Project Illuminated - XtremeSystems Tribute

[NaeKuh]

What you said makes sense but let me make sure it's in the context of what I intend to do:

Within Pedestal 24 (loop 1) = Reservoir > Pump > NB/SB & CPU > Rad (Monsta) 1 > Rad (Monsta) 2 > Reservoir

Are we speaking of the same/similar thing? Maybe I need to take a look at some of Martin's archive files... do you have any suggestions?
thx.

Where's your gpus?

LOL that looks like a normal loop but you only listed 1 pump... which is okey, but im wondering where the other 3 come to play :P

Lets see you have 3 d5's and 2 monstas...

EKres Assy + D5 -> Gpu 1, 2, 3 -> a regular D5 -> Monsta 1 -> EK ResAssy.

Loop 2:
EkRes Assy #2 -> board -> Monsta -> EKRes Assy.

(i would QD both loops)

You can also bring the tubing from the bottom up for a nice bling effect.

Loop3:

EK Spinner -> Ehiem -> Ram?? Hard drive?? -> Small radiator -> Ek Rad Assy.

On the monsta's i would push/pull yates, and remove the side covers so i could get more air though the radiator.

[xaxis]

... admittedly this would all be much easier to describe with a diagram/flow chart... perhaps I'll get to work on that? Then we can continue this conversation with a little more mutual footwork to rely on. Sound good?
I'll try and finish it by tonight.

[NaeKuh]

LOL.... sure...

[MomijiTMO]

O__O

OMG

[Prince21]

You are one rich guy...LOL all that gear must cost tones..the value of the PC must be extreme...

[xaxis]

I hope the below images will help make clear my intent as far as my GPU loop is concerned. Note that I turned everything sideways for ease of viewing and that the actual bits I'll be using will change by the time I install it. Also, if your thinking, "hmm those reservoirs sure look unstable..." they won't be after I've installed all the corresponding UN Designs and EK mounting equipment.







Loop Redundancy
I still intend on providing this 3 series loop with two hot spares for redundancy. This means I'll be purchasing 2 more D5 pumps. These spare pumps with about 50 microseconds notice will have the ability to take over in place of any failed pump. I'm implementing two just in case two of the pumps fail simultaneously so there is little chance the tri-series loop will falter. I've been working on the control logic which has been a breeze and will be housed in a tiny box and built on a tiny 1.5" x 2" PCB. It basically will be connected to three flow rate sensors that will be installed at the inlet of each pump. In the event that it detects a drop in flow rate that quantifies a dead or dieing pump, it will immediately activate a pre-primed spare. 2 valves will also activate, one directing water flow from the reservoir to the spare pump and 1 that allows water pumped though the spare pump to connect to the rest of the loop at wherever the failure point may be. ... Once I've had a chance to diagram, purchase needed parts, and finish the control schematic I'll post more details on this.

I intend on doing something similar for the second "dual monsta" loop as well as for the hybrid phase change/tec system I'm working on. (Not going to get into this now).

I'm always willing to take a critical beating related to my ideas so feel free to be frank, make suggestions, words of advice, ideas, etc.

[bianco]

wow... looking like a very nice tribute xaxis.

[dingdong555]

O - M - G!!!!

[xaxis]

wow... looking like a very nice tribute xaxis.

O - M - G!!!!

Thank you both.

I was wondering if anyone had any ideal sources related to Electrically actuated ball valves that are sold with the intent of water cooled computers in mind or are easily modifiable enough to be integrated into a standard high flow water cooling loop like mine shown in preview below. Finding the right part is crucial when it comes to the aspect of building in pump redundancy and is holding me back from completing the controller schematic...

Also if anyone has any ideas related to series pump setups + redundancy I'm all ears.

Anyway, I've been scouring the net to be sure, but if anyone has anything to say about this or anything else really it would be appreciated.

[Pedalmonkey]

My only thing would be to flip ur inlet/outlet layout. You'll get cooler water to the GPU's from the rad, but thats more of a personal thing b/c temps will probably be almost identical.

