16 Nov
2005
16 Nov
'05
3:46 p.m.
Subject: Re: removing parts from PCBs
From: William Donzelli <aw288 at osfn.org>
Date: Wed, 16 Nov 2005 10:27:05 -0500 (EST)
To: "General Discussion: On-Topic and Off-Topic Posts" <cctalk at
classiccmp.org>
It's the pumps that are connected with the chiller or fans if they used
a direct radiation to air I was refering too. The pump for the local
loop may be small or not depending of the length of the loop and
diameters.
Water cooling had one less obvious avantage. You can dump the waste heat
outside the building or at least outside the computer room.
Allison
Then there is an efficientcy problem as you end
up using power to move heat which adds heat..
I think that is a minor issue. Even a relatively small pump will move the
coolant around enough. The flow is generally not all that fast, nor is the
temperature gradient.
William Donzelli
aw288 at osfn.org 
4:45 p.m.
New subject: Cooling and leaks; was: removing parts from PCBs
On 11/16/2005 at 10:46 AM Allison wrote:
Water cooling had one less obvious avantage. You can
dump the waste heat
outside the building or at least outside the computer room.
At CDC Sunnyvale, we had a big cooling tower that handled both the chilled
water for the computers as well as the HVAC for the building. When I first
came to Sunnyvale, I remember that the first thing that I saw of the
building (sitting in the middle of an onion field) was the big vapor plume.
I can't ever remember seeing a water leak in the machine room. The Bryant
6603 disks (and one very memorable 808 drive) would leak hydraulic fluid,
however. Made a terrible mess.
By far, the worst problem was construction that was going on in the area.
It would cause the power to go out at unexpected times. Bringing a machine
back up after a power failure in those days was a major chore taking hours
and sometimes days. Normally, machines were never powered down, except
for moving or major maintenance.
As I remember it, both CDC and Cray used the same guy to design their
refrigeration--and he wasn't an employee of either firm, but a guy who used
to work for Amana.
Cheers,
Chuck
5:48 p.m.
New subject: Can anyone identified these old water chillers ?
I have been attempting to identify some early CPU coolers, At first I was led to believe
they were off of an IBM system, I have not found any IBMers that could identify them.
I am now thinking they may have been fabricated for any water chilled system,
they may have been a retrofit for an air cooled system in a water chiller enviroment.
From the size of the feed lines, this was a low
pressure high volume chiller, and there
were between 30 and 40 of these hanging on
the system doors.
Here is a link, if anyone recognize them, I am sure Lew (the owner) would love to know as
would several others who are equally baffled.
http://www.ibm-collectables.com/gallery/view_album.php?set_albumName=album07
Bob Bradlee
On Wed, 16 Nov 2005 08:45:33 -0800, Chuck Guzis wrote:
On 11/16/2005 at 10:46 AM Allison wrote:
>Water cooling had one less obvious avantage. You
can dump the waste heat
>outside the building or at least outside the computer room.
At CDC Sunnyvale, we had a big cooling tower that
handled both the chilled
water for the computers as well as the HVAC for the building. When I first
came to Sunnyvale, I remember that the first thing that I saw of the
building (sitting in the middle of an onion field) was the big vapor plume.
I can't ever remember seeing a water leak in the
machine room. The Bryant
6603 disks (and one very memorable 808 drive) would leak hydraulic fluid,
however. Made a terrible mess.
By far, the worst problem was construction that was
going on in the area.
It would cause the power to go out at unexpected times. Bringing a machine
back up after a power failure in those days was a major chore taking hours
and sometimes days. Normally, machines were never powered down, except
for moving or major maintenance.
As I remember it, both CDC and Cray used the same guy
to design their
refrigeration--and he wasn't an employee of either firm, but a guy who used
to work for Amana.
Cheers,
Chuck
21 Nov
21 Nov
8:55 p.m.
New subject: Cooling and leaks; was: removing parts from PCBs
On 11/16/05, Chuck Guzis <cclist at sydex.com> wrote:
I can't ever remember seeing a water leak in the
machine room. The Bryant
6603 disks (and one very memorable 808 drive) would leak hydraulic fluid,
however. Made a terrible mess.
