Subject: Re: Watercooling?
From: Scott Johnson (sjohnson@gci.net)
Date: Mon Jan 21 2002 - 23:57:46 AKST
Basically it was due to bad pumps. I had a dual pump setup going (figuring
that even if one failed I still had one pushing the water) but alas, they
just gave out. Unfortunatly, instead of getting one high-quality pump, I
got two medium/low quality pumps. Since there was no water circulating
through the system, the water that was sitting over the CPU (in the water
block) just kept getting hotter and hotter and finally just started melting
stuff. First the plastic tubing, then the CPU itself. It even messed up
the ZIF socket so the MB went too. I was partially able to salvage the vid
card (which took the brunt of the water damage) but it still gets flaky when
running 3d-intensive applications. So first off if you're considering doing
this, GET A GOOD PUMP. When I first did my setup, I went to Lowe's to see
their selections. Most of the pumps they have though are for sump pumps or
wells; not meant to be left going 24x7. I would look into a high-quality
aquarium pump. Alaska Aquarium comes to mind. When I checked with them
(after mine went belly up), they said they had a pump that could push water
up 20' vertically. I would imagine if you're going from your crawl space to
first floor, you'd need something like this. I really didn't have a problem
with loss of water as it was a closed-loop system. If you're not going to
have fans on your radiator, just make sure you get a radiator with a lot of
surface area. Putting it down in the crawl space will defiantly help too.
When I was doing my initial testing, I set my radiator outside my bedroom
window. This was in the middle of December (2000). I was able to get my
water temperature down to a chilling 34 degrees F, which got my CPU down
to -20F. I was able to get my Tbird 900 up to 1.2Ghz and still keep the
temperature around -5F. You really don't wanna do that though in a
long-term basis as after about 2 hours you start getting condensation on the
CPU & ZIF. Also, if you can swing it, I'd try to use nothing but distilled
water instead of tap. That way if you were to develop a leak, there's less
of a chance of shorting something out.
Anyway, any other questions just ask.
Scott
----- Original Message -----
From: "Mike Tibor" <tibor@lib.uaa.alaska.edu>
To: <aklug@aklug.org>
Sent: 21 January, 2002 23:27
Subject: Re: Watercooling?
>
> On Mon, 21 Jan 2002, Scott Johnson wrote:
>
> > I *was* doing water cooling on my comp until it had a melt down and I
lost
> > some hardware :(. I would defiantly NOT recommend doing this on a
server.
> > I really don't see the need unless you're overclocking. Besides being
> > pretty expensive, it's a LOT simpler to just get a good Heat-sink fan
combo.
> > For anyone who is interested though, email me and I'll tell you what to
do
> > (and not do.) I have pics of my former setup on my site at
> > http://coolcase.akghetto.com if you're interested.
>
> Can I ask how you toasted the hardware?
>
> The most recent server I built included a solid copper cpu heatsink (the
> "Millennium Glaciator" from www.pcnut.com) with a decent ball bearing fan.
> While I haven't really been interested in overclocking my systems
> recently, the overclocking community has some cooling methods that are
> nothing less than pure genius.
>
> The reason I'm looking at this is that most of the setups I've seen (I
> haven't looked at the URL you mention yet, so you may have already
> overcome this) seem to have a closed system with a small reservoir
> volume. This means that in a server environment when the admin takes a 2
> week vacation, the water level drops according to whatever loss rate there
> may be in the system due to either evaporation, or leakage (usually very
> unlikely), or possibly corrosion if differing metals are used (aluminum +
> copper is one example).
>
> What I was thinking was a very simple system where the reservior was a 5
> gallon bucket or something else of adequate volume placed in the crawl
> space of my house (about 40 deg. F. in the winter; no more than 60 deg.
> in the summer), with a toilet fill valve screwed into the lid, and no
> radiator--or possibly one without fans. The fill valve would draw water
> from a spigot that was constantly on, and would eliminate coolant loss in
> the system. I have personally seen many toilet fill valves last 30+
> years, so they should be a reliable way of ensuring adequent coolant
> supply. This plan would require a higher powered pump that most I've
> seen in order to overcome the height difference between the first floor
> level (computer) and the crawl space (water reservior).
>
> My goal with this would be to increase cooling reliability as well as
> efficiency while also reducing computer room noise (my server has a pretty
> loud cpu fan, as well as two 80mm case fans--with this setup I should be
> able to eliminate at least one, perhaps both, of the 80mm case fans in
> addition to the cpu fan).
>
> What kind of problems would you forsee in this kind of setup?
>
> Thanks,
> Mike
> --
> Mike Tibor Univ. of Alaska Anchorage (907) 786-1001 voice
> Network Technician Consortium Library (907) 786-6050 fax
> tibor@lib.uaa.alaska.edu http://www.lib.uaa.alaska.edu/~tibor/
> http://www.lib.uaa.alaska.edu/~tibor/pgpkey for PGP public key
>
>
>
> .
>
This archive was generated by hypermail 2a23 : Mon Jan 21 2002 - 23:57:01 AKST