Custom water loop - possible high temps?

[foolishlywise]

With the help of a few people on these forums, I basically have now started to enjoy water cooling - added in a newer pump and new rad into the loop. However, something that's been bugging me is the temperatures. Previously on air, they weren't fantastic (with a noctua NH-D15s) and they're not all that brilliant on water. Could it be a rubbishy CPU? 

 

Prime 27.9 large FFT is giving temps of 75-80degC, AIDA64 stress in default setting is giving around 77degC. 

 

Its a 4790k oc to 4.6 at 1.284v (manually set) 

 

Thermal paste applications aren't making a difference. If the temps/voltage is high, I assume it's the CPU at fault? 

[FloRolf]

Is your cpu block seated correctly? 

Are the fans spinning? 

How's the pumps waterflow? 

How many rads are you using? 

Airflow in the case? 

[foolishlywise]

7 minutes ago, FloRolf said:

Is your cpu block seated correctly? 

Are the fans spinning? 

How's the pumps waterflow? 

How many rads are you using? 

Airflow in the case? 

CPU block correct (analysing the previous TIM application shows it to be even) 

 

Fans spinning (3xNF-F12) at 900-1000rpm and 1x A15 at 800rpm

 

Pump is, er, pumping. It's a D5 PWM (built into the res - EKWB XRES D5 PWM). It's audible so at a high RPM

 

1x 240mm rad with 2 NF-F12's (intake), 1x 120mm rad with 1 NF-F12 (exhaust) 

 

Airflow is seemingly good in the case - in respect of positive air pressure 

 

Anything wrong with the above setup? 

[Skiiwee29]

10 minutes ago, FoolishlyWise said:

CPU block correct (analysing the previous TIM application shows it to be even) 

 

Fans spinning (3xNF-F12) at 900-1000rpm and 1x A15 at 800rpm

 

Pump is, er, pumping. It's a D5 PWM (built into the res - EKWB XRES D5 PWM). It's audible so at a high RPM

 

1x 240mm rad with 2 NF-F12's (intake), 1x 120mm rad with 1 NF-F12 (exhaust) 

 

Airflow is seemingly good in the case - in respect of positive air pressure 

 

Anything wrong with the above setup? 

GIve us a quick down and dirty on your loop. 

[foolishlywise]

1 hour ago, legacy99 said:

GIve us a quick down and dirty on your loop. 

Pump: EKWB X-RES 140 D5 PWM (ramps up to 100% PWM) 

Rad (120mm exhaust): CoolStream XE 120 (NF-F12, Push)

Rad (240mm intake): CoolStream PE 240 (2x NF-F12, Pull)

Block: EK Supremacy MX Acetal (Copper) 

 

---- may be useless info----

10/16mm EK ACF fittings

15.9/9.5mm EK ZMT tubing

EK Premix coolant (clear) 

 

And, to see it in its glory: 

 

 

[FloRolf]

2 hours ago, FoolishlyWise said:

CPU block correct (analysing the previous TIM application shows it to be even) 

 

Fans spinning (3xNF-F12) at 900-1000rpm and 1x A15 at 800rpm

 

Pump is, er, pumping. It's a D5 PWM (built into the res - EKWB XRES D5 PWM). It's audible so at a high RPM

 

1x 240mm rad with 2 NF-F12's (intake), 1x 120mm rad with 1 NF-F12 (exhaust) 

 

Airflow is seemingly good in the case - in respect of positive air pressure 

 

Anything wrong with the above setup? 

How long has the system been like this? Maybe you have some air trapped in the cpu block. Try moving the case around a little and let the loop run for a bit. 

Are the rads getting hot under load as well? (they take some time to heat up ofc) 

[Lurick]

I know this might seem like a stupid question but is your water flowing into the IN port of the CPU block and OUT of the out port or is it reversed?

[foolishlywise]

14 minutes ago, FloRolf said:

How long has the system been like this? Maybe you have some air trapped in the cpu block. Try moving the case around a little and let the loop run for a bit. 

Are the rads getting hot under load as well? (they take some time to heat up ofc) 

-System been powered on for about three/four days (including 24hrs leak testing with pump at 100%, 8 hours with OCCT running so pump again 100%) and then its been a mix of playing with speeds and leaving it idle. Idle temps are 28-34degC (ambients at 24~26degC)

-Radiators are getting noticeably warm (but not baking hot kinda temps)

 

10 minutes ago, Lurick said:

I know this might seem like a stupid question but is your water flowing into the IN port of the CPU block and OUT of the out port or is it reversed?

