Hutzy XS — Ultra Compact Gaming Case (<4L)

By the way the effect of (a bit of) heat on SSDs is probably the opposite than on HDDs.
They're said to charge the cells better when the temp is a bit higher (but not high enough to make the controller throttle).

(whereas heat is the number one killer for HDDs as shown in many studies by Google, Amazon, MS, Facebook and others in their data centers)

So SSDs and HDDs should be mentioned as different situations, not "SSDs and HDDs" together, when talking about heat.

Just a side note.
 
But what about a NVMe-based Samsung 950 Pro M.2 format SSD…?!?
 
KazeoHin said:
That would be on the mainboard, directly cooled by the CPU HSF.
Click to expand...

No, that would be on the BACK of the MB, sandwiched between the MB & the divider panel; so the question still stands…
 
I just finished running a stress test to get some numbers:

22.5 degrees Celsius ambient

Testing procedure:
1) Start Prime95 in Large FFT mode. Run 30 minutes
2) At 30th minute, Start Furmark to run in conjunction with Prime95 for 30 minutes
3) At 60th minute, take a snapshot
4) Turn off both stress tests and start a timer
5) When both CPU and GPU has dropped to 45 degrees Celsius, take a snap shot

Tested using:
i5-4440S
Gigabyte GTX 970 mini
Samsung 840 EVO 120GB

All of the following numbers are peak plateau numbers, meaning that these numbers were reached and continued to float in the neighborhood of (but under) for at least 20 minutes.
Therefore I assume that the system has reached a perpetual state.

Here's the results:
CPU: 70 degrees Celsius
GPU: 78 degrees Celsius (1 degree lower than my Unigine Heaven stress test)
SSD: 55 degrees Celsius

Cool-to-45 time: 3m25s



P.S. There was a clear noise difference between just having the CPU at load, and having the CPU+GPU at load. Also when I put my ear against the system, I hear the GPU fan louder than the PSU.


.
 
Boil, I do not have an M.2 SSD yet. I will be getting a new Skylake cpu + mobo + M.2 SSD when I get the metal prototype.

I'll make sure to perform the same tests and make sure that the M.2 is usable.
 
Did you log the gpu clocks? gpu will always run on load at ~78 degrees but the clocks are what matters.
 
Hahutzy said:
Boil, I do not have an M.2 SSD yet. I will be getting a new Skylake cpu + mobo + M.2 SSD when I get the metal prototype.

I'll make sure to perform the same tests and make sure that the M.2 is usable.
Click to expand...

Good deal, please make sure to get one of the Samsung 950 Pro SSDs, these are supposed to get hotter than the SATA-based M.2 drives, and I worry about what temps they might have when sandwiched between components & chassis panels & such (no matter which chassis that might be; Hutzy XS, A4-SFX, Osmi, Dune Case, etc.)…

Plus, if you are getting new Skylake gear, you might as well opt for the fastest of the M.2 SSDs that the new platform supports & go for the NVMe speed demon…!!!
 
Hahutzy said:
I just finished running a stress test to get some numbers:

(...)

Here's the results:
CPU: 70 degrees Celsius
GPU: 78 degrees Celsius (1 degree lower than my Unigine Heaven stress test)
SSD: 55 degrees Celsius

Cool-to-45 time: 3m25s



P.S. There was a clear noise difference between just having the CPU at load, and having the CPU+GPU at load. Also when I put my ear against the system, I hear the GPU fan louder than the PSU.


.
Click to expand...

Thanks for the thorough testing.

As I imagined, the drive gets toasty, and I believe it will get even worse once the case is made of metal. For me, 55ºC is unacceptable. You could try to make some holes on the bottom of the case to see if you can create some sort of convection airflow. Though I do not think it will help much, its better than nothing.
 
prava said:
For me, 55ºC is unacceptable. You could try to make some holes on the bottom of the case to see if you can create some sort of convection airflow. Though I do not think it will help much, its better than nothing.
Click to expand...

According to the specification of Samsung 840 Pro series SSD's, the "Operating temperature" ~55ºC is quite OK, because those drives can operate in the range form 0ºC-70ºC (link). It means, 55ºC is over 21% less then maximum temperature allowed by the producer...which isn't so bad with this SSD.
You have to also remember, that Samsung makes many other type's of drives, which also gets hot very quickly, like this M.2 : link .

