View Full Version : [SOLVED] Why isn't my watercooling cooling?
O.k. I've been reading the forum for a while and build my own water cooling system, but it doesn't seem to be doing as well as I expected. It consists of a 1.2 TB @ 1.46, a 4x4x4 reservoir with a submersible Rio 600 pump (120 gph at 2') the Swiftech MCW462-B waterblock with Arctic Silver thermal compound, and the Dangerden cooling cube with an Adda 120mm 55 cfm fan. My cooling mixture is 90% distilled water 10% Water Wetter. All tubing and fittings are 3/8". My only method for temperature measurement is the one reported by the temp probe on my KT7A (I know it's not that accurate). Now, normal use temps are pretty good about 30C. However, when I run a stressing program like Prime95 or Counter-Strike temps reach up to 50C. How does everyone else out there have his or her systems plumbed? For example: from the reservoir to the block, to the radiator, back to the reservoir. Has anyone noticed if this can make a noticeable difference? It seems for some reason my cooling system isn't transferring the heat well and I'm not sure why. I've reseated the block once and that didn't help. It's pretty hard to screw up with the Swiftech. Any comments or recommendations would be greatly appreciated.
I forgot, your plumbing order is correct.
Thanks for the response Colin. I had an all distilled water mixture that gave me the same temps? I figured I'd try something else.
Man I am stumped. You should be about 5 to 10C over ambient at idle with your setup.
Doh! I just looked at your normal use temps. FWIW, I had a Rio 800 pump running inline. When I switched to my current pump, my temps dropped 3 C.
The only other thing I can think of is your fan. With only 55 CFM to work with, it should be pulling the air through the radiator, not pushing. This bit of wisdom came from Dan at DangerDen.
dimmreaper
03-23-01, 01:52 AM
Correct me if I'm wrong, but isn't your waterblock the swiftech thats open on the inside, with the inlet placed dircetly over the CPU core, and it mounts via the four holes in the motherboard.
If this is the case, be sure thier is not an air pocket trapped in the open chamber. I have a TidalPool that doesn't work very well when their is an air pocket trapped in it.
Jeff,
Shouldn't bleeding the system overnight take care of the air pocket?
Shadow ÒÓ
03-23-01, 01:54 AM
what's the surface like on your block? Sanded well with fine grit?
What's the water temp......is this after running an hour or so and the water temp is actually warm?
so you're saying his pump is putting off too much heat submerged?
Jeff Evans (Mar 23, 2001 01:52 a.m.):
Correct me if I'm wrong, but isn't your waterblock the swiftech thats open on the inside, with the inlet placed dircetly over the CPU core, and it mounts via the four holes in the motherboard.
If this is the case, be sure thier is not an air pocket trapped in the open chamber. I have a TidalPool that doesn't work very well when their is an air pocket trapped in it.
Jeff,
Yes the block is the one with the inlet over the core, and mounts with the 4 holes in the motherboard. I'll play with the block in the morning to see if there is any air trapped.
Thanks
Rio's put out a fair amount of heat and this should figure into the equation but this can't account for 50C full load temps.
Shadow ÒÓ (Mar 23, 2001 01:54 a.m.):
what's the surface like on your block? Sanded well with fine grit?
What's the water temp......is this after running an hour or so and the water temp is actually warm?
Shadow,
The surface looked pretty good, but I noticed on the Swiftech site it mentioned not to lap the bottom because it had already been done. It doesn't shine? I may have to try a block from Dangerden.
Thanks
dimmreaper
03-23-01, 02:05 AM
Colin (Mar 23, 2001 01:53 a.m.):
Jeff,
Shouldn't bleeding the system overnight take care of the air pocket?I've found the best way to get all the air (small and large pockets) out of the block is to fill the system, leave the pump on circulating the water, and shake the waterblock like a half empty bottle of catsup (or is it catchup?). I was surprised as to how much air was trapped in my tidal pool! I noticed a 10C drop in temps afterwards.
Maybe try altering your setup. Instead of reservoir, block, radiator, reservoir, try reservoir, radiator, block, reservoir. This way, heat added by the pump is passed through the radiator before going back to the block.
I'm presently into the testing of radiators (by varying air and coolant flow and temp) and would observe that your air flow is low,
and set the fan with a 1in standoff
you need to know the coolant temps to understand what the system is, or is not, doing
I would be surprised if you have even 1/2 gpm actual flow, time filling a jug
I too have a MCW462, do bleed it, don't lap it - its NOT the problem
be cool
surlyjoe
03-23-01, 09:37 AM
stool has the ticket! ,if your running submersible pumps ,you wanna run the water from the res. to the rad. then the blocks,,the pump heat is dumped by the rad before it hits the block,I think water cooling the pumps kicks ass ,because it eliminates the need for fans to cool the pump ,(aka less noise)
Stool and SurleyJoe,
Very interesting, I guess I have an "inline" brain! Since all pumps produce some heat, this would make sense for many inline installations too.