Not stupid question, trust me I've nearly connected it in reverse myself! IN port (closest to the middle on the block) has water going in so that doesn't seem to be the issue. 

 

Flow is like this:

Pump/Res OUT -> CPU IN -> CPU OUT -> Rear Rad IN -> Rear Rad OUT -> Front Rad IN -> Front Rad OUT -> Pump/RES IN 

[Limecat86]

1 hour ago, FoolishlyWise said:

Pump: EKWB X-RES 140 D5 PWM (ramps up to 100% PWM) 

Rad (120mm exhaust): CoolStream XE 120 (NF-F12, Push)

Rad (240mm intake): CoolStream PE 240 (2x NF-F12, Pull)

Block: EK Supremacy MX Acetal (Copper) 

 

---- may be useless info----

10/16mm EK ACF fittings

15.9/9.5mm EK ZMT tubing

EK Premix coolant (clear) 

 

And, to see it in its glory: 

 

--SNIP--

 

I suspect a low waterflow. I have the same pump res combo laying around somewhere. I had trouble with it getting the water up to the rad mounted on the top of my case.

 

[foolishlywise]

1 hour ago, Limecat86 said:

I suspect a low waterflow. I have the same pump res combo laying around somewhere. I had trouble with it getting the water up to the rad mounted on the top of my case.

 

This may sound stupid but putting the pump on its highest setting should increase the flow? Would it hence be worth putting another pump in the loop after the rear rad to increase flow or? 

[Limecat86]

15 minutes ago, FoolishlyWise said:

This may sound stupid but putting the pump on its highest setting should increase the flow? Would it hence be worth putting another pump in the loop after the rear rad to increase flow or? 

Yes having it running on 100% should increase the flow. But in your setup it also must to push the liquid up to the rad mounted at rear of your case.

That gave me a flow issue in my setup. I also had my pump running at 100%.

 

I would get some clear tubing and fill the loop again. While bleeding you should be able identify flow problems if there are any.

[foolishlywise]

I found someone with some spare clear tubing that fits my system - getting it and fitting it later today. 

 

Although, I remember when initially filling the loop, the back did make quite a lot of 'waterfall' type noise (like there was a fair bit of air in the back rad). Bleeding the system for 15mins cleared it. Would've thought the D5 is powerful enough for the setup. 

 

If there is a lack of flow to the back, what can be done? Another pump I guess?

[Lurick]

I've got a D5 PWM pump in my setup and it's plenty powerful for everything you've got, plus two GPU blocks, and a bigger radiator, no problem.

[KE2012]

As Lurick said above, a D5 should be more than powerful for your rig.  Have you tried removing your OC and seeing how temperatures are at stock? 

[airdeano]

move the fans to 100% keep pump at 100% note temps.

as above remove any voltage tweaks (stock for testing) and note temps.

[foolishlywise]

4 hours ago, Lurick said:

I've got a D5 PWM pump in my setup and it's plenty powerful for everything you've got, plus two GPU blocks, and a bigger radiator, no problem.

I thought so - since the pump isn't driving much (cpu block and rads) but the point by Limecat got me thinking - weak flow? Head pressure seemed to be all fine but yes. 

3 hours ago, KE2012 said:

As Lurick said above, a D5 should be more than powerful for your rig.  Have you tried removing your OC and seeing how temperatures are at stock? 

 

3 hours ago, airdeano said:

move the fans to 100% keep pump at 100% note temps.

as above remove any voltage tweaks (stock for testing) and note temps.

Leaving EVERYTHING on Auto (ie as I got the motherboard stock), processor stays on 4GHz and never rises - running AIDA64 giving me 49degC. - 62 watts of power and the processor voltage at 1.0566v varying. 

 

Large FFT's at 53degC (again, 4.0GHz). 

 

4.4GHz at 1.18v manual was 62(ish)degC. This was with AIDA64 (as a comparison)

 

@KE2012, Nice location on your profile btw 

[KE2012]

18 minutes ago, FoolishlyWise said:

I thought so - since the pump isn't driving much (cpu block and rads) but the point by Limecat got me thinking - weak flow? Head pressure seemed to be all fine but yes. 

 

Leaving EVERYTHING on Auto (ie as I got the motherboard stock), processor stays on 4GHz and never rises - running AIDA64 giving me 49degC. - 62 watts of power and the processor voltage at 1.0566v varying. 

 

Large FFT's at 53degC (again, 4.0GHz). 

 

4.4GHz at 1.18v manual was 62(ish)degC. This was with AIDA64 (as a comparison)

With a 4670K at stock I hover around 49dc on prime large FFT's. I know you can't really compare chips like that temperature wise, but it sounds like you're in the right temperature range to me. 

