Concept Project Kappa: One Concept, Two Cases

theGryphon

Airflow Optimizer
Original poster
Jun 15, 2015
299
237
Hey people,
I've been around for some time now and if you recall me by name, you might remember me with some SFF case design ideas here and at [H] forum. I've had some very mainstream ideas as well as quite unconventional ones (at the time, that is). One of those unconventional ideas that didn't gather much attention then now finds itself being implemented by a very talented and successful designer these days... What I'm trying to get to is that SFF is evolving rapidly and an unconventional idea now will likely be in a product sooner than you think. Something to consider for everyone; those with the ideas, and those who are at the criticizing end...

This brings me to the realization that although I've had some great case ideas (most not shared), I've never attempted to see them through. Not due to laziness, mind you. I'm hard pressed with work, family and very limited financial resources, and I know that nothing less than full commitment can make an initiative successful, and I know that failure had huge stakes on my part. That is why I never could get myself to step into it.

That said, I observe that as the SFF became more mainstream in recent years, the manufacturing interests and capabilities also grew immensely. There are exponentially more small-scale names with awesome end products.

So, now I have this idea on not one but two cases (same concept, same design language, two sizes) that I've started working on, and I'd like to try and make it happen. However, I have some roadblocks on which I need input:
  1. To begin with, I still don't have substantial financial resources. So, I need some input on the cost of bringing a new case or two to life, and getting at least a batch production going.

  2. Closely related, I do not have access to a reliable and good quality manufacturing capability in the country where I live. After living most of my life in the US, I am now in Turkey. Awesome country, great people, sensitive politics and economy, and unfortunately pretty much zero know-how on PC case manufacturing. So, I'd like to get the manufacturing done elsewhere, and I need input on where it could be, or who (possibly among the household SFFN names) could/would do it.

Now on to the concept: While it's not completely outlandish, AFAIK it was never done before.
  • Form-factor is mini ITX slim tower with minimal foot print as possible. As little visible perforations as possible, and no windows.
  • The concept relies on liquid cooling for CPU and also preferably for GPU.
  • Two sizes: 10lt and 7lt

More on 10lt version:
  • Cooling: Up to TWO 240mm radiators, both with thin 120mm fans on them. Max fan+rad thickness support is 46mm (max recommended is 43mm). Max radiator length is ~280mm. Both AIO or custom liquid cooling setups are to be supported.
  • Cooling: Waterblock height limit is ~33mm when both CPU and GPU are liquid cooled.
  • Power: SFX or FlexATX or internal bricks (up to two HDPlex-size units) are supported.
  • Cooling/Power/GPU: FlexATX PSU (or a single internal brick) is supported with long GPUs (up to ~300mm PCB length) that are dual-slot (founder's edition type, stock fan left intact).
  • Cooling/Power/GPU: SFX PSU (or two internal bricks) is supported with long GPUs (up to ~300mm PCB length) only when the GPU is reduced to single-slot width (stripping off all stock cooling).
  • Cooling/Power/GPU: Custom loops with internal pump-res units (max 70x70mm cross section) are supported with FlexATX (or a single internal brick) and short GPUs (up to ~190mm PCB length).
  • Cooling/Drives: Two 80x15mm fans (OR two 7mm 2.5" drives) are supported at the bottom along with two other 7mm 2.5" drives (up to four 7mm 2.5" drives).
  • Cooling/Power/GPU/Drives: Two 80x15mm fans at the bottom AND four 7mm 2.5" drives (two on the bottom, two vertical) are supported with FlexATX (or a single internal brick) along with GPUs with PCB length up to ~280mm.
  • Cooling/GPU/Drives: If the GPU is not liquid cooled, up to two 3.5" drives are supported. In this case, the supported rad+fan thickness for the CPU increases to 63mm (supporting push-pull with thin fans).
  • Material/construction: 2mm aluminum construction with 1.5mm aluminum cover and brackets. Three main parts, disregarding the small brackets: two structural parts and a U-shaped cover. Main body is single-piece, which ensures rigidity while saving space. Second structural component is the radiator bracket.
    • The U-shaped cover is currently intended to screw on from both sides using a total of 8 screws. I will look into using another mechanism (push-pins?) and eliminate all screws.
  • Dimensions: 322 (height) x 311 (front-to-back depth) x 100 (side-to-side width) = 10 lt
  • Pictured below is the Sketchup draft that sufficiently depicts the main points. I am currently working on a detailed model using Solidworks.