But i understand Your point of view. Higher temperature usually equals higher error rate or it can shorten the SSD's lifetime. And with HDD, a 55ºC can be a limit, like for some of the Toshiba drives (link, link, link).
I'm sure Hahutzy will make some additional tests, and then we'll know everything. He's still at the prototyping stage, so he can make some improvements if needed.
 
Last edited:
I love the idea of this case. It's the logical next evolution for tiny powerhouses with the increasing popularity of very small powerful GPU's ala the Nano. I also think the idea to allow for use of either a flex-ATX or a DC DC board is a great idea for those willing to use a brick.

My only issue is aesthetic: the racing stripe. I'm not a huge fan of it, personally. Maybe make it optional?
 
prava: I think that 55 degrees Celsius peak is okay. It was not running at 55 the whole time. Regardless, I will be trying different ventilated side panels this week to test the hole configuration. One of them has ventilation all the way down.


illram: the problem with removing the highlight piece is that there will be screw holes showing through. I am considering adding a color option where the body is also black anodized aluminum so the whole case is monotone color. That might suit you and others that do not want the highlight piece to stand out.


SaperPL: I did not, because it never said that it hit a thermal throttle. For the sake of completeness, next time I run the tests again I will keep track of the clocks too.
 
Last edited:
Care to elaborate on that? What's supposed to tell you about hitting a thermal throttle?

The thing is, you've got a card that has base clock and turbo clock.

With great cooling performance (like in the cases pumping a lot of air through or when water cooling your gpu) you'll even get higher clocks than turbo clocks described in card's specs. With normal cooling or in open case you should have clocks between base clock and turbo clock advertised in the specs. With bad cooling gpu will throttle which means it's clocks will keep dropping until the temps are not raising over their target which is around 80 degrees for maxwell gpu's.

Note that you should check that while the gpu is fully loaded and not bottlenecked. There's two main bottlenecks that might lower the load on the gpu: either slow cpu (which shouldn't be a problem with your unit) or running with vsync limiting the framerates because that will end up with lower clocks and lower teps since there's no requirement for more power.

Finally, I'm not up to date with current gpu architectures but I think they also might be running using only a fraction of their stream processors/compute units with full turbo clocks if the application isn't supporting as much spread of data as the card could handle simultaneously and that could end up being misleading as you'd get low temps and full clocks while the gpu is not really fully loaded.

I've been logging my temps and clocks on the last prototype with msi afterburner which is quite nice - you can get almost everything logged except for hard drive temperatures.
 
My way of telling if the GPU is getting thermal throttled at any given moment is to use GPU-Z while doing stress tests

In the Sensors tab, there's an entry called "PerfCap Reason", which gives you a timeline of what is the current reason for performance cap.

At idle, the reason is "Util". it's capped by low utilization

At load, the normal reason is "Pwr", "Vrel" or "Vop". This means your card is limited by the power it's willing to receive, whether its the current, average voltage, or peak voltage.

When thermal throttled, it's "Thm", meaning your card got too hot and it's forced to downclock to stay in a temperature range that doesn't kill the card.

In my old case (Hutzy HS) when there was no ventilation at the top, my 980 Ti lit up a bar of purple with the label Thm because it was choked in hot air.

In all the stress tests I've performed with the XS so far, I have not seen a single strand of purple in the PerfCap Reason.

That is my reasoning when I say that "I have never seen it thermal throttle".

And related to this point, while recording the clock speed of my card is a worthwhile effort, in this scenario even if it's not Turboclocking any higher than spec, the reason is not thermals.

The reason is either:
- The system is drawing power at the limit of the PSU, so the GPU doesn't have extra juice to take.
- The GPU is at its limit and chooses not to draw more power.
 
Last edited:
Good to know about that indicator - thanks!

As for the reason - if the gpu is capped by power limit you might want to check out the riser you're using - with too long, bad quality or damaged riser you might starve your gpu not even knowing about it. We've had damaged riser early in the prototyping stage and the R9-270X and GTX750(without PEG connector) were blackscreening due to power starving while the GTX970 was running stable. This might be your reason for hitting power draw limit and that could also mean you're not testing the gpu on it's full potential.

Anyway thanks for the tip about gpu-z
 
Hahutzy said:
illram: the problem with removing the highlight piece is that there will be screw holes showing through. I am considering adding a color option where the body is also black anodized aluminum so the whole case is monotone color. That might suit you and others that do not want the highlight piece to stand out.
Click to expand...

I see. Are those the screws for the middle mobo plate?

