• Jens (unregistered)

    IMHO, TRWTF is that you could fit a power connecter into a sound connecter in the first place. And secondly, since it didn't burn out the card/computer there must have been some kind of "buffering" or surge protection or what not in place. QED.

  • Radu (unregistered)

    Yes, actually this will not hurt the board, and I will explain it to you why: Both power and audio connector has ground wires on the pins in the middle (second and third pin, counting from either way), so this one is no problem. A potential problem will be the connection of +12V and +5V lines on Left and Right audio input pins, but these are usually "buffered" with capacitors which won't let the DC voltage going into the soundcard input. So the board will not go out in smoke.

    By the way, did you know that smoke is build into computer boards? And when the smoke escapes the boards will stop working? :)

  • Vollhorst (unregistered) in reply to CrazyBomber
    CrazyBomber:
    3: I didn't witness this one, for it was told to me by a friend. When his PC fried, he took it to the local shop where he bought it. There, the tech opened the PC, took out the graphics card and pointed it towards the light, looking at it like it was a $100 bill, trying to find out if it was fake or not. "What are you doing", the custumer asked. "I'm checking to see if the RAM chips are ok" 0_o
    That totally makes sense.

    I once helped a friend to build his computer. When we started the machine everything went weired on the screen. It started to smell strange too.

    After a small check we realized what was the problem: The memory chips on the graphic card were mounted the wrong direction (you can the small notches which show in which direction they belong) and they started heating like wild which in turn melted the glue which was a hell of a smell. But thanks to the melted glue we could get the memory out and put it back in correctly. No problem afterwards.

    But that was about 9 years ago...

  • Jens (unregistered) in reply to Radu

    Ah excellent clarification. Funny with the smoke, but an excellent fail-safe i guess. Alternatives would be visible color changes in the board.

  • this webcomic is a wtf (unregistered) in reply to EE
    EE:
    Mike D.:
    This is a case of a power supply being connected to analog inputs. The buffers are likely unity-gain op amps, or a similar analog conditioning circuit that cleans up the signal for the analog-digital converters on the sound card.

    What the tech got wrong was the fact that sound cards can't stand 12 V on their inputs, and might not be able to deal with 5 V either.

    Line level audio inputs are usually AC-coupled and have a ~10 k resistor in series, good luck trying to get through that with 12 V DC.

    Buffers are a general concept also in EE, for example radio frequency oscillators that need to be very stable have a "buffer" as a following amplifier stage that is meant more to isolate the oscillator from following stages and loads than it is to actually amplify.

    Also, if you have a 10k resistor in series with the input that ultimately leads to a field effect transistor gate, the relatively low ratio of resistance of the 10k to the gate of the transistor (could be 100k on up) will effectively allow most of that 12 volts to be applied to the transistor gate. With mosfets (metal oxide semiconductor field effect transistors, this gate impedence is extremely high at lower frequencies (and can be blown very easily by static electricity).

    On the other hand, if there is a capacitor in series also, the resistor has no effect at dc as the capacitor by definition blocks it (although it will pass any changes in voltage to a varying extent).

  • test (unregistered)

    The Real WTF is english language. In russian technical terminology there is not problems with different meaning of word "buffer".

  • Kuba (unregistered) in reply to dams
    dams:
    Actually, what is funny is that you guys with your purely SW background do not understand that this is just HW jargon. Even if the sentence is quite lame, it is still correct:

    In HW, "a buffered I/O" means you use a "buffer" which is generally an open collector "YES gate" which provides a protection against short circuit thanks to the pull-up resistor which prevents the ground from being directly connected to the Vcc.

    First of all, the "buffered I/O" you're talking about went out of style with bipolar (TTL) logic. Most CMOS drivers I know about are inherently current-limited. Take any old CMOS gate, drive its inputs, and short its output via an ammeter to the ground or to the supply line. You'll get a few tens of milliamps flowing at best, and something may get warm, but that's about it. I routinely use this current-limiting to drive LEDs directly from CMOS outputs without any current-limiting resistors. Works like a charm, where the output drivers were designed to take such abuse of course (in most gates and MCUs that's the case).

