Production Aquanaut Basic/Extreme - Ultra Low Profile CPU Block & Pump Mount Combo

Nouvolo

Creator
Original poster
Silver Supporter
Sep 8, 2018
759
1,711
www.nouvolo.com
While waiting for the site to allow for orders, I came up with a question (suggestion?) for you, @Nouvolo:

I can't see if this has been said explicitly, but the cold plate looks to be entirely CNCed, including the microfins. Have you looked into getting the cold plate skived instead? I'm curious as to the reasoning behind going with a CNCed solution - is it because of accessibility, price, or performance? A lot (if not most) great water cooling cold plates seem to have skived microfins. Would that be a possible change for a future revision? At least on paper, skiving allows for thinner and denser fins, i.e. more surface area without significantly increased flow restriction.
Because it is one of the better quality cold plates that I can source and at reasonable price. I am sure there are other kinds of plates out there (better or worse). I am just offering what I am able to obtain and with a stable supply going forward for repeated orders. Btw, if anybody has a good source for certain components/parts, please feel free to let me know, may be I can make use of those ?.

Took me 4-5 months to bring this from concept to market. I spent about 3 weeks sourcing coldplates (and not easy with covid lockdown). Once I found an acceptable source, I stopped actively look for others. I am sure I can find others if I keep looking, but I have other things to look after in development of this product...
 
Last edited:

duynguyenle

Airflow Optimizer
Aug 20, 2019
328
330
While waiting for the site to allow for orders, I came up with a question (suggestion?) for you, @Nouvolo:

I can't see if this has been said explicitly, but the cold plate looks to be entirely CNCed, including the microfins. Have you looked into getting the cold plate skived instead? I'm curious as to the reasoning behind going with a CNCed solution - is it because of accessibility, price, or performance? A lot (if not most) great water cooling cold plates seem to have skived microfins. Would that be a possible change for a future revision? At least on paper, skiving allows for thinner and denser fins, i.e. more surface area without significantly increased flow restriction.

Where exactly did you get this information from? I did a cursory look through the most recent blocks from the major manufacturers, they all have coldplates with CNC milled fins. Here are the blocks I looked at:
EK Magnitude
EK Velocity
Optimus V2
Alphacool XPX
Heatkiller 4
Koolance CPU-400I

All of the above looks to have CNC milled blocks. The only open-loop CPU block I saw that may possibly have skivved fin is from the Phanteks C350i (based on this fin profile picture here: https://www.techpowerup.com/review/phanteks-glacier-c350i/images/block-19.jpg )

Typically, you most often see micro-skivved fins on AIO coldplates, not open-loop custom blocks. Also, flow restriction is a function of both fin thickness, height of stack and density. If you have higher fin density, you'll also have higher restriction, even if your fins are thinner.
 

Nouvolo

Creator
Original poster
Silver Supporter
Sep 8, 2018
759
1,711
www.nouvolo.com
... I did a cursory look through the most recent blocks from the major manufacturers, they all have coldplates with CNC milled fins. Here are the blocks I looked at:
EK Magnitude
EK Velocity
Optimus V2
Alphacool XPX
Heatkiller 4
Koolance CPU-400I
...
Yea, the water block Aquanaut uses is pretty similar to that of EK Supremacy (as below). That's why I picked it... I assume EK is pretty good benchmark? ? This thing costs USD49 on amazon

.. Now you have made me done some more market research maybe I should raise the price.?

 

fabio

Shrink Ray Wielder
Apr 6, 2016
1,885
4,325
Yea, the water block Aquanaut uses is pretty similar to that of EK Supremacy (as below). That's why I picked it... I assume EK is pretty good benchmark? ? This thing costs USD49 on amazon

.. Now you have made me done some more market research maybe I should raise the price.?

Or just change the mounting holes to match it, so people can use it! :) BTW, I wil be back in denmark next week and I will test your Aquanaut! ?
 

