1. Computer problem? Tech Support Guy is completely free -- paid for by advertisers and donations. Click here to join today! If you're new to Tech Support Guy, we highly recommend that you visit our Guide for New Members.

Solved: What PSU for 8+ HDDs & constant up-time?

Discussion in 'Hardware' started by Obinice, Sep 12, 2010.

Thread Status:
Not open for further replies.
Advertisement
  1. Obinice

    Obinice Thread Starter

    Joined:
    Sep 7, 2004
    Messages:
    294
    Hey guys! As I type this I can smell the remains of my OCZ StealthXStream 700W PSU, that up until now has powered my home server. As you can see from this photo it's no longer in working order, and has to be replaced.

    I don't know what caused the coil to heat up so much it burnt through the wires, but I can't take the chance that it was me overloading the PSU.

    I have 8 SATA HDDs that Windows powers up & down when necessary, simple graphics card for HDMI to the TV & a SATA controller card. I'll probably need to add two more HDDs early next year, when these ones are full to capacity (I store all my media on them to play back on the TV, as well as all my music, pictures, software, documents, basially all my important files). There are 8 HDDs because 4 are for storage, and the other 4 are for backups.

    Furthermore, I usually turn the server off when I go to bed, but I'll be moving away to university in a few weeks time, and will need the server at home to be on 24/7 so that I can access it from uni when I need to.

    Anyway I need a new PSU now, and I'm not sure what to go for. Like I said, I'm concerned the PSU may not have been up to the job (even though at the time I thought the power requirements were within the PSU's limits, my maths could have been wrong), and I need to know what will not only be up to the job of comfortably powering these HDDs now, but also up to the job of powering several more HDDs in the future.

    Thanks in advance :)
     
  2. win2kpro

    win2kpro

    Joined:
    Jul 19, 2005
    Messages:
    11,669
    Probably the best power supply made is one you never heard of, i.e Zippy. Zippy's are primarily server PSU's extremely heavily buit and designed for constant operation.

    Here are a couple of links to have a look at Zippy.

    http://www.bjorn3d.com/read.php?cID=985

    http://store.myaopen.com/ziposu.html

    $339.00 at the 2nd link.

    If you are interested in the 850 be sure and check the measurements.
     
  3. Obinice

    Obinice Thread Starter

    Joined:
    Sep 7, 2004
    Messages:
    294
    That PSU looks great, and it had never occurred to me to look for server-quality PSUs before. It is really expensive though (the cheapest I could find it was £226). Is there nothing closer to the £100 mark that would do the job safely? Something from here for example? I mean granted, this PSU is the cheapest of the lot from that page, but it still looks okay to me...

    But what do I know. My last PSU blew up :p
     
  4. win2kpro

    win2kpro

    Joined:
    Jul 19, 2005
    Messages:
    11,669
    Personally, I would purchase a Corsair unit over an Antec.
     
  5. Obinice

    Obinice Thread Starter

    Joined:
    Sep 7, 2004
    Messages:
    294
    Okay, one more question. My last PSU had 4x 12v rails, this one only has 1x 12v rail. That makes me wonder if something happened to one of the rails, in the other PSU it may only have damaged 1/4 of my components, wheras with this single 12v rail PSU, it could damage everything (which would be catastrophic for my data).

    What do you think?
     
  6. win2kpro

    win2kpro

    Joined:
    Jul 19, 2005
    Messages:
    11,669
    In my opinion a single +12v rail power supply is superior to multiple +12v rails since you don't have to worry about power distribution. All I use in my builds are single +12v rails.

    Most all the higher end power supplies, i.e. Zippy, Cosair, PC Power & Cooling, etc. all use single +12v rails.

    In my opinion what caused your meltdown was OCZ likes to play games with output numbers. I'm sure you didn't have as much output power as you thought you had. Read post #17 in this thread.

    http://forums.techguy.org/hardware/947171-new-build-problems-2.html

    It also possible that you had a fan(s) failure in the power supply that caused the meltdown. The manufacturer would have to disassemble the unit to determine where the problem occurred.
     
  7. Obinice

    Obinice Thread Starter

    Joined:
    Sep 7, 2004
    Messages:
    294
    Ahh interesting stuff, I'll certainly take that into consideration in future. Also I just checked after you mentioned the fan, and it's working okay. The exhaust isn't usually much warmer than room temperature anyway. I'd had that PSU for many months. I think I'm going to get the CMPSU-850TXUK PSU tomorrow!
     