    The biggest WTF in your comment, though, is that you're talking about buffering logic signals, which doesn't even apply in this case. The poor tech was talking about the sound card inputs (analog inputs!) being buffered. This wouldn't have mattered much, unless the buffer circuits were designed to take 12V on their inputs (that's the yellow wire on the power connector). Some AC-coupled soundcard inputs will take such abuse, some won't.

    The real reason why "the card and the slot were fried" is wholly different though: the 5V/12V supply could have coupled, via some protection diodes, to the local 3.3V bus on the card. Alternatively, the 5V line, which will happily provide tens of amps worth of short-circuit current, got shorted to the ground on the sound card. It's easy enough to look at what could have likely been shorted to what else, by examining the photo, which I'll leave to the dear reader.

    Kuba

  • Brady Kelly (unregistered) in reply to dams
    dams:
    The computer science does not revolve around WebDev, I would even say that it is not part of Computer Science. Don't forget that you use something you don't really understand...

    Says who?

  • Motty (unregistered) in reply to dams
    dams:
    Actually, what is funny is that you guys with your purely SW background do not understand that this is just HW jargon. Even if the sentence is quite lame, it is still correct:

    dams rocks, purely because he didn't start his post with "The real WTF is..."

  • Motty (unregistered) in reply to Zylon
    Zylon:
    CrazyBomber:
    1: When a friend of mine's old PC went MIA, with some usual beeping sounds, I figured it would be one of the components who had just died.
    Unless your friend's computer was stolen, I can assure you that it didn't go MIA. KIA, maybe. MIA, no.

    You're so pendantic

  • Kuba (unregistered) in reply to Merijn Vogel
    Merijn Vogel:
    The real WTF is the knowledge of electronics, the 'victim' of this article was right about the buffering; the inputs and outputs of an audiochannel are usually provided with a small capacitor (aka buffer) to correct for a possible DC-component. So inputting +5V or +12V DC into the input of a decent audiochannel (like the input of this soundcard) does not do any harm.

    Nothing more than a collision of jargons.

    Since I have a couple old PCs gathering dust in the closet, I've just pulled them out. They were all win 95/98 machines. On each one of them, I have started the recorder and fed some music from the iPod -- it worked. That's the baseline.

    Now I've got a 12V, 5A bench supply and hooked that up between each L/R line input and ground in turn. Out of 5 machines, 1 won't boot anymore and judging by the smell must have emitted magic smoke, two didn't smoke but won't pass POST (they beep), one works without sound input nor output (the sound card gets detected, though), and yet another one shows no signs of problems. On all of them, the motherboards are different, and soundcards (if no on-board sound) are different too.

  • nat42 (unregistered) in reply to Motty
    Motty:
    Zylon:
    CrazyBomber:
    1: When a friend of mine's old PC went MIA, with some usual beeping sounds, I figured it would be one of the components who had just died.
    Unless your friend's computer was stolen, I can assure you that it didn't go MIA. KIA, maybe. MIA, no.

    You're so pendantic

    I had a vision of the dieing PC crawling into the woods to die alone - beeping all the way, and right in the middle of it's work too. How sad for CrazyBomber's friend

  • Karl (unregistered) in reply to OperatorBastardusInfernalis

    I wouldn't have believed it if I hadn't been standing right there. There was a crater left in the chip that was about (as I recall) 1/4" long and 1/8" wide, all of which burned up in the air (nothing was left anywhere of whatever had been ejected)

  • FIA (unregistered) in reply to Opie
    Opie:
    ...

    Ever held on to a telephone wire pair while it's ringing? You might be shocked at the amount of fun you'll have performing that experiment...

    ...

