Crazy high WC temps, what gives?

D

Daemas

Gawd
Joined
Apr 11, 2009
Messages
791
Just set up my first WC loop and the temps are insane.

Old setup:

H100 with default fans in push config, highest pump speed
24.5 Ambient
31-33 Idle
75-85-85-81 highest load temp via IBT maximum.

New Setup:

2x XSPC EX240 rads each with 2x Scythe Kama 3000RPM 120x38mm fans in push config
1x Swiftech Apogee HD Waterblock
1x Swiftech MCP35x
1x Swiftech Mini res v2
7/16 ID 5/8 OD Primochill Tubing
Various compression fittings, only 1 90deg and 1 45deg
Distilled Water and Killcoil

Pump > Rad > Rad > CPU > Res > Pump (Crudely drawl picture of loop)

28.5c Ambient
37-40 Idle
Upon starting IBt, temps jump immediatly to 55 and 2-3 seconds later immediatly to 68-77-78-73. Temps will eventually reach 82-85 ish after 2-3 passes of IBT.


The temps from the new loop are at BEST 5deg cooler than the h100. From what I have been reading, they should be roughly 12-15deg cooler.



What gives? bad CPU block mount? pump broken (tach reads 2000+ RPMs in BIOS)? Too lofty of goals??


thanks in advance
 
What CPU? What kind of clocks/voltage?

If you're sure the pump is running, I'd guess the block is badly mounted. My 3570k stays far cooler than that even when I'm pushing over 1.3v, with an XSPC Raystorm block and a single Ex240 with 2krpm fans in push/pull. I idle at just a few degrees over room temp.

Core temps do shoot up quickly when I fire up IBT, but not to 85c in a matter of minutes. The only time I've seen near 80c was when I was pushing 1.4v, and even then temps dropped back down to idle levels within a few seconds of core load being reduced. In the case of Ivy, heat just seems to do a bad job of getting from the cores to the heat sink, but your high idle temperatures and hitting 85C immediately suggests a bad mount unless you're running a TON of voltage.
 
GroktheCube said:
What CPU? What kind of clocks/voltage?

If you're sure the pump is running, I'd guess the block is badly mounted. My 3570k stays far cooler than that even when I'm pushing over 1.3v, with an XSPC Raystorm block and a single Ex240 with 2krpm fans in push/pull. I idle at just a few degrees over room temp.

Core temps do shoot up quickly when I fire up IBT, but not to 85c in a matter of minutes. The only time I've seen near 80c was when I was pushing 1.4v, and even then temps dropped back down to idle levels within a few seconds of core load being reduced. In the case of Ivy, heat just seems to do a bad job of getting from the cores to the heat sink, but your high idle temperatures and hitting 85C immediately suggests a bad mount unless you're running a TON of voltage.
Click to expand...

2500k @ 4.7Ghz @ 1.35v

Is there anyway to check and see if the pump is for sure working besides taking everything apart? The pump obviously worked when i was filling the loop, although I was using a AC/DC convertor plugged into the wall instead of my PSU. I have it powered via a molex adapter from a fan (that is currently spinning) with the 4pin attached to the mobo for PWM function.

thanks in advance

EDIT: I guess I could unplug all the 300RPM screamers and that'll probably allow me to hear it. i'm also thinking of unplugging the 4pin from the mobo, it may be like throttling it down to 50% or something

EDIT2: I can also see the water at the top of the res rippling, so the pump must be working.
 
Last edited:
GroktheCube said:
Yeah, it's almost definitely a bad mount then. Too much goo? Not enough pressure?
Click to expand...

the Appogee HD has a mounting system that pretty much tells you when to stop tighening the screws (they bottom out). So it can't be that. and I only used a pea of goo. But I'll give it a go tomorrow and update.

the only other thing with the block could be that there is a dedicated inlet and outlet and maybe they got switched around, but I'm 99.99% sure I did that right.

hmmmm:confused:
 
I always take a pen in my hand, and just start from the beginning, and trace my lines from beginning to end, and look where my flow is headed, i have had times i mixed up in the middle before :)
 
Make sure you have the optimum holes in and out on the HD block.

I think the HD works better with 1 in and 2 out puts to help speed up flow.

Also make sure you got a good prime in that block so there is no air bubbles stuck in the block.
 
SonDa5 said:
Make sure you have the optimum holes in and out on the HD block.

I think the HD works better with 1 in and 2 out puts to help speed up flow.

