Question Is there'll be any benefit to run the loop this way?

[CodeZ1LLa]

I need advice.

I have the next parts:
"D5" from Alphacool
XSPC Raystorm Pro
EK-FC1080 GTX TF6
HWLabs 480GTX
HWLabs 420GTS
Alphacool XT45 360.

They are currently connected in series: Pump ===> XSPC CPU ===> EKWB GPU ===> Alphacool XT45 360 ===> HWLabs 420GTS ===> Res.
GPU OCed to the limit but running pretty cold and never exceed 40C which is a 15C delta over the ambient
CPU is 3700X and it reaches 70C under AVX load which is 45C over the ambient air and 42 over the water that comes to the blocks.

Water from the rads equalizes at 3-4C over the ambient air under the GPU and CPU stress test running simultaneously.

Water outcoming from the blocks is basically 1-1.5C over the chilled water after rads.
I have an HWLabs 480GTX that I'm planning to install in the next few days.

Is there'll be any benefit from running the loop in parallel?
Also, what would you improve? Should I leave just two rads to improve the flow?

I would be very grateful to hear some recommendations.
Thank you in advance!

P.S.: Answering the question from rubix_1011: I working on the project which has a goal to maximize the PBO performance of Ryzen processors. We are modding BIOSes, Power profiles, testing different SMUs, etc. For testing purposes, I need my system as cold as it can get without LN2. This is the reason why I need so a lot of radiators. Thanks for your response!

 

[CodeZ1LLa]

Here is the sequence in which I want to build it. Is there'll be any benefit compared to the regular "series loop". Also, is there any difference in which order the rads will be placed to utilize them more efficiently? The D5 running at ~1.5GPM
Thanks!

 

[rubix_1011]

Having too many radiators doesn't mean you can get lower than ambient, though. Your coolant temperatures are always bound by ambient room temp.

In order to go colder than room temp, you need dry ice, compressor chilling, Peltiers or LN2.

 

[CodeZ1LLa]

Having too many radiators doesn't mean you can get lower than ambient, though. Your coolant temperatures are always bound by ambient room temp.

In order to go colder than room temp, you need dry ice, compressor chilling, Peltiers or LN2.

That I can't go lower than ambient is understandable
I want to shave that 3C delta over the ambient during the stress test.
Chiller would be great, but I'm paying for my education to get a degree in computer science so I have a limited budget, to be honest.
What is the best I can do having the parts described above? What would you change if it would be your build? Thanks!

 

[rubix_1011]

You could do it that way, yes. I don't think I've seen a loop where the radiator (Alphacool) has the purge/fill port used like that, but there isn't anything dictating that you can't.

Other than that, decent idea on the parallel setup, might help a bit with restriction on the CPU and GPU blocks, but likely won't matter much on the radiators.

 

[USAFRet]

Once you get past a certain radiator size in sq in, the actual limiting factor then becomes the size of the CPU face.
About 1.5 sq in.

There is only so much surface area you can present to move the 'heat' into the next medium, the liquid.
More rad surface area doesn't get it any cooler.

 

[CodeZ1LLa]

Once you get past a certain radiator size in sq in, the actual limiting factor then becomes the size of the CPU face.
About 1.5 sq in.

There is only so much surface area you can present to move the 'heat' into the next medium, the liquid.
More rad surface area doesn't get it any cooler.

I thought that while I still have delta between water and ambient air there is room to go with a radiators

 

[CodeZ1LLa]

You could do it that way, yes. I don't think I've seen a loop where the radiator (Alphacool) has the purge/fill port used like that, but there isn't anything dictating that you can't.

Other than that, decent idea on the parallel setup, might help a bit with restriction on the CPU and GPU blocks, but likely won't matter much on the radiators.

Thanks! Should an increase of the flow improve the temps, or I'll have a pressure drop after the y-splitter before the blocks, which will actually make cooling worse?

 

[USAFRet]

I thought that while I still have delta between water and ambient air there is room to go with a radiators

No.
The CPU is generating heat. That heat can only be moved to a different medium (the liquid) from that 1.5 sq in surface area.

The only way to have it at or near ambient is to turn it OFF.
Or, have some active cooler.

 

[USAFRet]

You could use an olympic size pool as a reservior and 1,000 GPH pump, and there will still be a delta between ambient and CPU.

 

[CodeZ1LLa]

No.
The CPU is generating heat. That heat can only be moved to a different medium (the liquid) from that 1.5 sq in surface area.

The only way to have it at or near ambient is to turn it OFF.
Or, have some active cooler.

Right. And that heat already moved to another medium(water) since water warmed up over the ambient temp. Now, I need to cool down that water, and for this purpose, we have radiators.
I think that there is a misunderstanding. I'm talking about the delta between the ambient air and chilled water that comes from the radiators, not between water and CPU

 

[USAFRet]

Right. And that heat already moved to another medium(water). Now, I need to cool down that water, and for this purpose, we have radiators.
I think that there is a misunderstanding. I'm talking about the delta between the ambient air and water, not between water and CPU

And there will still be a delta.
That water is being heated by the CPU.

 

[CodeZ1LLa]

And there will still be a delta.
That water is being heated by the CPU.