[NaeKuh]

nonononono....

never have a pump push into a res bro

Also why do you have so many radiators.. i wanted to ask that..

any noise advantage you had by having all those rads will be taken away by your Single Stage.

[xaxis]

nonononono....

never have a pump push into a res bro

Also why do you have so many radiators.. i wanted to ask that..

any noise advantage you had by having all those rads will be taken away by your Single Stage.

I've read not to push a pump into a res but I can see no reason not to, especially when I'm not actually concerned with increasing the flow rate. Maybe you have some insight on this?

Noise is irrelevant. The machine along with my other computers will be placed in a sound proof server/computer closet with environment controls.

Also, I've decided I'm going to build a hybrid tec/phase change system... It will make more since when I have time to organize everything and demonstrate the schematics I have so far. Basically I'll be able to switch from TEC to Phase via a switch... two seperate linesets will be attached to the same "condensor" block, which... ah, I'll talk more about that later..

Anyhow, as far as pump into res... reasons please?

[SNiiPE_DoGG]

how the hell are you gonna fill with the pump outlet on the reservoir? you cant. the pump will try to draw water from its inlet and have nothing there...

also, why are you using 3 pumps for a GPU loop? FC blocks are not restrictive and I doubt you will be voltmodding so ~1gpm is perfectly adequate. - in fact as hondacitys (maybe HES but I dont think its him) recent tests have shown FC blocks can perform 95% the same with only .5gpm per block. Also d5's dont die often, in fact it is extremely rare, so I might understand a second pump for redundancy but a third, no.

I understand that you are trying to be "extreme" but 3 pumps in series for a loop with FC gpu blocks is just plain stupid.

[exemay]

nice looking build there bud! the only critisism id have is same as NaeKuh

why have pumps pushing into res??? its utterly pointless, because u wont get any gain atall.....if u want 3 pumps then either put them in series, or put the pumps between the radiators to stop such a big pressure drop!

having pumps feed into res's is COMPLETELY pointless!!

[xaxis]

how the hell are you gonna fill with the pump outlet on the reservoir? you cant. the pump will try to draw water from its inlet and have nothing there...

also, why are you using 3 pumps for a GPU loop? FC blocks are not restrictive and I doubt you will be voltmodding so ~1gpm is perfectly adequate. - in fact as hondacitys (maybe HES but I dont think its him) recent tests have shown FC blocks can perform 95% the same with only .5gpm per block. Also d5's dont die often, in fact it is extremely rare, so I might understand a second pump for redundancy but a third, no.

I understand that you are trying to be "extreme" but 3 pumps in series for a loop with FC gpu blocks is just plain stupid.

If my understanding or intuition on this matter is correct, a system first primed (filled to full volume) with fluids, then connected as I've suggested, will have no trouble drawing liquid from the reservoir (please explain in greater detail why you feel/(perhaps know) this won't work. It seems to me if water is returning to the reservoir inlet in a loop that is fully occupied with liquid it cannot magically have "gaps" in the loop where liquid is not present. So why can't the water be drawn out downwards via the res outlet. I'll test this soon perhaps.

Your idea of stupidity is perhaps my idea of harmless fun. One thing, even if this idea is stupid it's certainly not "plain". Example of my kind of stupidity: (fun) ... the odds of any of my 4 16 hard drive (4x4x4[raid5]) fileservers linked with some extremely high speed infiniband cards to operate as a RAIN that sits on an array of batteries with enough juice to power the systems to rewrite and check parity on all the drives a million times over (... well 12 hours, pretty impressive for a UPS), which itself is powered entirely on solar so the only real chance of outage is at night... Also, it's contained in a faraday cage with a number of other safety precautions taken and has no internet/ethernet connection of any type... All backups take place via manually walking to the servers with the encrypted data medium, entering in a 30+ digit keycode within 4 minutes or else a home brew thermite compound will ignite and destroy every drive in the systems in about 27 seconds, not to mention the biometric authentication.... Now, I have calculated the risk metric using a couple of standards and the probability of me losing any data (excluding natural disasters as a cause or a memory lapse) is somewhere around 1: (insert number of atoms in the universe) ... however this doesn't stop me from doing it... I know it's excessive, I know it's "stupid", but it still manages to place a smile on my face every single time I think about it. You see, sometimes a lack of good reasoning creates something unexpected and if lucky, even something new. One might summarize: my reason based on reasonless reasoning resonates raisins and.... oops, guess I last track on that one... I tend to ask myself... why not first, without ever taking the time to answer the question.