Also make sure you got a good prime in that block so there is no air bubbles stuck in the block.
Click to expand...

how can i check to make sure there are no air bubbles in the block besides tapping on it with a screwdriver handle and tilting the case?
 
you got it, repeated tilting under operation will eventually drive bubbles out, try different directions
 
Just got finished taking apart my loop (really just took off the cpu block while the rest was still connected :eek:)

Took apart CPU Block, no gunk in it, Took a tooth brush to it anyway. made sure inlet was IN this time (it was last time)

Same amount of goo but I used AS5 instead of the Swiftech supplied stuff (martin or skinee said it was good)

After removal: http://imageshack.us/photo/my-images/59/wp000041p.jpg/
http://imageshack.us/photo/my-images/827/wp000040.jpg/

new TIM: http://imageshack.us/photo/my-images/14/wp000050i.jpg/


I feel like this time when I bled the loop A LOT more bubbles came out. I had the res filled up enough that I could lay the case on it's side. Didn't get much of anything from tilting in the other three directions

in the process of tilting a caught my hand in one of the 3000RPM kazes and broke off a fan blade. Hopefully I can run like 15min of stress testing for you guys to see if the temps went down any. I'd rather not have the fan explode and send fan blades into my components though, heh.

will edit with temps in about 15-20 minutes.


EDIT:

26.6 ambient
32-36 idle
after 10 rounds of IBT, highest the cores ever got was 70-80-80-76
If I go off like the first 2 or 3 passes like the temps in the OP, it's like 66-76-76-72ish.
maybe a degree or two better, but not the 5-7 I was looking for.


Anything else anybody can think of?

I'm starting to think I'm just getting to the point of where the CPu block just can't wick away the heat fast enough. then again, i haven't tried going any higher
 
Last edited:
Check for flatness on both your heatspreader and your block... I have seen some VERY warped heat spreaders come of out of Intels factory and that makes for some horrible heat transfer.

In fact, my AMD heatspreader right now is not flat and it effecting my temps. I just havent had the time yet to disassemble and lap the block and CPU.
 
Have you got both the rad's fans set to blow air into the case?
If so then you have a very high pressure inside the case and the two rad fan systems are kinda fighting one another and you're not getting the air flow through the rad's that you would be getting if you had the front set to blow in from the front and the top ones set to pull air out the top .
 
I'm really starting to think you have a low flow problem.

I think single CPU would be fine with only 1 of those radiators with the push pull monster fans that you are using.

2nd Radiator would come in handy if you were running a GPU block.

Again I mentioned this before. If you go 1 in and 2 out on the HD block the flow will increase. The HD block has 1 in and 3 out puts. I think the flow will increase if you have open up some of the outputs provided your pump is strong enough to push it.

I ran the HD block with 1 in and 2 output to gpu block and another out put as bleeder to reservoir. My temps were decent with 2500k.

Photos.


ApogeeHD.jpg




BnBday1.jpg
 
I'll echo the flow/block problem.

I'm running a 130w i7-920 @3.8 with a GTX480 with a solo triple radiator, and IBT doesn't get my cores above 70C with my fans at 1100 rpm.

You have more fans, more radiator, lower thermal CPU, and no GPU in your loop, so you should at least be keeping even :cool:

Disassemble, clean out, reinstall, rebleed.
 
It might be the split off from the cpu block. You got one line going from the cpu to the res and one to another 120 rad then to graphics card then to res. Or maybe move the top left hose to the top right output.
 
Nettwerk said:
It might be the split off from the cpu block. You got one line going from the cpu to the res and one to another 120 rad then to graphics card then to res. Or maybe move the top left hose to the top right output.
Click to expand...



:rolleyes: You are looking at my photos that I posted to help OP.

My configuration works great.

I think the Apogee HD block works better when using two outputs from the cpu. The photos are to help OP see this.
 
First: Your fans are SILLY. You are dumping a lot of hot air into the case - way more than any amount of exhaust fans can remove, which means lots of turbulence and bad airflow. In fact, the pressure is probably pushing back on the rads

Make the fans at the top pull.

Second, try putting the CPU block as the first thing after the pump. Laing (the OEM for your pump) recommends this config for best results.
 
AS5 is mediocre at best compared to the best TIM now, that should net you 1-3c (Prolimatek PK-1, and Phobya HeGrease).

The flow is fine unless you messed something up.

Are your fans all set as inlets, so they draw fresh air over the rads? If not, set them that way. It will be an overpressure case, which is fine.

Pump before CPU block means nothing, proven at skineelab/martinliquidlab to be at best a .25c gain....

Run your radiators in parallel, so go

...........................>Radiator|
Pump>CPU Block|...............> T fitting>Res
...........................>Radiator|

That configuration will net you max flow rate and temps.

Remember, if you have 1 Set of radiator fans as an inlet and the 2nd setup as outlet fans, the 2nd rad is pulling heated air over it, and will do nearly nothing to help cool and may actually hurt your temps.

Aside from that, if your temps are still a little too high (and they are currently) more pressure, figure out a better mounting setup and remount tighter.

For comparison, I have an i7 3820 (much hotter CPU), 2x6870s, and the VRM in my loop with 1xRS360, 1x200mm rad, 1x140mm rad and way slower fans, and I hold about 60-65c at your voltage/speed level. At 4.9 and 1.528v, CPU maxes at 73.
 
M249SAW said:
4c difference in ambient temps make a difference.
Click to expand...
First, this.

Second, I'm surprised nobody else pointed out that your pump should be running ~4500RPM, not 2000RPM. Unplug the PWM header and it should jump up to max RPM.

Also, you have an upside down rad, and a horizontal rad with the barbs pointed down... Even though you think you got all the air out... You probably have a very large air bubble in at least one of the rads. Remember, bubbles float 'up' so you need to orient your radiators such that they can escape, otherwise you'll _never_ bleed them out(unless you've got a rad with a bleed screw on the bottom).

Lastly, running your rads in parallel is a horrible idea for you. A single MCP35x will not have enough pumping power, especially since your rads are in different orientations. You would end up with significantly higher flow through one radiator.
 
Well you are pumping all your water thru the gts360 first, that would be your flow problem....

The gts is very restrictive, and the hd is a high flow block,.

You should reroute your tubing so that you are getting all your water thru the hd without restriction and only some of the water thru the gts.
 
the op doesn't have a gts360, and sonda5 isn't the one having issues.
lol@u

Sonda, looks like you found a legit fc waterblock that fits your 560ti 448?
 
Zoson said:
Lastly, running your rads in parallel is a horrible idea for you. A single MCP35x will not have enough pumping power, especially since your rads are in different orientations. You would end up with significantly higher flow through one radiator.
Click to expand...

You are correct about the Pump RPM, it should not be so low.

You are misinformed concerning the pumping configuration and the ability of the MCP35x however. It is one of the best pumps ever made for water cooling, and no I don't have one, so I am not a fanboy.

At 100% duty cycle and 12v, where you should run it (not like it's loud anyway), your getting over 2.5GPM unrestricted. So considering the parts the OP has in his system if configured in series the OP will still be well north of the 1GPM guideline for minimum flow in a quality loop.

Radiators are the least restrictive component in the loop. You also forget that the larger the Delta from water to air temp the greater the amount of heat that is removed from the water, even though the percentage remains the same. By running them in series your taking what is already cooled to fairly close to ambient, and then trying to cool it again. You will still remove a static percentage of heat, however the absolute amount removed is less. Parallel radiators are superior, and scientific testing verifies what I know to be true, all of this data can be referenced via Skineelabs and martinsliquidlab if you do some reading AND comprehending vs just scanning the reviews.

About the vertical vs horizontal, for bleeding yes its not optimal, but is totally possible if you simply turn the case 90 degrees while bleeding, air will escape just fine. Concerning pumping power due to difference in orientation, it is irrelevant. The system is a closed loop system, by having to pump up to the top, hes gaining gravity assistance on the way down, its a net result of 0 energy difference if the tube length remains the same.

I guarantee if you set your loop up the way I stated earlier you WILL have a temperature drop vs 100% series loop if all else (fan speed/orientation, loop order, pump speed, ambient temps, thermal load, and CPU mount) remains the same. Primarily your gains will not be due to the effect of the cooling power of Parallel vs Series rads, but the gains will come from the increased flow rate over the block when using 2 outlets and the reduction in pressure drop of the parallel instead of series rads.


OP choose who to listen to, but I am right in this particular case, and supported by page after page of documentation on reviewer sites if you simply look around.

