My PC has died once again. This has happened before but it turned out to be a video card failure. I was able to identify the problem by listening to the beep-codes, ordered a new card and I was back in business. This time it seems like it is something a tad more serious.
I was talking about it this morning on Twitter so to avoid repeating myself over and over, let me embed my rant here:
My desktop won't boot. Fans spin up to the max, and stay that way. Doesn't even get to POST. No beep codes or blink codes. ಥ╭╮ಥ
— Luke Maciak (@LukeMaciak) August 5, 2014
When I had issues with it last year, it was a video card issue and I got a clear beep code. http://t.co/X3Bhc1j0tR This seems more serious.
— Luke Maciak (@LukeMaciak) August 5, 2014
The video card is brand new so it shouldn't be an issue. I tried swapping it with a spare that I know works, but it still won't even POST.
— Luke Maciak (@LukeMaciak) August 5, 2014
Took out all RAM sticks to see if I can make it beep. It does not. All I get is howling fans. :(
— Luke Maciak (@LukeMaciak) August 5, 2014
When I replaced the video card last year I noted the bottom PCI Express port would not work. I'm thinking the mobo finally died on me.
— Luke Maciak (@LukeMaciak) August 5, 2014
That or the PSU. But the PSU doesn't seem dead – I see the network card LED's blink, the case gets lit and the GPU fan is spinning normally.
— Luke Maciak (@LukeMaciak) August 5, 2014
From what I have gathered so far, the culprits can be:
- The motherboard – when I replaced the video card last year I noticed one of the PCIe ports was not working at all, so it is likely that something on the board shorted out back then and I was simply really lucky to bypass it and get a whole another year out of the hardware. Replacing it wouldn’t make my sense because I would also likely need to replace the PSU and the CPU, expand memory, and with a new PSU I could actually get a new video card so I would essentially be building a brand new machine.
- The PSU – I am told that a failing power supply can cause such symptoms. If the CPU and/or video card are not getting enough juice the system won’t even POST, but the chassis lights, fans and LED’s which only require low voltage may still be fully operational. Unfortunately I don’t have a spare PSU lying around to test this theory.
- The CPU – someone said to take the fan and radiator off and re-apply thermal paste which may work, unless it’s already to late and it fried itself. You technically ought to be able to see/smell when a immolates itself, but other than eyeballing it I don’t really have a way of testing it.
I kinda hate these situations where it is really hard to tell what is going on. I honestly kinda hate working with hardware. I can tinker with software all day, but with hardware I always feel like I’m just throwing money on the problem and I can never be sure if I:
- Replaced the wrong part (and the real problem is actually elsewhere)
- Replaced the right part but somehow attached it or configured it wrong
- Just damaged a brand new component by doing something stupid
At this point I am seriously tempted not to be frugal and just irresponsibly splurge on something I can have shipped to my house within a week, that I will be able to take out of the box and start using right away:
I am suddenly really tempted to just go and spend an irresponsible amount of money on a pre-made brand-name gaming rig with a warranty. :P
— Luke Maciak (@LukeMaciak) August 5, 2014
Anyone ever had a similar issue? Any troubleshooting steps that I’m missing? Any recommendations?
Troubleshooting Update: Friday, Aug 8
As per some of the great suggestions here and on twitter I did some troubleshooting. First I bought this PSU tester:
Using this contraption to test a desktop PSU. It's flashing LL and beeping. I have no working PSU for base comparison pic.twitter.com/dLGhqf9QNC
— Luke Maciak (@LukeMaciak) August 8, 2014
It did give me some interesting readings that I am not entirely sure how to read. The manual that came with the device is more than useless:
Anyone can tell me what this output of a PSU voltage tester means? The "manual" is useless 1 page POS. pic.twitter.com/qMRPuEKno1
— Luke Maciak (@LukeMaciak) August 8, 2014
The LL symbol under the +12V2 is flashing. The 1 page manual claims it either means no voltage or voltage lower than acceptable minimum. Greg on Twitter suggested my PSU might be missing that rail so I checked it:
This is the spec block on my PSU.
@Vanderdecken it looks like it's supposed to have 2nd +12V rail pic.twitter.com/wUWmWBEhb8
— Luke Maciak (@LukeMaciak) August 9, 2014
The +12V2 rail is listed on the PSU spec block so it does seem like it is supposed to have it. So am I correct interpreting this as my PSU having developed a fault on that rail and needing to be replaced?
I don’t know of any trick to see if there is a CPU vs a mobo failure, however telling if something is wrong with the power supply should be feasible if you have some voltmeters or even just a good eye for components that have reached their end of life.
It would either be CPU or motherboard I would imagine, the only real way to determine which it is would be to find a CPU that’s compatible and put it in. If it works, CPU is the problem. If it doesn’t, motherboard is the problem.
If it’s motherboard and the system’s more than a couple of years old, finding a suitable motherboard’s going to be a pain so may be new system time. Lenovo have a deal on their Erazer gaming machine (yes it is a strange name) at the moment that seems pretty well priced as far as prebuilt gaming systems go… I’ve been pretty impressed with their stuff, might be worth a look if you do head down that path?
