Searching \ for '[OT] liquid pump for liquid cooled computers' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: www.piclist.com/techref/index.htm?key=liquid+pump+liquid
Search entire site for: 'liquid pump for liquid cooled computers'.

Exact match. Not showing close matches.
PICList Thread
'[OT] liquid pump for liquid cooled computers'
2011\12\10@211017 by V G

picon face
Hi all,

Generally, when one builds a liquid cooling system for his computer, what
is the mechanism of operation of the liquid pump? For example, is it a
piston pump? Centrifugal pump? What kinds are normally used

2011\12\10@212437 by Carl Denk

flavicon
face
Check out these people. I have dealt with them several times, and we good.
http://www.xoxide.com/watcoolkit.html

I would think, one would buy a pump made for the task.

On 12/10/2011 9:10 PM, V G wrote:
> Hi all,
>
> Generally, when one builds a liquid cooling system for his computer, what
> is the mechanism of operation of the liquid pump? For example, is it a
> piston pump? Centrifugal pump? What kinds are normally used?
>

2011\12\10@214458 by Djula Djarmati

flavicon
On 11-Dec-11 03:10, V G wrote:
> Hi all,
>
> Generally, when one builds a liquid cooling system for his computer, what
> is the mechanism of operation of the liquid pump? For example, is it a
> piston pump? Centrifugal pump? What kinds are normally used?

I know almost nothing about liquid CPU cooling systems but I would be surprised if they used anything other than the simplest pump type which is a centrifugal pump with sealed permanent magnet rotor - like a CPU fan turned inside-out.

Djul

2011\12\10@215913 by RussellMc

face picon face
> Generally, when one builds a liquid cooling system for his computer, what
> is the mechanism of operation of the liquid pump? For example, is it a
> piston pump? Centrifugal pump? What kinds are normally used?

The  most consistent requirement seem to be that they are made of expinsiveum.
I would not be surprised if you can buy then withcrystal rotors and
OFC windings.
But  ...

Expect centrifugal in almost all cases to get easy highish head.
Peristaltic possible.
Piston also possible but tend to be extra cost when you can sell
centrifugal at high $ and get extra profit instead.

_______

Here's a slew to allow their methods to be examined.

          http://www.xoxide.com/wacopu.html

Of those, this fine example at $US80

   http://www.xoxide.com/swiftech-mcp655.html

.... is centrifugal, has its own microcontroller,
Maximum 10 foot head and maximum 50 psi, so we can already see the
OFC/crystals at work - ie 10 foot head around here is 10/32 x 14 =~ <
5 psi.
And no - the 50 was not a 5.0 typo as they say also 3.5 bar. Maybe
they are bah-cubs?
317 gph/1200 lph.
12v DC 2A nominal.
Hmm. 1200 lph x 9.8 (G) x 3.3M head  X 1/3600 h/s =~ 4 Watt-hour  SO
24 Watt-hour ~= 16% efficient may be about right.

Quick check 1200 lph = 333 cc/second
Water stores 4.2 J/g/K so for a 10K rise wattage able to be handled =
333 x 4.2 x 10

=~ 14 KiloWatt.

That should be enough for most cpus   :-)

_____________

Also centrifugal
More $
Less Watts
Less flow
Hald the pressure max
50% more head max.
This is fun.
Only good for 14 kW.

_______________

Slightly mote realistic

Thermaltake spare part.
$50

7.2W nom
500 +/- 50 litres/hour
Lift 1.8m - probably AT rated flow as they tend to be somewhat real.

5.8 kW cooling at 10C rise for perfect mixing.

"Adequate".

____________________________

Water heat capacity = 4.2 J/cc/K

So cooling capacity =  4.2 x  CC/second x degrees K (or C) temperature rise.

So eg 100 cc / second at 1 degree rise handles 420 Watts IF you get
all water rising by that amount. (ie if average rise is that)




       Russel

2011\12\11@112806 by M.L.

flavicon
face
On Sat, Dec 10, 2011 at 9:10 PM, V G <spam_OUTx.solarwind.xTakeThisOuTspamgmail.com> wrote:
> Hi all,
>
> Generally, when one builds a liquid cooling system for his computer, what
> is the mechanism of operation of the liquid pump? For example, is it a
> piston pump? Centrifugal pump? What kinds are normally used?

Centrifugal. When I was in college I water cooled my CPU -- don't ask why.
I used an aquarium pump that I submerged in a rather large reservoir.
It worked pretty well, but there wasn't much of a real necessity to
any of it.
Newegg sells this pump:
http://www.newegg.com/Product/Product.aspx?Item=N82E16835108063

That sort of special-purpose pump didn't really exist several years ago.

