Bitcoin Forum

Could you share your experience with a failing PSU?  Note, I mean failing, not already failed (that's obvious).

What were the symptoms leading up to a failure?  any unexplained behaviors observed?

Thanks!

A failing PSU can produce random problems. Say if the failure was in the +12V output, causing it to sag every now and then, it could cause your PC to shutdown, or a hang because the CPU couldn't operate correctly, or a hard disk spin down. If it was the +3.3V then you might get random memory errors which would also cause undeterminable problems. If it was spiking, it might kill the VRegs if it goes too high.

I had a Huntkey 550Watt PSU once that wasn't made for the job...

12 Volt rails dropped slowly to 10.8 Volts and then the PC just rebooted... It ran again for 30 minutes and dropped again, slowly...

I used it for Seti@home (Intel Q9550, 1x GTX9800+ and a GT9500).

After that I bought an Antec 650 and it ran fine... Now I'm using the same 650 for one of my 2x 5870 rigs :)

My rig has two 5830 ...

second card getting only 260 Mh ...first has 300-310 ....

working for about 2 week ..

then  almost no name power supply 750w start smelling, PC shut down ...



replaced PSU with coolmaster gtx 750 ...

both card running stable at 300+

750w are not equal 750w ...

I had an ocz 850w go bad on me in my F@H rig and almost cause some serious damage.
I was pulling 625w from the wall and everything was running fine for around 6 months until one day I noticed a pretty funky fishy type smell in the room. I didn't think anything of it at the time and just opened a window. As I was walking back across the room I heard this terrible cat/bat type screeching sound come from the back of the pc so I walked over to it and got about 3 steps away and BANG, followed by a puff of smoke.
Luckily the fuse blew on my socket and no damage was done to the property but after further inspecting the power supply its outer casing was terribly deformed, almost as if someone had hit it from the inside out with a sledge hammer, breaking the fan & grill. It looks as if one of the larger capacitors (sorry I'm not an electrician so it may be called something else, transformer unit?) had exploded and shot through the psu case.
I was a little shaken up and called ocz that afternoon to arrange a replacement, which has been running fine for 12 months now though I'm not sure how far away it is from doing the same or if the previous one was a manufacturing defect, ocz was rather quite after the replacement was sent.

I had a very similar experience I bought a 850w Kingwin PSU and I had just 3x5850's on it OC'ed. It was on a sunday not too long ago where I was watching the bitcoin prices on that same computer when all the sudden LOUD POP, burning smell and I looked down saw the PSU actually had a little fire come out of the back then a puff of smoke. I ran over to the breakers shutdown the breakers, grabbed a fire extinguisher and ran back to see that the PSU was smoking a tiny bit but was not on fire anymore. When it all cooled down took out the PSU all I saw was black char'ed components inside the PSU. Nothing else got damaged amazingly nevertheless I will never buy from Kingwin again the experience was not pleasant. Tried to return the PSU but they said it was my fault no refund, that I did something wrong. Now I only run with Antec's all 1200w Silver / Gold and never had any problems since and I have 4 rigs.

Kingwin PSU that started smoking: http://www.microcenter.com/single_product_results.phtml?product_id=0341646
Antec PSU that I use now: http://www.microcenter.com/single_product_results.phtml?product_id=0336040

So yes I can see how bitmining can lead to serious accidents, if I was not home I could only imagine what could have happened.

My experiences with bad power supplies:

1.Computer shuts down, will not turn on, there is a smell of something burning coming from the PSU.
2.Computer randomly reboots, but a reboot is more likely if I play a game. Over time reboots become more frequent. The power connector on the motherboard is a bit burnt, but still works with a new PSU.
3.The computer hangs randomly when accessing one hard drive. I hear that the drive spins down and back up when I try to access it (and computer stops responding while the drive spins up). With my multimeter I find out that the +5V line voltage is more like 4.6V and sometimes sags to 4.2V, which causes the drive to spin down. After replacing the power supply, the hard drive works normally.
4.Tape drive complained (showed error code) about out of spec voltages. New power supply solved that problem.
5.Computer randomly shut down, turns out the PSU fan was bad and overheat protection was kicking in. With new fan, the PSU works normally.


