see below, plz.
Dick
----- Original Message -----
From: "J.C. Wren" <jcwren(a)jcwren.com
To: <cctalk(a)classiccmp.org
Sent: Friday, June 07, 2002 12:56 PM
Subject: RE: [CCTECH] Interesting tidbit on 6502
You're actually taking the "catch fire"
*literally*? Silicon doesn't burn
*that* well. Well, except for an Athlon (and if you haven't seen the
smoking Athlon video on
www.tomshardware.com, I recommend checking it out.
It's entertaining). However, much like 3 voltage EPROM socketted backwards,
the 6800 would get boiling hot for short time before it failed.
What normally catches fire is the PCB. However, lots of parts got plenty hot
in normal usage. Many wouldn't run on an extender board because of the added
capacitance and path lengths, but, often enough, it was because, out of the
normal cardcage forced-air cooling, the parts got so hot they no longer worked
within the limts to which the system was designed.
Many NMOS parts would get pretty hot,
particularly if you considered simply
hot-to-the-touch as being hot. The 2147's in my early fast 6502's were seldom
cool enough to touch without making a sound, both vocally and from the sizzle,
once they'd been powered up for more than 10 minutes and exercised. We even
had a routine in the memory diagnostic that was intended to heat the memories.
It certainly did that. There were four boards of these 2147's and you could
tell which one was being exercised simply by holding a finger over the board.
70 degrees Celsius (the upper limit of "industrial" temperature range) is
pretty hot!
> It wasn't like there was actual flames
jetting out of the thing. That'd be
> silly.
> I have a record of it somewhere, I'm
sure. Accessibility is another thing.
> Living on a houseboat, 90% of my stuff is in boxes in a warehouse. Since
> I've been inclined to try and fire up the IMSAI lately, I'll try to find the
> notes from the period.
> --John
> -----Original Message-----
> From: cctalk-admin(a)classiccmp.org [mailto:cctalk-admin@classiccmp.org]On
> Behalf Of Richard Erlacher
> Sent: Friday, June 07, 2002 14:44
> To: cctalk(a)classiccmp.org
> Subject: Re: [CCTECH] Interesting tidbit on 6502
> HORSEFEATHERS!
> I've got one of the first few dozen
6800 parts that were ever allowed
> outside
> the plant, and it's not even marked XC. It's simply marked SAMPLE and is
> otherwise unmarked. However, based on when I got it, it has to be a very
> early part.
> I don't for a moment doubt that there
were things that could have caused an
> internal failure. However, I do very much doubt that there was a specific,
> if
> undocumented, instruction that could bring such a failure about. If there
> is,
> then please tell us all about it. Then tell us how that would cause the
> device to catch fire, will you?
> Dick
> ----- Original Message -----
> From: "Jim Battle" <frustum(a)pacbell.net
> To: <cctalk(a)classiccmp.org
> Sent: Friday, June 07, 2002 9:21 AM
> Subject: RE: [CCTECH] Interesting tidbit on 6502
> > At 11:00 AM 6/7/02 -0400, you wrote:
> > > Why don't you learn that you don't always know what
you're
> > > talking about?
> >
> > >
Indeed, the first step level of the 6800 had an undocumented
> > > opcode that
> > >results in 3 particular transistors turning on that connected Vcc to GND.
> > >The damage caused by these three transistors shorting damaged the silicon
> in
> > >the immediate area, rendering the CPU useless.
>
>
> > The R2000 MIPS CPU's had a similar problem. The TLB was implemented
via a
> > small CAM. If more than one entry matched the virtual address, there was
> a
> > fight between two different words attempting to drive the output
> > bitlines. This wouldn't be a problem other than the fact that the TLB
> > contents were managed by software. Granted, the code would be supervisor
> > code and not user code, but when developing the supervisor software, there
> > was always the danger that buggy software could damage the chip.
>
> > I believe this was fixed in R3000 and
later parts.
>
>
>
> > -----
> > Jim Battle == frustum(a)pacbell.net
>
>