-- Martin K

2011\12\11@123417 by Electron

flavicon
face

You may want to consider to seal the cabinet and immerge it into mineral oil
(which, note, is of course not electrically conductive) then, if necessary,
cool that whole oil mass. Cheap and efficient.


At 03.10 2011.12.11, you wrote:
>Hi all,
>
>Generally, when one builds a liquid cooling system for his computer, what
>is the mechanism of operation of the liquid pump? For example, is it a
>piston pump? Centrifugal pump? What kinds are normally used?
>

2011\12\11@152215 by V G

picon face
On Sun, Dec 11, 2011 at 11:27 AM, M.L. <.....mKILLspamspam@spam@lkeng.net> wrote:

> Centrifugal. When I was in college I water cooled my CPU -- don't ask why..
> I used an aquarium pump that I submerged in a rather large reservoir.
> It worked pretty well, but there wasn't much of a real necessity to
> any of it.
> Newegg sells this pump:
> http://www.newegg.com/Product/Product.aspx?Item=N82E16835108063
>
> That sort of special-purpose pump didn't really exist several years ago.
>
>
Ah. Going to build a system with a radiator mounted outside my apartment,
so need a pump that has enough pressure and flow to push through many
meters of tubing. It's pretty cold here in Toronto. I'll move the radiator
to inside my (surprisingly quiet and very cold) chest freezer during the
summer.

My current setup is using a CoolerMaster Hyper TX3 fan on the CPU which
keeps the i7 under 55 degrees under full load. Around the same temps when
overclocked to 4.4 GHz. The liquid cooling isn't necessary at all - just
very fun

2011\12\11@181310 by RussellMc

face picon face
Most if these sound like they have enough pressure for quite a lot of
tubing - if specs believable.
Also, as my figures hopefully showed, you don't need much flow if
water mixes fully.

42 watts per (cc per second) at 10C rise.

Long tubes allow you to add unreliability and risk of coolant loss and
death of cpu which adds spice to life :-)

>From page of pump which Electron recommended.


http://images17.newegg.com/is/image/newegg/35-108-063-S02?$S640W$
http://images17.newegg.com/is/image/newegg/35-108-063-S06?$S640W$


22 psi
60 C max temp.
Noryl material (glass filled nylon afair).



   Russel

2011\12\11@182638 by peter green

flavicon
face
RussellMc wrote:
{Quote hidden}

>

2011\12\11@202203 by RussellMc

face picon face
> > Also, as my figures hopefully showed, you don't need much flow if
> > water mixes fully.
> >
> > 42 watts per (cc per second) at 10C rise.

> 10C rise seems rather high to me.

That deep-ends on what you think is rising relative to what.

And you CAN get very small water  temperature rise at the IC but still
get a terrible result overall.
Hopefully most people who design such systems will design all parts in
a balanced way. One can hope :-).

To get heat transfer you must get SOME rise.

The rise of 10 degrees I mentioned is in the water at the IC  - it's
saying that the water comes in 10 degrees cooler than the IC. This
doesn't say ANYTHING about how hot the IC is (except that it's
probably under 110C :-) ).

There are 4  temperatures iinvolved, assuming each is homogeneous
/uniform for substances involved.

Ta    Local ambient
Twc Water cold temp
Twh Water hot temp
Ths  Cooled heatsink temp

If water energy carrying capacity  at say a few degrees rise is large
compared with energy to be carried then Twh assymptotes to Ths BUT
what Twc does compared to Ta depends on how you get the heat out of
the water on the cold side.

eg if you had the very large flows I mentioned with the sample pumps
that could carry say 2 kW per degrees C rise then delta T at heat sink
will be small. ie Ths-Twh very low. As you suggest.

BUT if you do not do a good job of getting the heat OUT of the water
then Twc-Ta can be very high. You can end up with say 20 C ambient,
60C heatsink and 59C hot water and maybe 57 degrees cold water. Water
cooling is doing a good job of getting energy out of IC but you need a
large delta T to get it out of the water.


 Russell McMaho

2011\12\11@205630 by peter green

flavicon
face
RussellMc wrote:
>>> Also, as my figures hopefully showed, you don't need much flow if
>>> water mixes fully.
>>>
>>> 42 watts per (cc per second) at 10C rise.
>>>      
>
>  
>> 10C rise seems rather high to me.
>>    
>
> That deep-ends on what you think is rising relative to what.
>   My understanding was that by "rise" you meant Twh-Twc

>Ths  Cooled heatsink temp
What do you mean by "cooled heatsink"? do you mean the waterblock or the radiator?


