Hmm, that's an idea. Next time I'm by a Radio Shack, I'll grab a power
resistor (
http://tinyurl.com/7rqpf) to test this out. It seems odd that
the 3.3V line could provide 21 amps without there really being a need
for this voltage.
Some boards do indeed use it.
Consider that the 3.3V amperage is a function of the
available 5V current and the current handling ability of
that subcircuit. For it to be able to supply 21 amps
doesn't "take away" anything from the rest of the supply to
the extent that no concessions are made to the 5V or 12V
peak to implement it.
Already did that. Besides keeping all the cables secured off to the
side, I also drilled out the 3-1/2 bay drive blank to let in more air
through the front. I did this because I noticed air was getting sucked
through the 3-1/2 floppy drive's slot. Often a front fan doesn't help
and actually hurts. Even AMD recommends not using a front fan: "A front
cooling fan is not essential. In some extreme situations, testing has
actually shown that these fans can recirculate hot air rather than
introducing cool air." (
http://snipurl.com/ym8). Experimentation works
best to figure out if it helps or hurts, and if it helps then by how
much versus the added noise.
It all depends on the implementation.
If a front fan is used, and mounted on the case wall, the
front bezel needs be of a design (or a partition added) that
prevents drawing the air OUT of the top of the chassis' main
compartment then down through the chassis-bezel chamber and
then into the chassis main compartment again, a
recirculation of the air already inside.
Even when a case causes this, the solution is simple- add a
partition, whether it be a piece of plastic glued above
where the fan intake stops, or hot-gluing on a piece of
dense foam, or whatever- just a prevention of this
recirculation.
The benefit of the front fan is more that of assisting the
exhaust but also to direct the flow past the hard drives,
southbridge, and more flow under the video card. While your
3-1/2 bay plate solution could be used on some cases and
situations, on others it may reduce HDD and lower-chassis
cooling. It can need be taken on a case-by-case basis.
Also keep in mind that AMD, like Intel, provides specs that
emphasize keeping the CPU cool with their retail 'sink.
Towards that end, the lowest cost cooling implementation for
an OEM is minimal number of fans possible, such that a
single fan has more of a positive effect on CPU cooling by
being placed below the power supply on the rear case wall.
That IS a good placement for a fan, is usually the first
place a fan should be added to a system, but properly
implemented the bottom-front fan is the second place one
should go... almost never a side-fan without a bottom front
as that drastically reduces the flow past the drives, unless
the side-fan were an exhaust which then drastically reduces
flow past (most components including CPU, motherboard, and
power supply exhaust.
I'll can't find the old one so I'll have to pickup another nibbler tool
(
http://snipurl.com/fm2k) though tough on the hand after cutting awhile.
I don't like the idea of using a wire cutter to snip the metal between
the outside ring of holes because that will leave a bunch of snaggies.
I've never see a hole punch that big. Any idea of how to make a clean
hole other than nibbling away at it?
Depends on the thickness of the case. I don't use ANY cases
thinner than 1.0mm myself, and find that I can use a
fine-toothed sabre saw and finish with some medium-fine
sandpaper, only along the cut-edge to smooth it, not sanding
on the finished surface of the panel. Depends a bit on
where you're cutting too... in some situations I've dressed
holes in side panels with vinyl automotive "door edge guard"
molding, cheap/small type that costs about $1.50 per yard,
which starts out black but can be painted with vinyl dye to
match many color schemes. It has a glue strip in it that,
after being heated with a heat-gun or hairdryer, will melt
and secure the trim onto the metal edge pretty well. Using
such trim, the edges could be quite crudely cut and still
look good, though I sand them smoot first regardles of this,
but the whole cutting/sanding process is always done prior
to building a system, never with parts inside.
For thinner metal you can use metal shears, a bit like a
nibbler in some respects but motorized. Even tin snips can
work well if the're decent and you cut a rectangular hole
with semi-rounded edges rather than trying to get it
circular. The easiest way to get a pattern for this is to
scrap an old fan, cut it in half lengthwise with a hacksaw.
Then you're left with a template for the square/rounded-edge
holes and for marking new screw holes.
The other option is a 3" hole-saw (for 80mm fan holes).
They often look good if you can put a block of wood behind
the region being cut, BUT most fans slope outward towards
the edge of their fans from the center, so a 3" hole is a
little more impedance than a square/rounded-edge hole.
In some cases where there was already an appropriately-sized
stamped-in-metal grill, I'll just use tin snips to cut the
grill out, being conservative to leave a few mm of material
behind so the cutting doesn't bend the portion of metal I
intend to have remaining when finished, then I remove the
remainder of the (mm of material) with a dremal or drill
plus grinding stone.
The central theme here may be that it varies, based on the
case and the specific goal, in addition to what tools you
have.
Actually mine was indicative of bearing wear with the cold-start rumble,
turning into a buzz as it warmed over, and eventually getting much more
quiet after many minutes.
Oh, now I see.
The fan might be salvageable for a less-critical use by
lubing it with moderately thick oil/grease lube.
When the box went into standby mode, heat
output reduces and the thermistor in the PSU reduced the fan speed and
it would get noisier (but not quite as bad as during a cold start). If
you've heard CD-ROM drives (with a CD-ROM in the drive) revving up on
power startup and the extra noise therefrom then you'll have an idea of
the level of noise coming from the defective fan until I replaced it.
If the box was on the floor to the side of my desk then I probably
wouldn't have bothered until it was always making a buzzing noise even
after warmup; however, it sits on the desk next to the monitor because
that's the only room that I have for it. So the noise is in my face all
the time.
Alright, I was under the impression it was also spinning
faster than average, but now realize the situation.
I figure the high 3.3V reading really isn't a problem because the box
has been working well for over a year. I figure, like you mention, that
it isn't loaded and that might why it is high and causes no problems.
It's one of those gotchas that perks up your eyeballs when you see it
and wonder why nothing bad has happened so far (that you know of).
We used to see memory and AGP using 3.3V, but often the
video was stepped-down from 5V. Even further back in
history, more things running from 3.3V, but these days all
the core chipets are using less than 3.3V, and while some
subcircuits can use 3.3V, many (like audio codecs) work
better at 5V. 3.3V is just too low, higher percentage
losses when remotely supplied rather than stepped-down from
5V. Now we're seeing same thing with 5V, more and more CPU
and video cards make use of 12V for similar reasons.
