Cooling Petition for short tube sff specific 120mm AIO

[Thehack]

Hello, I'd like to tag @AseDen for this thread.

So when designing chassis, some of it is a bit of a chicken and egg problem. We had a few chassis built around specific coolers in mind: namely the Big Shuriken 2 and the NH-L12. The Dancase A4 had also been modified to fit the Asetek 92mm in its latest revision.

With that said, I'd like us to see a 120mm product in the market.

A 120 AIO is about equal to NH-U12S. However, it is much more compact and can actually be mounted above the motherboard.

I myself have a few design ideas for a 120mm AIO above the motherboard but the issue in actual usage, it is very difficult to install current AIO on the market. It more difficult than it is worth recommending.

We already see a a few cases that can support a short tube 120 (with maybe some specific fitting suggestions).

- RVZ01 and RVZ03 family
- NCase M1
- Ghost S1 ( not actually above it, but does benefit from shorter tube)

My suggested specifications:
- standard asetek block height
- 27mm radiator
- short flexible tubing
- elbow on radiator, but on the sides instead, similar to new 92mm cooler
- shorter length.
- fill port to refill radiator

Standard LP ram is about equivalent to 40mm cpu cooler clearance. Add 27mm radiator + 15mm fan you get 82mm clearance.

I hope others can chime in.

Here is a quick drawing:

So you need about minimum 85mm tubing. Accounting for clearance and arcing, it is like 100-110mm. The swivel elbow would make it easy to reduce the height by curving it. If it is about 110mm, it should be able to be used inside the Ghost S1, set to push. I can probably build a set up to test this if Asetek is really interested in this spec.

A regular AIO is about 300mm in length, like so users can install it the front of a tower.

 

[AseDen]

I'll be following along here, to hear what people has to say.
Also, I might chime in with questions every now and then

 

[robbee]

You'd loose one of the main advantages of watercooling though: being able to dissipate heat in a different place than where it's being generated. Copper heatpipes already transfer heat very well but can't be shaped as easily as plastic tubing.

This is an interesting idea though!

 

[VegetableStu]

so something like the masterliquid maker 92mm AIO from CM? o_o
it'll have to beat all C-shaped 120mm air coolers out there. otherwise it's pretty pointless ._.

(note noctua's 92mm air coolers right above the maker 92 ._.)

 

[Thehack]

You'd loose one of the main advantages of watercooling though: being able to dissipate heat in a different place than where it's being generated. Copper heatpipes already transfer heat very well but can't be shaped as easily as plastic tubing.

This is an interesting idea though!

For that reason, the many long tube versions already serve that purpose on the market.

 

[Thehack]

so something like the masterliquid maker 92mm AIO from CM? o_o
it'll have to beat all C-shaped 120mm air coolers out there. otherwise it's pretty pointless ._.

(note noctua's 92mm air coolers right above the maker 92 ._.)

The 120mm size I'm referring to is the actual radiator size.

It's be the equivalent of a corsair h60 but with modifications for sff. The main modification is sff optimized elbow and tubing length.

It'd be install like in the NCase M1 (parallel to the motherboard).

 

[VegetableStu]

yeah THAT 120mm AIO in the configuration you mentioned has to beat all C-shaped 120mm air coolers out there. otherwise it's pretty pointless ._.

 

[Thehack]

yeah THAT 120mm AIO in the configuration you mentioned has to beat all C-shaped 120mm air coolers out there. otherwise it's pretty pointless ._.

Ah okay. It will.

Here is guru3d review.

Give or take a few. If you use 25mm fan, the H60 should beat the U9S definitely at less than 100mm height.

Generally a 120mm AIO is about equal to a single 120mm tower. As far as I know, no C type, at least under 100mm comes close to a single 120 tower.

 

[VegetableStu]

mmmmnnnn.... might need some proof of concept first ._. even with thin rads I kinda have my reservations
(thinking in an A4-SFX for the time being)

 

[Thehack]

mmmmnnnn.... might need some proof of concept first ._. even with thin rads I kinda have my reservations
(thinking in an A4-SFX for the time being)

The A4 is definitely better design for 92mm. I installed a 120mm and I can't image it would be easy regardless of the tubing length.

 

[VegetableStu]

so for something like an Ncase M1... I guess it just needs a board-standing frame?

EDIT: oh OH i get what you were wanting ( /)_(\ ignore my posts. sorry

 

[Thehack]

so for something like an Ncase M1... I guess it just needs a board-standing frame?

EDIT: oh OH i get what you were wanting ( /)_(\ ignore my posts. sorry

All good. I updated original post to help further illustrate.

 

[AseDen]

All good. I updated original post to help further illustrate.

