My ultimate dream Watercooling case idea.

Power5 said:
I have knocked it down to 20 fans. 12 on the quad rads and 3 on the triple rad. Then I have 3 intake fans for the rad areas, just to help out a bit. The majority of the air will be sucked in through the top and bottom mesh areas. I also have a typical 140mm exhaust and 200mm HDD fan. Least I could cut it down to would be 17 fans.
Click to expand...

I'll ask again, because 'ultimate dream watercooling case' is a very vague concept. What are the results you want to get? What variables are we trying to maximize and what are we trying to minimize? Each time you double the ability of the entire system to cool, you only reduce the temperature by at most half the difference between the hottest part in the system and ambient temperature.
 
I find it hard to believe that 4 fans could suck air through 2 rads without running louder than 4 fans for each rad. Rad compartments are sealed from main area of case. Pumps will be in top and bottom section as well so also not visible. HDD would be installed so the connections are on the back side, so they would be easy to hide.


I thought I mentioned it before. This is attempting to cool 3x 5970 and CPU+chipset.
 
Last edited:
most ppl run a single 3 fan rad for tri sli gpu's motherboard and cpu.
fair point on the hdd's
to the fans look at the angle of the blades you will soon relies what i mean.
bearing in mind that the cooling system will only go as low as the room temp.
best idea is have to sensors one inlet of rad and the other outlet of rad to see what temp difference is
and let me point out ppls have run tri sli 8800gtx on a single 3 fan rad and the 8800 run way hotter than the 5970.
 
good god, if you only accomplish one thing in your life, it better be building that case!
 
Still not quite sure about the airflow problem. I would love to be able to run 3 5970 at 30c. That would be 5c above my ambient. hell even 40c would be great. I just dont see how that would be possible on a sinlge 360rad. If it was that easy guys would not be running 480s and 560s. I do appreciate the opinions and since I have no real watercooling experience I take it all under consideration. I have to believe that trifire 5970 will generate a considerable amount of heat.

Going to pick up some angle this week and get started in construction. I know I am going to run into problems when I try to start making the side panels as I have no bender and need a piece of acrylic precision bent to match the aluminum side panel.
 
most use a heating element for bending heat the plastic so it gets warm then bend
ok it you run 4 rads just run 4 fans bottom pulling air in and out the rads and same on top trust me that many fans is going to be like tonado outside your window and the humm will drive you insane
trust me i have 2 120 fans and i have to were headphones
 
I'd call it the "I-Beam." :D

I like it for the most part, but I hate knobs. Lose those and replace the fan controller with something digital. If you must use knobs make them smaller and less gaudy than those in your concept. Or use just a couple less of them. I'm also not sold on the feet. I'd go with either a roll bar design like Corsair and Apple use or home audio/component style feet like Silverstone uses.
 
Dan_D said:
I'd call it the "I-Beam." :D

I like it for the most part, but I hate knobs. Lose those and replace the fan controller with something digital. If you must use knobs make them smaller and less gaudy than those in your concept. Or use just a couple less of them. I'm also not sold on the feet. I'd go with either a roll bar design like Corsair and Apple use or home audio/component style feet like Silverstone uses.
Click to expand...

I use the NXZT Sentry LX to controll all my fans. The only problem you may run into based on how many fans/how much juice they use it can only deliver 4 watts per channel. If you are using low rpm late yoon's or something like that you may be able to daisy chain them to the same channel and not have a problem.

Just a suggestion however as some of the knobbed ones are pretty cool too
 
EndersShadow said:
I use the NXZT Sentry LX to controll all my fans. The only problem you may run into based on how many fans/how much juice they use it can only deliver 4 watts per channel. If you are using low rpm late yoon's or something like that you may be able to daisy chain them to the same channel and not have a problem.

Just a suggestion however as some of the knobbed ones are pretty cool too
Click to expand...

I love both of those fan controllers. My buddy has the NZXT one in his HAF 932 and it works amazing, and it's easy to use to boot. I also have always like Zalman stuff, and I like how it kinda goes with the color of the case.


