'Howard Goldstein' wrote, in part:
| Thank you for mentioning this and saving me a lot of time and
| frustration with an approach that would be worse than doing nothing.
| I think folks who come into froups who don't mention that they're new
| at the stuff miss out on all of the great expertise that's so freely
| offerred. In this part of the country I'd be fighting condensation 9
| months out of the year so maybe it's for the best that it isn't
| practical.
_____
I'd say that, at least from me, what you can find here, as a new
overclocker, is the benefit of my experience and mistakes rather than any
particular expertise. I do try to keep my comments based on things I have
actually done and/or researched pretty thoroughly, and then checked for
accuracy.
This is a pretty long post, so I'll list a few points up front:
* use more grades of grit rather than fewer
* don't expect water cooling a CPU to solve an air circulation problem
* electronic systems are designed to operate reliably at 60 to 70 C
* big cases make overclocking a lot more fun and cooling a lot easier
In no particular order -
I. You mention "In this part of the country I'd be fighting condensation 9
months out of the year..." What country and what area? I lived in the
Miami area 1984 - 2004 and never had a problem with condensation in computer
systems (in fact, now that I am in mid-Georgia again I find something that I
had forgotten; it is MORE humid here than in the Miami area, especially in
the Winter! I've also used laptop and desktop computers in the Middle East,
Southeast Asia, South and Central America, the Caribbean, the South Pacific,
Africa, and all over the USA. I've found that while high humidity may be
uncomfortable for humans, it certainly reduces static electricity. In fact,
when I build a system, or change component around, I always turn any air
conditioning OFF.
II. I'm glad you re-read the lapping directions. Try to get emery paper
rather than sandpaper for as many of the grit sizes as you can. Emery paper
is more for metal (and high gloss acrylic finishes) and sandpaper is more
for wood. A grit size of 40 is probably too coarse as a starter size for
the ridge you describe; 40 grit is a particle size of about 1/2 mm.
Particle size is reduced by approximately 1/2 each time you double the grit
number:
40 0.425 mm
80 0.190 mm
150 0.092 mm
320 0.036 mm
600 0.016 mm
1000 0.010 mm
The more intermediate grades you use, the faster the job will go. One 8.5"
X 11" sheet waterproof emery paper can be cut into at least four pieces
large enough for lapping, and each piece will last for several projects. To
save time, I suggest you get one sheet each of
40
60
80
100
120
180
240
320
400
600
800
1000
1500
2000
The total cost for the 14 sheets shouldn't be much over $15 US. You could
probably eliminate a few intermediate sizes, but the lapping will take
longer. With the awkward shape of the fins and heatpipes attached, you must
be very careful to keep the bottom surface of the heatsink level, else you
will get a rounded surface. Also, the lapping surface (the size of the
piece of sandpaper/emery paper) MUST be more than twice the dimensions of
the heatsink bottom else the surface will turn out concave. Be particularly
careful to avoid digging the edge of the heatsink or the edge of the ridge
you describe into the sandpaper/emery paper - it will tear the paper and
will tend to give a convex surface to the heatsink base. Lighter pressure
and faster motion helps avoid this problem. Water base lapping compound (in
graduated grit sizes, used, for example, to lap valves for combustion
engines) used directly on a flat glass surface with water also works well,
but is messier.
III. Water cooling your CPU will take out only about half the heat
dissipated inside the case. Good air circulation inside the case and good
air flow through the case is still going to be necessary even if you move to
cooling your CPU with water. Check your hard drive temperatures. You may
be surprised to find that hard drive temperatures can be over 60 C unless
you have good air flow across the drives.
A good quality motherboard has capacitors that can operate at 60 C without a
problem, and 60 C or 70 C is no problem at all for the switching FET's. In
fact, most electronic equipment is quite happy at 60 or 70 C if the
electrolytic capacitors (the ones that look like a 3/4" segment of a pencil)
are of high quality. CPUs, memory modules, and chip sets ('Northbridge'
plus 'Southbridge') are quite happy at 60 to 70 C. ATX power supplies are
quite happy at 60 to 70 C (check out a Hewlett-Packard or Dell desktop - you
will find the temperatures of components inside the case to be in the 60 to
70 C range. In fact, Intel DESIGNS and manufactures CPUs to operate
reliably at these temperatures AND adds and extra temperature safety margin
(think of all the office PCs with great wads of dust bunnies clogging the
heatsink fins.) I've see quite a few Dell computers that have only ONE fan
for the entire system, ONE fan that draws air through the case, circulates
air within the case, AND draws air through the CPU heatsink!
When you are overclocking you try to trade operating temperature safety
margins for higher clock speeds. Only three components REALLY need extra
cooling for this trade-off; the CPU, the chipset, and the memory. The rest
of the system is hardly affected at all. Extra cooling for the CPU (plus a
higher CPU core voltage, if necessary) alone will very likely get you 90% of
the overclock possible; extra cooling for the memory modules and chipset
may get you another few percent while extra cooling for the entire case and
power supply may get you a final few percent.
