OK. Do it the scientific way: Set it going and see what gets hot. You
have fairly sensitive infa red detectors on the ends of your fingers :-). This
fault description raises the possibility of signal degradation; you may have
something barely passing signals, then gradually going to the point where it
doesn't do so acceptably;
TTL voltage logic level limits for each line are:
Less than 0.8Volts - definitely low
Over 2 Volts - definitely high
0.8-2 Volts - indeterminate.
As you see, you want to stay away from the 0.8-2Volts area. What happens is
that at the very ends of the indeterminate area (0.9?V and 1.7-1.9V), things
switch. Everywhere else in the indeterminate area, you retain the last valid
value. Faulty hardware may drift in here, and stop switching when the lows
don't go low enough or low fast enough, or the highs don't go high enough, or
high fast enough. This is impractical to test, unless you have an oscilloscope,
and great patience. But you can find what's doing it, and replace that
hardware. Slow the system clock right down, and see if the problem persists -
that will limit it to the laptop if it does.
In this scenario, I'd ask - What's in the other pcmcia socket? Does the network
last longer with the other socket empty? It will be pretty obvious which chip
is driving the sockets - does that get uncomfortably hot? Is the network
putting extra load on the system and causing other heating? ELIMINATE things,
and you'll solve it
On Mon, 09 Apr 2001, Wolfgang Bornath wrote:
> On Mon, Apr 09, 2001 at 14:01 +0000, Declan Moriarty wrote:
> >
> > As a hardware guy, let me say that the main heat sensitive device is
> > the cpu. PCMCIA stuff uses no current worth talking of, hence no heat is
> > generated. Other heat sources are battery, and power supply. What should happen
> > is that the cpu is heat protected, and will slow down, then cut out in time to
> > save itself, which will mess up the interrupts on a thinkpad. If something else
> > is heating up big time, or showing heat sensitive behaviour, it's faulty :-{
>
> Generally speaking you are quite right. But I don't get any errors in
> any processes running, which would be the case if the cpu gets too
> hot or shuts down. It's just that the networking gets shaky (first 10
> or 20% packet loss then 40 to 90%). During that time all other
> processes run without probs.
> The other test I made was: When the time came and networking began to
> falter I removed the pcmcia card and stuck it back in after one hour.
> After that networking ran perfect for some time then faltered again.
> I removed the card, let it cool down, stuck it in again and
> everything worked again.
> So you say the card may be faulty. Fine. I'll try to turn it in at
> the shop but I have not much hope. THey'll put it into a notebook,
> test it and say that it's the fault of my notebook...
>
> > Try a session running from mains psu with the battery removed (It can
> > get hot if overcharged) and the case open; it's usually enough to remove the
> > keyboard. Make sure it is on a flat surface (not a bed!) and that the fan
> > works. See how it lasts then. If there's a metal plate covering the cpu, paint
> > it black. That alone may cure it! Matt Black heatsinks run cooler than any other
> > colour, strange as it may seem.
>
> As I said, everything else is running perfectly so it can't be the
> cpu getting too hot.
>
> Thanks anyway.
> wobo
> --
> GPG-Fingerprint: FE5A 0891 7027 8D1B 4E3F 73C1 AD9B D732 A698 82EE
> For Public Key mailto [EMAIL PROTECTED] with Subject: GPG-Request
> -------------------------------------------------------------------
> ISDN4LINUX-FAQ -- Deutsch: http://www.wolf-b.de/i4l/i4lfaq-de.html
--
Regards,
Declan Moriarty
Applied Researches - Ireland's Foremost Electronic Hardware Genius
A Slightly Serious(TM) Company
Without the optimist, the pessimist wouldn't know how happy he isn't.
