If it doesn't burn right away, just try harder



  • One Saturday morning I looked at my PC's semi-transparent case and concluded that cables inside made a pretty big mess. So I decided to rearrange the cables a bit.

    I removed all cables going from the power supply, and then started putting them back in a more orderly fashion. One thing I overlooked was how do I hook the cables back to the power supply.

    Now, that was a big, shiny Chieftec power supply unit with modular design — peripheral cables could be strapped on and off. There were six sockets in it, and they were color coded, four of them black and two red. By that time, I forgot what exactly red and black meant, but as they were otherwise identical, I plugged the cable which powered my three hard drives, into a red one. Unfortunately, there was no one there to whack me with a cluebat, so I plugged in the rest of the cables, and powered the monster on.

    It didn't turn on. LEDs on the motherboard blinked once, and it went off.

    So I tried harder.

    The third time it gave up and powered on. Instantly, there was a cracking sound and smoke went from one of the hard drives. Unfortunately, to find out what it could be, I powered the thing on once again.

    Another hard drive went in flames.

    “Oh crap, it must be serious,” thought I.

    I remembered that red sockets on the PSU had “PCI-E” engraved next to them. A thought wandered through my mind, so I googled up the voltages. Of course, any decent piece of electronics expecting 5V will go up in smoke if you give it full 12V.

    To sum it up: one cable fuckup. A new power supply. A new motherboard. A new hard drive.
    I really should not play roulette if I cannot tell red from black.
    A software engineer shouldn't mess with hardware.
    And a small piece of hate towards Chieftec, for making voltage-incompatible sockets plug-compatible.



  • I believe the correct terms are electrically incompatible but mechanically compatible...

    Burn, hope you didn't lose anything.



  • Jeez. We make $80,000 motion simulators and have every data and power connector in them (and it's a lot of them) physically incompatible, and the ribbon connector pinouts are set up so one pin offset either way, or a 180 degree rotation, won't cause problems.

    So, yeah, big WTF for Chieftec, but a pretty big WTF for you, too for plugging the thing back in the second time, let alone a third. Seriously?!



  • @PeriSoft said:

    a pretty big WTF for you, too for plugging the thing back in the second time, let alone a third. Seriously?!

    Seriously, it was plugged in once; it's the power button which I fiddled with for the second and the third time.



  • @nexekho said:

    Burn, hope you didn't lose anything.

    Well, not quite so...

    • A 250 GB Maxtor SATA drive
    • A 250 GB Seagate PATA drive
    • A 320 GB Seagate PATA drive
    • A MSI P45 Diamond motherboard — some of its SATA channels were screwed after that
    • A 700W Chieftec power supply unit which hasn't quite recovered from the experience and shut itself off for 4 seconds and back on every 0.5-1 hour of work

    Lucky me, two 1 TB Samsung drives were connected to the right socket and didn't go in a puff of smoke. Three dead drives were replaced with a single 2 TB one.

    TRWTF was that this happened right before I had to move to another city and additional expenses were very unwelcome.



  • @PeriSoft said:

    physically incompatible

    Ages ago I worked in a 1-hour photoshop. For you kids, this was back in the day of analog photography when you coulnd't see pictures immediately on the back of your camera but you were eager to see them asap.

    Anyway, these shops had big expensive machines to process the pictures, using three kinds of chemicals which had to be applied in a certain order. The shop manager decided to clean the machine, also cleaned the tubes that went from the tanks of the chemicals to the machine but made a tiny mistake reconnecting them, because they all three looked exactly the same. The effect was that the pics came out looking good, but they degenerated quickly over the next couple days.

    It therefor took us a couple days to notice that something was wrong, and then WEEKS to deal with the dissatisfied customers that came in one by one during the next weeks. Of course the ones that were so dissatisfied they didn't even come back but instead told all their friends were even worse. I remember thinking: how the hell can they make all three tubes look the same when a mistake ca be this costly?



  • That's just seriously bad design. I've owned two different modular power supplies (the first one was shitty enough that it started killing my ram and making my computer otherwise unstable), and the modular ends of the cables have always been physically compatible with only the port they're supposed to connect to.



  • I dialed into this topic solely for the title, expecting some special sordid suppository story, maybe having to do with 3.5" floppies or rabid gerbils.

    Entertaining, okay, just not dirty enough.



  • @Yo-Yo Mo said:

    I dialed into this topic solely for the title, expecting some special sordid suppository story, maybe having to do with 3.5" floppies or rabid gerbils.

    Entertaining, okay, just not dirty enough.

    If you want to have a bigger laugh, I'll tell you that I'm not making it up.

    You see, flipping the on/off switch and watching your HDDs and probably some small things on your mobo burn in flames, one by one, is definitely a WTF. But that's what happened.

    When I related the story to one of my colleagues, he told me a similar one. He happened to work at a company in Moscow, where there was a bunch of servers and a guy maintaining them. At that time, Molex supplied a whole batch of really WTFy peripheral cable adapters which had their 12V and 5V swapped. Unknowingly the company bought some of those, and the server guy did fireworks to three spare hard drives with such adapter. Instead of asking himself why the magic smoke goes out, that idiot just took next spare drive from the shelf, plugged it in and watched it go “Pfffzzztt!” I don't know what was done to him afterwards.