    I was once trying to fix a very long modem extension cable and used my teeth to strip the wires, whilst it was still plugged in to the wall phone socket..... didn't do it again.... :)

  • Maarten (unregistered) in reply to dams
    dams:
    Actually, what is funny is that you guys with your purely SW background do not understand that this is just HW jargon. Even if the sentence is quite lame, it is still correct:

    In HW, "a buffered I/O" means you use a "buffer" which is generally an open collector "YES gate" which provides a protection against short circuit thanks to the pull-up resistor which prevents the ground from being directly connected to the Vcc. This is not completely correct as the protection can be done using serial resistors (most usual) are optocouplers, but it is still a valid sentence from a HW point of view.

    The computer science does not revolve around WebDev, I would even say that it is not part of Computer Science. Don't forget that you use something you don't really understand...

    Your story doesn't quite add up either.

    An open collector buffer will be blown to pieces when connected straight to +5 or +12 and then driven low. The only luck in this case, was that the inputs of the sound card were actually connected through capacitors or high value resistors, which is common practice in analog circuits and has nothing to do with buffering whatsoever.

  • ape (unregistered) in reply to dams

    I'd have to agree, we just went over that the other day in my microprocessor class. our Hcs12's have "buffered" inputs, but not in the computer science sense of the word. The point is so that it wont break if some idiot plugs in a 12v (it only supporst 5v). Hooray for idiot proofing test boards.

  • (cs) in reply to dams
    dams:
    Actually, what is funny is that you guys with your purely SW background do not understand that this is just HW jargon. Even if the sentence is quite lame, it is still correct:

    In HW, "a buffered I/O" means you use a "buffer" which is generally an open collector "YES gate" which provides a protection against short circuit thanks to the pull-up resistor which prevents the ground from being directly connected to the Vcc. This is not completely correct as the protection can be done using serial resistors (most usual) are optocouplers, but it is still a valid sentence from a HW point of view.

    The computer science does not revolve around WebDev, I would even say that it is not part of Computer Science. Don't forget that you use something you don't really understand...

    Fantastic. I can hardly count the WTFs in your post for falling over laughing at you, but let me try anyway:

    WTF#1: You don't know what buffering is. Buffering is not any kind of overvoltage protection. Buffering is how you connect mismatched impedances without suffering loss of signal integrity and SNR.

    WTF#2: There's no such thing as a "YES" gate, but if there was, it would be something that outputs logic 1 in response to any input at all, or "the positive power rail", as us engineers like to call it.

    WTF#3: Protecting an output against being shorted to ground is hardly relevant to the case of plugging the positive supply into an input.

    WTF#4: You describe the output stage of a tristateable digital logic gate, but that is an analogue circuit we're discussing, and this design is not relevant to the situation.

    WTF#5: You conflate "buffering" and "isolation" (which are completely separate concepts each with their own entirely separate methods, purposes, and technologies) while understanding neither.

    WTF#6: You assume that anyone could be so stupid as to think "buffering" in this context referred to software buffers, despite all the earlier posts (such as e.g, WhiskeyJack and pitchingchris) making it perfectly clear that nobody was under that false impression. How smug, patronising and just plain wrong is that?

    WTF#7: You presume to lecture people for talking about stuff they don't understand, but in the process demonstrate that you are pretty ignorant of the subject yourself. A PKB of the first water.

    ... Enough. There's probably more, but I'm laughing so hard I have to go change my pants.

  • Poltras (unregistered) in reply to OperatorBastardusInfernalis
    OperatorBastardusInfernalis:
    > Then was nearly hit in the face (at 6' 4" that's no mean distance > from the bench surface) by a fireball shooting out of the CMOS chip > they they were trying to connect.

    Am I missing... an eyebrow?

    Seriously, CMOS chips can do that much?

    Well a 5th level CMOS chip can cast one fireball per day (assuming he studied it first, which seems to be coherent with the case at hand).

  • hachu (unregistered) in reply to Karl
    Karl:
    Then was nearly hit in the face (at 6' 4" that's no mean distance from the bench surface) by a fireball shooting out of the CMOS chip they they were trying to connect.

    A about 7 years ago, I was in a lab group with 2 other students. We were all fairly smart people so I assumed it'd be fine if we each simultaneously set up the breadboard lab in different areas so we could finish as fast as possible. (parallelism ftw?)