 

Just incase your interested I have a heatkiller IV CPU block and 2x360mm rads. A D5 pushes the water.  

[foolishlywise]

On 05/07/2016 at 0:11 AM, KE2012 said:

With a 4670K at stock I hover around 49dc on prime large FFT's. I know you can't really compare chips like that temperature wise, but it sounds like you're in the right temperature range to me. 

 

Just incase your interested I have a heatkiller IV CPU block and 2x360mm rads. A D5 pushes the water.  

Pump is powerful enough for my, in scale, minuscule needs! Although, here's another thing - when the liquid gets hot and reaches equilibrium temps (eg: on a stress test, lets say core running at 77degC), when stopping the test, the temps hover around 49degC for at least a few hours and stay there. Only if I turn off the system for a couple of hours do the temps go back down to 31-35degC. 

 

For example, I ran a stress test 11am this morning. Stopped it at 12. Idles straight after were 47-49degC. Left it on up until now (11pm) and temps are 40degC with ambients at 24degC. It seems to be that the coolant isn't getting around enough, idk? Rads are warm to the touch though. 

 

Any ideas? Voltage is hovering 0.8v (offset mode) and because there was literally no load, probably stayed around there for the most part 

[0ld_Chicken]

3 hours ago, FoolishlyWise said:

Pump is powerful enough for my, in scale, minuscule needs! Although, here's another thing - when the liquid gets hot and reaches equilibrium temps (eg: on a stress test, lets say core running at 77degC), when stopping the test, the temps hover around 49degC for at least a few hours and stay there. Only if I turn off the system for a couple of hours do the temps go back down to 31-35degC. 

 

For example, I ran a stress test 11am this morning. Stopped it at 12. Idles straight after were 47-49degC. Left it on up until now (11pm) and temps are 40degC with ambients at 24degC. It seems to be that the coolant isn't getting around enough, idk? Rads are warm to the touch though. 

 

Any ideas? Voltage is hovering 0.8v (offset mode) and because there was literally no load, probably stayed around there for the most part 

did you put that clear tubing into the system?  If not I'd say try to bleed any air out of the system by tilting the case while the pump is running (you may want to power off mechanical drives) 

try increasing your fan speeds to (~1200rpm or something)

Sounds like you already tried reapplying thermal paste, were you getting a good mount with lots of pressure? are you turning the thumb screws all the way down on the block?

 

im running a different prime version (28.5 i think) and I hit 55°c with large fft, 75°c small fft running 1.33v on my 4790k.

are you still running 44 @ 1.18v?  I can try those settings and see what I get.  I've got the same cpu block but a lot more rad space (since I have 2 gpu's in the loop too).  My water temps never go up if I'm only loading the cpu so my temps will be like your temps in the first few minutes of the test, yours will probably rise after that and that will tell you about how much your water temps are raising.

edit: I delidded my chip so disregard all that

[foolishlywise]

On 10/07/2016 at 0:01 AM, 0ld_Chicken said:

did you put that clear tubing into the system?  If not I'd say try to bleed any air out of the system by tilting the case while the pump is running (you may want to power off mechanical drives) 

try increasing your fan speeds to (~1200rpm or something)

Sounds like you already tried reapplying thermal paste, were you getting a good mount with lots of pressure? are you turning the thumb screws all the way down on the block?

 

im running a different prime version (28.5 i think) and I hit 55°c with large fft, 75°c small fft running 1.33v on my 4790k.

are you still running 44 @ 1.18v?  I can try those settings and see what I get.  I've got the same cpu block but a lot more rad space (since I have 2 gpu's in the loop too).  My water temps never go up if I'm only loading the cpu so my temps will be like your temps in the first few minutes of the test, yours will probably rise after that and that will tell you about how much your water temps are raising.

edit: I delidded my chip so disregard all that

The tubing a friend of mine had was a different diameter. Offered to use cable ties as temporary hose clamps but the barbs are too big on mine so it would've been a pointless exercise. It is worth noting I've ALWAYS had high temps on this chip, only slightly reducing down with a custom loop vs the H100i and the NH-D15S that was on it. However, when starting the system, the back rad did make a waterfall-type noise (where water was trickling or something) but that has since been resolved by tipping the case around everywhere. Tapping the tubes has zero bubbles going into the res and the water block is full with water - can see using the acetal bit on it. 

 

Thermal paste application, all thumbscrews are tightened to the block and inspections of previous applications show an even, thin spread. 

 

In terms of delidding, you are a brave brave man.