  8. Obinice

    Obinice Thread Starter

    Joined:
    Sep 7, 2004
    Messages:
    294
    Bought and installed, though I can't tell much yet. I can access all the drives via UBCD4Win, but Windows itself on the HDD is acting very strangely. I know this question might belong in another thread, but it is directly connected to what caused me to get a new PSU:

    Do you have any idea why XP SP3 would slow to a halt? It got as far as the desktop, mouse wouldn't move, and keyboard commands lagged by something like 20 seconds, most of the system stopped responding entirely, so I rebooted. This time it got as far as a black screen showing the mouse, and then the next time it didn't even get that far. Oh, I also got a missing hal.dll file error at one point, but that didn't appear again.

    I'm running Memtest86+ now, then I'll be running Samsung's Hutil (HDD-Diag boot cd) to test my boot drive (the only one I have connected now, until I figure out the problem). If I find no problems then I'll reformat and reinstall Windows to that HDD and see if the problem persists.

    As the power cables that were compromised during my PSU meltdown were leading to my motherboard, could there also be the possibility that this is the mobo or CPU? Oh and if you've used Memtest86+ before, do could you tell me if the time elapsed (WallTime) counter is supposed to count seconds quicker than real seconds? It counts normally, then decides to speed up, then counts normally, then speeds up....weird....

    Sorry for all the questions there, lol.
     
  9. win2kpro

    win2kpro

    Joined:
    Jul 19, 2005
    Messages:
    11,669
    I haven't used Memtest in quite awhile and have never really noticed how it counts seconds. I just start it running and generally leave the machine until the test completes. You need to test each stick of RAM individually with a minimum of 4 full passes. If you get an exceptionally high number of errors it may not actualy be the RAM at fault. I have found that defective board capacitors can cause Memtest to report exceptionally high numbers of errors that are not really there.
    If Memtest indicates a high number of errors I always verify the finding by installing the RAM in a properly operating machine, run Memtest again, and see if the errors re-appear.

    When the power supply melted down it could have shot high voltage throughout the components. You'll just have to check and prove each component individually. It's possible that in the meltdown capacitors were damaged and also possible that the drive was damaged.

    I've given up on HDD manufacturers diagnostics. In the last 6 months I've tested a Seagate and a WD drive where the diagnostic test indicated both drives were good, and both proved to be defective.

    I've personally never cared for Samsung drives and never used their diagnostic utility so I can't tell you anything about their diagnostic utility.

    I've just seen too many instances with a power supply failure where the board and possibly other components are damaged.
     
  10. Obinice

    Obinice Thread Starter

    Joined:
    Sep 7, 2004
    Messages:
    294
    I like Seagate, I have to admit that after doing a lot of research into what drives to get, and deciding on Seagate over Samsung for various minor reasons (power consumption, heat & reliability I think they were, I'm not sure now), I went out and accidentally ordered the Samsung drives instead. Not my brightest moment.

    I've been running a tool called Emsa DiskCheck on two of my drives (my Windows drive and it's backup drive). Both are giving me a lot of "Warning: Can't get attributes:" and "Error accessing drive" errors. One of the errors pointed towards my system32 folder. I browsed to it (I'm running UBCD4Win at this point, Windows won't boot up) and the whole folder's corrupted, as are many of my other files it seems across both drives I've checked.

    I've never seen so many corrupted files. It's really strange. But then I've also never had a faulty HDD.
     
  11. Obinice

    Obinice Thread Starter

    Joined:
    Sep 7, 2004
    Messages:
    294
    Well it took time, but I've finished using HD Tune to check all 8 of my drives. This is what it found... (Note the only "bad" S.M.A.R.T value I found was vendor specific "(B8) (unknown attribute) Current: 1, Worst:1, Threshold: 99, Data: 600". Who knows!)
     

    Attached Files:

  12. jiml8

    jiml8 Guest

    Joined:
    Jul 2, 2005
    Messages:
    2,634
    Your OCZ supply experienced an internal short. Maybe a diode failure or some such. Most probably wasn't due to being overcapacity. All I can say is that sh*t happens. The PS evidently failed safe because it didn't carry out any of the rest of your system.

    This machine I am typing this post on has a quad core Phenom-II 955, 6 internal hard drives (2- 15K RPM SCSI drives, 3- 10K RPM SCSI, and one WD Green SATA.) with 8 gigs of RAM, a GT-240 video card,and a total of 11 fans.

    This system runs 24/7/365 in an environment that during the hot months hovers around 28-29 C, and is frequently quite busy in the middle of the night (backups, etc).