Correct. And GPU in my case.
I'll try to make it more clear:
Water_out (measured on the exit from the blocks) is 29-30C under the stress test
Water_in (measured between the pump and before entering blocks) is 28C under the stress test
Meanwhile, the ambient air is 2-3C below the "water_in". That delta between the ambient air and water_in could be minimized further in radiators before it enters the blocks. Hope it makes sense now

 

[Karadjgne]

You are right, there is a slight misunderstanding, but it's on your part. Actual temp of the water going in is the same as that going out. What you are seeing as temp, isn't. It's the energy absorbed by the coolant as it goes through. Some very small amount is absorbed by the hoses, some very small amount is eventually absorbed by the coolant, which is why stress tests on liquid cooling should be done over ½ hour, to give the coolant enough time to absorb the limit of energy it can.

That excessive energy leftover is carried to the rads, where it's easily transfered through the tubing, siphoned off by the fins. But the coolant itself suffers extremely little change.

Inline temp monitors, IR spot checkers, any temp reader of any sort is not reading actual coolant temps. The metal plate in contact with the coolant does nothing more than act like the radiator, absorbs heat. IR testers read hose temps in/out differences, so will always read slightly higher on output, reading all that excess energy absorbed by the hose.

You will not get a Delta change to be anything but different, always higher on output than input. Radiators are not perfect on transfer, there's always energy leftover, that travels around the loop, and gets further absorbed by other tubing.

So anywhere you read temps, it's going to be higher than ambient, because the coolant itself is not only carrying energy, but has also absorbed some of that energy so will always be higher than ambient.

There was a video created by Jayz2cents, showed 4x AlphaCool (I believe) 600w capacity 480mm rads, so that's 2400w worth of capacity, in a parallel loop cooling an i7-8700k. Idle temps still higher than ambient.

Without chemical intervention, you are at the best you can get, right now. There is no fix or cure.

Easiest way I can explain it is with a pan of water on the stove. Even on high, that water is constantly absorbing the energy transfered to the pan. Takes a good long while for that water to absorb enough energy to even raise the temp 1°C. If you took that water and shoved it through a rad to siphon off that energy and returned it to the pan, the water would never boil, wouldn't even get that hot. A burner runs @ 1500w on high, your cpu is at 200ish?

 

[rubix_1011]

Radiators do not chill water.

Radiators simply equalize the warmer temperature medium with the lower one.

If the ambient room temperature is warmer than that inside the cooling loop, the cooling loop temp rises, which is why higher ambient temps allow for higher registered CPU die temps (reported temps).

Watercooling and air cooling only work as long as there is an ability for the cooler to maintain this unequal temperature difference. This is why you do not want to use radiators when you are using a chilled loop; the radiators actually warm the coolant with ambient room air rather than removing any thermal load from it.

A radiator is a 2-way thermal equalizer...it always will be.

 

[CodeZ1LLa]

Radiators do not chill water.

Radiators simply equalize the warmer temperature medium with the lower one.

If the ambient room temperature is warmer than that inside the cooling loop, the cooling loop temp rises, which is why higher ambient temps allow for higher registered CPU die temps (reported temps).

Watercooling and air cooling only work as long as there is an ability for the cooler to maintain this unequal temperature difference. This is why you do not want to use radiators when you are using a chilled loop; the radiators warm the coolant with ambient room air rather than removing any thermal load from it.

A radiator is a 2-way thermal equalizer...it always will be.

Thanks! I do understanding that and experienced it before. I think I used the wrong word "chilling".

An interesting explanation about radiators as an equalizer between mediums rather than "cooler", I truly like it.
What I want to achieve, is an equal temp between the ambient air and water that comes after radiators. At least to make temperatures as close as possible, say within 1C under the load.

 

[CodeZ1LLa]

You are right, there is a slight misunderstanding, but it's on your part. Actual temp of the water going in is the same as that going out. What you are seeing as temp, isn't. It's the energy absorbed by the coolant as it goes through. Some very small amount is absorbed by the hoses, some very small amount is eventually absorbed by the coolant, which is why stress tests on liquid cooling should be done over ½ hour, to give the coolant enough time to absorb the limit of energy it can.

Thank you for such a great and unfolded answer.
See, when I have a low load, let's say with a browser and a Visual Studio, my water_in and ambient temp could be equal sometimes. Water_out is almost always 1C higher than water_in, unless I'm running the fans full speed when PC is idling.

The "ambient air" sensor placed right in front of the radiator fan, so it's measuring the temperature of the air that being intaked, not a case temp or anything else. Also, the ambient air sensor has been calibrated with a Kestrel 5500 anemometer which is very accurate. The only takeaway, that my sensors measure the temp without decimals, rounding temp to next or previous number if above or below .5, so half of the degree is a margin of error.

When I'm running fans full speed, the delta between the ambient air and water_in shrinks and CPU temperature drops accordingly, which as I understanding, means that there is some room to reduce the coolant temp a little bit further. Please, correct me if I'm wrong, and bear with me, because thanks to you and Rubix I just start to rethink my life in general

 

[CodeZ1LLa]

Thanks! You are absolutely right, there is still will be some delta, even if it will be so low, that it will be impossible for me to measure. What I want to do, is to minimize it.
I still think that with enough radiator surface I can achieve better results. Let's imagine that the radiator surface will be a mile(just for example). Water that passing trough it will be as close to the ambient as the closer it gets to the end of the rad. By this logic, I still hope to get a smaller delta between ambient air and water that comes after radiators. I just hope I'm not going to need a square mile or the radiator surface to achieve that
Anyway, I really appreciate all your input, guys, I'll add another 480 rad since I already have it and just in the name of science, and I'll post my results