Valantar

Shrink Ray Wielder
Jan 20, 2018
2,201
2,225
Where exactly did you get this information from? I did a cursory look through the most recent blocks from the major manufacturers, they all have coldplates with CNC milled fins. Here are the blocks I looked at:
EK Magnitude
EK Velocity
Optimus V2
Alphacool XPX
Heatkiller 4
Koolance CPU-400I

All of the above looks to have CNC milled blocks. The only open-loop CPU block I saw that may possibly have skivved fin is from the Phanteks C350i (based on this fin profile picture here: https://www.techpowerup.com/review/phanteks-glacier-c350i/images/block-19.jpg )

Typically, you most often see micro-skivved fins on AIO coldplates, not open-loop custom blocks. Also, flow restriction is a function of both fin thickness, height of stack and density. If you have higher fin density, you'll also have higher restriction, even if your fins are thinner.
Hm, I was sure there were more skived open-loop parts. Corsair definitely uses skived fins for their XC9 at least, but I might have been looking at too many AIO cold plates recently. It makes sense that not many open loop component makers have skiving machines as they are massive, expensive, and essentially unitaskers, but they also allow for much cheaper and faster production of parts of arguably higher quality (not to mention the difficulty of CNC milling thin channels into soft copper). Of course I would imagine the MOQ from any cooling OEM with a skiving machine likely being in the thousands (if not tens of thousands), so for anything on this type of scale that's likely an impossible goal.

As for flow restriction, what you're saying is obviously true, but it doesn't take away from the fact that thinner microfins allow for either more fins or wider channels in the same area, with the reduction in thermal transfer from thinner metal being essentially negligible at scales this small (especially as the channels tend to be quite shallow). Skiving also more easily allows for taller fins than milling, as there's no chance of tear-out. As such, depending on the exact configuration of your design it is always possible for a skived block to match the flow of a milled block while having more surface area, or for it to have better flow with the same surface area. The difference is likely not large, but it's there. But as I said, I would imagine this comes down to economies of scale and the rather extreme high cost threshold of getting one of these machines in the first place.

Because it is one of the better quality cold plates that I can source and at reasonable price. I am sure there are other kinds of plates out there (better or worse). I am just offering what I am able to obtain and with a stable supply going forward for repeated orders. Btw, if anybody has a good source for certain components/parts, please feel free to let me know, may be I can make use of those ?.

Took me 4-5 months to bring this from concept to market. I spent about 3 weeks sourcing coldplates (and not easy with covid lockdown). Once I found an acceptable source, I stopped actively look for others. I am sure I can find others if I keep looking, but I have other things to look after in development of this product...
That was pretty much what I was expecting, but it's nice to get it confirmed :) I can't even imagine how difficult the process of ordering parts like this is (and I would imagine a lot of OEMs won't even talk to you unless you're ordering thousands of units), so settling on a good solution and moving on to other parts rather than spending significantly more time looking for a perfect one makes complete sense. Chasing perfection is a .... ahem, perfect way of burning out without ever getting anything done.
 

thelaughingman

SFF Guru
Jul 14, 2018
1,413
1,566
got the Aquanaut in hand today and assembled the pump block combo with the Bitspower DDC pump (integrated heatsink). took measurement from bottom of cold plate to the top of the pump and it's 55mm, without the pump heatsink it was 46mm. Basically for the T1 in 3 slot mode it needs a pump without the integrated heatsink, 2 slot mode is fine with any DDC pump. really impressive work!

@Nouvolo my heavy hands over-tighten the screws on the last step (attaching cold plate to the block) and there appears to be tiny cracks around the screw holes but not reaching to any of the water channels / pathways. Should I be worried? I'm still gonna do pressure testing with the EK leak tester kit later on anyway but just want to know now if I kinda fuck up already ?
 
  • Like
Reactions: DrHudacris

duynguyenle

Airflow Optimizer
Aug 20, 2019
328
330
Hm, I was sure there were more skived open-loop parts. Corsair definitely uses skived fins for their XC9 at least, but I might have been looking at too many AIO cold plates recently. It makes sense that not many open loop component makers have skiving machines as they are massive, expensive, and essentially unitaskers, but they also allow for much cheaper and faster production of parts of arguably higher quality (not to mention the difficulty of CNC milling thin channels into soft copper). Of course I would imagine the MOQ from any cooling OEM with a skiving machine likely being in the thousands (if not tens of thousands), so for anything on this type of scale that's likely an impossible goal.