OP what is that LCD on the top for? Is it going to be used a temp readout or did you plan to use it as a readout for anything such as programs also?
 
EndersShadow said:
I use the NXZT Sentry LX to controll all my fans. The only problem you may run into based on how many fans/how much juice they use it can only deliver 4 watts per channel. If you are using low rpm late yoon's or something like that you may be able to daisy chain them to the same channel and not have a problem.

Just a suggestion however as some of the knobbed ones are pretty cool too
Click to expand...

I really like the NXZT Sentry, but I don't know that I'd ever use one. I'm not into fan controllers. It does set the standard for cool looking fan controllers though.
 
Dan_D said:
I really like the NXZT Sentry, but I don't know that I'd ever use one. I'm not into fan controllers. It does set the standard for cool looking fan controllers though.
Click to expand...

I use it due to having a massive amount of fans in my case (3 230's in the 840, 6 120 Noctua P12's). It keeps them all running quiet enough I dont hear any of them (besides a slight hum from the GPU fan). For me it was about making sure they were running fast enough to cool my massive setup, but not full rpm all the time.

It has been a worthwhile investment for me. I am not sure how it would handle a watercooling enviorment however as the temp of the water may not correspond to the temp within the case.
 
Power5 said:
Still not quite sure about the airflow problem. I would love to be able to run 3 5970 at 30c. That would be 5c above my ambient. hell even 40c would be great. I just dont see how that would be possible on a sinlge 360rad. If it was that easy guys would not be running 480s and 560s. I do appreciate the opinions and since I have no real watercooling experience I take it all under consideration.
Click to expand...

Have you thought about the possibility that they share the same predicament, and that's why they ended up buying 480's and 560's? One-upsmanship is a powerful force at the top end of any hobby, and your 'ultimate dream' will only be ultimate until some other guy builds one 25% larger. 5C is entirely within your reach, though. It's just not gonna make a huge difference in overclocking versus 10C or 15C.

It's my understanding that 3x 5970 is not even possible, if you want them to all work correctly. Crossfire takes a huge massive performance hit on the 4th GPU (often going into negative returns over three cards), and anything over 4x Crossfire starts to disable in drivers - people were reporting that 3x 5970's results in 3 cores running full speed and 3 cores running at 1/3 speed.
 
Last edited:
yep, trifire 5970 is not possible, since that would involve 6 gpu's... what i was trying to get at in my previous post was that my friends rig was completely over the top, especially since he was only running a single GTX295...

i would say that one set of fans on each rad would be plenty, particularly in the top compartment... i agree about sealing the top and bottom compartments, so that the main chamber has positive pressure, and cables are easier to manage... i would think that 13 fans would be easily possible, with intake/ exhaust for the main chamber and 1 set of fans on each rad...
 
Sucks about 3x 5970 doing no good. Well, how about 2x 5970 and a triple rad cooled gtx 295 physics card and the other quad for the CPU+everything else. Kidding of course. Hmm, I could shrink the case to have 3x triple rads + 1x double. Not sure how that would look proportionally though. Or just go with 5890s when they come out and be happy that they would never get hot at all. Or according to some questionable reviews, 3x fermi will need some massive cooling to get above idle.

I like that zalman, and that was the one I was thinking of using. The renderings were just an idea if there was a 8x option. If 4w would power 4 fans, that digi is an option though it only has 5 controls it looks like. Anyone got pictures of actual units installed?

Rads I planned to use are the high airflow ones, not the super restrictive types. And only the 1.5" thick ones to keep the case from being excessively wide. Not sure how loud 20x 800rpm fans would be, but that was the idea. Wondering how people have custom installed filters as well. I havent thought of a clean way to have filters that are easy to remove yet.

Anyone know how to upload something to sketchup? I have modeled the case in the program, but it wont upload the component for some reason. Its a 24mb file.
 