IV. Condensation inside your system case can NEVER be a problem so long as
the component temperatures inside the case are HIGHER than the dew point of
the air OUTSIDE the case. Example: if the air you pull into the case is at
30 C (86 F) and the relative humidity of that air is 80%, then condensation
will not occur on any surface hotter than ~ 26 C (79 F) {see
http://hyperphysics.phy-astr.gsu.edu/hbase/kinetic/relhum.html } Water
cooling using a radiator with room ambient air circulation to cool the water
CAN NOT CAUSE CONDENSATION because the water CAN NOT BE COOLER than the air
used to cool it! Peltier arrays or phase change cooling (think refrigerator
or air conditioner) CAN produce areas inside the case that are cooler than
ambient air temperature, so condensation can be a problem inside the case,
but ONLY if any exposed surfaces are cooler than the dew point of the room
ambient air. Condensation inside the case can be prevented if the case has
sufficient turbulent air flow over all the surfaces and if the components
that are cooled below the dew point are insulated (thermal insulation around
fluid circulation tubes, thermal enclosures over cooled heatsinks and around
edges of Peltier arrays). Finally, you can use resistors to warm the
outside of insulated components to take care of any insufficiency of the
thermal insulation.
V. If you have a tower type case, then you can cut a hole in the removable
side panel and install a large, slow fan (>120 mm). on rubber grommets. The
hole does not even need to neat; you can place a square filter frame large
enough to entirely hide the hole (round or square.)
What I suggest is a very large case like the Enlight 8950 server cases I
use - only two 80 mm chassis fans come with the case, but there is plenty of
room for two 120 mm fans behind the louvered front panel, with three 5.25"
and one 3.5" bays still exposed for removable drives. The back edge of my
DVD drives are 7" from the front edge of the EVGA 680i SLI motherboard, so
there is even room to mount a water-cooling system completely INSIDE the
case.
I will admit that very quiet operation is not a driving goal for me; I
listen to music with the computer while I work, but I don't use it for DVD
movie playback. If I want really good sound, I just use a Beyerdynamic DT48
stereo headset. But as it is, with the air circulation system I described
earlier in this thread, the two 80 mm fans in the case back panel produce
most of the noise, and I can just switch them off with very little
temperature rise inside the case and negligible CPU temperature rise for the
CPU since I installed the Thermaltake 7i heatsink/fan.
I have a 12" X 12" radiator for a water cooling system that I can use with a
10" box fan. But I don't really want to reinstall water-cooling; it makes
it so hard to fiddle with the system B^( ...and I don't really need more
computing power at the present - I'd rather be able to change hardware
around easily than squeeze out the maximum possible performance.
Phil Weldon
"Howard Goldstein" <hgoldste DeleteThis @mpcs.com> wrote in message
news:1193856630.1158@news.queue.to...
| On Wed, 31 Oct 2007 00:03:53 -0300, Phil Weldon
<not.disclosed DeleteThis @example.com> wrote:
| : 'Howard Goldstein' wrote:
| : | I hear what you're saying but the ridge is huge. It's like the Rocky
| : | Mountains running straight across the midde base for the entire length
| : | and 1/4" wide, made of stuff that's far tougher than Core 2 heat
| : | spreader material which compared to the 120x's base is like
| : | chalkboard. I'm afraid that to get it flat it's either dremel, lathe,
| : | or garbage; I don't really want to trash it. Perhaps if I clamp the
| : | dremel such that it can only grind down that high ridge without
| : | touching the rest of the base which is nice and flat, then finish it
| : | off as usual.
| : |
| : | The Monsoon II lite looks interesting (and hopefully *that* is at
| : | least flat), combination peltier and fanned heat sink...?
| : _____
| :
| : No, a Dremel tool with a grinder is not the way to go. Neither is a
lathe.
| : A fly cutter tool on a vertical milling machine is probably the way to
go
| : considering all the heatpipes and fins are already fixed to the
heatsink
| : base, but I'll bet you don't have one (and you'd still have a problem
to
| : solve in clamping the heatsink. I'd say your reasonable choices are to
| : return the heatsink as defective or lap it down. The surface can't be
as
| : hard as you claim, else it is steel, not copper. Waterproof emery
paper
| : might be what you need. And if the task is taking you longer than a
couple
| : of hours to get the ridge level you are doing something wrong (read the
| : directions again.)
|
| I really did printed out those directions but set them in with the
| sandpaper and glass since I was so discouraged by the 120x's hump.