More on 7lt version:
  • Cooling: Up to TWO 120mm radiators both with thin 120mm fans on them. Max fan+rad thickness support is 46mm. Max radiator length is ~175mm (virtually all are supported). Custom liquid setups are not supported internally, but a pump+res combo can be installed on the outside (although I haven't fully worked that usecase out).
  • Cooling: Waterblock height limit is ~33mm when both CPU and GPU are liquid cooled.
  • Power: FlexATX or a single internal brick is supported at the bottom.
  • GPU: Up to ~190mm PCB length dual-slot GPUs are supported.
  • Cooling/Drives: Two 7mm SSDs are supported (vertically, on the front) along with radiators up to ~160mm length (virtually all).
  • GPU/Drives: Two more 7mm SSDs are supported on the front along with GPUs up to ~175mm PCB length.
  • Cooling/GPU/Drives: If the GPU is not liquid cooled, a single 3.5" drive is supported. In this case, the supported rad+fan thickness for the CPU increases to 63mm (supporting push-pull with thin fans).
  • Material/construction: 2mm aluminum construction with 1.5mm aluminum cover and brackets. Three main parts, disregarding the small brackets: two structural parts and a U-shaped cover. Main body is single-piece, which ensures rigidity while saving space. Second structural component is the radiator bracket.
    • The U-shaped cover is currently intended to screw on from both sides using a total of 8 screws. I will look into using another mechanism (push-pins?) and eliminate all screws.
  • Dimensions: 348 (height) x 201 (front-to-back depth) x 100 (side-to-side width) = 7 lt


I believe three facts with this concept might cause friction, so I'd like to comment on those:
  1. Reliance on liquid cooling. This is why this concept would never see the light of day couple of years ago. However, AIO liquid cooling became increasingly common, and pretty much mainstream these days. Plus, I don't think anyone can argue against the cooling chops of two 240mm radiators compared to anything else you can fit in 10lt (or two 120mm radiators in 7lt).

  2. Inclination towards FlexATX (in case of the 7lt version, it is rather "reliance on FlexATX"): FlexATX now is what SFX was 3 years ago. You see the trend? Better FlexATX units are bound to come as demand for "smaller" keeps rising, and as PSU technology keeps improving, and as power efficiency of PC components keep improving. More and more FlexATX units are being introduced every year at CompuTeX etc... So, believe me when I say, FlexATX is the tomorrow of mainstream SFF, and it will come sooner than you think.

    Note: SFX PSU is supported just fine by the 10lt version (read details above). I currently intend to support the truly SFF power solutions with internal bricks, but if the usecases will be too narrow (especially in the 10lt version), I might not pursue that.

  3. Using two radiators "like that": No discussion about it, using two radiators independently from each other would give better thermal results. However, there are best and worst application scenarios on this matter. Namely, 1) it is easily verified that the GPU will dump far greater heat to the radiator than a CPU, so arranging the CPU side as the intake and GPU side as the exhaust will give far better thermal results than the other arrangement, and by experience (mine and many others') I can confirm that the delta compared to "independent radiators" is very little, 2) Using two sets of fans instead of a single set of fans (sandwiched) will give significantly better thermal results. Hence, the arrangement in the picture above...

    FURTHERMORE, this case allows for both radiators to exhaust, creating a negative pressure in the case, in which case there is no heat transfer between the radiators. As soon as I get my hands on a prototype, I will test for multiple scenarios and I suspect that the negative pressure setting will give the best thermal results.


Like I said above, I am looking forward to getting any and all input and guidance, as well as other thoughts on successfully seeing this project through, and on the case design as usual. Please keep it coming through comments on the thread or through PM...

Many thanks in advance!
 

Gautam

Cable-Tie Ninja
Sep 5, 2016
148
123
I like any concept that takes cooling seriously, as there is a dearth of it in the SFF community. My system also has a flex ATX unit, and I like the flexibility that they offer.

My issues are going to be unusual, but your designs are, too, so, shrug. I'm commenting on the 7L for now.

I dislike slim fans altogether because of their noise pitch. I would rather go with a 38mm rad (or similar) with 25mm fans, rather than 2 with slims, even if they performed worse. That's an easy change, though.

FlexATX is annoyingly loud. The 40mm fan standard is their biggest impediment to being more widely adopted. The only way I get by with mine is by using a 92mmx25mm fan to cool it, making it quieter than every SFX on the market. But supposedly some of them are quiet with loads much lighter than their ratings. Gutting a flex and running it (semi) passively might be possible in some scenarios. I do think that they will grow in popularity, and not enough people are experimenting with them, leading us to get nothing but DAN clone after DAN clone... There are vastly superior designs possible (like yours could be) which aren't being explored unfortunately.

Also, the strongest air coolers are on par with entry-level single rad AIO's. I've been testing a Rajintek Pallas vs a Corsair H60 and the former is pretty much identical, whilst being quieter and occupying (a lot) less volume. The towers can get even better still. Not completely relevant to what you're working on, but something to think about. You're exploiting water cooling's true advantage, though, which is that you can place the radiator wherever you want.

I also think your dimensions are overly optimistic. You'll probably have to increase each of them to make it viable in the real world. I think you've touched upon a lot of interesting ideas though. Allocating the amount of volume you are towards cooling will allow for much more power and quietness than most of the stuff out today if configured properly.
 

theGryphon

Airflow Optimizer
Original poster
Jun 15, 2015
299
237
^ Thank you for your input!

I am not sure under which setting you are comparing the Pallas to the H60 but H60 should perform significantly better. If you check the reviews, a single fan 120mm AIO cooler performs pretty much equivalent to a single fan 120mm tower cooler, which should normally perform better than a top-down cooler similar to the Pallas. I do own a Pallas and similarly sized Cryorig C1, and I can tell you that they do not perform at their best in a cramped SFF case (as opposed to open-air). The radiators in the presented cases here will get unrestricted fresh air from the outside of the case and dump it out directly without circulating the hot air inside the case.

I hope everyone can see that the cooling design in this concept is as good as it gets: the heat is carried away from the hottest components as quickly as possible and exhausted outside of the case. No case in this class is able to do so AFAIK, and I am very confident in the cooling performance of these cases. Tests will be done and presented as soon as I get a prototype, but physics is soundly there.

I intentionally avoided going into component specifics in the first post but we have to go there inevitably:

About thin fans, it's established that Noctua NF-A12x15 represents the state-of-the-art in slim 120mm fans,and it is definitely the recommended fan here. It's not any noisier than other 120mm fans at same RPM, and performs admirably. Most other thin 120mm fans will not give desired cooling and noise performance.

FlexATX. As I said above, while FlexATX is not at its prime yet, better and better units are showing up. Consider the new (not yet available) Enhance ENP-7660B: A 600W 80+ Platinum unit that should be very reasonable with noise up to 500W:



In a year or so, other brands should follow suit is my expectation.

Finally, I'm not sure what you mean by "... your dimensions are overly optimistic." If you mean that the larger the case, the better the cooling? If so, that should be obvious to everyone. If you mean the components that I claim to be supported won't fit, than you're mistaken.

