Do you have a question? Post it now! No Registration Necessary. Now with pictures!
- Posted on
- Cursed Computers - When to give up
May 31, 2007, 8:57 pm
rate this thread
identify a hardware problem, order the parts and have it fixed in a couple
Last summer I got this one computer that had experienced a surge during a
No boot - tried new power supply - still not working
Ordered mother board - put it all back together with old power supply.
It doesn't boot - try new power supply boots but can't find hard drive
Stick hard drive in another old computer now this computer doesn't work.
Really freakin out and pissed
Order a couple of POST Code readers from Ebay. One doen't tell me anything,
one tells me everything is rosy.
Put in another hard drive and computer stops booting again, just the CPU fan
I return and order another motherboard(Feel kinda bad about this because I
probably fried it, but Newegg is great)
Put it semi back together with other hard drive it works
Put the modem, CD,DVD back in and it stops working.
Start to pull components when I notice the modem is a couple of hundred
Pull it and it's working again
Get ready to bring it over, just downloaded all XP updates.
Reboot again for a final check and it will not boot up again.
Give up and tell them I cannot revive it, the very first computer that I
have not been able to fix. With all of those shorted components I just
threw the whole mess in the trash.
Just last week I got another computer and I'm headin down that road again.
How do I stop myself? What steps should I take to safely and logically
diagnose a computer? Do those POST Code readers really do anything usefull?
Re: Cursed Computers - When to give up
useful data - numbers. But unfortunately, you added variables to the
problem by shotgunning. For every addition variable, then a failure
becomes exponentially more complex - which is why shotgunning is not
performed by better techs.
For example, if a problem was not in a power supply and you replaced
the power supply with a new one that was either marginal or defective
(either way, it is neither good nor bad - it is unknown), then you
have exponentially complicated the failure.
In CSI, they 'follow the evidence'. You must do same. First, the
foundation of a computer is its power supply. Marginal operation by
a power supply will make any problem appear everywhere or anywhere.
Don't diagnose a power supply by swapping. A 3.5 digit multimeter is
even sold in Kmart (as well as Wal-Mart, Radio Shack, Lowes, Tru-Value
Hardware, Home Depot ... ). Meter is an essential tool for anyone
doing electric work - as important as a screwdriver.
Second, procedure to analyze the power supply 'system' (a 'system' -
not just a power supply) as summarized in "When your computer dies
without warning....." starting 6 Feb 2007 in the newsgroup
Notice that others in that discussion would even give up. But
Valentin restarted by using a meter to learn which signal was
reporting what. Included would be important numbers for any one
orange, red, purple, and yellow wires: above 3.23, 4.87, or 11.7
Third, a power supply 'system' is unknown or defective until the
procedure puts it into a 'definitively good' category. Only then is
one suspect eliminated - you have finally accomplished something - are
ready to more on to other suspects. Notice after all that swapping,
what have you accomplished? Nothing. You still don't know if
anything is 'definitively good' or 'definitively bad'.
Remember everything is either known good, known bad, OR unknown.
The world is not binary as so many shotgunners assume. You must
establish each 'suspect' as either good or bad (in a ternary world)
before replacing or even disconnecting a single wire. Long before
disconnecting or swapping anything, knowledgeable techs first identify
a suspect(s). Even changing or disconnecting one wire may only
exponentially complicate the failure.
Using a meter means numbers. Even if those numbers are 'good or
bad', still post them here because numbers can result in other useful
information from more experienced help. Another reason to use the
meter? Without numbers, then more informed posters may not reply.
Your replies will only be as useful as facts you provide. That means
Until a power supply 'system' is known good, then post codes will
report numbers that are uncertain. Power supply must be 'known good'
before numbers in the post code reader can be trusted.
Re: Cursed Computers - When to give up
guess it was just dumb luck that all worked until that one.
Of course in most cases it's obvious, bad hard drive, cd etc.
I have a multimeter that I've used for other projects and want to learn more
about using it to diagnose problems. I want to diagnose my current problem
computer correctly and methodically.
Thanks for the link, looks like a lot of good info in that thread. Any
other resources you can point me to(websites/books) to learn more about it?