Regarding the 55 CFM fan, I can dial mine down that low. Unless you are dissipating a whole bunch of heat as in a pelt, this can’t account for the temps. I will repeat that with one low CFM fan, you want to draw the air through the Cube. Pushing the air into the Cube causes turbulence. That’s why there is a gap before the fins start on one side. If you push are through the Cube, a plenum would help.
O.k. guys I beat my block until it begged for mercy. I didn't notice any air bubbles that got released from the block. I reapplied the thermal compound and temps seem to have lowered a couple of degrees C. When I get home from work tonight I will crank up Prime95 and take a temperature reading of the cooling liquid at full tilt. I'll buy some more hose today and try changing things around plumbing wise. I also switched my fan from pushing to pulling air off the cube. I'll keep you all posted what happens this evening. Thanks for everyone’s input, it's greatly appreciated.
surlyjoe (Mar 23, 2001 09:37 a.m.):
stool has the ticket! ,if your running submersible pumps ,you wanna run the water from the res. to the rad. then the blocks,,the pump heat is dumped by the rad before it hits the block,I think water cooling the pumps kicks ass ,because it eliminates the need for fans to cool the pump ,(aka less noise)
I must lend support to the radiator, block, resivior pumping order. This configuration worked far better than any other configuration I tested. BTW, I used a sumberged pump
Here is what I did. I redid the plumbing to go from the pump, to the radiator, to the block, and then back to the pump. I first tested the flow of water through my system by timing it for one minute. The setup had a flow rate of .5 GPM. I then tested the ambient temperature of the cooling solution, It read 25C. I then proceeded to run Prime95 and the CPU temp rose to 46C from 32 C and the cooling solution rose to 29C. Air coming from the radiator never really seems to be that warm. Do you think the block could be the problem or is the temp probe on the MB just off by that much? I don't know what to do next!
Thanks
In my experience with the Cube, 29C water temps will not yield noticeably warm air are exiting the radiator. 38C water temps will cause a hot desert Ghibli to come off the Cube.
How warm is everyone elses cooling mixture under full load with a similar setup?
William
03-24-01, 02:56 AM
have you tried using a larger fan on the radiator. It could help. Also, if you made your resevoir bigger, the water would cool down in it as well as in the radiator. Other than that, i don't know why you are having temps so high.
Stan, correct me if I'm wrong, but you seem to be describing a 4C rise in the coolant temp.
By most "standards" that's not so bad - particularly in light of the low flow rates (both). Increasing the flow rates will decrease the temp, but not much.
(As you approach ambient the curve is becomming asymptotic, Joe Citarella's "law" of diminishing returns.)
Do some investigation of the MB probes, I dont use one but there are NUMEROUS informed articles and posts to the effect that the numbers are all over the map.
Watercooling systems are designed empirically as they consist of elements whose specific performance is often poorly defined (or understood, or applied).
Cooling "back" to ambient temp necessitates substantial overcapacity and the "design process" is often a serial selection of components to identify, and then replace, the limiting element.
be cool
What I have done to keep track of this is to merely keep an eye on my coolant temps. I purchased an indoor/outdoor themometer from the Shack, and placed the external remote into my reservoir.(It is waterproof, BTW) As the CPU heats up, I keep track of ambient and coolant temps, as opposed to the core CPU temp. I use a mix of 90% water, 5% WaterWetter, and 5% anticorrosive. Under a very heavy load, coolant temp doesn't exceed 8-10c over ambient, and is always better than temps I used to record with air cooling. I am also using 2 120mm 120v Orion fans on my radiator(mounted and powered externally away from the case)
Hey Stool, sounds like an aggressive - and effective - cooling setup.
You mention "5% anticorrosive", apart from WaterWetter.
What is this ?
be cool
BillA,
Could the lack of coolant temperature also be attributed to the fact that the block may not be removing the heat effectively from the core? Thus causing my higher core temps.
Its hard to believe the block is the problem.
I have one and don't know what might be better.
(After the current radiator heat dissipation project, the next is to be on waterbolcks.)
So what are the possibilities:
inaccurate MB probe - my vote
poor transfer between cpu and block - ck goop, not too much
too little pressure - should not be a factor with the MCW
poor finish on block - NOT a factor with the MCW
what else ?
be cool
KeyboardCowboy
03-25-01, 05:03 PM
do you have a pielter on your water cooling system????
Stan, a thought from another thread -
are you sure, quite sure, that the block's setting flat ?
be cool
BillA (Mar 25, 2001 08:26 a.m.):
Hey Stool, sounds like an aggressive - and effective - cooling setup.
You mention "5% anticorrosive", apart from WaterWetter.