As for flow restriction, what you're saying is obviously true, but it doesn't take away from the fact that thinner microfins allow for either more fins or wider channels in the same area, with the reduction in thermal transfer from thinner metal being essentially negligible at scales this small (especially as the channels tend to be quite shallow). Skiving also more easily allows for taller fins than milling, as there's no chance of tear-out. As such, depending on the exact configuration of your design it is always possible for a skived block to match the flow of a milled block while having more surface area, or for it to have better flow with the same surface area. The difference is likely not large, but it's there. But as I said, I would imagine this comes down to economies of scale and the rather extreme high cost threshold of getting one of these machines in the first place.


That was pretty much what I was expecting, but it's nice to get it confirmed :) I can't even imagine how difficult the process of ordering parts like this is (and I would imagine a lot of OEMs won't even talk to you unless you're ordering thousands of units), so settling on a good solution and moving on to other parts rather than spending significantly more time looking for a perfect one makes complete sense. Chasing perfection is a .... ahem, perfect way of burning out without ever getting anything done.

I agree, I think it mostly comes down to MOQ since waterblocks are relatively niche items and move miniscule unit sold volumes compared to AIOs. I also agree (for the most part) that thin fins typically is slightly better (though there are so many more variables affecting thermal performance, that this might very well be a wash). I somewhat disagree on the point of taller fins being better. Typically, the deeper you go, the worse it gets due to boundary layer effects, so it's not always better to have deeper channel, since you have progressively slower flow rate gradient from the top of fin to the base of the fin, and to keep a more uniform flow gradient, you may need a more restrictive jetplate.

It's been a while since university days when I was dealing at fluid dynamics (and especially since the working fluid I typically deal with back then was air, not liquid, but the principles should still stand, especially since we're talking incompressible flow at relatively low velocities here).

In any case, you don't really need skivved fins to achieve REALLY thin fin profiles, just look at the rather impressive 0.102mm thin fins on the Optimus V2 block:





Fin thickness and stack height isn't all there is to it either, as how close the base of the fin stack physically is to the bottom surface of the CPU block also maters (i.e. how much material is between the bottom of the fin to the contact surface with the CPU), obviously thinner is better here, and I am not sure if micro-skiving has a definitive advance in this area compared to conventional milling (CPU blocks are expensive and I don't want to buy a whole bunch just to saw them in half and measure them!)
 
  • Like
Reactions: Valantar

Nouvolo

Creator
Original poster
Silver Supporter
Sep 8, 2018
759
1,711
www.nouvolo.com
From what I have read so far there are many opinions on a few things about custom cooling, namely flow direction and cold plates (fins), ranging from myths to enthusiasts pushing for the best. The main points I got are:
  1. flow direction affects thermals within margin of error (<1°C)
  2. cold plate fins affects thermals to within 5°C vs a plain flat plate
  3. once a certain level of flow rate is achieved, >0.7gpm, flow rate is not a major factor
  4. assuming other things are all properly setup, radiator/fan size is the major factor in overall thermal performance (rate of taking heat away from the system)
I am in the camp of go for bigger rad, the easiest way to get better thermals.
 

samyope

Chassis Packer
Aug 16, 2019
16
15
@Nouvolo my heavy hands over-tighten the screws on the last step (attaching cold plate to the block) and there appears to be tiny cracks around the screw holes but not reaching to any of the water channels / pathways. Should I be worried? I'm still gonna do pressure testing with the EK leak tester kit later on anyway but just want to know now if I kinda fuck up already ?
Not sure if POM/Acetal is more expansive than acrylic, but I am always afraid of cracking and take the Acetal option when available. Hopefully, if there is enough demand, an Acetal version can be made in the future ?
 