    To screw your own hardware is one thing; screwing your company's property is completely, completely different thing. I hope those drives weren't SCSI or SAS or other expensive acronym.



  • Worst I've ever done is remove a Voodoo II video card and install a Voodoo 3 card... while the computer was powered-on!

    Both cards survived, and the computer just needed to be hard-reset.

    (Cool story, bro.)



  • I had a roommate who was in the habit of hotswapping PCI cards in his desktop.  I don't care if it actually worked, I told him every time that he was crazy.

    Also, once I helped a friend build a relatively high-end computer, but some of the mounts on the motherboard weren't set right.  Hit the power button, nothing happened.  Fiddled around a bit, hit the power button, nothing happened.  Fiddled around again, hit the power button, and a couple of elements on the motherboard lit up in an intense orange flame. It was pretty.  Always check those motherboard mounts.

    I also had a power supply burst into flames after the caps exploded, but I guess those sorts of things just happen from time to time.



  • Several years ago, I bought a RAM upgrade for my PC. I pulled the plug, opened the case and began installing the first memory module. When I had it half inserted, my PC suddenly turned on. It was alive for merely a second, but that was long enough to fry my newest aquisition. Apparently "off" and "unplugged" do not imply "discharged" or "safe" any more these days.



  • Everyone looked at me funny for always powering down boxes to replace PS2 mice and keyboards. It was a habit I developed after watching a one-in-a-million event. I was helping a guy at a data center where the rack didn't have a KVM. We were using the data center's monitor/keyboard cart and jumping from machine to machine while troublshooting and there was a horrible flash when he plugged into the backend SQL server - the live, production SQL that ran a very important part of the system. I have no idea what he did and technician out on the warrantee repair (4 hours later) had no idea what he did but the motherboard never turned back on again.

    Now that USB is the norm, that will probably never happen again...

    Another "cool story, bro" moment



  •  Was this (OP) a case of "if at first you don't fail, try, try again"?



  • @fatbull said:

    Apparently "off" and "unplugged" do not imply "discharged" or "safe" any more these days.
    Caps in modern power supplies don't drain immediately if you power off the machine properly (that is, do a Start -> Shutdown and after it's off pull the plug). So if you do any hardware work, either just pull the plug or (better) hit the power button after unplugging the computer.



  • I had a housemate who once added a floppy drive to his computer (this was back in 2003). He plugged it in but didn't notice the power plug was off by one. You do know the middle two wires are short-circuit on the device with solder pads, right? On system turn-on the 5V wire glowed red hot, burnt the IDE cable and melted itself and the floppy drive's power socket, before he leapt to the wall socket to turn it off. Then he removed the floppy drive, cut out the burnt wires and replaced the IDE cable. The HDD had died but the PSU still worked fine, albeit without a floppy connector. No fuses! The motherboard survived too.



  • @Zemm said:

    I had a housemate who once added a floppy drive to his computer (this was back in 2003). He plugged it in but didn't notice the power plug was off by one.
    I managed to do this with RadeOn 9500 (which had a floppy power connector). One of the resistors glowed orange and let out some magic smoke, but after it was replaced, the card worked fine.



  • @ender said:

    @fatbull said:
    Apparently "off" and "unplugged" do not imply "discharged" or "safe" any more these days.
    Caps in modern power supplies don't drain immediately if you power off the machine properly (that is, do a Start -> Shutdown and after it's off pull the plug). So if you do any hardware work, either just pull the plug or (better) hit the power button after unplugging the computer.

    Most modern computers have a little green LED somewhere on the motherboard. When it goes out, then it's safe to do stuffs to it. If it's glowing still, it's not.



  • @blakeyrat said:

    When it goes out, then it's safe to do stuffs to it. If it's glowing still, it's not.
     

    Keyboard indicators and sometimes a transparent optical mouse are also fine charge indicators.



  •  I use the green light on the ethernet port - though I suppose with wireless now, it might not be there. Then I'd just unplug and hit the power button.



  • Once upon a time, soundcards had a second duty of acting as CD-ROM controllers. At some point in time, this changed.

    I was building a computer sometime after that point, apparently.

    The soundcard had what looked like the familiar IDE bank of pins on it.... I fried two IDE cables before I convinced myself that I was wrong.



  • @dhromed said:

    Keyboard indicators and sometimes a transparent optical mouse are also fine charge indicators.
    While I've seen computers where keyboard LEDs were lit and sometimes the mouse diode was powered on while the computer was shut down, none of my own computers did this (even though I've been using keyboard power-on ever since I found out about it). Various LEDs on the motherboard are a good indicator though.



  • @RTapeLoadingError said:

     Was this (OP) a case of "if at first you don't fail, try, try again"?

    No. It was a case of “if at first you don't fail miserably enough, try again harder.”