    I am powering the whole thing with a Coolermaster 500W supply. I estimate that the supply is running at about 85% capacity when the CPU is fully loaded and the drives are all busy. It shows no tendency to overheat and this system is solid; absolutely no evidence of power starvation.

    The point is that with 700 watts, you had enough capacity.

    Edit: Regarding your evident HD problems, the PS might have spiked your system when it died. If so, the drives could have taken some "soft" damage - meaning corrupted write operations due to voltage transients - and might be OK after a reformat. Alternatively - and this is a procedure I recommend HIGHLY given your situation - run Spinrite on them. Spinrite isn't free, but it is VERY effective.
     
  13. Obinice

    Obinice Thread Starter

    Joined:
    Sep 7, 2004
    Messages:
    294
    Thanks for the info jiml8, it's been educational :)

    Okay, one more question. Some very odd things are happening. Firstly, When Windows would still boot after this issue, it gave me a "This device cannot start" error for the onboard LAN. Since then Windows has screwed up so I've booted from UBCD4Win, which gives me the same piece of information. I've connected a PCI network card and that works fine. I've also reformatted and restored an image of my boot drive.

    The second odd thing is that Windows won't boot at all. It's not that the BIOS insists there's no boot disk. Instead, after trying to boot from CD, the flashing underscore just remains, flashing, and nothing more happens. If I hit CTRL+ALT+DEL, the system restarts and reaches the same point, but this time the keyboard (USB) freezes up. This was happening before I restored the image of my boot drive, so I don't believe it's a problem there.

    The network adapter issue isn't so bad, I was going to use the PCI network card anyway (the onboard LAN has always kept changing its MAC address to something screwed up, which is very annoying). But not being able to boot from the hard drive, that is a big problem. I think if I can find a solution to that, all will be back to normal!
     
  14. jiml8

    jiml8 Guest

    Joined:
    Jul 2, 2005
    Messages:
    2,634
    Your onboard network card probably didn't survive the power supply failure. It would appear that the OCZ supply did NOT fail safe; your system DID get spiked. I'll put that into my long-term memory storage as a reason to not purchase an OCZ supply...

    The hard drive issue probably is corruption of the master boot record or the boot loader, probably due to the power supply failure. It needs service. If you don't have/can't afford Spinrite, then you need to do a surface scan and map out bad blocks, then reformat and reload it. I don't know anymore how to do that without Spinrite...that is how I always approach such problems and have done so for years now.

    Edit: I agree with Win2kPro about multiple 12V rails; marketing hype that complicates installation and can lead to unnecessary problems.
     
  15. Obinice

    Obinice Thread Starter

    Joined:
    Sep 7, 2004
    Messages:
    294
    Well the coil was in direct contact with the wires leading to the mobo, so it doesn't surprise me.

    I've got my hands on Spinrite, and it gives me this message "Invalid partition for drive size. This partition exceeds the size of this drive as defined by the system's BIOS or BIOS extension. You should not proceed to use SpinRite on this drive until you have verified and corrected the disparity between this drive and the BIOS' or BIOS extension's understanding of the drive size".

    This is after I reformatted and used DriveImage XML to restore a backup made of the same drive from several months ago. The drive was accessible after I restored the backup, although now that I've replaced the motherboard I'm plagued by blue screens (the details of which are posted below) and other errors while trying to load UBCD4Win (and also it's onboard Windows Recovery Console, which gave the error "cpqarray.sys is corrupted") while the hard drive is connected. Without the hard drive, UBCD4Win loads fine. I also tried loading UBCD4Win with a different hard drive connected, and got another blue screen (details below).

    Problem after problem it seems... :(


    Blue screen details:

    STOP: 0x0000007E (0xC0000005, 0xF731D40B, 0xF7912B44, 0xF912840)
    acpi.sys - Address F73AD40B base at F73AD000, DateStamp 41107d27

    Another blue screen when trying to load UBCD4Win with a different hard drive connected:

    IRQL_NOT_LESS_OR_EQUAL

    STOP: 0x0000000A (0x00000001, 0x0000001E, 0x0000001, 0x8061AC16)
     
  16. Sponsor

As Seen On
As Seen On...

Welcome to Tech Support Guy!

Are you looking for the solution to your computer problem? Join our site today to ask your question. This site is completely free -- paid for by advertisers and donations.

If you're not already familiar with forums, watch our Welcome Guide to get started.

Join over 733,556 other people just like you!

Loading...
Thread Status:
Not open for further replies.

Short URL to this thread: https://techguy.org/949623