  • Like
Reactions: sos and DrHudacris

Qzrx

Buried under radiators
Dec 29, 2019
90
219
While waiting for the site to allow for orders, I came up with a question (suggestion?) for you, @Nouvolo:

I can't see if this has been said explicitly, but the cold plate looks to be entirely CNCed, including the microfins. Have you looked into getting the cold plate skived instead? I'm curious as to the reasoning behind going with a CNCed solution - is it because of accessibility, price, or performance? A lot (if not most) great water cooling cold plates seem to have skived microfins. Would that be a possible change for a future revision? At least on paper, skiving allows for thinner and denser fins, i.e. more surface area without significantly increased flow restriction.

Skivved fins are not inherently better than CNC’d fins; scroll about 2/3 of the way down:

 

Valantar

Shrink Ray Wielder
Jan 20, 2018
2,201
2,225
Skivved fins are not inherently better than CNC’d fins; scroll about 2/3 of the way down:

It goes without saying that if you're able to CNC fins as thin as skived fins you'll obviously get all the same benefits - but that generally isn't the case, not by a long shot (though that clearly has gotten better inrecent years with other manufacturers too). I don't buy their marketing spiel of their thin fins being immune to deformation or collapse that other similarly thin fins are subject to though - going by their text, if the deformation is caused by pressure, there's no reason their fins would be significantly stronger than other fins of a similar thickness. What they're talking about there sounds more like a consequence of higher QC standards (that's mass production vs. premium niche production for you) and a higher willingness to reject sub-par parts. So while you're right that there's no inherent superiority, there is an (admittedly minor) advantage in practice due to the realities of manufacturing.
 

DrLeroy

Noob Saibot
May 15, 2020
186
117
Sorry if it has been mentioned but I cannot seem to find it. . Has anyone seen if this fits the Asus Strix X570-i? looking at the photos on the website, it looks like the Plexi part of the block will interfere with the VRM Sink on the Strix X570-i board?
 

Nouvolo

Creator
Original poster
Silver Supporter
Sep 8, 2018
759
1,711
www.nouvolo.com
@DrLeroy
Just finished some quick tests and the results are looking good! Just keep in mind that they are in no way super scientific and should be treated as anecdotal:

The Loop was stressed using Cinebench R20 Looping for 1 hr to reach steady-state conditions.

System:

CPU3950x (Locked at 4.0ghz 132W)
PumpEKWB DDC 3.2 (PWM 100%)
RadiatorXSPC tx240 (25mm Slim rad)
Fans2x NF-A12x15
MotherboardAsus Strix X570-I
Noise Floor36 Db
Room Temp23.5 C
CaseNcase M1 v 6.1 Stock Panels

Results

Apogee Drive iiAquanaut
Noise39 db40 db
Pump Body Temp41.1 C42.5 C
CCD1 Temp61.25 C61.75 C
CCD2 Temp60.25 C61.25 C
Tdie62.38 C63.25 C
Cinebench Score8891 CB8904 CB

From my perspective, the temps noise and temps look well within the margin of error and could be down to mount variance or some unbled air in the loop. As for user experience, its mostly good. The instructions were easy to follow and the machining on the block looks very good.

One suggestion is to look at a different mounting system. It is by no means difficult, but the washers sit just above the last threads on the stud, so if they aren't perfectly aligned, the can slip off easily when screwing in the thumbnuts. It would also be nice to have thumb nuts with a Philips head on the top. ITX systems are tight, and even on an open bench, it can be tough to push down and tighten the thumbscrews at the same time to engage the threads. Finally, when screwing the studs into the backplate, it will leave some space between the motherboard and the backplate. This isn't a problem because the thumbscrews and springs create mounting pressure, but to a new user, it may seem like something is wrong.

Here are some quick and dirty pics of the setup: Pic 1 Pic 2
 

thelaughingman

SFF Guru
Jul 14, 2018
1,413
1,566
Sorry if it has been mentioned but I cannot seem to find it. . Has anyone seen if this fits the Asus Strix X570-i? looking at the photos on the website, it looks like the Plexi part of the block will interfere with the VRM Sink on the Strix X570-i board?
this is what you are looking for
Here are some pictures showing motherboard clearance with the Strix X570-I. There's about 5mm between the heatsink and the water block

 

fatmandandan

Efficiency Noob
Aug 29, 2020
7
24
Could you please list all the fittings that were used. Most of all I am interested in what Koolance QDC fittings are used here, for what are the diameter of soft tubing.