    We managed to get everything wired in quickly but as soon as power was turned on, the opamp chip squealed for a second and then blew apart in a puff of smoke.

    Turns out, the guy who went to go get the power supply cables to wire to the breadboard didn't know the red/black positive/negative convention and wired the supply backwards.

    The teaching assistant eventually walked by, stopped in our area, sniffed, and asked, "Um, is something burning?" to be responded with "I don't smell anything, do you? Nope. Nothing." :P

  • hachu (unregistered) in reply to Kuba
    Kuba:
    Since I have a couple old PCs gathering dust in the closet, I've just pulled them out. They were all win 95/98 machines. On each one of them, I have started the recorder and fed some music from the iPod -- it worked. That's the baseline.

    Now I've got a 12V, 5A bench supply and hooked that up between each L/R line input and ground in turn. Out of 5 machines, 1 won't boot anymore and judging by the smell must have emitted magic smoke, two didn't smoke but won't pass POST (they beep), one works without sound input nor output (the sound card gets detected, though), and yet another one shows no signs of problems. On all of them, the motherboards are different, and soundcards (if no on-board sound) are different too.

    That's okay, your one working machine probably felt like the black sheep when the 4 Dells were around.

  • anonymous (unregistered)

    That's what's wrong with my sound. I was thinking it was a driver

  • biziclop (unregistered) in reply to dams
    dams:
    Actually, what is funny is that you guys with your purely SW background do not understand that this is just HW jargon. Even if the sentence is quite lame, it is still correct:

    In HW, "a buffered I/O" means you use a "buffer" which is generally an open collector "YES gate" which provides a protection against short circuit thanks to the pull-up resistor which prevents the ground from being directly connected to the Vcc. This is not completely correct as the protection can be done using serial resistors (most usual) are optocouplers, but it is still a valid sentence from a HW point of view.

    The computer science does not revolve around WebDev, I would even say that it is not part of Computer Science. Don't forget that you use something you don't really understand...

    Thanks, for a moment I thought I missed something in the story.

  • Chris (unregistered) in reply to dams

    If you're going to get techy about it, in an audio card the protection wouldnt be with resistors, they are more likely to use capacitors to block DC. The only time I ever heard someone talk about electronics being buffered was when a guy in a shop sold me a memory chip when I was younger. He had it in his hand and I asked him whether that would damage it because it was CMOS. He told me it was okay because it was buffered. - He lied.

  • (cs) in reply to hachu
    hachu:
    We managed to get everything wired in quickly but as soon as power was turned on, the opamp chip squealed for a second and then blew apart in a puff of smoke.
    Heh, I managed to do that to a 741 once, it was just carelessness, I wasn't paying enough attention when I wired it up.

    Still, 741s are like what, ten cents these days? Cheap enough to blow up for fun!

    Plugging an LED across the full mains voltage is an amusing way to get showered in little fragments of plastic shrapnel, too. That one wasn't an accident. I meant to do it.

  • Alex (unregistered) in reply to dams
    dams:
    Actually, what is funny is that you guys with your purely SW background do not understand that this is just HW jargon. Even if the sentence is quite lame, it is still correct:

    In HW, "a buffered I/O" means you use a "buffer" which is generally an open collector "YES gate" which provides a protection against short circuit thanks to the pull-up resistor which prevents the ground from being directly connected to the Vcc. This is not completely correct as the protection can be done using serial resistors (most usual) are optocouplers, but it is still a valid sentence from a HW point of view.

    The computer science does not revolve around WebDev, I would even say that it is not part of Computer Science. Don't forget that you use something you don't really understand...

    I couldn't explain it better... The comment "They're buffered" makes perfect sense from HW point of view.

    If the tech didn't work there for long, he probably found a better job, since he knew better than "jjeff1" what was going on.

  • Billy (unregistered)

    After installing a new video card, my CPU temperature started to raise a lot. I couldn't play CS Source for more than 2 minutes before the PC shut down.

    I was ready to buy a new cooler, but first decided to give the old one a last shot and would clean it.