A multimeter and a scope can help find out if the PSU is bad. ATX spec has defined voltage and ripple tolerances for each voltage. A voltmeter can help find out if the voltage is always too low (if it is near the edge there can be a problem) while a scope can measure ripple and show if the voltage goes out of spec for a few ms.

Havnt had one fail since I was back on XP.
But what happened were random blue screens under high load.
Then it started freezing on desktop randomly (thought it was ram at first so changed that)
Worked for a few more weeks that started BSOD at start up.
Then just didnt turn back on.

Had another one fail where the fan stopped then it released the white smoke and turned off.

And back in the win 98 days had one go pop, And it electrocuted me when I touched the case.
Wasnt serous voltage was more like those zap toys you can get.

If you want, look for a program called Everest, It can show voltages for 12, 5, 3.3 , etc - if they are 10% off or more you may want to get a b/u PSU

sometimes it can find problems before they even begin.

it's one of the most common problems and often one of the harder to diagnosis from just the computers temperament, because a bad power supply can have so many wide variety of symptoms.

computer manufactures and often regular people tend to go cheap on the PSU, this is NEVER a good idea.

I agree with Palmdetroit, you can use a program like everest and monitor voltages. I would also put your computer under as much load as you can while watching the volts.. I would put on a hd movie, maybe scan a full cd for a file.. defrag a spare hd.. play a game.

Put it under load and then see how your voltages are.

Ahh very informative going to try out Everest, Thanks Palmdetroit

Recently my Corsair VX550 died. First I had random reboots every 12-24h, then one day it died completely, 2 pins on the 24pin cable fried (plastic had melted too), the motherboard socket where the 2 pins were connected is also a bit burned, but fortunately it works fine with a TX 850 V2 that I just bought.

I think the overload might've killed it - hd6850 and 2x hd5850, all 3 overclocked by 10-15%, 2 hdd, q9300 @ 3.25 GHz, 1.25V
but anyways - this shouldn't have happened, I thought better of Corsair, and the PSU was only 2 years and 8 months old (mining 24/7 only for a month).

Ya that program is great (and much easier/safer than using a multimeter to check voltages, hehe). I did forget to mention to slowly increase load from vid/hds, cpu, and monitor the voltages, as joules pointed out. I always run everything at 100% so forgot that part  ;D

It is less accurate than a multimeter and especially a scope.
The sensors on the motherboard can be reporting the voltage incorrectly (had that a few times). The voltage on other lines/rails may be different. The voltage may sag for a time, shorter than the measuring interval of the sensors on the motherboard.

I learned this the hard way. A hard drive was acting up, but everest showed 4.6-4.7V instead of 5V, which is (barely) within tolerance. So, I almost bought a new drive when it hit me to measure the voltage at the power connector to the drive. Guess what? The voltage there is 4.6V but sags to 4.3 or 4.2 sometimes which causes the drive to spin down then spin back up again.

A scope can show ripple and very short voltage dips, so it is even more accurate than a multimeter.

wait, so if a PSU is either underpowered or bad quality, it will lower your mhash/s?? i thought that your video card will work or it doesn't if there's not enough juice to it, instead of working but lower the mhash/s. is this confirmed?

I would hazard to guess that a poor quality PSU under load would output low or unstable voltages that could cause various behavior in a GPU card, of which, low performance could be one.

It's good to know that having a solid PSU gives you a better chance of getting best performance. I suppose you could test the cards one by one and see if they perform better than when together and that may indicate you can improve with better power.

I've had an old TAGAN ITZ Series 900W PSU, which was a refurbed replacement for the orignal failed TAGAN PSU, for several years and thought I'd build a caseless miner with 4 x 5830 GPUs. While the TAGAN boots and runs, the GPUs seem to come and go when querying with AMDOverdriveCtrl -i. I disconnect one card and it will run 3 at a time but one will not hold an overclock.

Being the impatient type and having the funds, I now have an Antec High Current Pro HCP-1200 1200W coming next week ;-)

random lockups and BSODS. boot failures on initial cold boot requiring several restarts to finally boot stable.

if you can run furmark (on all GPUs) and prime95 at the same time for an hour your PSU and system cooling are fine as that will be max power draw. *** CAUTION *** you aint seen temps till youve dont that so BE AWARE you can possibly damage your PSU, GPUs, and/or CPU if youre not careful monitoring temps on GPU and CPU during the test. *** CAUTION ***