Re: Cursed Computers - When to give up
Don't fix computers while drunk or sleepy. Don't get
"stuck" on insisting you find a solution at one moment
(unless an emergency / job related), if you get stuck, set
something aside and your subconscious mind may work out a
? Depends on the problem, often just replacing the part is
akin to giving up, though time is valuable too, some
problems aren't worth solving. Of course other times the
part is just damaged/dead/ or bugs that would never be
Before new PSU did you unplug the present from AC, then
clear CMOS, and replug/try again? Always unplug PSU and
clear CMOS before changing parts, especially since it's
quick, easy, free.
Why so? A surge travels every path it can, though some
lower impedance will be (more) destructive. If the surge is
bad enough it becomes a scenario of taking a working system
and trying each questionable part in turn in that system.
Even worse is that some forms of surge damage are not
immediate or obvious failure, a chip can have some burns and
still more-or-less work, it's that "less" part which could
eventually be a problem... so stability checks are very
important like Prime95, memtest86+.
I don't understand why you did this, at this point. Based
on what was written (and if you had unplugged from AC and
cleared CMOS) it would seem the PSU, board and HDD are
damaged. You report no other problems with the board *yet*
so perhaps you left something out, but this is one area
where having another drive or boot device to run memtest86+
might not hurt, to see if the basic functionality of the
board is in tact. IOW, if the board had finished posting
onto the point of detecting drives and that fails, AND the
drive in question seems nonfunctional in other system, you
had indication the board was working up to that point... and
you had no real reason to POST code check the new board with
the new psu since neither were subjected to the surge...
unless again you left out some details.
We dont know if it was stable. Where does it stop booting?
you are leaving out the key troubleshooting and
system-validation-as-stable details and only describing what
it won't do, not what it does do successfully. Both must be
considered when finding fault.
Maybe, if the old PSU were outputting bad power so before
trying the old PSU the new PSU should have been tried or at
least the old checked with a multimeter before attempt to
power anything valuable... but you only wrote that system
didn't boot with old PSU, again leaving out key details of
exactly what it WAS doing... so we have no way to know if
your next step was reasonable or not.
Don't bother trying to build up a whole system then try it.
Start out with minimum parts - CPU, 1 memory module,
heatsink/fan, video card (least power hungry card you have,
preferribly not the one in the system previously during the
surge event). See if it can then do what is expected given
the parts attached. Attach a boot device that can load and
run memtest86. If it is stable at this point, attach hard
drive. See if drive is detected, if it starts to boot the
OS. Forget about modem/opticals/etc until bare minimum
parts work to run more diagnostics or OS, then diagnostics
(which could be anything, using the required features of the
system and checking basic stability.
Unless it was inserted in the slot wrong, (which is hard to
do, but leaving no possibility unaccounted for) throw it
away, do not put in this or any other system it's not worth
the time or risk.
At this point, you have identified a component which if
previously installed when the system wouldn't work, might
have been at fault. Now the other parts might be rechecked
without the modem if it was always installed. Another
reason to start out with minimal build and work upwards to
final config instead of subtractively. Also with parts
subject to surges, it might not hurt to inspect them under
strong light before trying, applying power again.
While the odds of everything (or at least much) being
damaged are low, it happens... some failures are harder to
fix than others.
The most useful thing is to have spares of compatible parts,
preferribly a whole system. Since you are yet again fixing
another suspect system it could be good to maintain a small
stock of such parts and in your spare time, instead of
throwing away the questionable parts (besides those with
obvbious faults like the modem) retest the other parts as it
is convenient, when an opportunity presents itself with
least work or time required to do it. This is more feasible
for maintaining stock of a spare drive, PSU, modem, etc,
than for parts changing frequently like motherboards, CPUs,
and memory... but even those are not so hard to stock in
some situations. For example if you are getting at least
some kind of just compensation for the time, part of that
could be keeping the whole suspect system and to the
(customer?) it is as good a value to just pay for
replacement parts to achieve a quick build, minimal extra
bench time, instead of diagnostic time just to end up with
parts known already to be subject to damaging surge
conditions and that were potentially not very valuable
anymore as they depreciated over time and use/wear.
In summary, surge damage is one of the biggest PITA to
troubleshoot, particularly if you don't have a whole 2nd
system in which to swap out parts. The POST code reader is
most valuable when you need to check what a motherboard is
doing if you have reasonable expectation the CPU, memory and
PSU are still good.
- » Newest MemTest86, MemTest86+ incompatible with Sandy, Ivy chipsets?
- — Next thread in » Computer Hardware