[Tfug] Pizza-box power problem

Bexley Hall bexley401 at yahoo.com
Fri Dec 14 11:28:22 MST 2012


Hi Erich,

On 12/14/2012 10:30 AM, erich wrote:
> OK,
> I have a Gateway E-3400 box. In it's heyday the manufacturer must've made
> zillions of these things. I've seen so many around.

First clue:  age of machine
Second clue:  "Gateway"

> It has trouble at power-up. I hear relay chatter on the riser card and the
> mother board may or may not power up. By that I mean the case fan in the
> front runs, and the fan inside the power supply runs, but the CPU fan is
> idle, and I see no boot up screen.

Third clue:  at power-up

No idea why there would be a (electromechanical) *relay* anywhere in
the box!  And, if it is being used to switch *power*, chatter is not
going to be a Good Thing (since most things don't like their power
applied and removed repeatedly in rapid succession!)

Power supply fan is "hard wired" to a (12V?) supply inside the PS.
The same is true of other large-ish fans in the case.  The fan on
the CPU ("active cooler") is often "controlled" by something
(i.e., something "smart")

> If I apply heat with a hair drier to the back of the box, (the relay
> on the riser card is toward the back), I hear less chatter and the box
> starts up.

Less chatter doesn't tell you if the relay is staying engaged or
staying *open*.  I.e., an odd number of clicks means it has changed
effective state from "open" (when powered off) to "closed".  An
even number means it has returned to its *initial* state.

[And, of course, a "click" need not mean the armature has moved
completely -- it may have just *tried* to move (similar to the
chatter when a car's bendix chatters!)]

Do you know if the box can be operated in the *absence* of the
riser card?  (if the card is largely passive, this is often the
case)  Also, can you remove as much stuff from the machine as
possible (second disk drive -- heck, even the *first* disk
since we don't care if it doesn't boot... we just want to see
if the processor can put something on the screen:  "Disk not
found", etc.).

> There is always relay chatter. There is LESS relay chatter when
> 1) the box is warmed up, and 2) when the motherboard powers up.
> In other words less relay chatter isn't a guarantee that the motherboard
> will power up.
> Any thoughts?

Given all the "clues" <grin>, I'd strongly suspect bad caps.
Start in the area proximate to the processor.  Look for devices
that either have "brown crud" (almost looks like *old* rust)
on their tops (the caps have vents through which this stuff
can ooze) or around their bases -- though usually much harder
to see.  If the tops appear bulged (upward), that's another sign.

If possible, try to note the name of the manufacturer(s) on the
devices (often the same manufacturer throughout the machine but
try to verify this, if possible).  Some are more notorious for
bad parts than others.

Heat is the killer (and poor quality components exacerbate this!).
Manufacturers often only specify devices with low temperature
ratings (85C instead of 105C) *or* rated at just 1000-2000 Po-Hrs.
Server-grade machines usually have more robust components (on the
other hand, they tend to be *on* for a lot more hours!!) and
almost always far better cooling.

However, heat might not come from *outside* the device (i.e., by
being located near a heatsink, etc.).  Rather, the "ripple currents"
that the caps are intended to filter generate heat inside the
device.  Heat is heat, regardless of its source.

Note that failing to see crud or bulges is not a conclusive
indication that the caps are good.  You actually have to examine
them electrically to pronounce them "known good".  So, what you're
looking for is something that tells you "very likely BAD"!  :-/

At the end of the day, if this proves to be the case, you will have
to decide how valuable the machine is, to you.  You can replace the
defective caps (though you often want to replace more than you
"suspect") but it's tedious, not cheap and, if you're not reasonably
skilled with a soldering iron/puller, you are liable to damage the
circuit board in the process (lift a pad/foil, etc.).

OTOH, don't just *assume* it's bad caps!

Regardless, you can almost always recover the data on the disk
drive by temporarily installing it in another machine (one that
has support for whatever file system resides on the disk).

[Hint:  I keep a "cosmetically compromised" external USB disk
enclosure around as an expedient when I need to attach a "loose"
disk to a computer.  That way, I don't have to disassemble the
computer's case and possibly remove an existing/useful drive
for the duration.]

Good luck!
--don




More information about the tfug mailing list