So... I think having the fittings mount like that might pose some issues (I'd have to talk to some of the engineers). Reason being, I don't think I've ever seen a radiator made that way... (correct me if I'm wrong).

@VegetableStu - as for the 92mm ML Maker. We've made something like that looooong ago. I think it was used for workstations, and based on 120mm. We still have one around, I can upload a picture if you'd like to see it

 

[Sashby]

@AseDen Here are some examples of AIOs with 90 degree swivel fittings on the radiator:


The Cooler Master Glacer 240L v2 used a Swiftech MCR-H220 radiator with 3/8" Swivel Barb Fittings.
Review of CM Glacer 240L by tweaktown

Spoiler: Picture of CM Glacer 240L

Corsair One Gen 1 and Gen 2 (i140/160/180)
Review/teardown of Corsair One i160 by techpowerup

Gen 2 AIO is more inline with what @Thehack is asking for but in a 120mm fan configuration

Spoiler: Picture of Corsair One i160 (Gen 2) CPU side

 

[AseDen]

My immediate concern with something similar to this, would probably be air getting trapped in all the wrong places.
Considering the nature of SFF builds and how different they are, I could see potential for untimely failures, as it is very hard to predict use cases.
And as you all probably know, air in pump = bad!

That being said, I think moving the connection could potentially be interesting. Although not quick job to do, as we'd want to make sure that longevity etc. was kept.

Oh also, just re-read the post. 90 degree bends are not an issue. We do it on the 645LT. It's their orientation within the hex that worries me.
Actually sometimes, in specific cases, we even do 135 degree bends (which allow the tubes to run flush with the cap).

 

[Sashby]

Exactly. I think it's possible but definitely not an easy task as you said. Trying to finalise a product that suits upcoming case designs or "pizza box" cases, ensure people don't install them incorrectly and end up with a faulty product well before it's intended lifetime.

The Swiftech MCR-H220 radiator had a built-in reservoir

Spoiler: (Bottom left of drawing

to combat air permeating over time from the tubing and ending up in the pump. Correct orientation of the radiator and CPU block/pump below the elevation of the reservoir was documented in the manual.

Spoiler: Similar to what EKWB did with the Predator and Phoenix

(those too can only be mounted a certain way to prevent air entering into the pump)

 

[AseDen]

Exactly. I think it's possible but definitely not an easy task as you said. Trying to finalise a product that suits upcoming case designs or "pizza box" cases, ensure people don't install them incorrectly and end up with a faulty product well before it's intended lifetime.

The Swiftech MCR-H220 radiator had a built-in reservoir

Spoiler: (Bottom left of drawing

to combat air permeating over time from the tubing and ending up in the pump. Correct orientation of the radiator and CPU block/pump below the elevation of the reservoir was documented in the manual.

Spoiler: Similar to what EKWB did with the Predator and Phoenix

(those too can only be mounted a certain way to prevent air entering into the pump)

Issue with the Swifttech solution is, what happens when the user turns the unit upside down... air right to the pump
An AIO will always have a little air left. Which is what can become an issue

 

[Sashby]

Issue with the Swifttech solution is, what happens when the user turns the unit upside down... air right to the pump
An AIO will always have a little air left. Which is what can become an issue

The manual stated it's correct and only acceptable orientations (otherwise warranty voided )

 

[Thehack]

Bubbles entering the pump, usually through mishandling, shipping, or a bad install is something that affects AIO correct? The issue is whether elbows allow to bubbles collect and may cause these type of malfunctions more than a regular AIO? If I may interject, we should also weigh in the benefits of the design that may outweigh the increased failures.

Also, isn't the 645lt installed with the radiator lower than the pump?

 

[AseDen]

The manual stated it's correct and only acceptable orientations (otherwise warranty voided )

I realize, but that we do strive to make user friendly designs

Bubbles entering the pump, usually through mishandling, shipping, or a bad install is something that affects AIO correct? The issue is whether elbows allow to bubbles collect and may cause these type of malfunctions more than a regular AIO? If I may interject, we should also weigh in the benefits of the design that may outweigh the increased failures.

Also, isn't the 645lt installed with the radiator lower than the pump?

You are absolutely right. Statistically, it's likely to fail earlier - which of course is something we want to avoid. The longevity of our products is a key feature However, I do also follow your argument, that there should always be reasoning behind pro's and con's

Yes, the 645LT is indeed installed in a less than ideal way. And we do hear people in the SFF community talking about failures of the pumps on the old 545LC. My understanding, so far at least, is that the SFF community quite highly specialized and very aware of the fact that they are making compromises - among others longevity.
That being said, I think we've got a pretty solid solution, even with the slight issues that are posed in the A4.