• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Loop tweaking...

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

torin3

Member
Joined
Dec 25, 2004
Well, since the rumor is that the EK full coverage GTS waterblocks will be out this week I figured I'd ask for a little help on tweaking my loop.

I currently have two different sized radiators (one is 1x120, the other is 2x120 fan sized), and a maze 4 CPU block, and it is powered by a Eheim 1250.

The other components I have on hand or I am getting that I'm planning to add are a chipset block and (obviously) the 2 EK full coverage GTS blocks, plus a DD Mag II pump.

The loop I have is pretty much all plumbed with 3/8" ID tubing with 1/2" fittings. Well, since I use the worm-screw hose clamps I finally got tired of dealing with the very tight fittings. So I started to get 3/8" fittings for my new stuff, but on second thought, I'll go ahead and get some 1/2" ID Tygon and upgrade my whole loop to 1/2" ID.

So, here is where I'm looking for the help in tweaking. I'm wanting to split up my components into two loops. One for the video cards and one for the CPU/Chipset blocks. Which radiator, and which pump should go on which loops? I'm willing to switch my DD Maze 4 CPU block for another block if it will be signifigantly better. I will be overclocking the cards, and the CPU is currently overclocked by about 1 Ghz.

I know the chipset block won't really help my overclock, but I'm adding it mainly for reliablity. I've lost one chipset fan already and didn't notice it for a while.

Thanks for the tips!
 
I'm still a novice at this but in my opinion i would run the 2 GTS blocks of the dual 120 radiator and keep your CPU / chipset on the single. I would imagine the 2 GTS's are going to put out alot of heat...
 
I would honestly not waste your money on a second loop. The benifits will not be that much difference. Your ability to cool remains constant. And your ability to heat is also constant. So the only factor is the flow rate, and your current pump has lots of flow. The overall result will be the same overall temp of the water.

OCers forget how much energy it takes to warm up and cool water. It stores allot of heat, so it also takes alot of cooling and heating to change temps. I would figure that your overall temp is pretty good.

You can test this. Compair the temp of water going into the radiator with the temp of the water going out. You will probably see that it is less then 1c difference. This means that one loop of the water will cool 1c, and heat 1c. That means that if you add more flow, this ratio would get closer together. You might end up with .5c difference. Now how much difference does these two temps mean on overall cooling and heating... almost nothing.

See the heating and cooling remains constant, unless the temps difference of the hot and cold surfaces move closer or farther away from 0.

Your money would be better spent on adding a radiator, either in the middle of the hotside loop, or just inline with the cooling loop. I would change the loop to have the smaller radiator after thoose CPU blocks, and before the videocard block, then drop the dual radiator after the Radiator. Additional Radiators will add more cooling, and thus will change the temp of your water.

However, my machine runs 90f MAX water temps. I have a 3 120mm radiator, and I have a high flow pump like yours. I had noticed that by increasing my fan voltage from 5v->12v that the temps went from 100f max to 90f. I also run a Radeon X1900XTX, which is a VERY hot video card. Much hotter then your Geforce.

I switched from a low flow (pump 1 l/min, and CPU block on my HTPC) to a high flow system (gaiming rig), and I saw a decrease in temps of about 1 or 2c... That change cost me around $300. WOOPIE:(

I actually contribute most of my change to using a different (better, and more expensive) WaterBlock.

If you REALLY want to test the performance of your machine... do this... Put a Temp Reader on your cooling or heating side... then heat up the water with prime95... after it steadies out... shut off prime and watch how slow the water temp drops. (got to run, more later)
 
GreenJelly said:
I would honestly not waste your money on a second loop. The benifits will not be that much difference. Your ability to cool remains constant. And your ability to heat is also constant. So the only factor is the flow rate, and your current pump has lots of flow. The overall result will be the same overall temp of the water.

Ok, you've lost me here. When I added in my 2 6800GTs into my old cooling system, I got about a 5°C bump in the temperature of my CPU. Now I understand that given a fixed amount of heat sources and a fixed amount of radiative capacity, I won't get a net overall difference in my system. But my radiative capacity is in unequal sized units, and the heat sources do not have the same operating temperature ranges. It makes sense to me that it would be possible to get some better performance on the loop that is more sensitive to the final temperature.

