The dead home desktop returns from the dead


I brought my dead home desktop computer to the office so I could fiddle with it after work while surrounded by large quantities of geek equipment that could step in to assist.

I tried to use a power supply from another computer in a sort of computerish version of jump-starting a car, but it was a hopeless endeavour because the cables on the power supply assume that the motherboard is right there. They barely reach the motherboard when the power supply is installed in the case; trying to connect one machine's power supply to another machine's motherboard is hopeless. I didn't want to unscrew the motherboard from the case if I didn't have to, so I left the power supply theory off to the side as a backup.

I yanked the memory from the machine—no help.

I yanked the video card from the machine—bingo. The BIOS made its happy booting sounds, but naturally it couldn't boot since it had no video card, no keyboard, no mouse, no memory, and no storage devices. But at least it went beep.

I went into my neighbor's office looking for a spare video card and found one on a side table. Plugged it in, hooked up the monitor cable, and... dead computer.

Oh, so close yet so far. It seemed that merely having a video card plugged in is enough to make the computer stop working.

I left my investigations at that for the evening. The next day, I told my neighbor about my "I thought I had it solved but I was wrong" close call the previous evening.

"Which video card did you use?"

That one.

"Oh, I don't think that card works. Here, try this one."

Just my luck. I had run a video card test with a broken video card.

I returned to my office, plugged in the new video card, hooked up the monitor, and—Hallellujah—I got a POST screen and some angry BIOS messages since it didn't have a keyboard, mouse, memory, or storage devices.

Okay, plugged in the memory and storage devices, booted up the machine again, bingo, everything started up. Well, okay, I couldn't type in my password since I didn't connect a keyboard, but everything was working well enough that I was confident that I found the problem. Busted video card.

Thanks to all the commenters for your suggestions, even though I didn't act on most of them.

Bonus puzzle: When I ran the video card test that first evening, I unplugged a monitor from an existing computer and plugged it into the dead computer. After the test failed, I plugged the monitor cable back into the original computer. And the result was weird.

The picture on the monitor came back just fine, but it appeared to be frozen. I wiggled the mouse and banged on the keyboard, but the cursor didn't move on the screen and my typing didn't appear either. The computer was still running, however; I could ping it from another computer in my office, and I could even use Remote Desktop Connection to connect to the computer and use it. But the screen was frozen. It displayed a snapshot of my desktop at the point I originally yanked the monitor cable, sort of like Grandfather's Clock. The image finally disappeared when I powered the computer off.

When I turned the computer back on, the monitor showed just a blank screen. The computer itself ran fine, but now instead of showing a frozen image, it showed nothing at all.

The puzzle: Explain the exhibited behavior.

Comments (39)
  1. Mark says:

    Is it a DVI connection or VGA? If it’s VGA then the problem is likely a problem with the video card or motherboard. Something hardware related.

    If the connection is DVI (or HDMI) where a digital connection, HDCP, and other nuisances can interfere there are several possibilities.

  2. Spike says:

    Wild guess at the bonus puzzle:  Had you borrowed that monitor from a PC with dual monitors?  Did that deprived PC disable the monitor when it was unplugged but not clear the screen?  When you replugged the monitor was it now disabled but displaying the legacy screen image?  Which disappeared when you switched the machine off?

  3. shf301 says:

    Bonus puzzle:  The video card died (for some reason) when you unplugged the monitor.  When you plugged the monitor back in it was still displaying what had been in VRAM.  Once you turned the computer off the VRAM was cleared so it showed a blank screen.

  4. John says:

    The puzzle: Explain the exhibited behavior.

    The answer: gremlins.  It’s always those damn gremlins.

  5. Stan says:

    I like where Spike is going. If it was a dual display, perhaps the card detected when this (second) monitor was removed and reconfigured for a single monitor. For some reason, reenabling this second monitor is not automatic even when being plugged back in or rebooted. Maybe it requires an explict  software action of some sort — this would explain why it was still blank on a complete power cycle. Makes me wonder if this other computer was a laptop.

  6. Steve says:

    Here is a tip for the next time : usually video cards left on a table don’t works well (the one inside a computer have a better probability).

    I bet you put it back on a table ?

    1. Monitor unplugged from B
    2. Monitor plugged into toasted videocard on A

    3. Monitor plugged back into B (still running)

    4. Restart B, Monitor blanks

    I would assume the monitor is some sort of CRT VGA with an internal buffer that’s kept as long as it doesnt receive a new vsync (a sony i had displayed that behavior). I’d assume computer B deactivated its output when it detected the screen was gone, and for some reason or another didn’t detect the reconnection and didn’t send a new signal, hence not clearing the monitor’s buffer.

    When restarting, a signal was detected at post (probably some detection of which screens are plugged in) but the output is still deactivated, probably as pointed out because it’s a dual screen port and default to DVI, or maybe because the card in the absence of any monitor reassigned the default to another port.

