.jpg[/img][/QUOTE] Those are capacitors, and them venting their electrolyte like that means they've failed. It's also why your computer is crashing, because they can't output proper voltage anymore and are probably outputting so much ripple current that you might as well call it AC power. Don't use it again, just strip the 80mm fans out of it for projects and throw the rest away. If you're concerned you need more power, you can get a 500W or something. I have a 550W that runs an i5-750, two SATA drives, two optical drives and an HD5870 and it works fine.

[QUOTE=bohb;31191418]Those are capacitors, and them venting their electrolyte like that means they've failed. It's also why your computer is crashing, because they can't output proper voltage anymore and are probably outputting so much ripple current that you might as well call it AC power. Don't use it again, just strip the 80mm fans out of it for projects and throw the rest away. If you're concerned you need more power, you can get a 500W or something. I have a 550W that runs an i5-750, two SATA drives, two optical drives and an HD5870 and it works fine.[/QUOTE] Sounds good, those fans could come in handy some time. I was surprised when one of those power supplies you linked me too said it was intended for i7s. I would have thought that, platforms that use Core i7 processors would, in the end, require more power. I've got a friend who's willing to let me use an old 585w power supply of theirs today. Can't wait to see what the brand is... :P

[QUOTE=tgp1994;31191457]Sounds good, those fans could come in handy some time. I was surprised when one of those power supplies you linked me too said it was intended for i7s. I would have thought that, platforms that use Core i7 processors would, in the end, require more power. I've got a friend who's willing to let me use an old 585w power supply of theirs today. Can't wait to see what the brand is... :P[/QUOTE] i7s only pull 130W under maximum nominal load, the most power sucking thing in a system these days are GPUs.

Yeah, that's mainly what I was getting at. I can't image someone wanting to get an I7 Extreme, and then a Geforce 8600 GT or something like that.

I really need to find an oscilloscope and test mine. It looks kinda fishy to me. I only bought it because I was lazy. Fun fact: The warranty doesn't cover acts of god.

[QUOTE=moesislack;31192032]I really need to find an oscilloscope and test mine. It looks kinda fishy to me. I only bought it because I was lazy.[/QUOTE] An oscilloscope would be awesome... I'm surprised they're not cheaper, seeing as how they've been around since the turn of the 20th century.