Concept Nouvolo Aquachill Res❄️- Small size pump/reservoir combo (maybe with active cooling)

[Nouvolo]

Aquachill Res - Small size reservoir with (optional) active cooling (TEC modules) (external mini-compressor unit)

Current status : Concept Stage / Community feedback welcome!
Dev Log :
[2021-9-30] : research feasibility of using TEC modules, on the technical issues, such as power draw, efficiency, extra heat, etc
[2021-10-04] : research feasibility of using mini compressor, i.e. an external cooling unit

*Project may not proceed if there is no suitable solution for active cooling. Creator has no interest in pursuing project as a plain reservoir product

Aquachill Res configuration (community suggestions welcome)

Compressor Cooling

Spoiler: TEC cooling - dev halted

Features

• small size reservoir - comparable to a 120mm fan (with mounting brackets), may not fit the smallest SFF cases (Steck, FormD T1) but target SFF cases of Meshlicious class
• reservoir with 2 separate, back to back, compartments (1 cold side, 1 hot side)
• no extra pump required
• active cooling liquid with TEC modules (non sub-zero, but down to about ~10-20°C), using PSU 12V supply
• copper plates with high density fins to maximize heat/cold transfer
• separate controller to manage TEC modules temperature (Windows compatible)
• lowering liquid temperature to give more room for overclocking, or just to be "cool"
• water loop continues to operate passively when active cooling modules are turned off

Active Cooling Operating Modes

• Off - no active cooling (no power consumed), operates as reservoir, zero effect on the water loop
• Ambient - cold plate maintain at ambient temperature
• Dew point - cold plate maintain above dew point to prevent condensation
• Manual - full manual mode (with power limit)

Further investigation required

• appropriate TEC modules for heat transfer and acceptable efficiency
• power electronics to drive the TEC modules
• temperature sensors / controllers

 

[Nouvolo]

[Dev Halted] - Aquachill Res (TEC) flow process in steps:

1. radiator feeds to reservoir cold side inlet (passively cooled)
2. [Temp. step down] liquid temp steps down further (actively cooled) and feeds out to the system
3. [Cooler loop temp.] loops runs through the whole system, and warm liquid then feeds to reservoir hot side inlet
4. [Temp. step up] liquid temp steps up further (waste heat generated by active cooler), and feed to radiator
5. heated dissipated by radiator and fans
6. back to step 1 and repeat

 

[JDuval]

It's interesting!

Is there an estimate of how much cooling could be expected?

Right now I can't envision a use-case as I'd rather my components just barely fit in my case lol - just can't see anywhere to put it in any build I have in mind.

But if it's knocking 5C+ off the water temps then maybe I would have to reevaluate....

 

[Goatee]

Aquachill - Small size reservoir with (optional) active cooling (using TEC modules)

Current status : Concept Stage / Community feedback welcome!

Appreciate your participation in research and input about the feasibility of using TEC modules, especially on the technical issues, such as power draw, efficiency, extra heat, etc.

How active cooling works here (steps) [diagram later]

1. radiator feeds to reservoir cold side inlet (passively cooled)
2. liquid temp steps down further (actively cooled) and feeds out to the system
3. loops runs through the whole system, and warm liquid then feeds to reservoir hot side inlet
4. liquid temp steps up further (waste heat generated by active cooler), and feed to radiator
5. heated dissipated by radiator and fans
6. back to step 1 and repeat

Features

• small size reservoir - comparable to a 120mm fan (with mounting brackets)
• reservoir with 2 separate, back to back, compartments (1 cold side, 1 hot side)
• no extra pump required
• active cooling liquid with TEC modules (non sub-zero), using PSU 12V supply
• copper plates with high density fins to maximize heat/cold transfer
• separate controller to manage TEC modules temperature (Windows compatible)
• lowering liquid temperature to give more room for overclocking, or just to be "cool"
• water loop continues to operate passively when active cooling modules are turned off

Active Cooling Operating Modes

• Off - no active cooling (no power consumed), operates as reservoir, zero effect on the water loop
• Ambient - cold plate maintain at ambient temperature
• Dew point - cold plate maintain above dew point to prevent condensation
• Manual - full manual mode (with power limit)

View attachment 1355 View attachment 1356
View attachment 1358 View attachment 1359

Further investigation required

• appropriate TEC modules for heat transfer and acceptable efficiency
• power electronics to drive the TEC modules
• temperature sensors / controllers

Risks

• Project may not proceed if there is no suitable solution for active cooling. Creator has no interest in pursuing project as a plain reservoir product

Can you build a space for a pump into it?

