Production Aquanaut - Ultra Low Profile CPU/Pump Water Block

Valantar

SFF Guru
Jan 20, 2018
1,544
1,405
But I can agree its kind of unfortunate that you cant pay additionally for another mounting mechanism, whats the problem with that really? That was one of the reasons I got rid of mine, fortunately without any loss.
For a small operation that kind of flexibility creates a lot of overhead - separate packaging for individual mounting brackets or re-packaging orders by hand for individual orders with multiple brackets, plus the need to have a stock of spare brackets of both types that might never sell. It's a lot more complicated than it might sound.
 

einmannbude

Caliper Novice
May 22, 2016
24
16
For a small operation that kind of flexibility creates a lot of overhead - separate packaging for individual mounting brackets or re-packaging orders by hand for individual orders with multiple brackets, plus the need to have a stock of spare brackets of both types that might never sell. It's a lot more complicated than it might sound.
How about reducing the complexity of two separate products even down to only one by including both brackets by standard?
 

Nouvolo

Creator
Original poster
Silver Supporter
Sep 8, 2018
508
1,020
www.nouvolo.com
I am STILL waiting for the screws!
They should be on its way. Since it is such a small value item, we can only use standard registered air mail to send them out (not as fast as FedEx). There may be delays due to the latest covid out break. We will check with post office to see if there are any issues also. Sorry again for the delays.

Email us with your order number and details. We will check whether delivery has been attempted (or failed).
 
Last edited:

Allhopeforhumanity

Master of Cramming
May 1, 2017
507
492
They should be on its way. Since it is such a small value item, we can only use standard registered air mail to send them out (not as fast as FedEx). There may be delays due to the latest covid out break. We will check with post office to see if there are any issues also. Sorry again for the delays.

Email us with your order number and details. We will check whether delivery has been attempted (or failed).

Mine just showed up over the weekend. With all of the christmas shipping/mail I'm sure the USPS has had their hands full. Now if only I could get my hands on a 5xxx series processor to put it on.
 

NRG

Cable-Tie Ninja
Oct 30, 2015
213
245
How does one go about acquiring the new stand off screws? I gave it a crack with my 9900K and temps seemed a little high, unsure if it’s getting good contact. Also have a crack on the DDC pump mount screw. I suggest you be careful with the screws used to mount the pump. It (the block as a whole) didn’t leak though, so that was positive :)
 

Post-Newt

Average Stuffer
Oct 12, 2020
87
211
I tested EK Velocity and Aquanaut at the same coolant temperature and pump speed while running AIDA64 stress test to simulate 100% load.

Setup:
  • Ryzen 5900X
  • Laing DDC-1T Plus PWM
  • Aquacomputer Quadro
  • Thermal Grizzly Hydronaut thin spread
CPU blockCPU ℃ averageCoolant ℃Pump speed
EK Velocity AM4654025%
Aquanaut AM4734025%

Aquanaut is a truly unique product but there is certainly room for improvement when it comes to thermal performance.

I think these results can at least partially be explained by the smaller fin area and shorter fins on Aquanaut. FWIW according to benchmarks on GN, typically AIOs with the largest fin area, tall and dense fins have the best thermal performance.
Hope this is something that will be improved in future revisions.
I'm aiming to do this comparison as well in a couple weeks. I've got an EK Velocity, Aquanaut, and just got the AM4 adapter for my ApogeeII. QDCs ready for some quick swaps. Currently googling best ways for a controlled/consistent experiment and any advice is appreciated. Also looking around for a good flow meter that doesnt cost a fortune. Your results are very surprising and has me completely second-guessing my build plan. Starting to think I might go D5 + Velocity in the winter case after all.

Would you mind checking if your Aquanaut cold plate is flat? Perhaps you have poor contact.
 
Last edited:
  • Like
Reactions: eedev

biopunk

Cable-Tie Ninja
Bronze Supporter
Sep 24, 2020
168
229
I'm aiming to do this comparison as well in a couple weeks. I've got an EK Velocity, Aquanaut, and just got the AM4 adapter for my ApogeeII. QDCs ready for some quick swaps. Currently googling best ways for a controlled/consistent experiment and any advice is appreciated. Also looking around for a good flow meter that doesnt cost a fortune. Your results are very surprising and has me completely second-guessing my build plan. Starting to think I might go D5 + Velocity in the winter case after all.