@ fundamental:
They even make specialized testers for PSUs; have one myself, dead simple to use. If you know any hardware-centric folks you might be able to borrow one to quickly narrow down that particular possibility.
Yeah, I absolutely dread troubleshooting computer hardware too. Troubleshooting software is interesting because if I can’t solve the problem immediately, I usually learn something new, interesting, and valuable. Troubleshooting hardware is time-consuming, expensive, non-deterministic, and generally doesn’t teach me anything useful. To avoid it, I just go the brand-name route and pay a little extra to not have to deal with the headaches. The risk is that I might buy something hostile to Linux (or to re-installing Windows), which currently an issue now that everyone’s switched to their own half-assed implementations of UEFI. It can be difficult to impossible to boot install media with this. To avoid getting stuck on this I spend a bunch of time researching before making the purchase.
For example, in February my wife’s computer’s cooling system died. It would shut off from overheating after about 3-5 minutes. It was a cheap iBuyPower computer, which I strongly discourage anyone from buying. Their builds are total crap, they have no quality standards, and their computers are noisy. Stay away. Since it was a couple years old and was kind of junky we just decided to replace the whole thing. She’s even more of a gamer than me (in terms of hours played), so we got a Acer AG3-605-UR38. It’s listed at $1,400 now, but back in February it was only $975 (why?). Outside of disabling UEFI in order to “upgrade” to Windows 7 — and so I could boot my personal live Linux system on it in case I ever wanted to use it myself — everything worked out without a hitch. I plan to do this every single time I buy a new computer in order to avoid messing with hardware.
The ideal way to solve this type of problem is with identical hardware. One approach to get similar results is to bye all possible replacement parts from ebay and sell what you don’t need back when you are up and running again.
@ fundamental:
I think my dad has some voltmeters lying around somewhere, but I probably wouldn’t know what to do with them.
@ Douglas:
Nice! Thanks for the suggestion. :)
@ Jason *StDoodle* Wood:
TIL that PSU Testers are a thing. :P
@ Chris Wellons:
Uhgh, thanks for reminding me of the EUFI bull-crap. Do they only enable that for machines that ship with Win8? We’ve been getting laptops with Win7 at work and they all have fairly standard (albeit proprietary) DELL BIOS setup not unlike what I’ve seen shipping with XP and Vista. I’m guessing that I get a system that’s can be “upgraded” to Win7 on the store page, it will have it disabled?
@ Luke Maciak:
I’d imagine that you should be able to find some information online as to which pins should be at which voltage levels in some sort of guide format. If you do go doing this though be aware that (IIRC) power supplies will expect some pins to get shorted out for them to power on and that the tolerances on what an acceptable voltage is will vary quite a bit if you’re testing an unloaded power supply.
If you can see something like the 12V line is only registering 3V though and the capacitors within the supply are bulging, then you can at least know that that supply (or at least that one cap) is shot.
A power supply tester might manage all of this easily, but I’ve got at least 3 voltmeters sitting around and this is the first time I’ve heard of dedicated power supply testers, though their existence makes perfect sense.
I did have similar symptoms (nothing happening, just a light on and fans, I’m not sure about beeps, since that machine has no speaker). Unplugging and removing the BIOS battery for few minutes actually helped.
you should check to see that you havn’t grounded out your mother board as that could cause the effects you have listed…
Guys, I troubleshooted (thoubleshat?) some more.
I went out and bought this PSU tester device:
It did give me some interesting readings that I am not entirely sure how to read. The manual that came with the device is more than useless:
The LL symbol under the +12V2 is flashing. The 1 page manual claims it either means no voltage or voltage lower than acceptable minimum. Greg on Twitter suggested my PSU might be missing that rail so I checked it:
The +12V2 rail is listed on the PSU spec block so it does seem like it is supposed to have it. So am I correct interpreting this as my PSU having developed a fault on that rail and needing to be replaced?
So, seeing a reading that indicates that the 12V line is too low either means that it’s too low or that you don’t have it connected to the meter right. Your power supply brick does indicate that it should have 4 accessible +12V lines.
assuming that this is your problem:
Depending on your usage and the power supply’s pinout that might just mean that even if one of the +12V lines is down, it could still be usable.
I’d recommend looking up some pinout charts like http://www.smpspowersupply.com/connectors-pinouts.html for your connectors. (heck that one might even have the needed pinout for you)
As it looks like the pinout styles only end up having two +12 V lines per connector, you might be able to find one that works fine.
If you are able to eliminate the ones that have a damaged then you might be able to reconnect and boot away.
I’d still shove the voltmeter on the pins that the PSU tester indicates are bad to see what sort of voltage actually is there, as it does eliminate ambiguity due to bad documentation on the PSU tester’s part.
Pingback: Spectacular Computer Failures: The Next Generation | Terminally Incoherent