The Hardware: Continued...

As far as pumps go, there is a broad range of pumps available  that pump outrageous amounts of water.  This is not a job for a piece of crap Via Aqua 1300, or a low powered Rio, or some other knock-off $20 pump.  The radiator alone would bring a Via Aqua to a screeching halt.  I currently have a Danner 350 GPH utility pump that I'm going to do my proof of  concept runs with (just one computer), and will hopefully last me through the  end of Phase 1.  The pump I believe I'm going to use  come production time (Phase 2 and beyond) is this little jewel:.

fan guards, which prevents me from turning my fingers into luncheon meat ... which would be the certain outcome if you happened to get caught in this with the metal blades spinning at over 3000 RPM.  Finally,  something to discourage me from tinkering around inside once I'm done!  I  will be using two of these side by side in the back of the cooling chest, giving me a net flow of 1100 CFM through the cooler.  One thing is for sure:  I will have to start looking for a few paperweights, or I will  have to contend with my paperwork floating around on the air currents while I work.

Construction:

Now that I had all the parts I needed, it was time to finalize my plan for construction.  I wanted this system to be as easy to assemble  and disassemble as possible.  With the ease of draining through the air trap/reservoir, I just needed a way to remove one machine from the system.   What I decided to do (after flirting with quick disconnects of all kinds), was use two way hose barb fittings (1/2" to 3/8") to connect between the 1/2" ID  hose from the manifolds to the 3/8" ID hose from the internals of the cases.  When one machine needs removed, I just need to drain the reservoir, undo the hose clamps to the two-way hose barbs, and pull the hose off, keeping any dripping far away from my precious components (*gollum* *gollum*).  The incidental benefit of this is  not having to reseat your CPU block every time you mess with your computer's internals.

To do this, though, we need easy draining.  Since the tall vertical air trap is a nice gravity well, instead of having a 90 degree bend in the bottom leading inside the cooling box, we will use a T junction, or similar  arrangements, with one side going to the cooling box and the other going to a drain valve.  Since butterfly valves will be built into the internals of the cooling box (to help  prime the pump around the radiator via a bypass line), any  coolant left inside of the box will remain there, though all except maybe eight fluid ounces should drain handily.

Problem two is due to the restrictiveness of the radiator.  Even the Mak 4 can only pump a maximum of 3.6 GPM through the radiator, though if I was to bypass the radiator and only divert 2 GPM through it (the rest passing from the reservoir directly to the pump), I could get a net flow of almost 8 GPM.  That is a 4.4 GPM difference!  By putting a valve on the bypass, I will be able to tune the performance of my system to use all of the components to their maximum effectiveness (in this application).  Many people believe that this will be an utter failure, and that I will end up just having to leave it closed to get best results.  I hypothesize that while I will get higher idle temps this way, my peak temperatures will be lower since  the blocks will be working more effectively.  We shall see who is right  shortly.

ever do again (though I should never have to worry about a self-contained system again, either).  Open hoses with coolant in them do not  belong in a case with components.  Period.

Phase 1:

This was an adventure from the start.  Over a hundred dollars worth of Schedule 40 PVC was dumped onto my living room floor.  I whipped out the primer and glue and got to work. First off, I would like to detail how I glued the PVC.  Consider this a  how-to for those of you that think you know how this works.  I had two  types of PVC:  the white kind, and the black kind.

  The white stuff  requires two types of glue to form a seal.  First you apply primer, rubbing  the brush around until the PVC starts dissolving and looks milky (do this on  both parts on all areas that will be connected), and then you apply PVC glue until the primer and glue are completely mixed (on both parts), and then you slide the parts together and hold on for dear life.  It takes a couple of  minutes for the glue to set, and it will want to push apart, so you need to hold the joint in place.  The seal you just made is now permanent and should never leak.  In the places where the PVC and black PVC (ABS) is joined, you need PVC to ABS glue, or you will quickly start leaking.  This stuff takes  a while longer to set, so make sure the TV is on or something, but you don't need primer for this to work, just the glue.

Ed: Looks an awful lot like a spud gun I used to have :)

Before you glue PVC, though, just make sure you have thought it through, because once you glue it, you can't go back.  The seals you make are VERY  PERMANENT.

Okay, enough with that, lets show some pictures.  First, I  made two identical PVC manifolds that splits one line into four.  All of the internals of these manifolds are in 1" PVC.  Since neither pump (old or new) is designed for over 3/4" flow, this should reduce any resistance from this part of the system. In the branches that aren't use,  I'll use PVC plugs and Teflon tape.  When it comes time to add

more machines, I just need to unscrew the plugs and put in hose barbs.  The process of adding more machines should take less than 45 minutes per  machine.  How is that for efficiency?

More of the Phase 1 Construction