Going off of the 4th picture and working my way from the top right corner to the bottom left, the fittings are:

Koolance QD3 10mmx16mm, Koolance 10x16 Compression, Koolance 20mm Extension, EKWB 90, Koolance 4 way fitting body, Koolance QD3 BSPP 1/4, Koolance male to male swiveling coupler, Koolance Low profile 90 degree fitting

Koolance Low profile 90 degree fitting, Koolance male to male swiveling coupled, Koolance 4 way fitting body, Koolance QD3 BSPP 1/4, EKWB 90, Coolance 90 degree 10x16 Compression, Koolance QD3 10mmx16mm

The Tubing is equivalent to EK's Zmt, just ordered from mcmaster in bulk. I'm loving the QD3's so far, They have a judicious click, and there are only a couple drops of liquid on the connecter when disconnected.
 

Fitchew

SFF Lingo Aficionado
Feb 2, 2019
111
63
Going off of the 4th picture and working my way from the top right corner to the bottom left, the fittings are:

Koolance QD3 10mmx16mm, Koolance 10x16 Compression, Koolance 20mm Extension, EKWB 90, Koolance 4 way fitting body, Koolance QD3 BSPP 1/4, Koolance male to male swiveling coupler, Koolance Low profile 90 degree fitting

Koolance Low profile 90 degree fitting, Koolance male to male swiveling coupled, Koolance 4 way fitting body, Koolance QD3 BSPP 1/4, EKWB 90, Coolance 90 degree 10x16 Compression, Koolance QD3 10mmx16mm

The Tubing is equivalent to EK's Zmt, just ordered from mcmaster in bulk. I'm loving the QD3's so far, They have a judicious click, and there are only a couple drops of liquid on the connecter when disconnected.
Many thanks! It helped a lot. Since I just ordered fittings from Koolance yesterday, your photos were just right. And I had to understand if I did the right thing, that I bought QD3 10mm * 16mm (3/8in * 5/8in) fittings. The only thing I'm not sure now is that Koolance 90 degree Rotary 10x16 сompression fittings will fit directly onto the Aquanaut, as I have not ordered low profile fittings.

But I do not understand the purpose of Koolance QD3 BSPP 1/4in in your loop. Could you tell or show how the loop looks in the end?
 

fatmandandan

Efficiency Noob
Aug 29, 2020
7
24
Many thanks! It helped a lot. Since I just ordered fittings from Koolance yesterday, your photos were just right. And I had to understand if I did the right thing, that I bought QD3 10mm * 16mm (3/8in * 5/8in) fittings. The only thing I'm not sure now is that Koolance 90 degree Rotary 10x16 сompression fittings will fit directly onto the Aquanaut, as I have not ordered low profile fittings.

But I do not understand the purpose of Koolance QD3 BSPP 1/4in in your loop. Could you tell or show how the loop looks in the end?
No Problem! Glad to help. From my fitting, the 90 degree Rotary 10x16 Should fit directly on Aquanaut. Standard 10x16 do not, The compression nut is slightly too large and will hit the DDC. You should also be fine height wise, I don't think the normal 90 10x16 is too much taller than the low profile 90s. There are 4 QDCs in the picture, 2 are used for connecting the pump to the standard loop, and the other two are just there to make it easier to fill. I use an external Res and additional pump to help flush out all the bubbles.
 
  • Like
Reactions: Fitchew

Valantar

Shrink Ray Wielder
Jan 20, 2018
2,201
2,225
Hmm, I'm getting my head all turned around from reading this tread back and forth. Which DDC pump would you guys recommend I get? There doesn't seem to be any clear consensus on one that is the "best", but some at least seem noisier than others. I want something that a) beats my current EK-SPC PWM pump in terms of flow rate, while also running quiet. It'll be used for a dual rad CPU+GPU loop. I also ideally want PWM control, but I can live without it if there's a good cheap option. Any recommendations?