  7. Kemp says:

    I would have gone with the easy answer: despite people insisting on doing it, the video ports are not (afaik) hot-pluggable, just like PS2 mouse and keyboard ports aren’t (yes, it can and has fried parts of motherboards). I believe the behaviour when hot-plugging something that doesn’t include hot-plugging in the spec would be undefined.

  8. Kemp says:

    Oh, I am of course referring to PS/2 ports (or whatever the official name is), not Playstation 2 ports =P And VGA ports, not DVI (which I suspect, but am not sure, probably aren’t spec’ed for hotplugging either).

  9. Torkell says:

    Been there, done that. I wired a pair of network links (wall socket to rj45 plug) as crossover due to using the wrong pinout for something, then tested them with a crossover cable. That took a bit of headscratching to figure out why two apparently good cable runs refused to work.

    Re bonus puzzle: I’m going to go with the above posters (dual display card being too smart for its own good). The only other thought is that you’ve actually killed the graphics card, though I’d be surprised if plugging in a monitor did that.

    Do please let us know what the answer is.

  10. KanOosting says:

    You bent one of the pins of the VGA connector, possibly the strobe signal.

  11. Cooney says:

    Puzzle – you accidentally fried the video card partway on the second computer. The computer thinks it’s ok, but it won’t actually display stuff now.

  12. Aaron says:

    Told you so. ;)

    I can’t explain the behaviour from ground up, but I’ve had it happen to me before; sometimes video cards are just "mostly dead".

  13. Jonathan says:

    Bonus puzzle: Maybe you were going through a KVM and it kept the frozen picture?

  14. Nawak says:

    Puzzle: the computer is training you not to do that again. And it’ll work!

  15. MadQ says:

    Betcha this will fix it:

    1. Turn off PC.
    2. Disconnect power cable from power supply.

    3. Hit power button. Notice power LED briefly lights up and the fans may briefly spin, despite the disconnected power cable.

    4. Plug power cable back in.

    5. Turn on PC.

    Bonus bonus question: WTF?

  16. Kemp says:

    "Hit power button. Notice power LED briefly lights up and the fans may briefly spin, despite the disconnected power cable."

    Capacitors, big big capacitors :-)

  17. mikeb says:

    Jonathon’s got the answer – KVM.  The mouse and keyboard were not actually attached to the computer, so no matter how much you messed with them, it wasn’t going to change the screen (unless ,of course you hit on the KVM’s hotkey, if any).

    This sounds very similar to Sara Ford’s tech support puzzle:

    http://blogs.msdn.com/saraford/archive/2006/10/06/Pop-Quiz_3A00_–How-good-are-you-at-technical-support_3F00_.aspx

    Is there such a thing as KVM Hell?  Sounds like maybe there is…

  18. Rick C says:

    "Capacitors, big big capacitors"

    Which generally won’t hold more than a second’s worth of power, even on a beefy 600W PSU.

    The entire purpose of that is to discharge the caps.

  19. Kemp says:

    That was my point, MadQ’s post conveyed a certain level of wonder that there was power without a power cable.

  20. Mark says:

    Bonus: You fried the video card when hot-swapping the monitor. Anything that happened to shows up onscreen after the fry-daddy is a fluke.

    You should know better!

  21. Darkstar says:

    quite funny what people suggest as solution to the bonus part. "Internal buffer", "missing strobe signal", etc.

    LOL

    I had a similar problem myself a few times and I found out that in removing ("wiggling") the monitor cable from the VGA card, I somehow accidentally wiggled the card in its PCI slot and it got a quick disconnect from one of the pins, enough to make the video driver fail in some weird way (i.e. losing connection to the card or something). The image was still displayed because the card didn’t "notice" the disconnect and happily continued to display the last data it received from the computer. But as the driver thinks his card has vanished, it won’t send any more updates. -> static screen.

    You can try to reproduce it by wiggling on the card again (i.e. while a monitor is still connected). Chances are you either crash instantly, hit a bluescreen, or observe the same phenomenon again.

  22. Ian says:

    If the AGP/PCIe (whatever) host interface died, it’s possible the video card could continue to cycle its video-ram & maintain the "snapshot" image.  

    Most probable explanation is the clock signal used to decode the host interface went away, while the clocks used for the chip core/ramdac/etc… continued.

    Video cards aren’t atomic beasties.   They are complex systems with hundreds of component elements.  Even if one fails, the rest can continue functioning.   As with all things these days.

    — Ian

  23. J says:

    Darkstar is pretty close to the bonus answer, but didn’t explain why it was still blank after reboot.  

    Yes, the card had tilted slightly in its slot, such that at least one pin was disconnected.  The card still had power to maintain refresh until shutdown and amazingly didn’t cause a bluescreen or whatever.

    Reseating the card brought everything back. Nothing appears to have been permanently damaged.

    It was one of those %!#&* cases with a little flip down cover over the expansion cards rather than screws holding them in place.

    — Raymond’s neighbor

  24. Mike says:

    damn it, I was just about to suggest display card not attached properly!

    as someone who never screws anything in place (except the mobo, so it doesn’t fall out) I often have that problem when switching monitors around.