Let's get theoretical for a moment. I've picked out some convenient numbers and run the calcs for you. If you get 4x 120mm fans to put out 70CFM each against the restriction and run this efficiently mixed airflow through a perfect radiator, in dry air they can cool a 560W load down to 3C above ambient. Settle for 12C above ambient, and a single 120mm 70cfm fan will do that. That's just a trivial result of the amount of air being heated per second. Actual performance will probably be quite different, but insofar as radiator air heating is the limiting factor in the cooling system it will perform roughly proportional to this multiplier.
 
Haha, cool, but you are way over my head. So, a single 120mm fan will cool 560w load, or 4 will do that? Is this a single 120mm fan on each rad? 70cfm is high and most likely running any fan at near full speed. I was going for maybe 40cfm so the fans are as quiet as possible.

What would 15 fans do at 40cfm?
 
It's a shame you didn't mount the PSU vertically as so many have a 12 cm fan these days... couple that with a right angled power lead (connected on the base of the case) and that should make better use of the space.

Of course I've absolutely no idea of what effects mounting the PSU that way could have but to me it seems more logical to have the 12 cm fan on the outside of the case blowing air straight out and would serve as an extra exhaust... of course that could just serve to overheat the PSU (I'm certainly no expert).
 
PSU is on its side. Large top mounted fans on PSUs are intakes, they exhaust air out the back of the mesh. Some just have an exhaust fan on the back and suck air through the PSU case. Standing it on end will increase the height of the bottom chamber by at least 3". Either way, its sucking in cold air through the side mesh, and exhausts it out the back. The biggest problem this creates is due to the fact that there is an intake fan in the back to help supply the rads. So this could easily suck in the hot PSU air.
 
what he is saying insted of using 4 fans to move 70 cubic feet of air is to use one fan that can move the same amount
or thats how i have understood it
and 560w load aka heat is shifted by a 4fan perfect rad so real speak 2 of these rads would pull close to 800 - 1000 watt of heat
i think what he is saying is what i said you could have 8 fans pulling 70cfm each they wouldent cool any more then say 4 fans with the same speed because of the ambint air temp moveing this air faster will not reduce temps
 
you could put the psu on the bottom the 120mm fan pointing down as you say this would pull cold air in and out the back ideal this also works the same way you have it now
 
I have it on its side to allow using another quad in the bottom. I had it on its fan before, but when I first designed it, I had push pull fans so there was not enough room for a PSU between the rads. Now, there may be, but it would be very tight.

I have no doubt a single fan running at high speed will do the same work as 4 fans running at 1/4 the speed. The slower fans would be obviously much quieter though.
 
as i was saying befor i feel the 2 quads in the top with 4 high flow fans will be more then enogh to cool this system and as some one else said running 3 of the gpu's you wont will drop in performance
 
i was going to build an extaernal res pump and rad unit on my system and from what i found out 2 quad rads with 4 highflow fans would be more then enogth to cool my system and then some
but only to room temp if you wont colder then you have to put the rads in a colder place but the problem with this is condensation
 
the way i would do thios is build the case to fit everything you wont run a single rad and other parts so you can build it for less then slowly add the other parts as you get the money or find that it dose not cool as you wonted
 
Power5 said:
Revised. The idea was a quad for each GPU. Thinking this could handle 3x 5970 now.

newfrontw.jpg


newbackw.jpg


It is large though.

22" monitor and HP mini tower matx case in photo.

actualbackw.jpg


actualfrontw.jpg


I could lose the 92mm intake fans. I was just unsure if the rads would be able to pull enough air without help.
Click to expand...

were did you get the card and how did you do this this is bloody ace idea for mock ups
 
just another note i would move the psu to thew top and the rads to the bottom that way there closest to the colder air in a room
 
Power5 said:
The biggest problem this creates is due to the fact that there is an intake fan in the back to help supply the rads. So this could easily suck in the hot PSU air.
Click to expand...

Aaah I see... I usually keep all rear fans as outward flow and not intake myself but then again I use air cooling instead of water cooling as I don't overclock, although I do love the look of water cooling but for me it would increase my CPU temps unless I spent loads of money.
 