    For my surprise, when I removed the cooler, found a plastic label stuck on the top of the processor. The guy who assembled the machine found a clever way to advertise his company by nearly burning the customer's CPU.

    Needless to say I'm using the same cooler until today.

  • (cs) in reply to Alex
    Alex:
    dams:
    Actually, what is funny is that you guys with your purely SW background do not understand that this is just HW jargon. Even if the sentence is quite lame, it is still correct:

    In HW, "a buffered I/O" means you use a "buffer" which is generally an open collector "YES gate" which provides a protection against short circuit thanks to the pull-up resistor which prevents the ground from being directly connected to the Vcc. This is not completely correct as the protection can be done using serial resistors (most usual) are optocouplers, but it is still a valid sentence from a HW point of view.

    The computer science does not revolve around WebDev, I would even say that it is not part of Computer Science. Don't forget that you use something you don't really understand...

    I couldn't explain it better...

    Well, you could if you knew what you were talking about, but I guess you don't know electronics any better than "dams". So, let me explain it better for you: NOBODY thought the original post was talking about the CS meaning of buffering, but TWO ARROGANT IDIOTS thought that everyone else actually WAS that stupid.

    Alex:
    The comment "They're buffered" makes perfect sense from HW point of view.
    Really? You think when you shove 5 or 12 volts into a line-level input it's really important to properly couple the mismatched impedances, in order to get good signal quality?
  • Dr Ben (unregistered) in reply to dams

    I hate sticking my oar in, but the juxtaposition of a bombastic opening paragraph to a myopic understanding of the term 'buffer' forces me to rush to the defence of some unfortunate programmers.

    The buffer to which you refer is not one you'd use on an audio circuit, unless you wanted to sound like Cradle of Filth. In the analogue world, 'buffer' refers to a unity gain op-amp.

    The aim is similar: to follow and regenerate the input voltage.

    And actually, the guy was almost right. Generally you'll be OK under these circumstances, because all but the most frugal sound cards are fairly idiot-proof. The first thing the 12V rail would encounter would be a ~1k resistor, so there'd be a little gentle warming but no tragedy.

    That said, I've seen plenty of people blow up iPods by connecting them to the microphone input of a studio desk and inadvertantly sending 48V up them -- outputs are considerably less resilient. [/electronics monograph]

  • Darkjedi (unregistered) in reply to CrazyBomber
    CrazyBomber:
    I've got a few stories about PC "techs" too :) 3: I didn't witness this one, for it was told to me by a friend. When his PC fried, he took it to the local shop where he bought it. There, the tech opened the PC, took out the graphics card and pointed it towards the light, looking at it like it was a $100 bill, trying to find out if it was fake or not. "What are you doing", the custumer asked. "I'm checking to see if the RAM chips are ok" 0_o

    Actually, burnt out chips can show scuffing/blemishes on the surface. I've identified several bad RAM sticks this way. I'm sure it would work for chips on video cards too.

  • Antonio (unregistered)

    Someone I knew was trying to build his own computer but the motherboards he kept buying would instantly start smoking. After the second attempt he concluded the power supply was the culprit and I finally took a look for myself.

    He was bolting the motherboard directly to the metal case. As in full contact with the entire back of the motherboard. He didn't know what those standoffs were for that came with it.

  • (cs) in reply to Dr Ben
    Dr Ben:
    I hate sticking my oar in, but the juxtaposition of a bombastic opening paragraph to a myopic understanding of the term 'buffer' forces me to rush to the defence of some unfortunate programmers.

    The buffer to which you refer is not one you'd use on an audio circuit, unless you wanted to sound like Cradle of Filth. In the analogue world, 'buffer' refers to a unity gain op-amp.

    The aim is similar: to follow and regenerate the input voltage.