| Reading more closely, I was violating a number of rules, was lapping
| in straight lines for 2 dozen passes and then rotating exactly 90
| degrees instead of figure 8, did not start *anywhere* near coarse enough
| given your description of what 40 grit does compared to 320 (at para
| 7). Looks like cockpit error + wrong stuff. I'll try to find the
| 40-120 grit stuff somewhere else. The auto parts places go down to 180
| on the waterproof stuff.
|
|
| :
| : And, no the Monsoon II Lite would be another mistake. Your CPU just
puts
| : out too much heat (~ 130 Watts tpd) for any air-cooled Peltier product
like
| : the Monsoon II to be anything more than an insulator. For a Peltier
array
| : to be better than an insulator it has to produce more waste heat than
the
| : CPU, and then you have more than TWICE as much heat to get out of the
case.
| : What a Peltier array can give you, in combination with water cooling is
a
| : CPU temperature 20 C or more below room ambient temperature, but the
array
| : you would need for a Core 2 Duo quadcore would require about 400 Watts
power
| : so the water cooler would have to move a total of 530 Watts of heat. A
| : Peltier array was a great cooling solution when CPUs dissipated 25 to
50
| : Watts; you could get a decent temperature drop with air cooling. You
could
| : even get a bit below ambient temperature with a Core 2 Duo; but the
same
| : setup with a Core 2 Quad would function as an insulator.
|
| Thank you for mentioning this and saving me a lot of time and
| frustration with an approach that would be worse than doing nothing.
| I think folks who come into froups who don't mention that they're new
| at the stuff miss out on all of the great expertise that's so freely
| offerred. In this part of the country I'd be fighting condensation 9
| months out of the year so maybe it's for the best that it isn't
| practical.
|
| Heat in the case: Besides the 120x which is in the core2 a newer and
| larger problem I'm running into now is dumping all of this heat out of
| the case with the q6600 and striker extreme (goes back to one of your
| first notes to me). Antec P182 with one extra 12cm fan, too noisy on
| anything but low and between the actively cooled northbridge (aside to
| Ed: Pulled the SE's pipes and put two HR-09Us on the FETs and a HR-05
| on the southbridge and a Noctua I think it's a u12 on the northbridge
| -- the last two are hotter than you know what on Auto voltages. And
| the FET heatsinks aren't long enough to cover all of the FETs without
| modding) it's like a sauna in the case
|
| Maybe I should be moving over to water so I can get the heat out of
| the case and a large enough radiator perhaps can be cooled with
| multiple large and quiet fans.
|
|
|
|
|
|
| :
| : Phil Weldon
| :
| :
| : Phil Weldon
| :
| : "Howard Goldstein" <hgoldste DeleteThis @mpcs.com> wrote in message
| : news:1193795155.3316@news.queue.to...
| : | On Mon, 29 Oct 2007 16:16:47 -0300, Phil Weldon
| : <not.disclosed DeleteThis @example.com> wrote:
| : | : 'Howard Goldstein' wrote, in part:
| : | : | How would you approach a seriously out-of-flat base like this one?
Or
| : | : | just ditch it. I think I still have more work to do because when
I
| : | : | popped the heat sink off sure enough there was MX2 accumulating on
| : | : | either side of what's left of that cylinder protburance that I
thought
| : | : | I took care of after the first 2 hours of grinding a month or so
ago.
| : | : | Would you try dremeling it down?
| : | : _____
| : | :
| : | : Do NOT try a Dremel tool to grind the heatsink. The heatsink (and
CPU
| : | : heatspreader, which has also had a flatness problem in the past)
should
| : be
| : | : flat over a large scale (check with a straight edge placed at
different
| : | : angles across the surface), on a medium scale ( 1 mm ), over a
small
| : scale
| : | : ( 0.1 mm ), and at a microscopic level ( < 10 micrometers ).
Grinding
| : with : a Dremel tool could take off the high points, but is not
suitable
| : for
| : | : producing a flat surface at ANY of these levels. And if you should
| : slip,
| : | : well, one slip and you may be worse off than when you started.
| : |
| : | I hear what you're saying but the ridge is huge. It's like the Rocky
| : | Mountains running straight across the midde base for the entire length
| : | and 1/4" wide, made of stuff that's far tougher than Core 2 heat
| : | spreader material which compared to the 120x's base is like
| : | chalkboard. I'm afraid that to get it flat it's either dremel, lathe,
| : | or garbage; I don't really want to trash it. Perhaps if I clamp the
| : | dremel such that it can only grind down that high ridge without
| : | touching the rest of the base which is nice and flat, then finish it
| : | off as usual.
| : |
| : | The Monsoon II lite looks interesting (and hopefully *that* is at
| : | least flat), combination peltier and fanned heat sink...?
| : |
| : |
| : |
| : |
| : |
| :
| : >> Stay informed about: "Top Of The Line" HS & F ??? 
FAQ
Search
Profile
Private Messages
Log in
.