OCers forget how much energy it takes to warm up and cool water. It stores allot of heat, so it also takes alot of cooling and heating to change temps. I would figure that your overall temp is pretty good.

You can test this. Compair the temp of water going into the radiator with the temp of the water going out. You will probably see that it is less then 1c difference. This means that one loop of the water will cool 1c, and heat 1c. That means that if you add more flow, this ratio would get closer together. You might end up with .5c difference. Now how much difference does these two temps mean on overall cooling and heating... almost nothing.

See the heating and cooling remains constant, unless the temps difference of the hot and cold surfaces move closer or farther away from 0.

Maybe we are talking at cross purposes here. I'm talking about two physically separate loops, not placing components in a specific order in a loop to get the coldest water over the CPU.

Your money would be better spent on adding a radiator, either in the middle of the hotside loop, or just inline with the cooling loop. I would change the loop to have the smaller radiator after thoose CPU blocks, and before the videocard block, then drop the dual radiator after the Radiator. Additional Radiators will add more cooling, and thus will change the temp of your water.

Ok, it does look like you are talking about a continuous loop, or two separate loops with a chiller or something in it. Not what I was asking about. However I can't add another radiator. I was originally going to have 2-2 fan radiators but I messed up on my case design and can only fit in 1-2 fan and 1-1 fan radiator in my case. It is a custom designed acrylic system.

However, my machine runs 90f MAX water temps. I have a 3 120mm radiator, and I have a high flow pump like yours. I had noticed that by increasing my fan voltage from 5v->12v that the temps went from 100f max to 90f. I also run a Radeon X1900XTX, which is a VERY hot video card. Much hotter then your Geforce.

I switched from a low flow (pump 1 l/min, and CPU block on my HTPC) to a high flow system (gaiming rig), and I saw a decrease in temps of about 1 or 2c... That change cost me around $300. WOOPIE:(

I actually contribute most of my change to using a different (better, and more expensive) WaterBlock.

If you REALLY want to test the performance of your machine... do this... Put a Temp Reader on your cooling or heating side... then heat up the water with prime95... after it steadies out... shut off prime and watch how slow the water temp drops. (got to run, more later)

Hopefully I explained we aren't talking about the same thing. Sorry to have you write all that.
 
Assuming you're going to run two loops run the CPU & single rad in one loop and the GPUs and chipset in the other. That way IF you need more cooling to keep your CPU OC you can put a higher CFM fan on it ...
 
Last edited:
I was talking about 2 seperate loops verse 1... I wasnt talking about a peltier system... and triing to explain how disapointed your going to be when you spend the money to do this, and find out that you dont get any additional cooling compacity. I am telling you that if you take data (water temp data) and compair the systems, you will be disapointed. I am triing to strongly recommend AGAINST doing this. I am triing to give you a brief teaching in a complex world of physics.

I fear you will be spending allot of money, and you will be very unhappy. I beg you not to do this...

Now... If you find you have a very low flow, with an experiment (2 jugs, 1 jug with source water, and another jug for measurement) and you find you have LESS then 1 leter/min... hell lets say 2 or 3 l/min then you will see a change in the CPU temp, but not much, though you still wont see a difference in overall water temp.

So if you find that you have these flow rates, and you want to go to a second loop, then I would recommend that you add another radiator.

If you find you HAVE a high flow rate +10 l/min, then you can think about buying another radiator without ever adding a loop. However, if you measure your hotside water temp, and find it 90f then you will find that it is very hard to drop the temp of the water anymore.

However, you have NO data... so I would again, take my experiance... and the money I wasted, and NOT go buy MORE watercooling equipment.

3x120mm Radiator is good for allot of heat.

Also remember if you drop the temp of the water by 1c, you dont drop the temp of the CPU by the same amount, but significantly less.

I understand your desire to make a water cooling system better then what you have. But your still stuck with water, and without a cooler, you are facing limits that are extremely expensive to improve with itsy bitsy changes.

Mike
 
DD Mag II

I have recent experience with the DD Mag II LE and I had to get 3 of them after 2 rma's before I got one that is quiet. It actually isn't quiet, it is just not as loud as the other two.