Again, if you're correct about the inlet feeding issue I'll openly admit I had better flip some stuff about.

Appreciate your frankness.

[xaxis]

nice looking build there bud! the only critisism id have is same as NaeKuh

why have pumps pushing into res??? its utterly pointless, because u wont get any gain atall.....if u want 3 pumps then either put them in series, or put the pumps between the radiators to stop such a big pressure drop!

having pumps feed into res's is COMPLETELY pointless!!

One point I might poke into the vast realm of pointlessness... less tubing, at least when one insists on using the equipment I am which some say is needless... Perhaps it will give me room or motivation to make a need?

[NaeKuh]

Hard mount 2 pumps like so.



you can get a cleaner appearance and ditch the second res.

The reason why i heard, and someone can double check, is i heard pump into res and you lose head pressure and flow.

[xaxis]

Hard mount 2 pumps like so.



you can get a cleaner appearance and ditch the second res.

The reason why i heard, and someone can double check, is i heard pump into res and you lose head pressure and flow.

Best part is... I'll be able to do/try both. Ordered a number of different flow rate meters to experiment about with...

[exemay]

The reason why i heard, and someone can double check, is i heard pump into res and you lose head pressure and flow.

i dont think you loose pressure/flow overall....its just pointless feeding into a res because there is a massive pressure drop going into the res.... so your better off just using 1 pump

[xaxis]

i dont think you loose pressure/flow overall....its just pointless feeding into a res because there is a massive pressure drop going into the res.... so your better off just using 1 pump

Thought: MASSIVE overall reservoir. Continuation of thought: More liquid (bigger res) allows heat to diffuse much further before being reused again. No?
EDIT: btw, I agree with you, I don't think you loose pressure/flow overall either especially with the parameters I mentioned.

[SNiiPE_DoGG]

Wait am I seeing this right or does the top behave differently than I thought it does? res is on top of the pump outlet and the tube going to the pump inlet is at the side?

[exemay]

Thought: MASSIVE overall reservoir. Continuation of thought: More liquid (bigger res) allows heat to diffuse much further before being reused again. No?

well not really...there is a slight relationship between dissipation and res size, but its kinda pointless using 1 huge res when u have 3 120.3 radiators!!!

if i was u id go for the 3 radiators feeding into a res, then into a pump. then into a second pump and leave it like that,

RAD>RAD>RAD>RES>PUMP>PUMP

this way u will get the benifits from having pumps in series, u wont have a pressure drop from feeding from res to rest and u will have alot of very very nice cooling form the 3 rads!!

[xaxis]

Wait am I seeing this right or does the top behave differently than I thought it does? res is on top of the pump outlet and the tube going to the inlet is at the side?

Res outlet is directly on top of pump inlet. The (to be) tubes, etc on the side are indeed pump outlets.

[SNiiPE_DoGG]

OHHHH so the res is indeed connected to the INLET (water being sucked into the pump from the res) then that is perfectly fine - I have that on the 2 pumps in my system. - I thought the tops behaved like a normal D5 top where the outlet of the pump is at the top.

[xaxis]

well not really...there is a slight relationship between dissipation and res size, but its kinda pointless using 1 huge res when u have 3 120.3 radiators!!!

if i was u id go for the 3 radiators feeding into a res, then into a pump. then into a second pump and leave it like that,

RAD>RAD>RAD>RES>PUMP>PUMP

this way u will get the benifits from having pumps in series, u wont have a pressure drop from feeding from res to rest and u will have alot of very very nice cooling form the 3 rads!!

Hmmm... I'll give this a try however I'm going to attempt to merge the reservoirs into 1.. They'll have 1 inlet, outlet... just like a singular res, but with the extra volume. so Rad>Rad>Rad>Res>Pump>Pump>Pump(why not the third?)