Would you mind checking if your Aquanaut cold plate is flat? Perhaps you have poor contact.
The coldplate on mine wasn't bent, it had some minor scratches but I don't think they would make a noticeable difference.

I think for a fair comparison between CPU blocks the (only) conditions which matter are:
1) thermal paste application. GN said the application method makes only a minor difference. Thin spread is preferable as it's more repeatable.
2) flow rate - depends on pump speed, pump top, and restrictiveness of CPU block if the rest of the components are the same. With a flow meter in the loop, you can change the pump speed to achieve a similar flow rate. I tried changing the speed by 5-10% (more would make the pump noisy) but didn't notice much of a difference in CPU temperature.
3) coolant temperature - should be the same between tests. If you have Quadro you can just set a target temperature and let the controller adjust the fans automatically.
4) CPU load - it should be consistent throughout the benchmark, I find AIDA64 pretty good in this regard.

When you'll be testing would you mind also tracking the time that CPU takes to cool down? With Aquanaut my CPU seems to cooldown noticeably slower than with EK Velocity.
 
Last edited:

Nouvolo

Creator
Original poster
Silver Supporter
Sep 8, 2018
508
1,020
www.nouvolo.com
Please remember determining factor for temps : radiator size
240mm rad most likely not enough for high end cpu+gpu cooling, and you will get high temps.

Suggest quote your radiators size for tests.
 

Post-Newt

Average Stuffer
Oct 12, 2020
87
211
The coldplate on mine wasn't bent, it had some minor scratches but I don't think they would make a noticeable difference.

I think for a fair comparison between CPU blocks the (only) conditions which matter are:
1) thermal paste application. GN said the application method makes only a minor difference. Thin spread is preferable as it's more repeatable.
2) flow rate - depends on pump speed, pump top, and restrictiveness of CPU block if the rest of the components are the same. With a flow meter in the loop, you can change the pump speed to achieve a similar flow rate. I tried changing the speed by 5-10% (more would make the pump noisy) but didn't notice much of a difference in CPU temperature.
3) coolant temperature - if you have Quadro you can just set a target temperature and let the controller adjust the fans automatically.
4) CPU load - it should be consistent throughout the benchmark, I find AIDA64 pretty good in this regard.

When you'll be testing would you mind also tracking the time that CPU takes to cool down? With Aquanaut my CPU seems to cooldown noticeably slower than with EK Velocity.
Please remember determining factor for temps : radiator size
240mm rad most likely not enough for high end cpu+gpu cooling, and you will get high temps.

Suggest quote your radiators size for tests.
I think I should be able to "borrow" a Dwyer from work with +-4% error and manually adjust pump rpm to 16gph, adapt it to G1/4 for $6, Connect a 360 with push or push/pull if needed (lock fan rpm or run off coolant temp?), coolant temp with this, only water cool the cpu (3900x with fixed clock speeds), and run a torture test til loop has stabilized.

Only thing I have on hand capable of quantifying pump noise is this. Will shield fans and gpu from pump and see what happens. Probably won't give any meaningful data.

Unsure what I can do about HVAC. Will probably lay everything out on a table and maybe cover with a box or something.
 
Last edited:
  • Like
Reactions: thelaughingman

Allhopeforhumanity

Master of Cramming
May 1, 2017
507
492
Please remember determining factor for temps : radiator size
240mm rad most likely not enough for high end cpu+gpu cooling, and you will get high temps.

Suggest quote your radiators size for tests.

To this point: pump flowrate, radiator thickness, fin-pitch, and fan performance curve (flowrate vs pressure) will also play significant roles in the cooling capacity of the system. Higher pump flowrate increases the convection coefficient of heat transfer between the fluid and the coldplate and radiator tubes. Radiator thickness and fin-pitch both increase the surface area and the total cooling power available per CFM of airflow, but keep in mind that these come at the cost of fan pressure requirements to maintain said flowrate. Lastly, the fan performance curve will indicate the how well suited it is to the aerodynamic impedance of the radiator; a general rule here is that more pressure generated at a given fan speed will always be better for performance, but can come at the cost of noise induced by turbulence.