DD and Jabtech were not helpful at all in my ordeal. So I would suggest another MFG and another Retailer.
 
GreenJelly said:
I was talking about 2 seperate loops verse 1... I wasnt talking about a peltier system... and trying to explain how disapointed your going to be when you spend the money to do this, and find out that you dont get any additional cooling compacity. I am telling you that if you take data (water temp data) and compare the systems, you will be disapointed. I am trying to strongly recommend AGAINST doing this. I am trying to give you a brief teaching in a complex world of physics.

I fear you will be spending allot of money, and you will be very unhappy. I beg you not to do this...

First off, I'm not going to be spending a lot of money on this, other than for the EK GTS waterblocks. I already have the second pump, and was planning on switch over to 1/2" ID tubing anyway. At most I'm looking at the cost of adding an extra t-line to split up the loops. As far as being unhappy, I'm not expecting W/C nirvana to result from this.

As far as the physics go, if you weren't talking about two interdependent loops, why the heck were you going into debunking component order? I wasn't asking about component order. Given water's heat capacity, once the loop reaches equilibrium component order makes no difference. Also, I have had physics classes, and with my hobby blacksmithing, I am aware of how this impacts the real world. I haven't run the numbers, but I figured I would ask here in case somebody had already done the work. I can look up the numbers and run the calculations myself, but that is a couple of hours of looking up the information and running the numbers that I was hoping to avoid.

Now... If you find you have a very low flow, with an experiment (2 jugs, 1 jug with source water, and another jug for measurement) and you find you have LESS then 1 leter/min... hell lets say 2 or 3 l/min then you will see a change in the CPU temp, but not much, though you still wont see a difference in overall water temp.

So if you find that you have these flow rates, and you want to go to a second loop, then I would recommend that you add another radiator.

If you find you HAVE a high flow rate +10 l/min, then you can think about buying another radiator without ever adding a loop. However, if you measure your hotside water temp, and find it 90f then you will find that it is very hard to drop the temp of the water anymore.

However, you have NO data... so I would again, take my experiance... and the money I wasted, and NOT go buy MORE watercooling equipment.

Again, comments where you tell me not to buy more stuff, when I already mentioned that I have pretty much everything already on hand lead me to belive you aren't reading what I've written. The tubing is going to run me about $30-40 at most.

3x120mm Radiator is good for a lot of heat.

Ok, maybe some pictures will help....

finish3.jpg

finish4.jpg

finish5.jpg

This is my system. I spent a fair amount on the case, and I'm really not going to modify it. I admit I had a bit of a design flaw, and had to change out a 2x120 sized heater core to a 1x120 sized heater core as the PSU was taking up part of the space for the larger core.

So I DON'T HAVE THE ROOM to add in a 3x120mm radiator!!!

Also remember if you drop the temp of the water by 1c, you dont drop the temp of the CPU by the same amount, but significantly less.

I understand your desire to make a water cooling system better then what you have. But your still stuck with water, and without a cooler, you are facing limits that are extremely expensive to improve with itsy bitsy changes.

Mike

Mike, I understand that {h(vi) + h(vi) + h(ch) + h(cp)} - {(Cw) + 2(Cw)} = Tx,
And that ({h(vi) + h(vi)} - {2(Cw)}) + ({h(ch) + h(cp)} - (Cw)) = Tx

But that isn't what I'm asking.

I'm not sure how to get across to you what I'm asking, but since you seem concered by how much money I'm spending, you can rest assured that it won't be more than $40.

(Well, until I decide to redo the design on my case to increase it to a 24" cube rather than a 18" cube. That will run about $200 in custom laser cutting fees, but then I can fit in 3x120mm radiators.)
 
QuietIce said:
VERY nice case! The lightning bolt is an interesting addition to the yin-yang ...

Thanks. Other than making it a bit too small, I'm very happy with how it turned out. The lightening bolt/yin-yang symbol is from a historical group I'm associated with that focuses on Mongol recreation.
 
Yeah, great looking case there.

If it were me, i would just keep it as one loop, and convert it all to 1/2'' tubing. It will do the best job at distributing the heat load running things in series as you are. Anyway -- thats what i think i would do.
 