I've avoided the potential leak hazard by keeping the water systems
underpressured at the computer. I've got a pump that can push water 10 ft
vertical that's located about 15 feet below the computer. If I get a leak
more than 10ft above the pump, it sucks air in rather than squirting out
water. A slow leak shows up as bubbles in a clear section of the hose (and
as bubbles in the reservoir, since the surface tension reducer is soapy).
The drawbacks of this system are that you need to prime it, and that a bad
enough leak will disrupt the coolant flow, so you need to monitor the flow
rate or have a decent thermal shutdown mechanism. You need that in case of
pump failure, anyway.
Eric
9:40 p.m.
New subject: Cooling and leaks; was: removing parts from PCBs
On 11/21/2005 at 12:55 PM Eric J Korpela wrote:
The drawbacks of this system are that you need to prime
it, and that a bad
enough leak will disrupt the coolant flow, so you need to monitor the flow
rate or have a decent thermal shutdown mechanism. You need that in case of
pump failure, anyway.
Why not simply use a standard sump pump? Most such pumps can push a 15 or
20 foot head and don't need to be primed.
22 Nov
22 Nov
1:06 a.m.
New subject: Cooling and leaks; was: removing parts from PCBs
Chuck Guzis wrote:
On 11/21/2005 at 12:55 PM Eric J Korpela wrote:
I think that putting the computer higher than the pump can push is
the point. If the computer's on the low-pressure side of the circuit,
(the pump is *pulling* water through instead of *pushing* ) it would
take a very large leak to get the system wet. It'll pull air into the
water lines, instead of pushing water out into the electronics.
Doc
The drawbacks of this system are that you need to
prime it, and that a bad
enough leak will disrupt the coolant flow, so you need to monitor the flow
rate or have a decent thermal shutdown mechanism. You need that in case of
pump failure, anyway.
Why not simply use a standard sump pump? Most such pumps can push a 15 or
20 foot head and don't need to be primed.
2:07 a.m.
New subject: Cooling and leaks; was: removing parts from PCBs
On 11/21/2005 at 7:06 PM Doc Shipley wrote:
I think that putting the computer higher than the
pump can push is
the point. If the computer's on the low-pressure side of the circuit,
(the pump is *pulling* water through instead of *pushing* ) it would
take a very large leak to get the system wet. It'll pull air into the
water lines, instead of pushing water out into the electronics.
I have such a sump pump as part of my septic system--the l500 gallon tank
drains overflow into a smaller 100 gallon tank and a float-actuated
submerged sump pump moves the waste water to the drainfield (which is
located about 20 ft. higher than the tank.
The sump pump works by means of an impeller; i.e., it sits submerged in the
water and centrifugally pushes it out the drain port. There are no check
valves. My own pump is rated at 28' lift and it can fully drain the 100
gallon holding tank in less than 30 seconds. If the tank is empty and the
pump continues to run, there are no bad effects and no air ends up getting
pushed out the drain line. A very robust design; my own pump has been
operating for more than 15 years without attention.
Here's a photo of such a pump:
http://www.pumps-in-stock.com/images/6-cia_large.jpg
Cheers,
Chuck
11:26 p.m.
New subject: Cooling and leaks; was: removing parts from PCBs
On 11/21/05, Chuck Guzis <cclist at sydex.com> wrote:
On 11/21/2005 at 12:55 PM Eric J Korpela wrote:
The underpressure is a design feature (to minimize damage from leaks). I'm
willing to prime the system on those rare occasion that enough air gets into
the system to stop the flow in return for never needing to worry that a leak
inside the computer will short something out. If a hose leaks inside of the
computer the water ends up in the reservoir, not in the computer. The sound
of a leak is also pretty obvious.
The drawbacks of this system are that you need to
prime it, and that a
bad
enough leak will disrupt the coolant flow, so you
need to monitor the
flow
rate or have a decent thermal shutdown mechanism.
You need that in case
of
pump failure, anyway.
Why not simply use a standard sump pump? Most such pumps can push a 15 or
20 foot head and don't need to be primed.
6983
days inactive
6989
days old
8 comments
6 participants
participants (6)
-
ajp166＠bellatlantic.net -
aw288＠osfn.org -
Bob＠BRADLEE.ORG -
cclist＠sydex.com -
doc＠mdrconsult.com -
korpela＠ssl.berkeley.edu