    This is also one of those extremely rare cases where kicking it repeatedly until it fixes itself doesn’t work.

  25. GreaseMonkey says:

    Well, I might need a heavier hacky sack. But seeing as you have fixed the problem, I guess I’m stuffed.

  26. Name required says:

    Bonus puzzle: it was terrorists. It must be, since you haven’t reported otherwise, and I have noted that any time anything bad happens now they say "terrorist acitivity is not suspected".

  27. ender says:

    My first guess would have been that the monitor froze – that happened about twice a year with my first LCD (15" Hansol with D-SUB connector). It would freeze and either keep displaying the same picture until display mode change, or it would display weird pattern (partially based on screen contents) that faded to white in about 15 seconds. Turning it off and back on however solved the problem for me.

  28. Ian Horwill says:

    Re bonus: my first guess re the unresponsive good computer would be that you were still typing at the other, dead PC :-) Doesn’t explain blank screen on power up though.

  29. mh says:

    Bonus: The PC also has an integrated video card, and one of those nice friendly cheerful unhelpful BIOSs that doesn’t let you fully disable it?  (Been there, done that, and despite not being hot-swappable, hot-swapping the monitor connection after the apparent "stall" will get things back.)

  30. MadQ says:

    @Kemp: The "WTF?" was more of a reflection on the reaction most people have when I tell them to go through that procedure. I know all about capacitors; you get a certain respect for them after you’ve been hit with a few hundred volts a couple of times.

  31. The KVM solution is probably quite simpler…

  32. Actually…

    "After the test failed, I plugged the monitor cable back into the original computer. "

    back into the original computer would tend to confirm there was no kvm.

  33. Xepol says:

    I gotta admit, its definitely time to reseat everything. Combine age, fiddling with cables and hair pulling frustation as you slap hardware through the machine – chances any base problems you have have been aggrevated by a loose connection.

    Since there is no discussion of releasing the magic smoke, I will assume you haven’t done anything too severe, and since your screen is frozen instead of magenta or cyan, I assume that you didn’t bend a pin on the display cable.  I have seen TVs that can freeze frame, and that probably means there are displays out there that can do it, they are probably rare and not the type you plug into a junker you are trying to resurrect.  

    It sounds like you’ve fried yet another display card.  That’s 3 in one machine.  That’s probably a sign that its time to seek closure and move on.

    Might I suggest using it as ballast in a home made trebuchet?  And if you already have said implement, use it as ammunition.

    [ps: I assume you’ve already tried the industry standard beating on the machine and just neglected to mention it.  If you haven’t, might I recommend SLOWLY working your way up from bare hand to the 5 lb sledge, lest you "accidently" overshoot the required force?]

  34. Thom says:

    Thrice I’ve faced seemingly random and bizarre problems with computers and each time the source of the problem was something simple yet unexpected.  The first was a loose solder joint on the motherboard’s crytal oscillator.   The second was due to a spare screw which was wedged between the motherboard and the case.  The third was due to a slightly off position mounting post for the motherboard.

    In each case I’d experience bizarre symptoms, unexplained lockups, or wierd temporary glitches in system behavior.  These seemed random as could be sometimes coming minutes apart and sometimes days or weeks.  I think the timing of the make/break electrical contact all came down to bizarre confluences of temperature change, vibration, computer position, phases of the moon, and sheer mischievousness of the computer itself.  

    The first problem was most prevalent and after all else failed I’d pulled the motherboard to look for loose solder joints.  The other two were so unpredictable that I blamed Windows95/98 for most of the flakiness and only discovered the source of the problem when I removed those motherboards for other reasons.  The spare screw fell out in the one case and I said hmmmm.  In the other the problem was worse when I reassembled the computer so I went looking for the source and discovered the misaligned post.

  35. Just enough power to keep the VRAM alive.

  36. David Moisan says:

    In my experience working with perhaps a hundred different boards, all KB connectors tolerate hotplugging.  The KB may or may not function after being plugged or unplugged, but I have yet to kill a keyboard or its port.

    I have killed one PS/2 port, but that’s been at least 8 years ago.  That doesn’t seem to happen with good motherboards, since I have plugged and unplugged (inadvertently) and not fried a port since.

    The PS/2 *drivers* are not hotpluggable, unlike USB ports.

  37. Kemp says:

    "Yes, the card had tilted slightly in its slot, such that at least one pin was disconnected.  The card still had power to maintain refresh until shutdown and amazingly didn’t cause a bluescreen or whatever."

    I’d imagine that if the screen wasn’t being updated then the machine *could* bluescreen and you’d never know… =P

    David Moisan:

    It’s not very common to fry a motherboard or keyboard by hot-plugging, but it has happened. The drivers don’t mind you hot-plugging (depending on circumstances) but the hardware spec itself isn’t designed with it in mind. Ironically, the cheaper boards used to be more resistant due to being designed with looser tolerances than the more expensive ones. Much like expensive locks being more vulnerable to certain picking techniques due to having more exact tolerances.

Comments are closed.