    I was with you up to this point. Then ...
    Dr Ben:
    And actually, the guy was almost right. Generally you'll be OK under these circumstances, because all but the most frugal sound cards are fairly idiot-proof. The first thing the 12V rail would encounter would be a ~1k resistor, so there'd be a little gentle warming but no tragedy.
    Umm, excuse me, but what you've just said demonstrates that the guy was not "almost right" but was in fact talking garbage. He didn't say "It's ok - it's buffered, but that's beside the point, because there's a 1k resistor on the input to the buffer that'll protect it". He said it was ok /because/ it was buffered, which is a non-sequitur.

    And, you know, you guys still have to explain how come the card was dead, because according to you it should still be in perfect working order, since you feel that plugging the power into the line inputs is no problem. Do the maths: 12v minus one junction drop (0.7v for silicon) equals 11.3 volts over 1k resistance is 11.3mA flowing into the input of that op-amp. Input bias current is supposed to be 500nA absolute max for something like a 741 or 380, so how is dumping three orders of magnitude greater in there supposed to work?

  • bobo1on1 (unregistered) in reply to dams
    dams:
    Actually, what is funny is that you guys with your purely SW background do not understand that this is just HW jargon. Even if the sentence is quite lame, it is still correct:

    In HW, "a buffered I/O" means you use a "buffer" which is generally an open collector "YES gate" which provides a protection against short circuit thanks to the pull-up resistor which prevents the ground from being directly connected to the Vcc. This is not completely correct as the protection can be done using serial resistors (most usual) are optocouplers, but it is still a valid sentence from a HW point of view.

    The computer science does not revolve around WebDev, I would even say that it is not part of Computer Science. Don't forget that you use something you don't really understand...

    You're talking about a digital buffer but this is an analog input!

    Since it's an audio input, there is probably a capacitor in series with it, so the soundcard only gets a short peak of 5 or 12 volts at it's input.

    A buffered input means the audio signal is fed into an opamp which amplifies the signal, an opamp does not break when it gets 12 volts at it's input.

    I doubt however that the soundcard has a buffered input, this would be a stupid thing to do since the output from the cdrom player is also buffered, it also costs more and manufacturers usually go for the cheapest options, even when it saves 1 cent per product. The soundcard does have buffered outputs however.

    The input of the soundcard probably has a series resistor which acts as a current limiter, protecting the soundcard from people who put the wrong connector on the wrong header.

    It is completely normal that this doesn't break the soundcard, I would be more amazed if it would break. I'd like to defend the electrical engineering tech, he actually knew what he was talking about, the computer techs DID NOT! They expected the soundcard to release the magic smoke just because the wrong connector was put on the wrong header, they didn't have a clue what was going on inside the machine.

  • bobo1on1 (unregistered)

    Oh dear, I misread the article, the soundcard did break. A resistor and two zener diodes would have prevented this, but that would have cost a whopping 10 cents!

  • (cs) in reply to bobo1on1
    bobo1on1:
    Oh dear, I misread the article, the soundcard did break. A resistor and two zener diodes would have prevented this, but that would have cost a whopping 10 cents!

    Yeh, but there's a reason why they don't do that with cheap commodity sound cards.

    Invoice for repair of Sound Card.

    1 x replacement Resistor ... 5c 2 x replacement Zener diode ... 5c 1hr. engineer's time to desolder old parts and replace ... $120

    . . . or . . .
    Invoice for repair of Sound Card.

    Cost of replacement card (new) ... $15

    You decide!
  • A.K.T. (unregistered) in reply to CrazyBomber
    CrazyBomber:
    Buy a stick of 512MB RAM... that should make it compress better.
    Why not? More available memory does allow to compress better for some algorithms. The advice may actually work for some program :)
  • sciron (unregistered) in reply to dams

    Hot dog! Someone else that realizes most graduating with a 'computer degree' these days have no flippin' clue what really happens under the hood. Enter the people we all love to hate: Day Coders.

  • Chris (unregistered) in reply to CrazyBomber

    RE #3:

    I work for an industrial monitoring equipment manufacturer, and we have a repair center at my location. I was hired on as tech support, and worked in the repair center for a short time as training. When some chips burn out, there is a 'dull' looking spot in the center of the chip. It can be hard to see without reflecting some light off the chip. We used that as a quick way to tell if any of the chips on a board needed to be replaced. I'm sure that all chips don't do this, but it can be possible to see if a chip has failed just by looking at it.