Making it a Pump / Res combo with something like a DDC mounting option could make it very appealing.

 

[Nouvolo]

It's interesting!

Is there an estimate of how much cooling could be expected?

Right now I can't envision a use-case as I'd rather my components just barely fit in my case lol - just can't see anywhere to put it in any build I have in mind.

But if it's knocking 5C+ off the water temps then maybe I would have to reevaluate....

Aiming to cool water down to around 10°C !

It may not fit the smallest SFF (i.e. Steck , FormD T1), but target cases similar to Meshlicious.

 

[threestripevida]

Can you build a space for a pump into it?

Making it a Pump / Res combo with something like a DDC mounting option could make it very appealing.

I would like to see this too. A more affordable alternative to the what EK has with their FLT? line up.

 

[Nouvolo]

I would like to see this too. A more affordable alternative to the what EK has with their FLT? line up.

It may not be that much cheaper than EK, but EK's Cryo design has a big issue :

• Cold plate in direct contact with CPU, means cold plate MUST not breakdown. The TEC module is a very poor heat conductor itself, and it will (100%) cause fatal problem if the module breaks down / turns off.
• When the CPU runs so hot that the TEC module cannot handle, TEC module efficiency is greatly hampered, and may even generates more total heat into the system

Main advantage of Aquachill Res design is that if it breaks down or powered off it just sits idle and does not affect the existing loop. It will behave exactly as a normal reservoir. 100% safe by design.
Disadvantage: it may take longer to lower the loop temp, as it is bringing down the liquid temperature, rather than being directly applied on to the CPU surface.

 

[Nouvolo]

I would like to see this too. A more affordable alternative to the what EK has with their FLT? line up.

 

[Nouvolo]

Can you build a space for a pump into it?

Making it a Pump / Res combo with something like a DDC mounting option could make it very appealing.

it is certainly possible, but I may do it as a separate product or as an option to the base model

 

[REVOCCASES]

Interesting idea. There have been several attempts using Peltier elements to improve cooling in PCs over the years but most of them couldn't really keep what they promised.

One of the major reasons is the pretty poor efficiency of TEC elements. To get about 10W of "cooling power" on the one side you'll typically have to feed the TEC with 50W which you'll have to get rid off on the hot side. So if you try to cool the whole thing in one single loop temps will actually be higher than without the Peltier element.

What would work better from my point of view: using a separate loop to cool down the hot side of the TEC and then a second loop for GPU and CPU connected to the cold side of the TEC. Not very SFF friendly but should work.

Maybe this explains better what I mean:


The big rad is used to get rid of the TECs heat which then drops the temp on the actual loop with the small rad by about 10C.

 

[Nouvolo]

Interesting idea. There have been several attempts using Peltier elements to improve cooling in PCs over the years but most of them couldn't really keep what they promised.

One of the major reasons is the pretty poor efficiency of TEC elements. To get about 10W of "cooling power" on the one side you'll typically have to feed the TEC with 50W which you'll have to get rid off on the hot side. So if you try to cool the whole thing in one single loop temps will actually be higher than without the Peltier element.

What would work better from my point of view: using a separate loop to cool down the hot side of the TEC and then a second loop for GPU and CPU connected to the cold side of the TEC. Not very SFF friendly but should work.

Maybe this explains better what I mean:


The big rad is used to get rid of the TECs heat which then drops the temp on the actual loop with the small rad by about 10C.

Thanks for sharing your findings. I have also dug further into the subject, seems most TECs run at their (relatively) more efficient at a lower current than Imax, usually around 30-50% of Imax, which corresponds to about 60-35% efficiency, and with a range of Δ20-30K, and with about 50% of Qmax of heat(W) removed from the cold side.

e.g. For single TEC module
TEC1-12715 with Vmax = 12V, Imax = 15A, Qmax = 137W, good operating range will be at around 5A-7A, removing about 65W, and lowering temp. by about 20-30°C. But this also adding back another 45-150W back into the system due to (in)efficiency, i.e. the radiator will have a total load of ~200W.

I see many TEC tests trying to push max current (e.g. 12V, unlimited current over TEC module load), which I think is not a good idea. I think the most efficient way is to run multiple TEC modules and let them all run at around 40-45% Imax, for the least amount waste heat generated (due to efficiency), and also giving a reasonable temperature drop and heat removal.

