New watercooling setup completed, temps spike HIGH

[shibbs]

So I have a question. I've got a 4670K Overclocked to 4.7ghz at 1.3 vcore.

 

I just setup a custom watercooling loop - EKWB EVO block (with EK's recommended 1150 plate and jet) and I have a 140mm hardware labs black ice radiator, and a 280mm rad of the same. So PLENTY of cooling. D5 pump/res combo.

 

Now for my question - When I start AIDA my temp explodes from 23 idle to 70 under load. (FPU only - maximum heat generation) I don't have any data to confirm this, but my old Corsair H90 didn't make it spike like that that I can remember. Is the heatspreader lid THAT shitty on these? Is delidding worth it? 

[Mr.Ganja]

Are you sure that you have good contact betwen your cpu and waterblock?

[shibbs]

1 minute ago, Mr.Ganja said:

Are you sure that you have good contact betwen your cpu and waterblock?

98% certain. Same method I've always used. And the idle temps (2 degrees above ambient) tell me that it's good. But I may be wrong. It's happened before.

[stealth80]

I think its something to do with FPU on newer Intels, I just ran it on mine and it spike to 70C also, I'm running 360mm + 240mm rads on my CPU+chipset blocks (separate loop for GPUs)

[GhostHitWall]

so does the spike means that it jumps to 70c then back down or just sitting at that temp?
Since the previous cooler didn't perform this way, the chance is that the block and CPU IHS aren't contacting well.

Before you disassemble the loop, try change your D5 speed to increase your flowrate.  If increasing flowrate doesn't give you any change in temp, I would tear the whole loop down to inspect.

delidding isn't magical, I remember reading a report on anandTech  saying the IHS of Devil's Canyon is too thin so contact between die isn't ideal.  Not sure if it's the same story to Haswell.

[Dackzy]

Water Cooling does affect the cpu that much so this is probably normally, the part that water cooling affect the most are the gpu.

[shibbs]

5 minutes ago, GhostHitWall said:

so does the spike means that it jumps to 70c then back down or just sitting at that temp?
Since the previous cooler didn't perform this way, the chance is that the block and CPU IHS aren't contacting well.

Before you disassemble the loop, try change your D5 speed to increase your flowrate.  If increasing flowrate doesn't give you any change in temp, I would tear the whole loop down to inspect.

delidding isn't magical, I remember reading a report on anandTech  saying the IHS of Devil's Canyon is too thin so contact between die isn't ideal.  Not sure if it's the same story to Haswell.

Jumps to 70 and stays there. The instant I stop the test, it drops down to 23 again. 

 

I've played with the pump running it at max and 3300RPM and less, nothing made a difference in cooling. Leaving it at 3300RPM now since it's the quietest with good flow.

[Vudian]

That's some bad Temps for being water-cooled. I'd say bad contact. Back when I finished my 1st loop I kept getting Temps like that on my 4790k. It turned out I was not getting my Waterblock tightened down all the way. that's ekwb supremacy MX and if I remember correctly they use the same system for mounting. They can be very hard to get to tighten down all the way and are made so you can't over tighten them. Hence the tapered screw. If you could show me some pictures of the screw down points I could tell you. If that doesn't help you might have got a bad block.

[shibbs]

4 hours ago, Vudian said:

-SNIP-

You have a Haswell refresh, though. I'm worried that with a 1ghz overclock that I'm hitting the limits of the TIM under the heat spreader. Have you ever delidded your processor?

[Vudian]

Nope and never will. Refresh or not, the Temps look like bad contact with the block, I'd take a rag with pliers and make sure it's tightened all the way down. I have had 2 ekwb cpu blocks I had to do that with it get good Temps

Also you could be using the ports on the block I the wrong order, also another issue I had that gave me high Temps. Providing some photos could really help figure out problems, some of us might notice something you did not.

[shibbs]

The ports on the block are in the correct order. I tightened the hold-downs on the board with a pliers.

 

Still getting 27 degree idle, and spiking to 70-75 the instant I start Aida64's FPU load. Temps are around 60 when I do the full suite of CPU, FPU, and Cache. The second I stop either test, the temps instantly drop down to 28-30c again.

 

 

[EdInk]

48 minutes ago, shibbs said:

The ports on the block are in the correct order. I tightened the hold-downs on the board with a pliers.

 

Still getting 27 degree idle, and spiking to 70-75 the instant I start Aida64's FPU load. Temps are around 60 when I do the full suite of CPU, FPU, and Cache. The second I stop either test, the temps instantly drop down to 28-30c again.

 

 

Are you able to change the pump speed? increase and run test. see if it makes a difference on load temps, what sort of liquid are you using?

[shibbs]

30 minutes ago, EdInk said:

Are you able to change the pump speed? increase and run test. see if it makes a difference on load temps, what sort of liquid are you using?

Pump speed is running at max (4900 rpm) even though temps are identical when running at 3000rpm. 

Running Mayhems Aurora 2 Tharsis Red. (ricer, I know, but dammit it's pretty)

[EdInk]

1 hour ago, shibbs said:

Pump speed is running at max (4900 rpm) even though temps are identical when running at 3000rpm. 

Running Mayhems Aurora 2 Tharsis Red. (ricer, I know, but dammit it's pretty)

Possibly the liquid then as per Mayhem

 

Mayhems Aurora is a totally different kind of specialist coolant and should not be used for normal every day use. Consideration about your loop lay out is a must before proceeding to use this liquid. Mayhems Aurora is a photographic, demonstration, Modding and looks only liquid and is not recommended for long term use.

 

Mayhems Aurora works best in duel pass rads and tube reservoirs. The more complex the loop the less time the liquid will work. Many uses do use the liquid for long term and the most we have tested it use for is 1 year. How ever in complex loops were people have totally ignored our warnings it might only last 24 hours or so.

 

I'd suggest switching to their pastel (which i've used for about a year now) ans see how you fare

 

[shibbs]

5 hours ago, EdInk said:

Possibly the liquid then as per Mayhem

 

Mayhems Aurora is a totally different kind of specialist coolant and should not be used for normal every day use. Consideration about your loop lay out is a must before proceeding to use this liquid. Mayhems Aurora is a photographic, demonstration, Modding and looks only liquid and is not recommended for long term use.

 

Mayhems Aurora works best in duel pass rads and tube reservoirs. The more complex the loop the less time the liquid will work. Many uses do use the liquid for long term and the most we have tested it use for is 1 year. How ever in complex loops were people have totally ignored our warnings it might only last 24 hours or so.

 

I'd suggest switching to their pastel (which i've used for about a year now) ans see how you fare

 

Well damn. I ordered Aurora 2, as I was told it's safe to run daily (at least moreso than the original) but you're not the first to say that the Aurora was a dumb choice. Coolant was definitely an afterthought for me. Just my luck that the one thing I don't research about this loop is the thing that blows up in my face.

 

Posting screenshots of my testing anyway. 

 

The first picture is when I just got home from work. Cold room, definitely cooler than normal air.

 

Second test is about 5 minutes ago after watching a few YouTube videos, some light gaming and whatnot. See what I mean about INSTANT spiking to 70+? The graph is set to update every 3 seconds for CPU and temps.

[Treij512]

Im sitting here looking at all the information you've supplied and , for the life of me, I can figure anything out. The temps are highly out of wack imo. My 4790k @ similar OC doesnt run that high with my ek predator 240.  only thing i can think of is reseating the cpu block thermal paste and if needed go back to stock speed to test as well.  Additionally is there any change in cpu temps with max fan speed? 

[shibbs]

24 minutes ago, Treij512 said:

Im sitting here looking at all the information you've supplied and , for the life of me, I can figure anything out. The temps are highly out of wack imo. My 4790k @ similar OC doesnt run that high with my ek predator 240.  only thing i can think of is reseating the cpu block thermal paste and if needed go back to stock speed to test as well.  Additionally is there any change in cpu temps with max fan speed? 

Done everything. Reseated the block, tightened down the pins holding the block and the pins on the motherboard, tried 3 different methods of applying thermal paste, low fans, high fans, low pump, high pump, RODI water, Mayhems Aurora...... All the results are identical.

 

The only thing I can think of is it being the heat spreader and the junk TIM under it from Intel. Maybe 3 years of daily use on a substantial overclock have done it in? It's about the only thing I haven't changed.

[Treij512]

Are you still spiking at Stock speeds? If so then id swap out to a different cpu or upgrade if possible

[EdInk]

3 hours ago, shibbs said:

Done everything. Reseated the block, tightened down the pins holding the block and the pins on the motherboard, tried 3 different methods of applying thermal paste, low fans, high fans, low pump, high pump, RODI water, Mayhems Aurora...... All the results are identical.

 

The only thing I can think of is it being the heat spreader and the junk TIM under it from Intel. Maybe 3 years of daily use on a substantial overclock have done it in? It's about the only thing I haven't changed.

I guess it could be. Did reseating the block include rotating it? I've seen some test where it did make a difference.  http://thermalbench.com/2016/02/11/ek-predator-240mm-aio-cpu-cooler/5/

 

Page 4, the placement is called 'goofy'. Haha!

 

 

[shibbs]

10 hours ago, EdInk said:

I guess it could be. Did reseating the block include rotating it? I've seen some test where it did make a difference.  http://thermalbench.com/2016/02/11/ek-predator-240mm-aio-cpu-cooler/5/

 

Page 4, the placement is called 'goofy'. Haha!

 

 

Ugh, I really don't want to have to re-bend more lines just to test...  haha  Really thinking I should have done soft lines.

 

@EK Luc do you have any insight to any of this?

[RafaelSoaresP]

19 hours ago, shibbs said:

..

I`m getting similar scenario on my air cooled 4770k (4.5ghz 1.3v)

As soon as I start the stress test, it jumps to 80 - 90 C, and imediatelly as i stop the test, it goes back to 50C

 

[shibbs]

13 minutes ago, RafaelSoaresP said:

I`m getting similar scenario on my air cooled 4770k (4.5ghz 1.3v)

As soon as I start the stress test, it jumps to 80 - 90 C, and imediatelly as i stop the test, it goes back to 50C

 

Hm. I think tonight I'll play with the overclock and see how much better/worse it gets if I play with the voltage. I'll see what I can get out of it with a lower voltage somewhere around 1.25 is a nicer number than 1.320.

[KE2012]

My temps do exactly the same with two 360mm rads on a 4670K with a rather pathetic OC to 4GHz. 

[EK Luc]

20 hours ago, shibbs said:

Ugh, I really don't want to have to re-bend more lines just to test...  haha  Really thinking I should have done soft lines.

 

@EK Luc do you have any insight to any of this?

I don't think something is wrong with your loop.  I would most probably blame the head spreader and the TIM inside.   

[RafaelSoaresP]

Also did some research, and it seems the i7 series have a generally bad thermal interface inside the chip. 

Some people even goes as far as opening it up and replacing the whole thing.