Have you considered an interior shell design to fix the interior plate to, covered with the two side panels and a full size L shaped panel covering the top and front? It would change the look of the case somewhat, i.e. the middle piece would be more flush with the side panels and there would be no additional highlight piece. It would be a more minimalist look, obviously if that is not what you are going for then ignore this! :)

Edit: I guess that might also cost a few mm here and there, which might not work.
 
Last edited:
illram: it is as you said, that design would add to the overall size.

--------

Turns out I screwed up the previous stress test.

I have had my prototype parallel to the wall, with GPU side against the wall, with around a 2-inch gap for air intake.

Turns out that was choking the GPU.

I found out today because I was stress testing some new panels, and a screw fell off the panel, so I moved the PC away from the wall to retrieve the screw.

In that short time, the GPU dropped from the usual plateau of 77-78 degrees to 72 degrees.

I then rotated it completely perpendicular to the wall. So the front is facing me, and the back is facing the wall, and continued the test for the remaining 20 minutes (I was 40 minutes into the test when this happened).

The result is that the CPU result did not change (as expected, because it was the side that was away from the wall in the first place)

But the GPU resulted in a peak plateau of 71 degrees Celsius instead of the previously claimed 78.



Also, I was curious why the core clock speed was hovering around 850MHz.

Turns out NVIDIA and AMD both downclock their gpus when it is detected that you're running Furmark or OCCT (maybe also other stress testers).
 
Last edited:
Looking at the operational temperature range of SSDs, most should be 0-70C. It would be nice for sub 50C SSD temps, but 54C should theoretically be just fine.
 
Maybe we need a couple of extreme tests.

1) both side vents covered up
2) top vent covered up

and assume no one will be covering up all three vents...
 
CknSalad said:
Looking at the operational temperature range of SSDs, most should be 0-70C. It would be nice for sub 50C SSD temps, but 54C should theoretically be just fine.
Click to expand...

Careful with that... Where I work we saw plenty of early mortality in some Crucial drives we had in a test enclosure at avg temps not much higher than this (58C)

To expand on this: We build a enclosure for banks of SSDs to use for data capture in the field (LADAR type application) and we noticed some drive issues after a few months of use and determined that the SSDs near the end of the enclosure were a little airflow starved and were averaging around 58C but spiking to 70C or so under heavy use which caused 2 of the 4 drives in the rear to die early on us. So yes, there is a operational temp range but you'd be surprised how much drives under heavy use can heat up.

Samsung 850 Pro 512 GB drives btw.
 
Last edited:
Here's the problem with that statement. An average is an average. At 58 average, you know your max is actually higher than that. It could very well be that your crucial ssd's went above 70. (Operating temperature of M500 is also 0-70. I don't know if that's the one you're referring to.)

Hahutzy is talking about the ssd's max temp in a stress test, which means the avg should be less than 55.
 
Gaming temps especially for the SSD would be nice to see. I would test demanding games such as Crysis for up to an hour or so straight.
 
CknSalad said:
Gaming temps especially for the SSD would be nice to see. I would test demanding games such as Crysis for up to an hour or so straight.
Click to expand...

Or just give a test system to Kyle and let him run the gauntlet on the thing.... I'm pretty sure [H]ard wants these answers as much as we do.
 
Correct me if I am wrong, but the final materials used for the case will be Brushed Aluminum, Anodized as mentioned on Hahutzy's OP on this thread? If that is the case, it should be a couple of degrees cooler than the 3D-printed model?
 
CknSalad said:
Correct me if I am wrong, but the final materials used for the case will be Brushed Aluminum, Anodized as mentioned on Hahutzy's OP on this thread? If that is the case, it should be a couple of degrees cooler than the 3D-printed model?
Click to expand...

It would be possible to use thermal pads to let the SSD dissipate a bit of heat to the outside, which could be beneficial, but for the other components, there will probably not even be a measurable benefit.
 
iFreilicht said:
It would be possible to use thermal pads to let the SSD dissipate a bit of heat to the outside, which could be beneficial, but for the other components, there will probably not even be a measurable benefit.
Click to expand...

Well, for m.2 SSDs sure. But 2.5" SSDs would require removing the cover (w/a pentalobe screwdriver) and many/most of them have components on both sides of the PCB.
 
darrpara said:
Well, for m.2 SSDs sure. But 2.5" SSDs would require removing the cover (w/a pentalobe screwdriver) and many/most of them have components on both sides of the PCB.
Click to expand...