    Mind you, MSI P45 Diamond is an exceptionally good mobo. It would recognize the imminence of the disaster and instantly shut down. It was just that 1ms is sometimes quite enough.

    After the experience, all it sacrificed to my stubbornness and ignorance was probably a single failing SATA channel.



  • @ender said:

    While I've seen computers where keyboard LEDs were lit and sometimes the mouse diode was powered on while the computer was shut down,
     

    I meant that, even after the motherboard has given the green light for a shutdown, and the fans stop spinning, it takes a good 5-10 seconds before the keyboard drops out, similar to many amplifiers' power LED.

    Interestingly, my previous computer did that for the mouse as well, but on this one, the mouse turns off immediately.

    YMMV.



  • @fatbull said:

    Apparently "off" and "unplugged" do not imply "discharged" or "safe" any more these days.

    Equating "off/unplugged" and "discharged/safe" is a dangerous assumption that has never been true since the beginnings of electronics. If a device has the propensity for storing energy then it should always be assumed to currently contain stored energy.

    This is a lesson you would have quickly learned if you have ever poked around inside the back of analogue TVs not so long ago.



  • @dhromed said:

    I meant that, even after the motherboard has given the green light for a shutdown, and the fans stop spinning, it takes a good 5-10 seconds before the keyboard drops out, similar to many amplifiers' power LED.
    Not on any of my computers - when it shuts down, the keyboard LEDs also shut down immediately.



  • I can confess to my own plug in screwup. About 10 years ago I was working on an industrial upgrade project and installing some heavy industrial grade I/O units that were powered by an external 24V DC supply. When they were wired in and switched on they didn't go up in smoke, but they never worked either. So I swapped in some more modules and they didn't work either. A more detailed look at the devices showed that I had killed them (and in later board level analysis you could see where the smoke was let out of a chip). What bit me was that the 24V DC PSU I was using to supply these modules had a 12VAC signal superimposed on it. This is fine for dumb DC lights and switches but it surprised me that the industrially rated modules I was using were killed by this given that they were installed in their expected environment. My failed assumption killed several modules with a list price of around $2.5k each.



  • My failed assumption killed several modules with a list price of around $2.5k each.
    Oh shit. I hope you weren't forced to pay for them.


  • We're not alone screwing up systems this way. Hospitals do it too, screwing living people by plugging them to the wrong tubes. Now that's the real WTF.



  •  @shimon said:

    ... any decent piece of electronics expecting 5V will go up in smoke if you give it full 12V.

    Is a contradiction, because if the electronics were decent they would have circuit protection and an accidental connection like that would have no effect. I'm not even talking a fuse here, I'm talking solid-state circuit protection, and it's not even that expensive.

    Sadly, this is the difference between quality components and "just make it good enough." Actually I'll amend that: the sad ting is that most of the time you can't even find the quality stuff becuase nobody manufactures it any more.



  • @too_many_usernames said:

    Sadly, this is the difference between quality components and "just make it good enough." Actually I'll amend that: the sad ting is that most of the time you can't even find the quality stuff becuase nobody manufactures it any more.

    Kids these days, with their skateboards and their hula-hoops! Why, in my day...



  • @too_many_usernames said:

    Sadly, this is the difference between quality components and "just make it good enough." Actually I'll amend that: the sad ting is that most of the time you can't even find the quality stuff becuase nobody manufactures it any more.

    I have a keyboard made in 1995. Works as new. I have disassembled it, cleaned it thoroughly and assembled it back, and it still works.

    Now there's a funny bit about it. It was made by Acer, would you believe it.

    I, for one, could not give a shit that Acer, being as crappy manufacturer as it is, could ever produce a piece of technology which doesn't fail after 15 years of constant banging.



  • In any decent piece of electronics, overvoltaging it might trip a PTC or blow a fuse, but more likely, you'd fry your regulator (probably a LM317 or LM7805, both very cheap and easy to replace) or a diode (still cheap). In either case you can fix it, with the hardest part being the removal of the idiot security screws.

    In the typical thing that you'd buy, it just drops however many volts it gets right into the processor, or motor, or whatever, and you're screwed.



  • @shimon said:

    a piece of technology which doesn't fail after 15 years of constant banging.
     

    I wonder what the life expectancy of sex toys is.



  • @dhromed said:

    [quote user="shimon"]a piece of technology which doesn't fail after 15 years of constant banging.

     

    I wonder what the life expectancy of sex toys is.

    [/quote]

    He's supposed to be using it to type.

    As for the OP... I'm a software engineer who regularly tinkers with computers - all of my own are self-assembled and I've probably built over a dozen full systems since I first got into PCs when I was 14 - and I have yet to cause anything to blow up due to plugging it in wrongly. (In fact the only blow-up-related incident was when my non-surge-protected ADSL modem got struck by lightning.) RTFM applies to software engineers as much as it does to mere mortals, y'know.

    Also, I propose that a mod changes the title of this thread to "If it doesn't burn, fail harder".



  • @dhromed said:

    I wonder what the life expectancy of sex toys is.
     

    That depends on how your mum uses 'em.


Log in to reply