[Fisbwp283]

is this for 24/7 use? I ran a SS phase back on my 955BE, it ran well for a long long time, but I ended up getting pin rot, even with die-electric grease in the cpu socket. 4 pins fell off.

Yes 24/7.

Corrosion that caused your pin rot only occur with both oxygen and moisture (hydrolysis/electrolysis & oxidation)... so in a dry nitrogen atmosphere this issue is prevented. Hence my interest in anybody’s experience running a system under N2 desiccation.

 

[Cerberus]

Yes 24/7.

Corrosion that caused your pin rot only occur with both oxygen and moisture (hydrolysis/electrolysis & oxidation)... so in a dry nitrogen atmosphere this issue is prevented. Hence my interest in anybody’s experience running a system under N2 desiccation.

oh, you're not wrong, I ran -40c in a room with no a/c for nearly three years without disassembly. it ran fine even with those 4 pins gone (by fine i mean it was able to play cs source and browse the web)

 

[Fisbwp283]

I do agree 100%. But I'm an enthusiast and take interest in specialized cooling such as glycol chillers and SS.... probably because I've never used either cooling type apparatus.
But done plenty of other chilling methods, just about all the other chilling methods in fact.

Currently running 2700X on stock air. -30c didn't make any difference in max OC, but did allow the use of lower volts at X oc.

Just curious what flow rate you are getting at -30c and what fluid?
also, how many pumps in series are you using?

 

[Fisbwp283]

After doing some more flow testing and temp testing, i think i can control the pressure and temps down to -20c at 42-45% glycol for benchmarks, i think -12 will be the limit for full load continuous operation with 1 GPU.
But to get sufficient flow at G1/4 i will need to add pumps.

Option 1
- two pumps in series and remove CPU jet plate giving a fluid flow:
Chiller > Agitator pump > RAD > twin d5 in series > CPU > GPU(s) > Chiller
with Jet plate in, the pressures spike when flowing enough.

Option 2
- one pump for CPU with jet plate still in
- one pump for GPU
- common return to chiller.

Has anyone run Glycol like this before? what was the solution for jet plate restrictions?

 

[ShrimpBrime]

Just curious what flow rate you are getting at -30c and what fluid?
also, how many pumps in series are you using?

Flow rate is variable. However it's (water) not cooling a processor, it's cooling a TEC which cools the cpu to -30c.

I was merely stating that even at such a temperature, -30c, there was no more to be gained with a 2700X processor. Better results with Ryzen chips would be under an LN2 pot. sry for any confusion there.

 

[clannagh]

Spending the money on the xtreme cooling gives dimineshing returns
Compared to the fancy CPU w a cheap 120mm AIO chiller + OC

I belive overclocking is stupid if you want outright perfomance and you currently are water chilled with AIO water 120mm rads or better.

Overclocking older tech often has the same practical value as hot-rodding your dad's 1980's Mitsubushi Galant.

Whilst there may be street cred in having the fastest Mitsubushi Galant in the state, it is still going to be slower than your mum's stock 2019 model SUV.

 

[Fisbwp283]

An update of Sorts,

I was about to pull the trigger on an X570 Aorus Xtreme MB and wait for the 3950x...but given the delays, i'm waiting to see what launches with threadripper and look at thermal and OC results before i start this build.

in the mean time, i seem to have the pumping situation sorted, but i now need a way to have PWM control of 4 D5 pumps. which will lock me into a high end MB with lots of 4 pin PWM headers, which i'll need when not running sub zero.

I have one Glycol pump which gets flow to the radiator at -25c but it's not enough so i want to split the loop to keep pressures under control and keep running the jet plate on the CPU block.
so flow will be:
Chiller>GP>RAD>Y1&2
Y1>EkD5>EKD5>CPU>Chiller
Y2>EkD5>EKD5>CPU>Chiller

 

[Fisbwp283]

After much trial and error.

New plan.

Dual pathway cooling with pathway selected by simple valve

pathway 1
- 30% Glycol chiller bath with agitator and chiller coils in closed loop with PC
- Ethanol + water cooling fluid.
- negative 10 Celsius target

Pathway 2
- DICE reservoir with chiller coils in closed loop with PC
- Pure Ethanol as cooling fluid
- Negative 70-80 Celsius target

To-Do
-Build gas tight chill box that can be nitrogen flooded to prevent condensation.