Eisbaer 280 high CPU temps under load

Dorstlesser

New member
Main Specs:
  • Mobo: ASrock X470 Taichi
  • CPU: Ryzen 3900X
  • GPU: Radeon RX5700
  • Case: Fractal Design Define 7
Cooling
  • Eisbaer 280 CPU cooler (with DC-LT 2 Ceramic pump)
  • Eisblock Aurora for GPU
  • Quick-connect extension kit
  • Casefans: 3 x 140 mm intake, 2 x 140 mm outtake on radiator

Dear all,
I've been running the Eisbaer 280 on my CPU with additional GPU block for a while now and while GPU temps have been amazing the CPU temps were always a bit on the high side with max 78C under load on 'ECO'. I applied some more thermal paste and refastened the cooler and CPU temps dropped a couple of degrees, which suggested that initially I had been too conservative with the thermal paste and fastening...

Recently I tried Ryzen Clock Tuner for which I turned off ECO mode and noticed CPU temps going to 90C in Cinebench and even 95C in AIDA64 stress test. I'm running the CPU at default settings and PBO is not turned on.

I'm wondering if the cause could be mounting pressure? I have followed the manual for AM4 mounting, put both 2mm and .5mm washers in place and I can not apply more pressure because the springs are completely pressed. I hope you can help.

Water flow is from radiator > pump/CPU > GPU > back to radiator. To confirm: GPU load does not lead to increased CPU temps but vice versa it does. GPU temps suggest pump works normally; max RPM is around 2700. Temperature in the case is between 30-35C.

What would be your advice?

Thanks,
Paul
 

Eddy

Iceman
Staff member
Are you using the nuts? If so, read them away. We have deleted them from the new instructions because they can prevent proper contact pressure if used incorrectly (and this can happen quickly).
Eventuell hilft es uns, wenn du mal ein Bild von der Montage zeigst.
 

Dorstlesser

New member
Hi Eddy,
Thanks for your quick reply. I was using the nuts indeed. I reinstalled without the nuts. This improved idle temps, and temps during Cinebench dropped by +- 5C from 90C to 85C which is a clear improvement. During AIDA64 stress test temps still go to 95C; I guess it's thermal throttling at this point and clocks might be higher than before but I did not yet check this. I did not tune voltages yet so that will also shave off a few degrees still.

I've included pictures of the build. I know tube length is a bit excessive and maybe that restricts flow a bit as well?

Again thanks for any advice. Really love your product line; would not have been able to afford putting both CPU and GPU on water cooling if not for your modular system! It's my first water-cooled build so I'm still learning.

Cheers,
Paul
 

Attachments

  • 20220225_094502226_iOS-2016.jpg
    20220225_094502226_iOS-2016.jpg
    335 KB · Views: 17
  • 20220225_094511487_iOS-2016.jpg
    20220225_094511487_iOS-2016.jpg
    310.1 KB · Views: 19
  • 20220225_094546100_iOS-2016.jpg
    20220225_094546100_iOS-2016.jpg
    298.4 KB · Views: 17
  • 20220225_095008253_iOS-2016.jpg
    20220225_095008253_iOS-2016.jpg
    226 KB · Views: 17

Eddy

Iceman
Staff member
Before I get into the actual problem, a few general points. Behind the radiator are directly the hard drives, this hinders the airflow in the case and thus also the cooling performance of the radiator. A 280mm radiator for GPU and CPU is already a bit limiting when both are running at full load. You won't get the best temperatures that way.

But to the actual topic. The hose routing and the GPU cooler naturally create a high resistance and reduce the flow. But actually, this should still be enough for the cooling to work.
If you load the GPU (without CPU with e.g. Furmark), the temperatures are in the green area, right? Because if they were also much too high, then you could say that the flow is no longer sufficient.
Why the CPU jumps to such temperatures is hard to say. The assembly seems to fit. I would check the voltages of the CPU. Certain tools or even a Bios update or various settings can raise them to an unnecessarily high value. This can be read relatively easily with the free tool HwInfo.
Have you screwed in the screws so far that the springs are on the stop? Have you looked at the thermal paste print in general? You have to take the cooler off the CPU with as few rotations as possible. Here you can often see well if something has not worked properly during assembly. E.g. that on one side no correct impression is to be seen or similar.
 

Dorstlesser

New member
Sorry for the late reply; I was AFK the last couple of days.

If you load the GPU (without CPU with e.g. Furmark), the temperatures are in the green area, right? Because if they were also much too high, then you could say that the flow is no longer sufficient.