1/2 Inch tube is really not user friendly. I strongly recommend 3/8" tube, because of its structural strength, ease of use, and generally if you studdy the loop and individual parts in regards to restrictions you will find that 8mm usually provides enough flow for most systems.

However 1/2 inch tube does look cool:)

BTW; if you didnt know already, Id also run one loop;)
 
Last edited:
i think someone needs to run a temperture test of 3/8 vs 1/2 vs 7/16 tubing, i would do it, but im more into getting my w/c loop together and playing some computer games lol. i might test it later though.


yea, one loop does sound best and much easier to create.
 
darkcow said:
i think someone needs to run a temperture test of 3/8 vs 1/2 vs 7/16 tubing, i would do it, but im more into getting my w/c loop together and playing some computer games lol. i might test it later though.
I would think that would be very dependent on the components.

I think the larger tubing is mostly used to reduce pressure drop over the loop (and for the bling!) :) After looking at the resistance for 3/8" I'd never use it. I might try 7/16" for an in-box loop someday but my main rig will never see it - the external runs are too long for that ...
 
GreenJelly said:
1/2 Inch tube is really not user friendly. I strongly recommend 3/8" tube, because of its structural strength, ease of use, and generally if you studdy the loop and individual parts in regards to restrictions you will find that 8mm usually provides enough flow for most systems.

I hope one of our resident water cooling gurus comes by and proves that all wrong. I strongly disagree. I'm not going to waste my time digging up the data, because it's relatively well known, but suffice it to say, 3/8" is far to restrictive for most loops on these new high wattage systems. 1/2" is great, and 7/16" is still very good. I made the mistake of running 3/8" in my last loop instead of the 7/16" I've always used. Never again.
 
well if you do the math and lets assume 1/2" as 100%

1/2" tubing = 100%
7/16 tubing = 87.5%
3/8" tubing = 75%

Those are you flows. theres a 25% difference between 1/2 and 3/8, but who know if that translates into lower temps? It would be highly dependant on block, like strom vs apogee, flowrate/head matters.


Quick example: i recently changed from apogee with 3/8" tubing, to storm with 1/2" tubing with little to no extra cooling by temp mointor.
 
Last edited:
I went for a somewhat less restrictive loop.

1/2'' tubing -- heater cores in parallel -- and an apogee and MCW60 for the blocks.
I figured, i would give the loop the best chance possible towards highest possible flow rates given my modest MagDrive3 pump.


But this is not a thread for component picking... or discussion of best order of components in the loop.
So, i think were getting a tad off topic here. Although i think that may just be me.
 
Im not going to defend this position. Just test your system yourself for flow differences between the different tubes. I did, and found no benifit in flow from 3/8" over 1/2"

Even if their is a difference in flow, you can design a system for low flow high pressure. The europeans (including the germans) all use 8mm in their systems. The largest restriction on my setup is the waterblock. It is so restrictive it makes no difference in the reduction of the tube.

Your setups may be different, but I spent allot of time doing experiments regarding this specific situation on my system, and it is different then what others say here. I can run 8mm on my system with very little difference in flow. You got my specs, and I can assure you my tests where done very well.

Sure... 1/2" tube is less restrictive, thats a given... but a Dam with a pin prick hole in it wont get any more flow out of it despite what tube you put to feed into.out of this hole.

Mike
 
Phrenetical said:
well if you do the math and lets assume 1/2" as 100%

1/2" tubing = 100%
7/16 tubing = 87.5%
3/8" tubing = 75%
Cross-sectional area is based on radius squared not diameter:

1/2" = 100%
7/16" = 76.6%
3/8" = 56.3%

And pressure drop through the tubing is even worse than that. For every foot of PVC pipe (sorry, no data for PVC tubing) you loose (in PSI):

GPM 1/2" . 7/16" . 3/8" . 1/4" . Apogee Storm
1.0 0.015/0.029/0.059/0.408 ... 0.5 ..... 1.7
1.5 0.031/0.058/0.120/0.831 ... 1.0 ..... 3.4
2.0 0.051/0.096/0.199/1.382 ... 1.8 ..... 5.8

http://www.pressure-drop.com/


Admittedly, if pump pressure > loop resistance then you should be OK regardless of the tubing ...
 
Last edited:
Back