Using the configuration tool linked below, you will get roughly 1.75 GPM, and should see approximately a 3c delta of air/water temps under load.
References:
http://martinsliquidlab.org/pump-and-radiator-optimizer-spreadsheet/ (A nice guideline tool, it is fallible however)
http://skinneelabs.com/swiftech-mcp35x/
http://skinneelabs.com/2011-thermal-paste-review-comparison/3/
http://martinsliquidlab.org/2011/04/26/pump-setup-series-vs-parallel/ (Comprehend this, not just scan it, the data holds true for radiators as well as blocks Parallel does increase total flow, but reduces flow per radiator, while reducing restriction)
http://wc101.com/articles/vs/?page=vs
http://www.overclockers.com/forums/showthread.php?t=419064 (Final post of this thread explains it better than I ever could)
 
Last edited:
Your argument is flawed because to set up parallel he would need to increase the length of tubing. Which means more pumping power. Yes there is lower restriction, but you lose significant head pressure.

Also, I have an MCP35x2. I forgot to plug in one of the pumps, and at full rpm it was barely able to pump through my cpu and three 120mm rads. That's approximately the same amount of restriction in the loop as the OP.

Lastly, you didn't comprehend your own link. Martin discussed parallel pump vs series pump. Not parallel rad vs series rad. And no, decreasing restriction by making rads parallel is not even close to the same thing as adding pumping power.

My argument wasn't that you don't end up with better cooling efficiency, IF you can run parallel rads... Just that his setup isn't optimal. The tubing runs to each rad would be different, which would cause flow discrepancy between the two. Not to mention the overall height difference within the loop between the two rads.
 
Since tubing is the single lowest restriction in any system, adding an extra 2 feet is not an issue, and will not significantly effect his flow/pressure/temps. "Significant Head pressure" no, you don't. You loose significant head pressure in ONLY the radiators, the rest of the loop maintains it. Since radiators are not nearly as pressure/flow sensitive as blocks are it makes perfect sense to go Parallel instead of series.

If your single 35x could not pump through 3 120.1 rads + CPU block, it's broken or spinning too slow or you have a kink. My XSPC 750 res/pump, was able to easily push through 2 MCW-60s, a Rasa block, and 3 radiators, a 360, a 200, and a 140 with no issue. It has FAR less pressure and flow than the 35x. Proof is in the testing linked.

Actually I did comprehend it, which is why my 40 dollar pond pump + cheap XSPC pump are in series and pump FAR more than any dual DC pump setup available beside exotics over 12v, and was a total of 90 bucks..... It also applies to radiators if you read the whole thing and understand that 2 rads in parallel dont cut flow by 1/2 and you dont combing the pressure drop of each alone, they gain efficiency in both departments, just as 2 pumps in series which is why i linked it.

The tubing runs don't have to be different, just cut four tubes the same length and run them. The restriction to each rad does not have to 100% identical, and 1 foot of tubing to 1 radiator vs 1 foot 6 inches to the other is not going to make a significant difference in flow to each or in Delta T.

There are plenty of loops such as the one pictured with asymmetrical radiators being run in parallel and seeing gains doing so, balance is important but not as important as total flow rate over the block. What will happen is that one radiator will dissipate slightly more heat than the other, not a big deal whatsoever.

Also like to say that pressure is not flow rate. Flow rate is what you seek, not pressure (to a point). They are not mutually exclusive, and if you can significantly increase flow with a small drop in pressure you will see gains. That is exactly what I did in my loop order I posted for him. 14 years of Automotive Turbo system design and installation have proven this right every single time, and continue to do so daily.
 
Was the MCP35x at that speed at load? If so you might have plugged it into the one of the case fan PWMs instead of the cpu one.

Zoson said:
Also, I have an MCP35x2. I forgot to plug in one of the pumps, and at full rpm it was barely able to pump through my cpu and three 120mm rads. That's approximately the same amount of restriction in the loop as the OP.
Click to expand...

That's odd. I ran an MCP35x through a high restriction loop (Apogee XT rev 1, EK7970 full cover block, 2 rads, 5 sets of Koolance VL3N quick disconnects) and it gave me around 1gpm at full speed. Swapping the Apogee for a Raystorm gave me 1.6gpm though:rolleyes:.
 
Zoson said:
Sonda, looks like you found a legit fc waterblock that fits your 560ti 448?
Click to expand...


Custom rigged Swiftech Apogee XT cpu block and a Koolance VRM block.

GTX560ti448BattleGTX.jpg


It's legit because it works well.
 
SonDa5 said:
Custom rigged Swiftech Apogee XT cpu block and a Koolance VRM block.

GTX560ti448BattleGTX.jpg


It's legit because it works well.
Click to expand...

doh, I was referring to this picture:
BnBday1.jpg

But i guess that's not the same card?
 
You must log in or register to reply here.
Back
Top