With this case it actually won't work for M.2 SSDs because the divider part of the case would be made from plastic even in the final version, IIRC.
Wouldn't 2.5" SSDs cool down a little as well, even if their chips don't directly touch the cover/heatsink?
 
iFreilicht said:
With this case it actually won't work for M.2 SSDs because the divider part of the case would be made from plastic even in the final version, IIRC.
Wouldn't 2.5" SSDs cool down a little as well, even if their chips don't directly touch the cover/heatsink?
Click to expand...
Well the small heatsinks do help in my experience with the 850 pros I mentioned above but if they are starved for fresh air it still creates a hot zone around the component. We had them out of the case but I'd bet under normal use the little pad heatsinks + opening the SSD case would drop temps maybe 10C under light loads. It took a few minutes longer to reach high temps but in the end they did still hit 70C temps. It is more of an air circulation issue than pure heat dissipation from the component (usually the controller chip is the hottest).

We ended up going to Extended Temp (85C) commercial drives and redesigning the production enclosure because of this issue and IIRC even Intel had an issue with their NUC kits bluescreening due to high msata SSD temps.

edit: This is what I did for my 950 Pro to keep temps down. Overkill but these m.2 drives run WAY hot under load. Samsung 950 PRO [Official Thread]
 
I entertained the thought of making it slightly bigger than 4L this past weekend.

The result would be:
- 4.3L (mainly an increase in vertical height)
- Still Flex ATX, but will have room on top of it for an 80mm fan. I will then order a custom batch of Flex ATXs where the 40mm fan is removed, and the cooling is driven by the external 80mm fan instead. This means the case will be bundled with the PSU.
- Instead of 1x 9.5mm HDD slot, there will be 2x 7mm, one on each side (in roughly the same as the current position but also on the CPU side).

So there's many advantages to be gained. But of course it comes at the price of being over 4L.

Thoughts?

 
Last edited:
Hahutzy said:
I entertained the thought of making it slightly bigger than 4L this past weekend.

The result would be:
- 4.3L (mainly an increase in vertical height)
- Still Flex ATX, but will have room on top of it for an 80mm fan. I will then order a custom batch of Flex ATXs where the 40mm fan is removed, and the cooling is driven by the external 80mm fan instead. This means the case will be bundled with the PSU.
- Instead of 1x 9.5mm HDD slot, there will be 2x 7mm, one on each side (in roughly the same as the current position but also on the CPU side).

So there's many advantages to be gained. But of course it comes at the price of being over 4L.

Thoughts?

Click to expand...

There are plenty of other designers who will make compromises. You should be the one who stands out as making NO compromises. <4L or bust.




Also, any chance you could have the front face panel removable, exposing a mount for a 2x 80mm radiator?
 
I agree with KazeoHin that this PC case would really stand out at <4L. I feel that this was your main vision/goal of this project in the first place. The main thing to address would be the SSD temps with various games and provide graphical charts and screenshots of HW Monitor for each game.
 
Even if you decide to use an external 80mm fan on the PSU, which would be a great idea, having two 7mm mounts is not a good compromise in my eyes. 9.5mm is the minimum for 2TB of HDD storage, which is important if you need or want that much but can't afford to buy two 1TB SSDs.
 
Hahutzy said:
I entertained the thought of making it slightly bigger than 4L this past weekend.

The result would be:
- 4.3L (mainly an increase in vertical height)
- Still Flex ATX, but will have room on top of it for an 80mm fan. I will then order a custom batch of Flex ATXs where the 40mm fan is removed, and the cooling is driven by the external 80mm fan instead. This means the case will be bundled with the PSU.
- Instead of 1x 9.5mm HDD slot, there will be 2x 7mm, one on each side (in roughly the same as the current position but also on the CPU side).

So there's many advantages to be gained. But of course it comes at the price of being over 4L.

Thoughts?

Click to expand...


I will add my 2 cents:

- don't go into custom PSU. This will definitely kill Your project. Some people will be happy getting a PSU with the case, but in a situation when you will have to replace it, the cost of a single, custom PSU will be very high. And also You will be the only supplier for this PSU, while people like to have a choice.
- what is the speed of the fan in Your PSU? I'm asking, because You're telling that it's quiet. I was having several servers in our office and i never saw a quiet Flex ATX PSU. Maybe You have found the best one? :D
- why don't You just add some venting holes in the front of Your case, in front of Your PSU fan? Make some nice figures/shapes with the punched holes as made at the top of the case. If this will be the same element/part, then adding 10-20 venting holes shouldn't be expensive.
- 2 drives instead of 1 is a good idea, but only 7mm ssd's isn't good solution. You are "closing" Your customers market. Many people have their actual rigs packed in some SFF cases and would like to switch to Yours, and You're telling them they will have to additionally buy another drive/drives. This might by a problem, but I can be wrong.

