Secret of Water Cooling

[*V53*-=(VJDV)=-]

Hi,

My name is James D'Aeron and I'm an enthusiast. I'm currently one of the global leaders of 3D mark 03 and 05.

I'd like to Publish this information and anyone interested in the subject and helping put it into words someone will understand please contact @ james.daeron@gmail.com. The principal I'm trying to share, Once understood - will reveal the fact that almost all water systems by design are in fact designing in oposite of the principal.

Background: I purchased a swiftech system and after the experience of smoking 2 $450+ video cards and the water system in a box under the house, I gave it one more go and found the solution to properly tuning a water cooling solution. Initially i contacted Swiftech about my find and they denied what i have yet to completely define as a simple a principal of physics (if you could help with finding a name for the principal would greatly help ~ My solution began as research into the principal of "Cavitation" which lead to the revelation.) Essentially the principal is revealed in the act of a siphon, where the gravimetric pressure and momentum of the fluid at elevation - once started continues to push the fluid through the tube. Now what i experienced in hind site is that ALL Manufacturers design their systems in contradiction to what I'm about to share:

First it is best to place the pump in the lowest part of the case. That done, the focus then is to center one's attention on the center of the pump and consider the volume of water and resultant gravimetric pressure the volume of water exerts on each side of the pump for proper tuning.

Example:

The initial design placed the volume of water to the gpu water block and water block volume + to the CPU water block tubing and water block + tubing to the radiator + volume of water in the radiator = a quanta i will define as 80% of total water volume. from the radiator the tubing was a essentially a straight shot back to the pump = to 20% volume. What that = is a gravimetric back-pressure of 60%- oppositional force of the total volume of water and simply that was enough to stalll the pump and burn it out. *note I'm refering to the 300gph swiftech pump!

My solution was to purchase a 2 bay reservoir that increased the volume of water and thus gravimetric pressure towards the pump flow changing the ratio into essentially the reverse of 80% volume on the inlet side of the pump with a resultant total of 60%+ gravimetric pressure.

While i don't have any equipment to give flow rates - i took the pump into the bathtub before I re-began the water adventure and observed what amounted to a dribble of output from the pump amouting to the flow protruding from the pump in an arc downwards of approximatley 3/4 of an inch. Now, on the inlet of my 2 bay water reservoir, there is a "jet" of stream that reaches from the rear of the tank to the front of the bay a full 7 inches with such a force that it creates a large bubbling effect.

James D'Aeron

[MaxxxRacer]

Let me be the first to Welcome you to Xtreme Systems, where we kill our hardware with Xtreme effeciency.


um.... duh... no offense, but duh... there is a reason that swiftech uses a res and why its is preferable to use a res.. the more water you feed the pump the better.. and when therei sm ore backpressure on the feeding side of the pump it doesnt have to work as hard to get the water into the pump, thus facilitating better flow rates.

[bxa121]

he may have apoint.."best 3d marks" remember??

[MaxxxRacer]

huh?

[STEvil]

yup, positive pressure on the input side of the pump helps a ton for flow.

[Ancient_1]

The pressure would not change because of volume only height. The diameter of a colum of water 1" in diameter 10' high will have the same psi as one 100' in diameter and 10' high. The added volume had no effect on the flow, but as max said lowering the restriction to the pumps inlet can make a big difference.

But in most of our system we run a close loop where there is no real effect of gravity on flow any water that goes pu against gravity also will get pulled down with the same force and will cancel each other effects.

As far as having the pump on the bottom of the loop the only real advantage is it will be easier to completely bleed, with it at the top any air will try to get to it and since they are not self priming it can cause the pump not to work and if run not pumping will burn out quickly (but even that will not happen in a fully bled system).

[Craig]

Welcome to Xtreme!

If you'd like to see even better flow in your rig, here's a tip.

Dump the bay res.

Install a large DIY res on the case floor right at the pump's inlet with as oversized inlet connection (keep it very short) as your pump can be set up with.

A oversized inlet line that's very close coupled to the res is the best arrangement of all.

Got any hard 90 degree fittings in your loop? Dump those next and replace with long ninty copper fittings.....less restrictive=more flow.

Hope this helps.

EDIT:

All that res did for you was to shorten your inlet to your pump, so it didn't have to work so hard to get water.

If your runnig a 3/8" line arrangement then you'd also get some modest gains by increae of line size to 1/2", but it also makes things harder to route.

[TheDude]

I have my system setup exactly as Craig described and agree that it is the best way I have found after much experimentation. :thumbsup:

[Butcher_]

Centrifugal pumps really don't like pressure on the intake - adding a res with a wide freed to the pump reduces that pressure a lot so you get good results. That's nothing new though, people have known this (including swiftech) for a long time now.
I always plumb by res straight into the pump intake personally.

[MaxxxRacer]

The only prob with having the res at the intake of hte pump is bleeding.. the reason i love my res its becuase i can have it above all other components which makes bleeding INSANELY short. about 30 seconds to fully bleed the system...

[Butcher_]

I find the water flow is uaully enough to push any air out of the system pretty quickly and air collects in the res just fine even with it at the bottom.

[MaxxxRacer]

just easier.. lol.. and besides i have 5/8ID tubing with 3/4 barbs on the pump and 5/8 barbs on the res.. I think it is the LAST point of restriction in my setup.. lol

[Craig]

Yes having the res high up makes instantaneous bleeds easy.


Bleed time is still short in a system with the res right at/on the pumps inlet. Most of the air is gone in under 30 min. & remainder by the morning of the following day. Not much to ask for a gain in flow rate. And flow rates are higher in such a system when compared to one with the pump in the bottom of the case, but the res up high & still having the line going to the pump from the res. Most of all if there's a 90 right at the res's outlet, as so many do when installing those bay res. Just as the thread starter saw quick gains when he shortened his return line to the pumps inlet, in his case by adding a res.

The gains aren't often large Maxxx, though they sometimes can be as the thread starter found, but as I recall seeing in one of your posts not long ago, ".1 or .3C I'll take any gains I can...."

IIRC that was in regard to having the pump before the rad or after it.


Same with flow gains eh?

Try connecting up to a res with just a couple inches of 3/4" copper, then recheck your flow rate. And I suspect this may shorten lines for you as well. And if my Z likes this best, I"m sure your RL will to as it's got a equally large thirst.

[MaxxxRacer]

Craig, you had to go with the low blow didnt you. Using my own words againts me. lol.

Tell you what, I will try it first with the res up top like i usually do, and then i will put my res on the ground along with the pump with an inch or less of 5/8 ID tubing. Either that, or I will get a milk jug and attach it to the pump with some silicone sealant and put a hole in the other side of the jug and put the tube from the video card there.

I will have to use the crapy onboard temp sensor on the nf7-s for temp readings, so if there are gains, i may or may not be able to measure them.