02 Mar 2003, 16:11, JIM HOLSONBACK (1:123/140), wrote to MATT MC_CARTHY:

Hi JIM.

 JH> Hi, Matt.

 JH> Some weeks back, you posted about some glitches while trying to 
 JH> change CPU cartridges in a PII or PII/III board.  I can't find that 
 JH> post, so I'll ask if you will please refresh my memory.

 JH> BIR, when you tried to install a cpu of different speed, it 
 JH> wasn't recognized, and you had to put the previous cpu back in there, 
 JH> and reset the speed for the replacement in SETUP, then power down, put 
 JH> in the replacement, and then it would work upon repowering?

 JH> Do you recall what those cpu speeds were, and what the mainboard 
 JH> chipset was?  I presume that the mainboard CLOCKMUL settings were 
 JH> jumperless, and autodected by the system?
..................

In the particular case that started this, we had just gotten in a Celeron
333, PII style, and the 'word on the web' said that they could be really
overclocked reliably.  

Well, I plugged it into an Intel chipped MB (that was the only kind my
dealer ever had at that time), let it autodetect, and all was fine.  Then I
thought to try the "overclocking", went back into CMOS and set it
for 400MHz (not a very big jump).  When I exited CMOS and
"saved", the board was virtually 'dead'.  It _may_ have caused
the monitor to turn on, but I don't recall much more than that happening. 
After a little 'sweatting', I shut down and pulled the Celeron, and tried a
genuine PII-200 (it was there, and outdated, so I figured if anything
smoked my dealer wouldn't be too upset).  Nothing happened again. 
Remembering that I had set the CMOS to 400, I snuck a brand new PII-400 CPU
into it, and it booted.  I went back into CMOS and re-set it to 333, saved
settings, and powered down.  Snuck the new PII-400 back into stock, then
put the 333 back on the board, and all was again well.

My thoughts were that this particular 333 could NOT run at the 400 speed,
and the board would _probably_ also have booted with a 500 (or was that a
533?), or _any_ CPU that was _faster_ than the board setting, in effect
forcing an "underclock" just to be able to get back to the
settings.  While "Clear CMOS" wiped out the date and time
settings, it did not restore the CPU setting to any 'default'.

Over the years, I _have_ repeated this with different CPUs at different
speeds, on different brand MBs, all with the same results.  I _must_ start
with a faster CPU to be able to get back into the system to set things back
to normal.

The boards you get to work with probably have no RAM on them either to help
give you a clue, such as if the RAM were PC66 or PC100, etc., and would
introduce an additional variable guessing what might have been there
before.

I did note last week that the current ASUS P4PE boards DO sense a CPU
change, and force a default to the lowest speed, which would indicate to me
that they were aware of users getting locked out of their MBs by carelessly
playing with CMOS settings.

 JH> We've seen a number of boards with those "bulging tops" on the 
 JH> large caps near the cpu, and when those boards don't work right off, 
 JH> we give up on them especially quickly.

Changing those caps has become so common now, I've begun keeping notes.  I
am finding that only on the 400 - 500 CPU speeds.  350 and below seem to be
OK, and 533 and above seem to be OK.  Those 400-500 speed CPUs must put
additional load on _part_ of the power system, or put out noise spikes
resonant with those size capacitors to cook them.  Just a WAG, but a
peculiar condition nonetheless.


     Good luck...  M.

--- Msged/386 TE 06 (pre)