For those of you who tell Hahutzy to keep his case internal volume <4L "no matter what", i have to tell, that i think you don't feel his actual situation. We had similar situation in the past with our pc-case. If he's going to sell his box, then he'll probably need some MOQ(minimum order quantity). For example, to get a good price from the riser manufacturer, he'll need to order 100-500 pcs. 100-200 pcs for vandal switch, 100-500 pcs for PSU, etc. etc.

How will he get this MOQ numbers when he at start will say, that Hutzy XS is <4L but...you can't do this, you can't do that, oh... and you also can't do this, that and that..... ?? If this case won't be at least in some fields universal, then Hahutzy might have problems with selling this case. Of course some people will order it, but the price (for example) from $150 will rise to $300. It will shrink his market for sure.

This pc-case has a potential. Keep up the good work :)
 
Last edited:
Hahutzy, while I'm not going to be under 4L or bust, I will say that you shouldn't worry about FlexATX being too loud. Let the FlexATX manufacturers worry about that. Custom PSU is a worse alternative.

Instead you should look into picoPSU or the HDPLEX, etc.
 
I feel like > 4L is worth the tradeoff of not having to have an external power brick, since even the HP firebird bricks have already shown to be incapable of handling AMD nano power spikes. What kind of wattage can you get on a Flex ATX psu, and wouldn't it be better to have the 80mm fan at the bottom as a direct intake for the case and psu at the same time?
 
I guess what I was getting at was this project would differentiate itself if it were <4L vs. the >4L cases such as the Logic Supply MC600 and NFC S4-Mini. However, now that I think about it, this case at >4L would have the flexibility of a FLEX-ATX psu or external power brick/DC power board solution. In addition to this, it would be the only mini tower between 4-4.5L. I would definitely market these aspects hard when you do finalize your PC case and sell it! Keep up the good work, I am really looking forward to the end product.
 
Would be a cool case for an ultra compact gaming rig, if the temps could be kept under control.

I would like to see the ability to use a 15mm thick 2.5" SSD though; namely the Intel 750-series NVMe model. This, of course, would need decent airflow/cooling as well as the other components…
 
Thank you all for the responses.

I thought about it a little bit more, and got some important ideas out of this exercise:

1) For storage, there will most likely be 2 common use-cases: i) Boot M.2 SSD + Storage 2.5" 9.5mm mechanical, or ii) Boot 2.5" 7mm SSD

2) The 80mm fan would be nice, but it also puts the liability on me (or my eventual organization?) if anything happens to the PSU, because the FlexATX maker did not intend for it to be cooled in this setup.

3) Due to manufacturing tolerances of components, what I initially believed to be enough space for PSU + 2x 2.5" 7mm drives is actually not enough. I would need 2mm more in width, which drives it ever closer to a width that begs the question: "why not just fit a SFX there then?"

Because of these above issues, I've decided to abandon the idea of the taller design, and will continue with the initial design



For good time's sake, here's a pic of the printed bottom panel for the testing

 
Last edited:
ZombiPL said:
- what is the speed of the fan in Your PSU? I'm asking, because You're telling that it's quiet. I was having several servers in our office and i never saw a quiet Flex ATX PSU. Maybe You have found the best one? :D
Click to expand...

I dare not claim that "it is quiet" until/if I ever have solid numbers that indicate as such. To that end, I'm still putting together a test for testing system noise.

My reasoning for staying with the design has always been:

1) Yes, it is a 40mm fan. It gets loud when you ramp up your system's demand. But at that point, your CPU and (more importantly) GPU fans are also going to ramp up.
From my experience, the amount of noise added due to the PSU is not significantly more compared to the noise from the CPU and GPU.
So yes, it would be nice to reduce PSU fan's noise, but it will not be "the change" to create a 4L silent system, because the PSU fan is not a big audible bottleneck

2) For demanding applications such as gaming or using VR, the user is most likely audibly immersed into an experience that requires the performance in the first place.
From my experience, the whole system's noise in gaming situations has not deterred me from immersing into the game and its audio, even when I was only using my monitor's speakers instead of a headset.

In the end, it's about balancing trade-offs. You can't win them all (y)
 
Last edited:
Thanks for the updates on all this! How exactly does that PSU exhaust? is it front-to-back?
 
You must log in or register to reply here.
Back
Top