Hutzy HS: Top-End GPU Gaming Case under 5L (formerly HASSIUM)

Hahutzy

Limp Gawd
Joined
Sep 4, 2015
Messages
169
Formerly HASSIUM - Performance Densely Packed

---

Hey guys, some news on Leap Day!


Update #1: New Design for the next Prototype


Gathering the feedback from the initial responses, I noticed 3 main things that stuck out as issues:
1) Limited vent holes on the side, no vent hole on top
2) Power button on the bottom
3) Structural weakness

I decided to spend some time to do a redesign with these things taken into consideration, and I am happy to reveal the result of your input! (Sketchup pics for now, fancy Solidworks renders to follow later)

Few quick notes about the new design:
- It will be 2-tone, with some parts in clear/light gray aluminum, and some parts in black/dark gray aluminum
- Flanges added to structurally weak spots to greatly increase strength
- Power button in the front
- Opened up more vents on each side, and added vents on top

15X4PuD.jpg


r27zlCj.jpg


XTqtHyJ.jpg


---

Update #2: Acrylic middleboard replaced with custom 3D printed middle board

What used to hold the mobo in place was a piece of acrylic and 3 angle brackets. Which made for a very clunky usability, and non-scalable production piece.

I have since designed and 3D printed a plastic middleboard to replace it.

More good news: because it's 3D printed, it doesn't need to be flat.
I can embed nut slots into the board, making all nuts and bolts fasten much tighter onto the board.
I can even create a sunken rail for the PCIE extension cable to fit tightly into!

I have it in my prototype right now, and I have been running my machine to its max heat generation to test the plastic's resilience to the heat generated.

TgZlF3v.jpg


TUgzR2N.jpg


---

Update #3: A new challenger appears: The sub-4L case

This is an idea that came about while I was revisiting the design of Hassium, and I worked on it alongside the redesign. The new Hassium design actually borrowed a lot of elements from when I worked on the XS.

The XS is a minimized version of my initial Hassium. And instead of fitting the top-end GPUs, it focuses on meeting VR glasses specifications in as tight of a size as possible.

In the next few days, I will create a new thread detailing what exactly the XS is, what the specs are, etc.

Here's a preview of a super early prototype of XS (on the left):

hFVpKN3.jpg
 
Last edited:
Update History:

2016/02/29 said:
Hey guys, some news on Leap Day!


Update #1: New Design for the next Prototype


Gathering the feedback from the initial responses, I noticed 3 main things that stuck out as issues:
1) Limited vent holes on the side, no vent hole on top
2) Power button on the bottom
3) Structural weakness

I decided to spend some time to do a redesign with these things taken into consideration, and I am happy to reveal the result of your input! (Sketchup pics for now, fancy Solidworks renders to follow later)

Few quick notes about the new design:
- It will be 2-tone, with some parts in clear/light gray aluminum, and some parts in black/dark gray aluminum
- Flanges added to structurally weak spots to greatly increase strength
- Power button in the front
- Opened up more vents on each side, and added vents on top

15X4PuD.jpg


r27zlCj.jpg


XTqtHyJ.jpg


---

Update #2: Acrylic middleboard replaced with custom 3D printed middle board

What used to hold the mobo in place was a piece of acrylic and 3 angle brackets. Which made for a very clunky usability, and non-scalable production piece.

I have since designed and 3D printed a plastic middleboard to replace it.

More good news: because it's 3D printed, it doesn't need to be flat.
I can embed nut slots into the board, making all nuts and bolts fasten much tighter onto the board.
I can even create a sunken rail for the PCIE extension cable to fit tightly into!

I have it in my prototype right now, and I have been running my machine to its max heat generation to test the plastic's resilience to the heat generated.

TgZlF3v.jpg


TUgzR2N.jpg


---

Update #3: A new challenger appears: The sub-4L case

This is an idea that came about while I was revisiting the design of Hassium, and I worked on it alongside the redesign. The new Hassium design actually borrowed a lot of elements from when I worked on the XS.

The XS is a minimized version of my initial Hassium. And instead of fitting the top-end GPUs, it focuses on meeting VR glasses specifications in as tight of a size as possible.

In the next few days, I will create a new thread detailing what exactly the XS is, what the specs are, etc.

Here's a preview of a super early prototype of XS (on the left):

hFVpKN3.jpg

2016/02/12 said:
- Have been using my prototype for about a month now as my daily driver. Not a single issue so far. Although I do agree that going forward, I need more vent holes on the side and the top.
- Spent a good part of January/February building a 3D Printer. Using it to test out different structures and layouts.
- Also considering a 4L case for ITX-class cards (R9 Nano).

Started out small with a PCI extender shroud (with orange filament):

6zfBAvU.jpg


Here's a case core fresh off the printbed:

oCehf5m.jpg


After a little bit of love from a hobby knife:

cdJ1x3v.jpg


(I snapped one of the flaps and taped it green. Whoops.)

2016/01/17 said:
Switched CPU cooler from Rosewill LP (with poor TIM application) to Noctua L9i

3eyyCn5.png

2016/01/14 said:
Hi all. I want to present to you a project I've been working on in the past months. A sub-5L case that houses a GTX 980 Ti.

Here's the prototype I currently have (exterior design update coming in prototype v2):
DB4hxMp.gif


And here is the planned design for prototype v2:
6eUS8LV.png


Motivation
Early last year, I built a sff pc with the Jonsbo v3+. It's small, but it was very limiting in the length of GPUs that fit. I started getting into content creation (streaming, video editing etc) and the short 750 Ti left a lot to be desired. I wanted to upgrade to a 980 Ti.

Around this time, I found OCnet, and read about the 500w FlexATX PSU. That thing is tiny compared to a standard ATX power supply. From finding OCnet, I also started looking at different build logs people had going on; I did not know custom cases and scratch builds were a thing. Threads on OCnet quickly led me to [ H ], where I found the NCase M1, and realized that not only do indie pc cases exist, but there's also a strong support network behind peoples' efforts.

And so I set out to put my own spin on an optimal pc case. Few months later, today, I have a functional prototype, with a much clearer vision of where I want to drive this.

Design Criteria
From day 1, I had set myself a list of design criteria to follow:
  • ultra compact (<4L, later changed to <5L)
  • rectangular shape (so that it can slip into a backpack)
  • fits a 10.5" gpu (EVGA 980 Ti)
  • minimalistic look (absolutely nothing on the front face)
  • no external power brick (because that's cheating the size constraint)
  • affordable (because I want it to be scalable in production)

Specification
  • material: 1.6mm Aluminum
  • dimensions: 273mm x 96mm x 190.5mm [LWH], 4.99L
  • cpu: up to 65W cpu tested
  • hsf: designed for Noctua L9i or any shorter cooler (like the Rosewill LP I'm using)
  • mobo: mITX
  • gpu: dual slot, 10.5" (267mm), up to EVGA 980 Ti ACX2.0+ SSC tested
  • psu: FlexATX [note: do not use Athena Power!]
  • hdd: parallel, vertical 2.5" hdd mount. up to 3x7mm drives (mechanical drives not recommended)
  • fan: up to One 80mm*10mm exhaust fan on the bottom is planned

Demo Unit Spec / Performance
  • i5-4440S, EVGA ACX 980 Ti
  • avg 3Dmark Firestrike: 11700 (the i5 is dragging the 980 Ti down drastically)
  • max cpu/gpu temps with no case fan (after 5 consecutive Firestrike runs): 59/84 deg C
  • stability: over 40+ runs of 3Dmark Firestrike/Unigine Heaven, no crashes

Intention Moving Forward
Yes, eventually I do want to produce these in a scalable manner. And I believe that at that time, it would be best to include the power supply as the part of the product. Another direction I've thought about is to release them as custom prebuilts.

I learned quite a few things with the first prototype. The biggest takeaway: prototyping a single unit is not cheap. So in the near future, I want to do a small batch run of the prototype v2, either powered by crowdfunding, or convincing some of my friends to let me build them a PC ;)

But before that can happen, I need to make sure every component that goes into the case is scalable. And right now, there are still a few things that aren't. However, I do firmly believe that it is the right time to reveal Hassium to fellow enthusiasts and garner some feedback before I set in stone more design decisions.

For news about Hassium, you can follow this thread in your favorite forum. I will try my best to update them all in sync.

Talk soon,
Hahutzy
 
Last edited:
is that the power button on the bottom? seems inconvenient to lift the case up to turn it on/off
 
is that the power button on the bottom? seems inconvenient to lift the case up to turn it on/off

The power button is indeed located on the bottom. That is a compromise made to keep the front and top panels' minimalistic look.

I completely understand the drawback of this design choice, in that the user would have to lift the machine up slightly to reach the power button. After talking it through with several people, the general consensus we've reached is that it is a manageable compromise.

For people who rarely turn off their pc, and for people that travel and are used to handling it on a regular basis, this is essentially not a concern.

With that in mind, I still want to reduce the unwieldiness of the button's location by pushing as close to the front of the bottom as possible, so the angle of tilt needed to reach it is minimized.


Where does the hot air go? Down out the bottom?

Yes, the hot air is designed to vent out the bottom.
The GPU and CPU's coolers are both intaking air from the side, creating positive air pressure in the case. Because of this air pressure, wherever there's vent slots to escape, air will push on through.

The amount and efficiency is currently untested, as I am aware the flow from pressure gradient is in direct contention with the flow from convection flow (hot air rises).

I am anticipating that by adding a 80mm*10mm fan at the bottom, the flow of air will undoubtedly be dominated by the fan's direction, hence improving the bottom exhaust scheme.
 
This is essentially the same layout as Dondan's case, as I'm sure you're aware, the only substantial difference being the PSU. I've modeled this exact layout myself - using a flex/1U PSU - years ago as well. I have some reservations with your particular implementation:
  • The severe lack of ventilation. What does the GPU clock speed drop to during testing? GPU temp doesn't tell us much, since it's designed to stay below a target temperature (83c) by dynamically adjusting its performance. Also, you have a bottom exhaust but virtually zero clearance between the case and the ground for air to actually flow. Are there feet that aren't shown, or are you planning on adding some? The bottom exhaust is practically non-functional without clearance.

  • The flex riser you're using is one of those IDE ribbon style cables, which don't have much in the way of shielding, plus it's on the long side which increases the probability of signal loss. I haven't personally tested this type of cable, but I've tested shorter, more robustly shielded cables and one bad one I had showed no problems in 3dmark, but did show graphical glitches in some games, so you want to make sure you test them thoroughly. This is why Dondan is erring on the side of caution and going with an expensive 3m cable in his case.
 
This is essentially the same layout as Dondan's case, as I'm sure you're aware, the only substantial difference being the PSU. I've modeled this exact layout myself - using a flex/1U PSU - years ago as well.

I am aware that Dondan also uses an extension cable to put the GPU on the back side of the motherboard to save a lot of horizontal clearance. My design focus was to get everything to fit into the sub-5L constraint while still housing and running the desired components.

The severe lack of ventilation. What does the GPU clock speed drop to during testing? GPU temp doesn't tell us much, since it's designed to stay below a target temperature (83c) by dynamically adjusting its performance. Also, you have a bottom exhaust but virtually zero clearance between the case and the ground for air to actually flow. Are there feet that aren't shown, or are you planning on adding some? The bottom exhaust is practically non-functional without clearance.

Good point. I will start recording clock speeds during tests as well as other data. There are dome feet attached to the bottom flanges on both side panels, creating clearance. I am currently still testing to see if a 80mm*10mm exhausting through that clearance is enough.

The flex riser you're using is one of those IDE ribbon style cables, which don't have much in the way of shielding, plus it's on the long side which increases the probability of signal loss. I haven't personally tested this type of cable, but I've tested shorter, more robustly shielded cables and one bad one I had showed no problems in 3dmark, but did show graphical glitches in some games, so you want to make sure you test them thoroughly. This is why Dondan is erring on the side of caution and going with an expensive 3m cable in his case.

You're right, and I have thought about that and am working on combating it.
 
Apart from what Necere already said:

If you don't want to put the power button on the front you should place it at the back of the case in easy reachable spot like in top corner of the back. Putting it on the bottom of the case that has literally no clearance is not a good idea.

I don't like how it's designed mechanically - the back of the case has a ton of weak spots prone to bending and breaking and you haven't implemented or shown us the middle wall keeping whole thing together and supporting the motherboard.

FlexATX psu is not as quiet as sfx-l and it's more server grade product.

I also don't like how it looks - that's of course subjective but what I don't like here is that it doesn't have any style - it's just a metal box.
 
I'm interested to see your GPU clock speeds and noise levels. I don't imagine either one is very good.
 
That is the original A4 Dondan had made initially. With a flex psu instead of a external brick.

images
 
Thanks for all the feedback. The consensus so far seems to be:

1) The button on the bottom is more of a concern than I initially anticipated. It's going to be difficult, but I will look at moving it to the back.

2) Thermals, which I agree with and am actively trying to improve. I am waiting for my 80mm*10mm case fan and Noctua L9i to come in to check the system's thermals before I draw any more conclusions. For the sake of aesthetics, I really do not want to open up vents on the top face. However I will consider it if the bottom vents and exhaust fan are not enough.

3) The minimalistic style is disliked. We're going to have to agree to disagree on this one. The goal of the design is for the case to blend into the space without grabbing too much attention.
 
Thermals as you said, and structural strength (as mentioned by SaperPL) need addressing.

I would run tests with and without the covers on like dondan did with the A4.

The look doesn't bother me personally, and the power button can be moved fairly easily if I must have the case as-is.

The power supply choice is more likely to stop me.
 
Last edited:
You could add mounting holes for a HDplex 250W power board. People would then be able to choose between using flexATX and an external supply brick.
 
That is the original A4 Dondan had made initially. With a flex psu instead of a external brick.

images

False. The PSU you can see there is the HD-Plex 250W DCDC PSU, not a FlexATX one. The HD-Plex is running of an external brick. FlexATX was never even considerd for the A4 IIRC.

You could add mounting holes for a HDplex 250W power board. People would then be able to choose between using flexATX and an external supply brick.

I am contemplating this for my design as well, as there seems to be a lot of distrust in FlexATX, and feel like this is a good compromise indeed. The issue here is that a 980Ti can draw a lot of power, so even with the strongest 500W FlexATX PSU out right now, the noise will be significantly higher than when you're limited to an R9 Nano.
From what I can tell, the PSU doesn't have it's own intake either, which concerns me a bit.
 
Love the case. A few modifications as mentioned above plus support for the HDPlex 250w and i'm sold.
 
Exhausting a 980ti down? With no ventilation? This is a pure textbook example of asthetics over function - to the point where function is broken. Even if it does work the fans will be loud to keep it cool.

Exhaust out back will help -- and won't break aesthetics.

Having a lot of experience with HP slimline cases and their inherit limitations. I'm just not interested in a case like this any more. Too many battles to fight.
http://hardforum.com/showthread.php?t=1224011
 
Exhausting a 980ti down? With no ventilation? This is a pure textbook example of asthetics over function - to the point where function is broken. Even if it does work the fans will be loud to keep it cool.

Exhaust out back will help -- and won't break aesthetics.

Having a lot of experience with HP slimline cases and their inherit limitations. I'm just not interested in a case like this any more. Too many battles to fight.
http://hardforum.com/showthread.php?t=1224011

A reference design was used for the prototype, which exhausts out the back just fine with no issues whatsoever.

I don't see how exhausting downwards is worse than exhausting upwards if a fan is used. As you can see in the OP, the design accommodates a 80x10mm fan at the bottom, which is more than enough to vent all the heat out. This fan would have to be included with the case, no doubt about that, but other than that, I don't see why this wouldn't work. Natural convection is way too weak to have any influence on the airflow when a fan is used.

Right now the only problem is that there is not probably not enough clearance at the bottom for the fan to work efficiently, but how much clearance is required can easily be tested with the prototypes, that's what those are made for.
 
This PSU can't be slapped into any case and just work. You have to screw it to a passive heatsink in order for it to dissipate its heat. HASSIUM has no such passive heatsink, and there's no space to put it either. Good thought, but it's not really useful in this scenario.

I was more thinking of supplying your own active cooling, given the claim that it'd be rated 450W with such.
 
I had some time this weekend to take apart the whole build and swapped out the Rosewill LP cpu cooler to a shiny new Noctua L9i.

I also noted that the Rosewill's thermal paste application was really poorly done (woops), and basically the layer looked so thin and spotty that I believe some surface was exchanging heat either by metal on metal, or had air between them instead of thermal paste.

I wiped every thing down, applied a new layer of thermal paste that came with the Noctua. And I installed the Noctua.

Tried the tests again, and the CPU temperature dropped by ~20 deg C, both idle and load.

3eyyCn5.png



Some answers

RE: structural strength
It did concern me as well. However after assembly, the machine is rock solid.

RE: HDplex 250
I will consider this some time later. Current priority is to test thermals and to see if changes to the design is needed.

RE: zeroflex
As far as I know, that PSU is actually not built to Flex ATX spec at all. Its size is different, and it does not have standard FlexATX mounting holes.

RE: exhausting down kills function
The whole "top is sealed" design is not set in stone. Believe me when I say function comes first in my design. But I want to make sure to exhaust all other options before I decide to open up the top. But if it is needed, I will not hesitate.

RE: non-reference 980 ti
The reason all my tests are done with a non-reference 980 Ti is because I am testing my case with the worse-case scenario
If I were to test with something like a 970 or a reference blower 980 Ti, there's no doubt it gives the design more leniency.
But there's no point, because I need to make sure it can handle whatever people will throw into it.




Next up for testing:
I need to get all my testing done with the Noctua and the 80mm fan installed.
I have a tablesaw at home, so If the thermals are still not good, I will cut off a chunk from the side panels' top flanges to simulate a design where venting holes exist at the top.
 
Last edited:
I love the simplistic approach here. It would work well on a shelf and not stand out. My only real issue is the power button on the bottom. I would prefer a button on the rear, but that's already been beaten to death in this thread :)
 
I was more thinking of supplying your own active cooling, given the claim that it'd be rated 450W with such.

Which would completely defeat the point of the exercise as the PSU is already larger than FlexATX with the heat exchange mounting plate and will be even larger if you attached an active cooling system to it. At that point, you can just use the FSP500-50FSPT (which will be supplied with the case), which supplies 500W and already has its own active cooling.

I had some time this weekend to take apart the whole build and swapped out the Rosewill LP cpu cooler to a shiny new Noctua L9i.

I also noted that the Rosewill's thermal paste application was really poorly done (woops), and basically the layer looked so thin and spotty that I believe some surface was exchanging heat either by metal on metal, or had air between them instead of thermal paste.

I wiped every thing down, applied a new layer of thermal paste that came with the Noctua. And I installed the Noctua.

Tried the tests again, and the CPU temperature dropped by ~20 deg C, both idle and load.

http://i.imgur.com/3eyyCn5.png


RE: zeroflex
As far as I know, that PSU is actually not built to Flex ATX spec at all. Its size is different, and it does not have standard FlexATX mounting holes.

Next up for testing:
I need to get all my testing done with the Noctua and the 80mm fan installed.
I have a tablesaw at home, so If the thermals are still not good, I will cut off a chunk from the side panels' top flanges to simulate a design where venting holes exist at the top.

Those temps look good, but what processor did you use?

Now that you say it, it looks like you're correct, the zeroflex doesn't even have standard mounting holes. The size of the case is to spec, just the extension bit that you screw the case to makes it larger.

Just as a reminder, make sure to test the elevation of the case as well. It might be that raising it by a few mms drastically increases the cooling performance or decreases the noise of the bottom fan.
 
Noise would be a concern with the FSP500's 40mm fan. Consider the ZF240 is meant to work in a completely passively cooled environment, you might not even need a dedicated fan for the ZF240. The size difference is not dramatic and doesn't matter if they both fit a given space. You also might not need the heat exchange plate if not cooling it passively.

People in the FSP500 thread talked about modding it to use a bigger fan. The ZF240 shouldn't require any mods. I wouldn't say it's completely pointless.
 
Those temps look good, but what processor did you use?
In his OP he mentions he tested with a 65W TDP CPU and also that it is a i5-4440S.
Not the highest spec out there, but definitely not a CPU to sneeze at.

Although considering that this is a "slower" CPU I don't think this CPU hits the 65W TDP it is rated for. My old i5-4670 only gets to the 65W mark, as a 84W TDP CPU and that is a faster Non-S CPU.

65W TDP doesn't mean the CPU needs and takes 65W, it means the OEM or user has to account for a 65W TDP.
 
Noise would be a concern with the FSP500's 40mm fan. Consider the ZF240 is meant to work in a completely passively cooled environment, you might not even need a dedicated fan for the ZF240. The size difference is not dramatic and doesn't matter if they both fit a given space. You also might not need the heat exchange plate if not cooling it passively.

People in the FSP500 thread talked about modding it to use a bigger fan. The ZF240 shouldn't require any mods. I wouldn't say it's completely pointless.

The heat exchange plate is an integral part of that PSU, it is thermally bound to all components that give of heat. You can't cool it without that plate, at least not efficiently.

That PSU is normally attached to a passive heatsink about the size of a third of this case. I don't think there is a way to passively or actively cool it at such a small volume, but if someone can prove me wrong, I'll stand corrected.

Yes, the FSP500 could get rather loud under full load when you put in the best components possible, so a 980 Ti + Xeon 16 Core CPU, but that's the price you pay for this small volume. Also, there hasn't been any professional testing done on this PSU, so anything we're saying right now is more speculation than anything.

In his OP he mentions he tested with a 65W TDP CPU and also that it is a i5-4440S.

Whoops, I couldn't find that section somehow. Thanks!
 
Updated the front page with some pics of the 3D printed prototypes I made. Long story short, trying to get the dimensions and the structure/layout of the case just right. Also considering a 4L case for ITX-class cards like R9 Nano.
 
Updated the front page with some pics of the 3D printed prototypes I made. Long story short, trying to get the dimensions and the structure/layout of the case just right. Also considering a 4L case for ITX-class cards like R9 Nano.

Any updates on this?
 
Yup. I just finished upgrading the 3D printer this past weekend, so I'm printing out some remaining parts.

I haven't gotten my hands on an R9 Nano yet, but I'll probably snap a few pics of whatever progress I've made and post an update with images this weekend.

EDIT:

Got a lot done. More than I originally planned. I'll do the write up / upload the images tomorrow.
Stay tuned!
 
Last edited:
Hey guys, some news on Leap Day!


Update #1: New Design for the next Prototype


Gathering the feedback from the initial responses, I noticed 3 main things that stuck out as issues:
1) Limited vent holes on the side, no vent hole on top
2) Power button on the bottom
3) Structural weakness

I decided to spend some time to do a redesign with these things taken into consideration, and I am happy to reveal the result of your input! (Sketchup pics for now, fancy Solidworks renders to follow later)

Few quick notes about the new design:
- It will be 2-tone, with some parts in clear/light gray aluminum, and some parts in black/dark gray aluminum
- Flanges added to structurally weak spots to greatly increase strength
- Power button in the front
- Opened up more vents on each side, and added vents on top

15X4PuD.jpg


r27zlCj.jpg


XTqtHyJ.jpg


---

Update #2: Acrylic middleboard replaced with custom 3D printed middle board

What used to hold the mobo in place was a piece of acrylic and 3 angle brackets. Which made for a very clunky usability, and non-scalable production piece.

I have since designed and 3D printed a plastic middleboard to replace it.

More good news: because it's 3D printed, it doesn't need to be flat.
I can embed nut slots into the board, making all nuts and bolts fasten much tighter onto the board.
I can even create a sunken rail for the PCIE extension cable to fit tightly into!

I have it in my prototype right now, and I have been running my machine to its max heat generation to test the plastic's resilience to the heat generated.

TgZlF3v.jpg


TUgzR2N.jpg


---

Update #3: A new challenger appears: The sub-4L case

This is an idea that came about while I was revisiting the design of Hassium, and I worked on it alongside the redesign. The new Hassium design actually borrowed a lot of elements from when I worked on the XS.

The XS is a minimized version of my initial Hassium. And instead of fitting the top-end GPUs, it focuses on meeting VR glasses specifications in as tight of a size as possible.

In the next few days, I will create a new thread detailing what exactly the XS is, what the specs are, etc.

Here's a preview of a super early prototype of XS (on the left):

hFVpKN3.jpg
 
Last edited:
Back
Top