Taking above single TEC baseline:
no. of TEC module(s) used: 1x => 65W , 2x => 130W, 3x => 195W, 4x => 260W
BUT total heat to be removed, or required by PSU : 1x = 200W, 2x => 400W, 3x => 600W, ~~~~

Let me know if there is any flaw in the above analysis. Seems it is theoretically not feasible..

 

[REVOCCASES]

Let me know if there is any flaw in the above analysis. Seems it is theoretically not feasible..

Seems about right. Unfortunately, no matter what you do, adding Peltier elements in the loop will add more heat than you'll be able to remove. It can be beneficial if you have enough space for an additional big rad and PSU - then you could use it similar like EK is doing it with their QuantumX... but I guess this is more interesting for large ATX builds with a couple of 360mm rads

how about removing the TEC feature and making custom designed SFF friendly reservoirs or pump/res combos instead?

 

[GuilleAcoustic]

how about removing the TEC feature and making custom designed SFF friendly reservoirs or pump/res combos instead?

I second that. We need more compact reservoirs like the old and venerable Swiftech MCRES-Micro :

 

[Nouvolo]

Alphacool has an external cooling solution

Alphacool Eiszeit 2000 Chiller​

Alphacool Eiszeit 2000 Chiller - black

The Alphacool “Eiszeit” is a compressor cooler with a maximum cooling power of 1500W. Unlike most so-called Chillers or circulating coolers, the Eiszeit has a high-performance integrated pump. Highlights Integrated pump with...

www.aquatuning.co.uk

Wonder if anybody will be interested in something smaller, e.g. size of another SFF case, with pump (D5), reservoir, and compressor cooling. Size like something below:

 

[Gilles3000]

This concept simply wouldn't work in a single closed loop, the speed at which coolant cycles trough the loop is much faster than you'd imagine, and the radiator doesn't cool as much in a single cycle as you'd think.

Essentially there's no such thing as a cold and a hot part in a single loop, especially a fairly small one as you'd find in in a SFF PC. So all this concept would end up doing is adding heat to the loop.

And compressors are indeed awesome, but they're expensive and loud, not really suitable for most people.

I used the hang out in the extreme cooling thread on OC.net a lot, and while TEC's work, they're always much harder to implement than people think at first, there are only a handful of TEC builds i've seen that truly work well, and it wasn't for a lack of attempts.

 

[Nouvolo]

I second that. We need more compact reservoirs like the old and venerable Swiftech MCRES-Micro :

But I thought everybody is using those (3 way) flow meters as their mini reservoirs (removing the spinner)? That thing is selling everywhere and is dirt cheap (about USD5.0 shipping). If so, I don't think I can compete with that, or it's a worthwhile project business-wise... Maybe I am missing something here, let me know what you think. Maybe pump+mini res? Like the iceman's one?

 

[threestripevida]

I would vote for a pump/mini res. Something more reliable than the alpha cool dc-lt pumps. Not all custom loops will use it since they might opt for the aquanaut, but having the pump/res might be more enticing for some. I think I would go with that option over the aquanaut for smaller loops like in the T1.

 

[Nouvolo]

Some existing DDC pump x mini-res examples in the market, already pretty good?. Let me know what other shapes are you looking for, or share your idea of your ideal pump/res combo

 

[thelaughingman]

Some existing DDC pump x mini-res examples in the market, already pretty good?. Let me know what other shapes are you looking for, or share your idea of your ideal pump/res combo

View attachment 1372 View attachment 1370 View attachment 1374 View attachment 1375

Haven't seen the XSPC one before, most likely discontinued now?

My preference is the rectangular-shape like the XPSC one, so other example of such shape is the Iceman res for the Meshilicious and Alphacool DDC 1U res. The Alphacool is great with so many ports for flexibility, but as we all know it leaks. I feel the Iceman is almost as good as we can get - maybe find some way to slim it down and/or shrink the res part to reduce the volume further?

 

[REVOCCASES]

Haven't seen the XSPC one before, most likely discontinued now?

My preference is the rectangular-shape like the XPSC one, so other example of such shape is the Iceman res for the Meshilicious and Alphacool DDC 1U res. The Alphacool is great with so many ports for flexibility, but as we all know it leaks. I feel the Iceman is almost as good as we can get - maybe find some way to slim it down and/or shrink the res part to reduce the volume further?

To make it unique, maybe could base the design on the Iceman you showed above but make it modular so that you could use it either as a small res only or with an addon on the side as pump/res combo. Hope you get the idea