DIYbanter - View Single Post

Tony Hwang

wrote:
On Mon, 31 Mar 2014 19:57:54 GMT, (Scott Lurndal)
wrote:

writes:
On Mon, 31 Mar 2014 09:45:51 -0400, "Mayayana"

Both the Win7 dual CPU box and my new XP box, with
"mediocre" AMD A6 2-core, respond instantly. I keep them
clean. If you find you need a high-power machine for
speed to do things less intensive than video editing then
you probably have a lot of crap weighing down the system...
And you've probably been reading too many mainstream
media articles written by tech journalists who depend on
hardware and software companies for ad dollars. The world
of tech survives on a dizzying pace of forced obsolescence,
so if you go by what the media tells you you'll end up
replacing gadgets as fast as you buy them.

Computers of that age have another common failure mode that slows them
to a crawl - leaky caps.

Can you clarify how a leaky cap will "slow them to a crawl"?

If the PLL controlling the clock signal doesn't lock at the
target frequency, the processor will never leave reset. I suppose
that qualifies as "slow", for some value of "slow".
I'm not a computer engineer, but I have experienced computers slowing
to a crawl with bad caps, that came right back to life when I replaced
the caps. It's not just the processor clock - it's the IO from the
hard drive, the refresh rate on the RAM, and the output to the video
that can all slow down. The processor misses clock cycles if the
voltage goes off spec too, from what I've been told.

Some bad caps will also make the computer not boot. Or make the
computer crash when it gets warm.

Hi,
It all depends which part of the logic the cap is located. Until
you see some thing caused by any component going bad you wouldn't
believe things happening in the field(real world). Bad cap even scres up
critical rise and fall time of a clock pulse. My job as a Sr. systems
support specialist was looking at this sort of things with multi channel
logic analyzer set up to catch things when it happens. Some things
glitch once in a blue moon but we know it is happening and we have to
catch it to generate engineering mod. with design engineers.

Some problems originates from poor quality control. Bad batch of chips
or parts will incur wasted expenses. Purchasing agent at logistics has
big responsibility in this regard. Timing I was dealing with was nano
seconds or fraction of it. Ordinary O'scope is unable to display it.
Storage scope captured signals had to be displayed in sort of scaled
slow motion to analyze it. x86 PC was used as a diagnostic tool to
trouble-shoot large scale multi layer logic board down to component
level. Any one heard of checking logic circuits by serial bit shifting
method?