JFG wrote:
"JFG" wrote in message
newsuu5k.156704$TT4.5397@attbi_s22...
"Paul" wrote in message ...
JFG wrote:
"Paul" wrote in message
...
JFG wrote:
Hey, all,

I'm back with a problem that I hope is curable, but I fear the worst.
I was given a Satellite A45-S250 by a family member. It
intermittently booted up but didn't run for long. It had the problem
of BSD after memory dump on a regular basis. I looked up the error
message and many suggested bad RAM. I replaced the RAM with known
good RAM and the problem persisted. I disabled all the peripherals I
could and tried to load a fresh install of XP. This didn't work.
Long story, short: I thought I might have a burnt CPU so I took out
the P4-2.8 Celeron that runs at 533 and replaced with a P4 M 1.6 that
runs at 400. According to the Toshiba website the mobo should run
either. I tried booting the laptop with the 1.6 and it wouldn't boot.
So I replaced the 1.6 with the original 2.8 and now the laptop won't
boot at all with the 2.8. It does not go into the BIOS even though
the lights come on and I hear activity in the cd-rom. I fear that I
might have blown the motherboard.

Anyway, if anyone has suggestions as to what I might do at this point
beyond getting a new mobo I'd like to hear them. Again, thanks to any
and all who take the time to suggest possibilities. JG
Examine the socket and the processor carefully, for connection
problems.
If there is a lever, or a screw that clamps the ZIF socket into place,
make sure it is secure. Make sure the physical assembly was all put
back
the same way (thermal paste, heatpipe heatsink onto processor, and so
on).

The Celeron would not burn, because it would have THERMTRIP. If the
heatsink falls off, the THERMTRIP signal is available to kill power
to the processor. So in theory, you shouldn't be able to burn a
processor. Both modern Intel and AMD processors now have THERMTRIP.
You'd have to go back to some of the early Athlons, to find cases
where the protection is less than ideal. (My AthlonXP motherboard,
has a separate eight pin chip, which monitors the temperature and
does the equivalent of THERMTRIP.)

You can check for THERMTRIP, by downloading the datasheet for the
processor. One way to do that, is go through the list of Celerons
here, select an appropriate one, and then see of there is a
documentation
link.

http://processorfinder.intel.com/

I wouldn't panic yet, because I interpret your results to suggest
something isn't assembled right. Like turning the screw on one of
those small ZIF sockets, to lock it.

Paul
Great post, Paul. Very good information. The cpu socket is one of
those with a lever. I checked the cpu itself for bent pins, etc and it
appears ok. It is locked tight in place and I'm pretty sure I put the
cooler assembly back correctly. You've convinced me that I haven't
burnt the cpu and now I'm wondering if I didn't do some harm to the
motherboard. The laptop wasn't working properly when I started on this
project and now it doesn't go to BIOS. I haven't completely given up on
the one I have yet. I'll try reassembling it today. I'm thinking that
what I'll end up doing is buying an equivalent unit from ebay with a
broken screen and missing parts. I'll cannabalize the one I have and
make one good one. Thanks for your response. JG
To know exactly what you did, I'd want the SLxxx number off both
processors. One of the differences on a mobile processor, is
there are a couple signals for SpeedStep. But that shouldn't harm
it. For example, people buy SL726 processors, and install them
in desktop boards, to overclock them. You can raise an SL726 to
FSB1200, causing the core to run at 3.6GHz, and I'm told they
run pretty cool while doing so. So I don't necessarily see the
installation of a mobile processor, as a death sentence. But
to know more, you'd want to use the code on the processor,
and look it up, to see exactly what it is. From that generation,
I don't know if there are any ULV (ultra low voltage) versions
of S478 chips or not. But it is easiest to trace this info,
by using the SLxxx number.

This is the mobile that overclocks so nice. It runs at a lower
speed than the stated values on this page, when plugged into
a desktop board, which is why it ends up at FSB1200 and 3.6GHz core.
The reason I'm interested in that one, is it would fit perfectly in my
current motherboard (P4C800-E Deluxe), which can apparently
overclock that high. I've run into a few people who buy these,
and don't know what a good deal they got (they're puzzled by
the speed it runs at).

http://processorfinder.intel.com/Det...px?sSpec=sl726

Paul
Turns out that you were 100% correct. I had not put the original CPU back
in correctly. It was slightly cocked and rocked a little before putting
down the lever. After installing correctly, the laptop then booted. So,
no harm no foul on the CPU switch. However, the original problem
remains - that is, the occasional BSD with the quick message of memory
dump. I have a couple of new sticks of RAM coming in and I'll see if that
makes a difference. Meanwhile I've bid on an equivalent model on eBay
that has no peripherals and a broken screen. If the motherboard on that
one works, I'll be in high cotton if I win it and can cannabalize my unit
successfully. I have to wait one week for that since the auction doesn't
end until then. In the nonce, I'll continue to fool around with the one I
have here.

I appreciate your taking the time to find out all the helpful information.

Best, JG

Follow-up for those interested in this problem: It turned out to be a
defective cpu. I replaced the 2.8 with a known good 2.6 and the machine
behaves as it should. I guess that the original cpu was overheated and
burnt in some way. Best, JG


There are other ways to kill a processor besides heat. It will be interesting
now, to see how long the replacement processor lasts.

As an example, Northwood processors are sensitive to excessive Vcore, and
some overclockers noted their processor croaked instantly, at a certain voltage. If
something on the motherboard of the laptop, is not regulating power properly,
it could place a voltage on there, which shortens the life of the processor.

The Arrhenius equation provides a mechanism to relate temperature, to operating
life, but generally speaking, this doesn't seem to be an issue for the
average life of computers. Most people upgrade, before any thermal limit
seems to be reached.

One of the video card companies, did a reliability calculation of one of
their competitor's offerings, in the hopes of scaring customers away.
The calculation predicted a short life for the video card, like a year
or less. But I never noticed any rash of those products dying.

Good reliability calculations, use field data collected from previous
failures, to adjust the equations to give better results. But that
was not the first time I was scared unnecessarily by such calculations.
The first big system I worked on, the reliability calculation said the
Intel processor used, would be more unreliable, than any other piece of
hardware in the equipment. Which was pure B.S.

Paul