What is this ?
be cool BillA
I went to an auto parts store, and by the cooling additives I found a cooling system anticorrosive(supposed to prevent interaction between copper, aluminum, and brass from corroding the system. Small bottle will work for 2-3 setups.
I also considered the block wasn't perfectly level, but I've seated it multiple times and it looks good and I haven’t seen any change it temps. It's strange because the temperature rise is pretty quick to about 42C and then slows from there until it reaches 46C. However, when I stop whatever program I'm running it cools back down very quickly to under 40C (a matter of 30 seconds or less). I'm thinking it's probably the MB temp probe, but you'd think it would misread the lower temps also.
Phobos (Mar 25, 2001 05:03 p.m.):
do you have a pielter on your water cooling system????
Phobos,
Nope, It's a water only setup for noise reasons.
I had a similar problem earlier today as I was setting up my system. Temps would hover in the upper 40C range and occasionally hit 50C with my setup - Be-Cooling Block, dangerden cube, eheim 1046 w/ water, 10% glycol coolant.
I reseated the block several times but temps would still stay in the upper 40's. I finally took off the mobo, disconnected the hoses and reseated the waterblock on the chip & made sure that it was totally flush with the CPU. After reseating, I'm at 38C CPU temp at 29.2C ambient (boy it's hot in here).
It sure looks pretty hard not to seat your block properly but it's seems to be pointing to it now that you've pretty much eliminated everything else...
I didn't read to see if someone posted this suggestion before (I am lazy tonight), but wht don't you install a peltier. That would put your temps down quite a bit with the right size TEC.
far more advisable to resolve the heat transferrence problem with the block, BEFORE adding a TEC
Stool, the kind of junk I find is "pump lubricant", which is soluble oil - nothing more.
What is the name of the anti-corrosive ?
thanks
be cool
I'll take a look tomorrow, and let you know. I have already used mine up.
Well I tried reseating the block again with no hoses and the MB laying flat on the table. It appears to sit level because the thermal compound is spread evenly. I think I'm going to try a DangerDen block or something else if anyone has any suggestions.
Thanks
Well I purchased a new pump and block and both should be here in a couple of days. I'm soon to find out what the culprit is. I'm also building a computer for my wife so the extra parts can be used to water cool her computer.
BillA (Mar 26, 2001 08:09 a.m.):
.....
Stool, the kind of junk I find is "pump lubricant", which is soluble oil - nothing more.
What is the name of the anti-corrosive ?
thanks
be cool BillA, I checked three different stores today, and couldn't find the stuff I used. The closest I could find was Prestone Super Antirust. I'll try some more places tomorrow.
I have found my new water cooling system to be real sensitive to the amount of pressure there is on the waterblock. Just a little too much pressure on one side or the other and it doesnt sit flat on the chip and the temps rise. A very slight adjustment on one wingnut or the other will drop the temps 5 degrees. You might try playing with that a bit and see what happens. I know it worked for me.
Thanks TT I'll play with it some more when my new pump and the other block arrive. I'll test both setups and see which one works better for me.
Newbie_Doo
03-28-01, 08:51 AM
I also have a submerged pump. My temps (admittedly on an overclocked c2-600) run about 2C over ambient. The thing I noticed about your setup that might be a contributing factor is the amount of coolant you have to work with. You are using a 4x4x4 reservoir. I use a 6x6x4 which has more than twice the capacity. If you have less capacity, you need to have a higher flowrate on your coolant and fan. I use an 80GPH Little Giant fountain pump (Rated at 2') and have a large radiator with 2 120mm Sunon fans drawing air through it. What I lack in flow, I make up for with volume and exchanger capacity.
Try a larger reservoir, or try a faster pump.
Just my $0.02
Correct me if I am wrong, but won’t a larger reservoir just delay the time it takes for the coolant temperature to reach equilibrium?
Colin,
I was thinking along the same lines as you with the larger reservoir. It would just take longer to heat more liquid. I don't have any tested proof though.
Colin (Mar 28, 2001 10:53 a.m.):
Correct me if I am wrong, but won?t a larger reservoir just delay the time it takes for the coolant temperature to reach equilibrium?
Basicly right - thermodynamics say that to heat 1ml H2O 1 deg C takes 4.18 Calories or 1 Joule (1 J = 4.18 Cal) . Thus 1000mL takes 1KJ or 4.18 KCal to heat 1C, 50000 mL takes 50 KJ (209 KCal). More water means you need more heat to raise the temperature up....
Equillibrium point is a different matter though and will depend on not only the volume of water, but also the heat transfer efficency both from the waterblock to water and water to radiator, heat loss (gain) to the atmosphere, heat gain due to pump inefficency, ambient temp ect ect ect.....
vBulletin® v3.8.7, Copyright ©2000-2012, vBulletin Solutions, Inc.