  • Ken (unregistered) in reply to dams

    this is an audio output... it would use an op-amp and a DC bias block capacitor as a buffer not an optocoupler, open-collector buffer (the 74LS06, a hex OC buffer, is what you're talking about), or resistors. Also, "serial resistors"? I think you mean SERIES resistors. Get your EE right if you're going to pick on other people for it.

    Also, on the subject of EE, I can believe the card being dead - but the slot? come on, ISA is a multi-drop bus - all the slots are connected exactly in parallel. Kill one slot, kill them all, unless the connector was physically damaged by this mistake which I can't believe would ever happen. There is no way an ISA-based system would even boot if the bus was dead, as all the onboard I/O was generally hung off the same bus lines. I'm not calling BS on the whole article, but there is definitely some exaggeration here.

  • PietSkiet (unregistered) in reply to dams
    dams:
    Actually, what is funny is that you guys with your purely SW background do not understand that this is just HW jargon. Even if the sentence is quite lame, it is still correct:

    In HW, "a buffered I/O" means you use a "buffer" which is generally an open collector "YES gate" which provides a protection against short circuit thanks to the pull-up resistor which prevents the ground from being directly connected to the Vcc. This is not completely correct as the protection can be done using serial resistors (most usual) are optocouplers, but it is still a valid sentence from a HW point of view.

    The computer science does not revolve around WebDev, I would even say that it is not part of Computer Science. Don't forget that you use something you don't really understand...

    WTF?!

  • Jeff (unregistered)

    When I was 11 or so I tried fixing a computer with my cousins. One of them (not me) did this, and within minutes the room was filled with black smoke.

    On the plus side, they got a brand spanking new computer with an All-In-Wonder Radeon 7500. Oh, how I envied them.

  • Rasmus (unregistered)

    Well, the sound-connector is built like that to avoid problems when people do this... the two middle pins are ground, just like the power connector. The other two are the left and right audio lines, and it's no problem connecting 5V and 12V to those pins, might result in somewhat surprising sound though if you listen to the cd :)

    They are indeed buffered with about 75k input impedance. Not the same as 75kB buffer size for you software people :)

  • Some EE Guy (unregistered) in reply to DaveK
    DaveK:
    Do the maths: 12v minus one junction drop (0.7v for silicon) equals 11.3 volts over 1k resistance is 11.3mA flowing into the input of that op-amp.

    What?

    A 1k resistance is negligible compared to the input impedance of an op amp. You'd struggle to get 1 microamp flowing through the input terminals.

    DaveK:
    Input bias current is supposed to be 500nA absolute max for something like a 741 or 380, so how is dumping three orders of magnitude greater in there supposed to work?

    Input bias currents are not what you think they are.

    A 741 will happily accept +-12V inputs with no damage.

  • Axel (unregistered)

    OK, I'm very late to this party, but maybe I can clear things up a bit. 12V won't damage an op-amp's input--IF the op amp has a supply voltage that is greater than or equal to 12V. Sound cards run at low voltage: usually 5V. Assuming a d.c. path (no blocking capacitor) from the CDROM connector to an op-amp buffer, the smoke will definitely leak out.

    The input structure of almost all op-amps creates the equivalent of a pair of diodes from each input to the power rails. Take the input past one or the other rail, and the diode conducts. In our case, it would short the 12V supply to the 5V rail, until the high current blows the IC. That current surge may have fried a pin on the ISA connector, taking it out also.

    A pair of series resistors and external diode clamps on the sound card could have saved the card/motherboard, but oh, well.

    BTW, a 75k input inpefance doesn't mean there is a 75 k series resistor. It means there is (at audio frequencies) the equivalent of 75k ohms to ground.

  • Axel (unregistered)

    Sorry. "Impedance"

Leave a comment on “They're buffered”

Log In or post as a guest

Replying to comment #:

Log Out

« Return to Article