On 27 Feb 2001, at 15:33, Jeramy Ross wrote:
> Electron boil-over is only one aspect to mention in this problem. While
> older designs are more resistant to that aspect of overheating, older
> designs tend to have just as much of a problem to failure due to excessive
> heat as newer ones. One needs to look at all aspects of damage due to
> overheating.
Physically everything starts with electron boiling. The second order
effect is that semiconductors tend to have a negative coefficient of
resistance i.e. resistance falls with temperature, leading to more
current being drawn, leading to thermal runaway. In practise this
tends not lead to permanent damage because of the limited current
supply capacity and because the negative coefficient of resistance of
the semiconductor parts of the circuit are offset by the positive
coefficient of resistance of the conductors used to wire the active
components together inside the chip.
And, yes, however fat you make the components, they will stop working
properly once they get hot enough. Eventually the substrate will
melt.
>
> [Hot voltage regs] aren't much to worry about, but that is a fine line that you are
> walking here. They either work or cook themselves to death. The fact that
> the processor is OC'd may be playing a part here, and there might be more of
> a strain on these regulators than the motherboard designer planned for which
> may cause a problem in the future, AND if they do fail they tend to take out
> quite a few other components on your motherboard with them..INCLUDING your
> processor. Their health is a very important aspect.
If the M/B designer has done even a halfway decent job, the voltage
regulator will present a short circuit to the PSU if it fails
catastrophically. So the PSU will shut down & protect the expensive
components fitted to the M/B.
It's certainly true that overrunning components tends to reduce their
life. This should be part of your consideration when you decide to
overclock. Having said that, you could (now) replace a whole Cyrix
233 base unit for around $100 - so the risk is not that great.
My point is simply that a marginally overheating CPU can lead to
processing errors, whereas a marginally overheating voltage regulator
won't.
>
> The processor at issue here is a older design, BUT new enough to merit the
> conductive grease or tape. ESP. when the processor is OC'd.
If you were building the system NOW, I agree. If it's a question of
disturbing a system with a two-year record of reliable operation,
you're taking a bigger risk disassembling the system to add the
thermal transfer compound than you are by leaving it alone.
>
> > Don't worry about it. Start worrying
> > if and only if you start seeing symptoms which could possibly be
> > attributed to overheating.
>
> Well this is one way of looking at it. Not to get harsh, but most people I
> know prefer to keep things in good working order and PREVENT such
> problems...not simply wait until they occur. Always better to be safe than
> sorry... and when it comes to ones computer...the sorry bit can get to be
> quite expensive.
Oh, sure. But don't forget there is a risk every time you remove the
case cover and start fiddling with the innards. The time to think
about "preventative maintainance" is when you're initially assembling
the system - which automatically reminds you of the expense, as
you'll either just have paid for the components or had them added to
your credit card balance.
And I _certainly_ agree with the point about case fans - but don't
forget that you may need to provide an exit for the hot exhaust air
as well as an inlet for cool inlet air.
Regards
Brian Beesley
_________________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers