Sub 3Litre Silent Case for up to 65W TDP

[K888D]

For the past few months I have been working on a custom design Mini ITX case made from sheet Acrylic.

The objectives of this case project are:

• Sub 3Litre in volume
• 65W CPU TDP
• Virtually Silent cooling
• Multiple storage options
• Use a 92mm Fan low profile CPU cooler
• Low cost chassis

I am really hoping that AMD's upcoming ZEN processors will further improve upon the already impressive gaming capability of the A10 Kaveri APU's, a 65W chip capable of PS4 level graphics would be ideal for this type of case.

BACK ONTO THE CASE:
The design is based around preventing the recirculation of warm air by channeling 100% external cool air to the processor, warm air is then forced sideways across the motherboard and exhausted out of the case simultaneously cooling the system.

The challenge has been to design an intake that lines up exactly with the processor fan and can be adjusted to suit different processor positions on different motherboards. For this I have designed a reversible cover with a large opening that can be further aligned with an 'L' shaped bracket resulting in a range of intake positions. To help airflow and reduce turbulence noise the intake is covered only by a mesh type fan filter.

The first draft design was 3D printed to test the airflow principle and whether it was realistic to cool a 65W system properly without a case fan.

For testing the system an AMD A10-7800 (65W) APU with a Noctua NH-L9a was used as a worst case scenario. Typical usage tasks were carried out rather than stress tests, the results were:

• Under idle conditions the processor sat at around 35C with a system temperature of around 37C with the fan spinning at around 700rpm.
• Under 50% processor load (playing Grid 2 demo @ 1080p) the processor topped out at around 50C with a system temperature of around 45C with the fan spinning at around 1000rpm
• Under 90% processor load (streaming 4K youtube video) the processor topped out at around 53C with a system temperature at around 47C and the CPU fan spinning at roughly 1200rpm

An RPM of less than 1000rpm with the Noctua is pretty much silent and at 1200rpm it is barely audible, so for typical usage this setup works very well at staying virtually silent whilst maintaining reasonable temperatures.

After these initial results I decided to get the first prototype laser cut and hot wire bent from Acrylic sheet by a local company, the results can be seen below:


There were many flaws in the first design which was reliant upon the Acrylic having a degree of flex to allow the slot together system to work, but unfortunately I completely underestimated how brittle Acrylic is, this led to the clip features breaking off to get the cover on. But clips aside everything else fit together ok and I was impressed with the quality of the Acrylic sheets and the accuracy of the bends that the laser cutting company had achieved.

That's all I've got time to write about this evening, but I am currently up to revision 3 with many improvements to the design, hopefully I will post about these revisions tomorrow.

 

[iFreilicht]

Very interesting premise, I've thought about ways to duct all intake air from a CPU fan in the past, but couldn't come up with a solution myself.

Do I understand you correctly in that there are basically two top panels? So one with the very large filtered intake and a second one below that which can be mounted in various locations to allow perfect ducting for the fan? That's what it looks like in picture 3.
But in picture 1, I can see no indication of such an "inner" top panel. Am I missing something here?

Either way, the case looks pretty nice as it is right now, acrylic really has moments to shine.

Is there a reason why there's so much space to the side of the board?

 

[jeshikat]

Ok, something like this should be pretty reasonable to get made in sheet metal. Not as cheap as acrylic but it shouldn't be anywhere what it's cost to get Nova/Cerberus prototyped

 

[K888D]

Thank you for the comments, I will be posting some more pictures later today of the second and third prototypes.

Do I understand you correctly in that there are basically two top panels? So one with the very large filtered intake and a second one below that which can be mounted in various locations to allow perfect ducting for the fan?

Yes thats right, the cover has a large opening with the fan filter, there is then a second panel beneath it with a medium size opening, this panel can be flipped into 4 different positions. The opening in the lower panel can then be refined with a screw on 'L' shaped panel which can be positioned in 8 different orientations, this allows for quite a few combinations which should (in theory) support most cpu positions.

A 2.5" drive can be mounted to the lower panel in the gap that is created beneath the cover.

Is there a reason why there's so much space to the side of the board?

This space was for a 120W power board which was the original design direction, but I have since decided to change to a Pico PSU to save space, the case has been reduced in width in revision 3 to a foot print of 210mm x 190mm x 75mm, with sheet metal you could probably shave 5 - 10mm off each side.

Ok, something like this should be pretty reasonable to get made in sheet metal.

I will definitely look into how I can adapt the design for sheet metal, hopefully this will also help reduce the size of the case further as the current design uses 5mm Acrylic with slot features. I would like to try and keep some of the Acrylic for Aesthetics, maybe have some kind of hybrid metal/acrylic design.

 

[iFreilicht]

Yes thats right, the cover has a large opening with the fan filter, there is then a second panel beneath it with a medium size opening, this panel can be flipped into 4 different positions. The opening in the lower panel can then be refined with a screw on 'L' shaped panel which can be positioned in 8 different orientations, this allows for quite a few combinations which should (in theory) support most cpu positions.

A 2.5" drive can be mounted to the lower panel in the gap that is created beneath the cover.

Ok, very nice. Is there anything that prevents you from making the second panel mountable on slots, so it can be adjusted continuously?

Yeah that seems like a nice mounting spot for a 2.5" drive, maybe even two of them stacked depending on the gap between the two panels.

 

[K888D]

Is there anything that prevents you from making the second panel mountable on slots, so it can be adjusted continuously?

In picture 3 you can just about see the cardboard poking through the slots, is this what you mean? The lower panel sits in these slots, to adjust this panel you flex the chassis slightly to release the slots, the panel can then be flipped or reversed into the closest cpu matching position.

Yeah that seems like a nice mounting spot for a 2.5" drive, maybe even two of them stacked depending on the gap between the two panels.

There is a 9mm gap in the revison 1 design allowing for 1x 2.5" drive only, unfortunately there is not enough room to have 2 drives sat next to each other as they would partially block the intake vent.

But, in the latest revision I have had to completely flip the whole concept around due to unavoidable edge distortion next to the heat bends, you have to leave a 10mm gap from a heat bend to an cut out edge to avoid distortion, this was affecting the IO shield slot on the rear.

To solve the problem the motherboard has moved upwards and the hard drives are now sat beneath the motherboard allowing 2x 2.5" drives to be fitted. The adjustable panel system has been removed and instead replaced by an alternative adjustable system that screws directly into the top cover.

When I get home tonight I will post some more pictures of how the design has progressed in order to solve some of the limitations I have come across working with formed Acrylic.

 

[iFreilicht]

In picture 3 you can just about see the cardboard poking through the slots, is this what you mean? The lower panel sits in these slots, to adjust this panel you flex the chassis slightly to release the slots, the panel can then be flipped or reversed into the closest cpu matching position.

[...]

The adjustable panel system has been removed and instead replaced by an alternative adjustable system that screws directly into the top cover.

I was more talking about the lower panel having slots through which you can mount it to some mounting points inside the chassis, maybe using 45mm standoffs on the motherboard or something like that. That way, if your CPU socket is off by 1mm, you can adjust the panel to fit bettter, but it sounds like your next revision might have something like that.

There is a 9mm gap in the revison 1 design allowing for 1x 2.5" drive only, unfortunately there is not enough room to have 2 drives sat next to each other as they would partially block the intake vent.

But, in the latest revision I have had to completely flip the whole concept around due to unavoidable edge distortion next to the heat bends, you have to leave a 10mm gap from a heat bend to an cut out edge to avoid distortion, this was affecting the IO shield slot on the rear.

To solve the problem the motherboard has moved upwards and the hard drives are now sat beneath the motherboard allowing 2x 2.5" drives to be fitted.

If you were to go back to the way you did it in revision 1, make sure to make that gap 11mm to support 9.5mm tall 2.5" drives, that increases the maximum capacity from 1TB to 2TB.

That edge distortion thing would be something you could solve by using metal instead of acrylic How big is the gap beneath the motherboard now? How much did the volume increase by doing that?

When I get home tonight I will post some more pictures of how the design has progressed in order to solve some of the limitations I have come across working with formed Acrylic.

Great, I'm looking forward to it!

 

[K888D]

A lot of this has already been mentioned in the previous posts, but here are some photos of the second prototype.

The core design was kept similar to the first prototype with a few minor tweaks, the main focus of this prototype was to improve how the cover clipped onto the base.

Another area of focus was the rear IO cutout, having any kind of cut feature near a heated bend line results in distortion which was witnessed in the first prototype, the forming company advised the slot either needs to start before the bend line or at least 10mm after the bend line.

As I didn't want to move the motherboard upwards by 10mm at this stage I opted to move the slot to before the bend line, as you can see from the photos this exposes a small portion of the motherboard which isn't ideal.

Revision 3 (post to follow) I have now amended the design moving the motherboard upwards to solve this distortion issue.

 

[K888D]

Lifting the motherboard upwards led to a shift in the layout with the fan intake alignment system being incorporated into the cover.

This has meant that the case can remain the same height, there is also now room to fit 2x 2.5" drives below the motherboard.

Below are a few screengrabs of the CAD showing the intake alignment built into the cover. The cover itself is reversible allowing the main opening (140mm fan filter) to cover the entire motherboard space.

A smaller panel with a 92mm opening is then screwed into the underside of the cover, an arrangement of spaced screw holes allow the opening to be positioned anywhere within the main opening. The screen grabs show 2 example positions and an overlay of the screw hole arrangement.

The screw holes in the cover are positioned under the frame of the filter so they will not be visible from the outside of the case.


You may have also noticed that the case is now more of a square shape than before as the powerboard mounting area has been removed.

I will try and post some pictures of the prototype based on the above design tomorrow.

 

[Hahutzy]

Have you tested how much of a performance difference there really is on CPU/chipset temps with and without the channelling layer?

 

[K888D]

Have you tested how much of a performance difference there really is on CPU/chipset temps with and without the channelling layer?

I did some testing on the very first prototype and found that the channelling made only 1 or 2 degrees improvement to the processor, but it made around 5 degrees difference to the system temps, I will try and get some more detailed test results on the latest design and let you know.

Although a few degrees difference isn't major, the drive behind this design was total system cooling with just the processor fan. My main concern of putting a 65W processor into a small case without a system fan is the build up of heat within the case, especially on components which run hot under high loads.

The theory of the design is to allow cool air to enter into the center of the case and only be allowed to vent out the sides of the case, this forces the air to travel accross the full width of the motherboard rather than back out the top vents (and potentially back into the processor fan). This setup creates a 1 way system for the airflow with a flushing effect preventing re-circulation and build up of heat.

 

[K888D]

That edge distortion thing would be something you could solve by using metal instead of acrylic How big is the gap beneath the motherboard now? How much did the volume increase by doing that?

I will look into making the base out of sheet metal, it opens up a lot of possibilites for reducing the volume, but I am hoping to keep the top cover as Acrylic. Maybe even add some sort of LED lighting or illuminated logo with frosted Acrylic.

The gap from the underside of the motherboard to the bottom of the case is 20mm, no volume has been added to the case as the storage bay has been moved under the motherboard into the space and the alignment panel has been integrated into the cover.

maybe using 45mm standoffs on the motherboard or something like that.

I like this idea of using bigger standoffs, thankyou!

Currently the motherboard is sat on 6mm standoffs resting on a raised platform that the storage drives will screw into the under side of. But instead the storage drives can be screwed directly into the base of the case, this platform can be removed completely and the motherboard can sit on 16mm standoffs. This would shave around 4mm off the case height and as you say the standoffs can then be swapped out to suit different socket heights if deemed neccessary.

Do any components on the rear of a motherboard get close to the 6mm in height of the standoffs? i.e. if the gap was 4mm would it reduce motherboard compatibility? Does anybody know what the height of the M.2 connectors on the rear of motherboards usually are?

 

[K888D]

3D Print built up with some experimental LED's and a Steam controller for size comparison, the adjustable intake panel has not been fitted in this photo

Here you can see where the 2x 2.5" drives can be fitted, some of the case wiring can also be placed under this panel

A view inside with the motherboard fitted

[/IMG]

 

[Hahutzy]

Those are some really nice looking 3d prints. PLA or ABS?
Working my way to get my 3d printer to perform as well.

 

[K888D]

Thanks, they were printed on an FDM Fortus 360mc out of PC/ABS.

After putting it all together I think there is room to take another 15 - 20mm off the width

 

[Hahutzy]

Thanks, they were printed on an FDM Fortus 360mc out of PC/ABS.

Ah a production-grade printer. My mouth is watering.

The theory of the design is to allow cool air to enter into the center of the case and only be allowed to vent out the sides of the case, this forces the air to travel accross the full width of the motherboard rather than back out the top vents (and potentially back into the processor fan). This setup creates a 1 way system for the airflow with a flushing effect preventing re-circulation and build up of heat.

I don't know if you have done so already, but I would make sure the sum of surface area of all exhaust holes is at least equal to the surface area of the intake NH-L9i. Or else the exhaust will bottleneck your 1-way flow.

 

[K888D]

Thankyou for the tip, I will check to make sure.

You can get some reasonable finishes on the fortus printers, but nowhere near as good as the Objet printers which cost about half as much, the Objet uses simulant materials which mimic plastics and rubber properties, whereas the fortus is good for getting parts made from actual plastics to give you a good idea of how an injection moulded part will perform, they are still pretty weak across the layer bonds though even for a so called 'production' level machine, you have to make sure you print it in the right orientation to suit the function of the part. I wouldn't feel comfortable using 3d prints off a fortus in place of production parts.

 

[Soul_Est]

This is a wonderful case you have here @K888D. It really shows where many of us would like to go with SFF. Imagine using such a small system containing an APU based on the Zen micro-architecture for a CAD workstation. Have you also looked into using Shapeways for your 3D printing?

 

[K888D]

Imagine using such a small system containing an APU based on the Zen micro-architecture for a CAD workstation

Thanks, that was exactly what I was thinking, my hopes are pinned on Zen for this case!

Have you also looked into using Shapeways for your 3D printing?

Wow that is an awesome site, I didn't even know such a thing existed, a way to sell your 3d printable products is awesome, Thankyou.

 

[Soul_Est]

* snip *
Wow that is an awesome site, I didn't even know such a thing existed, a way to sell your 3d printable products is awesome, Thankyou.

You are most welcome. Should you put it up for sale, please let me know so that I can add your case to the wiki!