I work in this industry. I have plenty of corrugated paperboard that I can CAD cut on a table which uses a reciprocating exacto blade. Just draw in up in our CAD program, and the table cuts it. Can cut up to 5/16" thick double wall corrugate, posterboard, foam core, thin acrylic.
I had planned to use this technique when I got to the parts that will need to be CVC. I can cut the mockup and test it in the frame to see how it fits. Move the holes as needed and recut. When perfect, I can send the file to the CNC person. Too bad the table cuts with a knife not laser, huh? Table would cost you a few bucks though. Ours is a demo $90k unit. Can handle up to 66"W x 120"L sheets.


I may remove the rear fan as I hadn't thought of sucking in the PSU exhaust.
 
lucky for some im just downloading a trail of 3ds max to mess with hopeing i can draw up a case or wc case id like the card would be cut the old way with me a stanly and a metal ruler lol
i guess a large rolder printer be cool for printing the shaes on card
 
I don't use 3ds max for the Cardboard mockup, just for my renderings. For the cardboard we use a program called ArtiosCad. You may be able to find a trial of that, not sure. For you I would just use sketchup as I am sure you wont want to buy a $4000 program like 3dsMax. 3dsmax is not a fun program to design in. I only use it for rendering, but since this company only has 3dsmax, I have to model in it as well. I assume sketchup can export a line drawing from the ortho view but have not really gotten too far into sketchups abilities.

I have decided to remove all rear intake fans. Top fan would also suck in warm air from the case exhaust right below it. So that has taken the total fan count to 18.
 
What I was saying was -
if airflow is a primary limiting factor,
then whenever you halve the amount of airflow,
you double the value of dT, where coolant temp = ambient temp + dT.

If airflow is a limiting factor, and one fan results in a 20 degree dT, then two fans will result in a 10 degree dT, four fans will result in a 5 degree dT, eight fans a 2.5 degree dT, and 16 fans a 1.25 degree dT. 32 fans nets a 0.625dT, et cetera, et cetera. Keep in mind that as you add more fans, airflow becomes less of a limiting factor, and you start worrying about the thermal interface between the water and the radiator fins, or between the waterblock and the water, or between the CPU and the waterblock.

There is no such thing as "airflow necessary to radiate a certain number of watts", or "how many fans would it take to get to 5C dT," my point is that the most simplified equation you can come up with has three variables: wattage, temperature delta, and airflow.

With dry air and theoretically perfect equipment (Real Life is always more complex than Physicsland), based on my calculations 280cfm total will radiate 560 watts at a 3 degree dT. Come to think of it, that's 1 CFM per watt for a 1.5 degree dT.
 
I would move the radiators to the bottom as well, in case of leaks. Water at the bottom, electronics in the middle, data on top. Party in the back.
 
Yup, obviously fluid dynamics is way past my knowledge base. So 7 40cfm fans would be sufficient to cool 560w to 3dT above ambient, and my current number of fans on my rads would take it to 1.5dT-ish since 2 fans are not attached to the rads.



There are radiators on top and bottom of this case, so makes little difference if I switch them. I am going to flip the PSU to breathe from the other side since this would be the most direct wiring to behind the mobo.
 
i know 3ds past my knoladge base so is the google tool you told me about power5 google one ant good enogth and well the other i ant trained and the thing i was trying to follow makes no sense
 
Dasimpson said:
i know 3ds past my knoladge base so is the google tool you told me about power5 google one ant good enogth and well the other i ant trained and the thing i was trying to follow makes no sense
Click to expand...

Wow, stop posting from your phone. :confused: Your posts have become a headache to try to understand. Trying to follow your posts are making no sense. :D

Not sure what computer drawing experience you have, but google is going to be the easiest to use for a novice. Like any new program you have to still get past the learning curve of how its designed to be used before you are able to utilize it well. Took me about an hour to get decent at it, but I have extensive experience with 3d modeling programs.
 
You must log in or register to reply here.
Back
Top