AIUI there are more like seven temperatures involved

Ta: the ambiant temperature arround the rad
Trad: radiator temp
Twc: water cold temp
Twh: water hot temp
Twb: waterblock temp
Tcase: IC package temp
Tcore: IC core temp

AIUI for heat from the core to flow through the watercooling system the folliwing inequality must hold

Ta < Trad < Twc < Twh < Twb < Tcase < Tcore

Now if we assume Tcore is about 70c (about as high as you want to run a processor 24/7 afaict) and Ta is 30c (not unreasonable, particually if the rad is being cooled by air that has already come through the PC case) you have about 40c to play with for the complete cooling system. Usually you try to keep Twh-Twc very low because doing so gives you more headroom for Trad-Ta (which afaict is the hardest thing to reduce)

2011\12\11@210605 by William \Chops\ Westfield

face picon face

On Dec 10, 2011, at 6:58 PM, RussellMc wrote:

> The  most consistent requirement seem to be that they are made of expinsiveum.

I don't know.  I'm being impressed that they seem cheaper than the last set of centrifugal pumps I looked at for fixing the hottub and/or fruit sprayer ("flojet" things, at about $100.  Difficult to compare specs, though...

BillW

2011\12\12@000029 by RussellMc

face picon face
> >>> Also, as my figures hopefully showed, you don't need much flow if
> >>> water mixes fully.
> >>>
> >>> 42 watts per (cc per second) at 10C rise.

> >> 10C rise seems rather high to me.

> > That deep-ends on what you think is rising relative to what.

> My understanding was that by "rise" you meant Twh-Twc
> >Ths  Cooled heatsink temp
> What do you mean by "cooled heatsink"? do you mean the waterblock or the radiator?
> AIUI there are more like seven temperatures involved

I imagine that we agree reasonably well on what was involved and that
a real [tm] fluidynamicist [tm] may even identify 42 relevant
temperatures.

I was aiming at basic concept - hence my use of "weasel words" such as
" ...  if water mixes fully. ..."

and

" ... There are 4  temperatures involved, ASSUMING each is HOMOGENEOUS ..."

eg the radiator ( or whatever (I didn't specify) is used) section was
dumbed down to T water in, T water out and local ambient.
I agree that lots goes on in there that you can play with. That was in
fact my point in noting that water temperature rise across the hot
element was not liable to be the key figure of merit.

Similarly I allowed a homogeneously hot thing cooled by a stream of
well mixed cold stuff. Getting the heat from core to fluid boundary is
a necessary 'exercise for the student'.
My main point there was that using water you can cool  4.2 Watts of
heating if you have a flow of 1cc per second at a 1C averge fluid rise
or 42 Watts for 10C rise. The reason for saying 10C rise was to show
intuitively that you can get quit real Wattages at very low flows by
any normal standards. Saying 100 cc / second to coo, 420 Watts at 1C
rise makes the point less graphically that saying 10cc /second at 10C.
Those who get excited by the subject can then explore.

FWIW air flow needs ~~~~~~=  500 to 1000 x the volumes to achieve the
same result. litres per second of air are no great problem (usually)
but it's worth noting).

But, yes, there's more to it than my simplistic summary covers.
But, given the stated assumptions I thing it gives a good feel for
what is needed.


     Russell

..

2011\12\12@192222 by M.L.

flavicon
face
On Sun, Dec 11, 2011 at 3:21 PM, V G <x.solarwind.xspamKILLspamgmail.com> wrote:
> Ah. Going to build a system with a radiator mounted outside my apartment,
> so need a pump that has enough pressure and flow to push through many
> meters of tubing. It's pretty cold here in Toronto. I'll move the radiator
> to inside my (surprisingly quiet and very cold) chest freezer during the
> summer.

I'm sure you've thought of the consequences of having very cold
apparatus exposed to warm moist air. Insulate well.

-- Martin K

2011\12\12@195731 by V G

picon face
On Mon, Dec 12, 2011 at 7:21 PM, M.L. <.....mKILLspamspam.....lkeng.net> wrote:
> I'm sure you've thought of the consequences of having very cold
> apparatus exposed to warm moist air. Insulate well.

Thanks for reminding me. Need to do some dewpoint calculations

More... (looser matching)
- Last day of these posts
- In 2011 , 2012 only
- Today
- New search...