TLDR: Don't get something like an XSPC TX240 (20mm thick) 240mm rad and expect it to be sufficient to cool a 5900x and 3080 without some involved power tuning and extreme fan speeds.
 

biopunk

Cable-Tie Ninja
Bronze Supporter
Sep 24, 2020
168
229
To this point: pump flowrate, radiator thickness, fin-pitch, and fan performance curve (flowrate vs pressure) will also play significant roles in the cooling capacity of the system. Higher pump flowrate increases the convection coefficient of heat transfer between the fluid and the coldplate and radiator tubes. Radiator thickness and fin-pitch both increase the surface area and the total cooling power available per CFM of airflow, but keep in mind that these come at the cost of fan pressure requirements to maintain said flowrate. Lastly, the fan performance curve will indicate the how well suited it is to the aerodynamic impedance of the radiator; a general rule here is that more pressure generated at a given fan speed will always be better for performance, but can come at the cost of noise induced by turbulence.

TLDR: Don't get something like an XSPC TX240 (20mm thick) 240mm rad and expect it to be sufficient to cool a 5900x and 3080 without some involved power tuning and extreme fan speeds.
Sure all of that matters when it comes to overall cooling performance but when you are comparing two CPU blocks to each other and the rest of the components are the same, the conditions that I listed above are the same, I'd argue that it doesn't.

@Post-Newt I had a thought last night that CPU block orientation might affect the flow through the block, so I would suggest mounting the mobo vertically for testing. Not sure about the inlet and outlet orientation though, FWIW I have them on the right side in T1.
 
Last edited:

NRG

Cable-Tie Ninja
Oct 30, 2015
213
245
Please remember determining factor for temps : radiator size
240mm rad most likely not enough for high end cpu+gpu cooling, and you will get high temps.

Suggest quote your radiators size for tests.
Hey Mate. How does one get replacement screws for the first batch of blocks sent out? Thanks
 

Allhopeforhumanity

Master of Cramming
May 1, 2017
507
492
Sure all of that matters when it comes to overall cooling performance but when you are comparing two CPU blocks to each other and the rest of the components are the same, the conditions that I listed above are the same, I'd argue that it doesn't.

@Post-Newt I had a thought last night that CPU block orientation might affect the flow through the block, so I would suggest mounting the mobo vertically for testing. Not sure about the inlet and outlet orientation though, FWIW I have them on the right side.

I'm not sure which post you're referencing here. But in the case of comparing blocks only, the microchannels and chamber impedance are likely the only contributors.

A higher flow rate has a side effect of increasing the coolant temperature, meaning that fans have to work harder and the whole system becomes louder. Also above ~25% DDC-1T Plus becomes audible so most people tend keep it at 20-25%.

This statement confuses me though. Higher working fluid temps means that the same radiator, utilizing the same fan profile, will have better cooling performance since total heat flux scales positively with deltaT. If instead what you are implying is that the pump block's convective performance scales faster than the radiator's, and that the chip is still producing more thermal power than the cooling capacity of the system, then run-away temperatures at the chip are inevitable, and would only be exacerbated by reduced pump speed.
 

matique

Chassis Packer
Oct 28, 2020
16
39
Hey yo. I'm going to be putting a supremecy coldplate to the aquanaut. The supremecy plate is deeper though, so i'm not sure if the aquanaut acrylic would make contact with the fins though. I'll find out in about 3 hours.

Alright so this did not work at all, and upon closer inspection i'm kind of disappointed at Nouvolo. Your coldplate fins are pretty bad. I knew they weren't good, but i did not expect them to be this bad, the fins are uneven in depth and sometimes even thickness. Not as bad as Cennis's copy, but still pretty bad compared to other coldplates.

EDIT removed photo as per Nouvolo suggestion.

The reason why the supremecy won't fit is because it has some depth to it to account for the EK jetplate and protruded inlet. Here's a comparison of the two to show the difference in depth.


Nouvolo, if you are making a revised coldplate, I hope you could link up with other vendors that have been making coldplates for awhile or have the machinery to do so properly. Your fin area is smaller than other coldplates, and the density simply isn't there. I hope you guys manage to figure out a method to integrate a forward flow system + a better coldplate. I do not mind paying more if you need to oursource your coldplate to say, barrow/bitspower/bykski. I think it's a good first attempt at a AIO and it has allowed to make a very competent sff system, but there is much room for improvement.

EK Supremecy on the left, Aquanaut on the right.


edit: seems like this was from a bad batch as others have had good milling. Will proceed to ask nouvolo about it.
 
Last edited: