My loop has poor CPU thermals :(

[_TJCAT_]

Hey, I'm running a 3900x and a RTX 2080 in a loop with two 360mm radiators and an EK Velocity block on my CPU. I have made sure to rid all air from my loop and have tried multiple thermal pastes and paste applications but my CPU still runs at 50-60c at idle with audible fan noise and will hit just over 90c under a sustained 100% load in AIDA 64 after 5 minutes (with no GPU heat being put into the loop). My GPU is in the same loop and is totally thermally fine, hitting a max of 52c at full load after a similar amount of time. I am suspicious that plausibly my block is not flat? as the CPU thermals increase rapidly when under load but tend to show little variance once it has heated up (for a 3900x). Even  now whilst typing this message my system fans are definitely audible as windows works through some menial background tasks with my CPU at roughly 8% usage. 

 

I sent a message to EK customer support just to see what they say, and I wont try to do anything as drastic as sanding the block to flatten it out until I get a reply, but I was wondering if y'all had any ideas for what to do next? I'm quite surprised about this issue. I had my 3900x on some 360mm aio before going full custom loop and i dont remember specifically how thermals were, but I do know that the CPU never came anywhere near 90c. 

 

Thanks homies 

[Master Disaster]

Yeah, you 100% have an issue somewhere. Check the water is flowing cleanly into the block, check theres no kinked pipes restricting flow and I know its a really dumb question but I have to mention it, you did peel the plastic from the bottom of the CPU block before applying it? (I already know the answer is yes to that one)

[_TJCAT_]

9 hours ago, Master Disaster said:

Yeah, you 100% have an issue somewhere. Check the water is flowing cleanly into the block, check theres no kinked pipes restricting flow and I know its a really dumb question but I have to mention it, you did peel the plastic from the bottom of the CPU block before applying it? (I already know the answer is yes to that one)

There are no kinked tubes, I did have a problem at some point where I kinked a tube (soft tubing fyi) whilst bleeding any small air bubbles out of the loop initially, but I recut and reran the tube and now everything is free. And yeah, I did remove the plastic. To be honest, the first time I realised my thermals were so high that was what my mind went to, so I did check it but unfortunately my problem was not that easy .

About clean flow - how would I check that? I can observe that the loop is flowing steadily through the res as the liquid spins but because the transparent GPU block has no air bubbles the lateral flow can't visually be seen and the CPU block is all opaque and so I cant directly see flow. But when first taking the kinked tube that I spoke of off, the block did seem to be holding a fair amount of liquid, or a fair amount flowed out. 
But I'm not convinced its a flow issue, because the thermals increase to 90c within seconds of a load being applied, which is not enough time for the liquid in the block to heat soak if that liquid wasn't moving. 

 

Thanks for answering! 

[AngryBeaver]

13 hours ago, _TJCAT_ said:

There are no kinked tubes, I did have a problem at some point where I kinked a tube (soft tubing fyi) whilst bleeding any small air bubbles out of the loop initially, but I recut and reran the tube and now everything is free. And yeah, I did remove the plastic. To be honest, the first time I realised my thermals were so high that was what my mind went to, so I did check it but unfortunately my problem was not that easy .

About clean flow - how would I check that? I can observe that the loop is flowing steadily through the res as the liquid spins but because the transparent GPU block has no air bubbles the lateral flow can't visually be seen and the CPU block is all opaque and so I cant directly see flow. But when first taking the kinked tube that I spoke of off, the block did seem to be holding a fair amount of liquid, or a fair amount flowed out. 
But I'm not convinced its a flow issue, because the thermals increase to 90c within seconds of a load being applied, which is not enough time for the liquid in the block to heat soak if that liquid wasn't moving. 

 

Thanks for answering! 

My guess is it is a mount issue. I would say you had the inlet and outlet on cpu block backwords, but temps wouldn't be THAT high.

 

I would suggest removing the block and looking at the paste spread to see if it is a issue with the IHS/block and to make sure the mount pressure was enough to spread it well enough.

 

Normally when you have one device on the loop with good temps and the other with bad... then the issue is mounting.

[_TJCAT_]

49 minutes ago, AngryBeaver said:

My guess is it is a mount issue. I would say you had the inlet and outlet on cpu block backwords, but temps wouldn't be THAT high.

 

I would suggest removing the block and looking at the paste spread to see if it is a issue with the IHS/block and to make sure the mount pressure was enough to spread it well enough.

 

Normally when you have one device on the loop with good temps and the other with bad... then the issue is mounting.

Yeah I agree with you that there is some issue with the thermal conductivity between the block and the CPU. I checked the loop orientation and the inlet/outlet ports are all lined up correctly.

My loop order is as such: Res/pump -> radiator -> GPU block -> Radiator -> CPU block -> Res/pump

This is the block I'm using: https://www.ekwb.com/shop/ek-velocity-d-rgb-amd-nickel-acetal

In the manual it shows the block is to be mounted with springs and thumbscrews, I have tried mounting and remounting the block like this several times(and observed previous paste to be spread well across the CPU) and at this stage the thumbscrews are tightened as hard as I can get them, the springs can actually be heard scratching against themselves as I tighten the thumbscrews because they are fully compressed. So I don't ?believe? there is a mounting issue? Or at least I see no possibility of one unless the springs that are included with the block are too large?

[_TJCAT_]

57 minutes ago, AngryBeaver said:

My guess is it is a mount issue. I would say you had the inlet and outlet on cpu block backwords, but temps wouldn't be THAT high.

 

I would suggest removing the block and looking at the paste spread to see if it is a issue with the IHS/block and to make sure the mount pressure was enough to spread it well enough.

 

Normally when you have one device on the loop with good temps and the other with bad... then the issue is mounting.

Thanks for the reply btw! 

[_TJCAT_]

UPDATE:

I spent forever talking to some EK customer support dude, hes actually really nice just took forever to reply, and im getting the coldblock replaced for free so we will see how effective that is.

[EvilVir]

I have some issues with my loop, using same block, as well. Same temps, just higher CPU (but no GPU) Is there possibilty you would take a picture of springs compressed while block is attached? I want to see how far they went.