Surely, this is Sparta SFF, and component selection is extremely crucial. For instance, it should also be quite obvious to those who would be interested in these cases that AIOs with very rigid tubing will not work. Having worked with AIOs in very tight spaces before, I know that those with flexible tubing will work fine.

Again, once I get the prototype, I will test a range of components and validate the design (which is exactly what a prototype is for ;) ) and I will have a list of recommended components.
 

Gautam

Cable-Tie Ninja
Sep 5, 2016
148
123
I am not sure under which setting you are comparing the Pallas to the H60 but H60 should perform significantly better. If you check the reviews, a single fan 120mm AIO cooler performs pretty much equivalent to a single fan 120mm tower cooler, which should normally perform better than a top-down cooler similar to the Pallas.
I own both too, and just returned the H60 because I didn't like the pump noise.

The reviews are all over the place, which is why I had to buy all 3 and see for myself. I could write at length about this, but the bottom line is, constrained to the same fan at the same fan speed, they're all extremely close. There's a margin of about 70MHz separating the H60 from the C1 on my 6700k, which some would care a lot about, I guess, but doesn't concern me much. The pump noise and nuisance of having to route tubes are palpable.

Finally, I'm not sure what you mean by "... your dimensions are overly optimistic." If you mean that the larger the case, the better the cooling? If so, that should be obvious to everyone. If you mean the components that I claim to be supported won't fit, than you're mistaken.
Guess its merely my perception, but it looks like routing cables and tubing will be a pain, because you have everything packed together about as tightly as possible. E.g. the DAN A4 could be close to a liter smaller in theory, but it would be a bit impractical.

I hope everyone can see that the cooling design in this concept is as good as it gets: the heat is carried away from the hottest components as quickly as possible and exhausted outside of the case. No case in this class is able to do so AFAIK, and I am very confident in the cooling performance of these cases. Tests will be done and presented as soon as I get a prototype, but physics is soundly there.
I do. I came off more critical than I would have liked to. It really is in the right spirit IMHO. Targeting sensible form factors and cooling is preferable to me over maximum hardware compatibility, esp when the hardware doesn't jive. (A 700W PSU doesn't make much sense alongside an L9i which can't handle a 100W CPU, but its a combo you see all the time...)
 

theGryphon

Airflow Optimizer
Original poster
Jun 15, 2015
299
237
No worries my friend, you are entitled to your position and I did not take it in any negative way. I just feel the obligation to defend this concept against early and not-yet-justified judgment.

I certainly understand your concern on dimensional tolerances and believe you me, it is going to be a tight fit in both these cases. The thing is, that comes with the territory (SFF yo! ;)).

Just because you mentioned it: I don't see how Dan A4 could be any smaller without giving concessions from supported CPU cooler height, GPU max thickness, GPU max length, SFX-L support, or SSD support. It's all about trade-offs in SFF. Dan could make it a bit smaller but he decided not to, due to the aforementioned "features" he decided to keep. He could make it way larger too, obviously, supporting way larger components. You draw the line somewhere, and this is where he decided to draw it, and that's what gives his case the character it has.

In my case, this is where I'd like to draw it. Meaning, I will naturally not be able to support some hardware (e.g. involving tubing stiffness, waterblock height, etc.) or combinations of some hardware that I could if I had made the case larger. My argument is that (which is pretty much the gist of the argument for any SFF case designer) the supported hardware is going to give sufficient number of combinatorial options that make the case valuable in a set of usecases that the I would like to target.

Specifically, the 10lt version targets those who values cooling, performance and compactness to the highest degree at the same time (e.g. a mildly overclocked 7980XE coupled with an overclocked GTX 1080Ti in the smallest volume and footprint as possible); 7lt version targets those who puts compactness above the others (e.g. a mildly overclocked 2700X and a mildly overclocked GTX 1080 Mini in the smallest volume and footprint as possible).

I share the same sentiment as you regarding power supplies. They are behind the curve, and they're pulling SFF down in the sub-8lt territory. That is why I support FlexATX as that's where we can have the highest W/lt for sub-500W range, which is the relevant envelope for sub-8lt SFF territory. For true enthusiasts (like myself), there are other options with internal or external bricks but that is still too niche. For mainstream, FlexATX is the natural evolution from SFX, which was the natural evolution from ATX. So, yes, I feel the same cringe when I see those super-tight (I mean it as a compliment), super-handsome cases with sub-300W feasible power envelope but still having to use SFX because there is unfortunately no better option yet (because FlexATX has not caught up with the mainstream yet).

So, in that sense, I feel like the cases presented here strike a very good balance in terms of cooling vs power supply support. I am confident that the 10lt version is a 600W-enclosure cooling and power supply considered, whereas the 7lt version is a 400W-enclosure.
 

theGryphon

Airflow Optimizer
Original poster
Jun 15, 2015
299
237
Would a 120mm really be enough for a Ryzen 2700X? How hot are we talking?

2700X is rated 105W and its stock cooler seems to perform just fine at stock speeds. A 120mm radiator should match the stock cooler at much lower dBA, and give a bit of overclocking headroom at the expense of dBA. 2700X seems to pretty much overclock itself (with its Precision Boost 2 and XFR 2) and serious overclocking seems to still require heavy cooling hardware.

After going through the reviews, it seems like 2700X is holding a steady full 8-core clocks of 3.9MHz (at stock, using stock Wraith cooler) under even the heaviest of loads. Overclocking seems to hit a wall at around 4.2-4.3GHz. My feeling is that the a 120mm AIO should enable OC up to 4.0-4.1GHz depending on the chip, and depending on the noise threshold of the user. A 240mm AIO should help find its ceiling, so I'd recommend the 10lt version for such usecase.

I feel like (and it seems to be the consensus) with 2700X most users would not benefit much from OC, unless one needs to run full 8-core loads all the time, or unless one can OC it to at least 4.3GHz. This is because the chip hits single-core clocks of 4.3GHz and more without any overclock (thanks to the improved Precision Boost and XFR). In any case, it's a monster of a CPU and both cases here would support it at stock and at varying OC levels.
 
  • Like
Reactions: Phuncz

Ricky

SFF Lingo Aficionado
Jan 23, 2016
92
36
2700X is rated 105W and its stock cooler seems to perform just fine at stock speeds. A 120mm radiator should match the stock cooler at much lower dBA, and give a bit of overclocking headroom at the expense of dBA. 2700X seems to pretty much overclock itself (with its Precision Boost 2 and XFR 2) and serious overclocking seems to still require heavy cooling hardware.

After going through the reviews, it seems like 2700X is holding a steady full 8-core clocks of 3.9MHz (at stock, using stock Wraith cooler) under even the heaviest of loads. Overclocking seems to hit a wall at around 4.2-4.3GHz. My feeling is that the a 120mm AIO should enable OC up to 4.0-4.1GHz depending on the chip, and depending on the noise threshold of the user. A 240mm AIO should help find its ceiling, so I'd recommend the 10lt version for such usecase.

I feel like (and it seems to be the consensus) with 2700X most users would not benefit much from OC, unless one needs to run full 8-core loads all the time, or unless one can OC it to at least 4.3GHz. This is because the chip hits single-core clocks of 4.3GHz and more without any overclock (thanks to the improved Precision Boost and XFR). In any case, it's a monster of a CPU and both cases here would support it at stock and at varying OC levels.

Ok. Man, my dream case for a build, and it’ll take months. Love the tower look over things like Dan A4, or GHOST.
 

theGryphon

Airflow Optimizer
Original poster
Jun 15, 2015
299
237
Too bad GPU’s don’t get 240 radiators though

True that there is no GPU with a 240mm AIO right out of the factory (at the moment), but you can always DIY ;)

When the GPU uses a 120mm radiator, the remaining space (the 10lt version) can be used to install additional drives, or a 120mm fan to maximize cooling.