The GPU stays absolutely frosty. During Furmark stresstest (1440p / 2 x msaa) edge temperature peaks at 56C and junction at 77C with the fans set quite high (as for CPU). The temperatures are the same when I stress the GPU alone with AIDA64. When I stress CPU and GPU simultaneously, the GPU temps increase to 65C and 86C (AIDA64). However, CPU temps are the same when I load only the CPU or CPU + GPU simultaneously (87C). When I run my normal, relatively quiet, fan curve, the GPU temps peak at 63C and 83C.

Because of that I'm not really worried about enough airflow or radiator surface. There's quite some space between the HDDs and side panels of the case where air can flow, but that's not really clear from the pictures.

I would check the voltages of the CPU.

It was on the high side but for these latest tests I have been able to lower it to 1.136 V with an all-core boost to 4100 MHz. I believe this is not a particularly high voltage.

Have you screwed in the screws so far that the springs are on the stop?

Yes

Have you looked at the thermal paste print in general? You have to take the cooler off the CPU with as few rotations as possible. Here you can often see well if something has not worked properly during assembly. E.g. that on one side no correct impression is to be seen or similar.

I have always ended up twisting the cooler too much to really accurately see this. Initially, temps increased by applying a bit more thermal paste and increasing the mounting pressure, but I'm not sure if one of those things could be working against the other?
 

pdifolco

Member
The temps you have on a 3900X are enormous, I've used a 3900X with a 280mm Eisbaer and never had more than 75C with 23C ambient
Mounting pressure won't never give you 15C less, I've lost some washers and now don't use any, didn't see any difference

But from the pics it seems that you use only 1 280mm rad, mounted on the bottom ? Am I right ?
Then your CPU block get all the air of the circuit, and if there's too much air the block will fail to cool anything, did you refill the circuit ?
 

Dorstlesser

New member
The temps you have on a 3900X are enormous

Yeah I've been getting that impression. First I thought that the Ryzen IHS is just crap and the 3900X is a powerful processor so there you go, but I don't think my temps are normal, all things considered.

But from the pics it seems that you use only 1 280mm rad, mounted on the bottom ? Am I right ?
Then your CPU block get all the air of the circuit, and if there's too much air the block will fail to cool anything, did you refill the circuit ?

Only 1 280mm rad, mounted on the front of the case as outtake (picture should be rotated 90 degrees; left is up). The top of the radiator sits above the CPU block (the side *without* the cable attachments; those are at the bottom, below the CPU block). I tried following Gamer Nexus' pointers (YouTube video) and took quite a lot of care to make sure I got as much air out as possible after refilling.

However I had not considered that there might be some air trapped in the CPU block. Would it be safe to try and dislodge it by tapping the cooler gently?
 

pdifolco

Member
Yeah I've been getting that impression. First I thought that the Ryzen IHS is just crap and the 3900X is a powerful processor so there you go, but I don't think my temps are normal, all things considered.



Only 1 280mm rad, mounted on the front of the case as outtake (picture should be rotated 90 degrees; left is up). The top of the radiator sits above the CPU block (the side *without* the cable attachments; those are at the bottom, below the CPU block). I tried following Gamer Nexus' pointers (YouTube video) and took quite a lot of care to make sure I got as much air out as possible after refilling.

However I had not considered that there might be some air trapped in the CPU block. Would it be safe to try and dislodge it by tapping the cooler gently?
Ok I didn't rotate my head accordingly :)
Lack of coolant could still be an issue, you can try moving the air, but Alphacool made tutorial vids on YT to showcase how to refill an Eisbaer, which will be better
 

Dorstlesser

New member
So, there was a bunch of air in there. Either I had not filled it up properly ever (possible, although I tried), or water had evaporated. Temps dropped to balmy levels after proper refilling.

However I did notice some nasty-looking discoloration on my GPU block. Could anybody shed some light on what I might be looking at here and how to fix and prevent it? I've only ever refilled with distilled water from Alphacool themselves.

It's difficult to see on the pictures but the cooling fins have colored almost black but the rest of the deposit seems to be more greenish/brownish.
 

Attachments

  • 20220406_123606078_iOS-3024.jpg
    20220406_123606078_iOS-3024.jpg
    486.6 KB · Views: 13
  • 20220406_123610065_iOS-3024.jpg
    20220406_123610065_iOS-3024.jpg
    533.7 KB · Views: 13

Thomas_S

... the nice marketing guy next door
Staff member
Cleaning of the cooler is necessary. How long has the system been running like this?
 

Tived

Member
@Thomas_S are there a tutorial on how to clean the 280? If so would you mind share it or provide some instructions please

Thanks in advance

Henrik
 
Top