Critique and suggestions desired.

I was planning on an mATX build, but couldn't find a suitable case, so I'm designing one myself.
Priorities are heat dissipation, noise, and size this is going to be my primary rig, so It will be running 24/7 in my bedroom, unless I've hauled it to a LAN.
It's also going to be used for livestreaming games, so it needs to be able to cool a moderately overclocked CPU, probably a 3570k or 3770k, also necessary is the ability to fit a full length graphics card with an aftermarket cooler.

Let's start with the basic layout, front holds 2 TY-150 fans from thermalright, one optical drive, 1 3.5 bay worth of IO, and 2 3.5 bays worth of fan controller.
Top holds a 240mm radiator.
Back exhaust doesn't have room for a 120mm fan so I'm using 2x 92mm or no fan at all.
Here's the pattern, I have it just about ready to go to CNC:

from top left clockwise:
side panel, side panel, bottom, front filter frame top filter frame, drive cage bits, top(and PSU filter), mobo tray/cable management partition, back, front.



Do I need ducting to redirect some airflow, is the whole idea a waste of time?

Have you taken a look at the lan-gear.eu Braveheart case?

http://www.lan-gear.eu/langear-microATX ... oduction-1

Interesting, certainly LAN compatible, but it doesn't look too quiet, in particular the GPU looks starved.

Also I don't think I can buy one, their US retailer site appears bugged.

On another note, does anyone have experience with the TY-150? Would 2 of them intake enough air quietly?

1. I would submit, that you need to look to your components for quiet. If you get the right components, it will be quiet. If you do not, nothing you can do will ever truly make it quiet. See: viewtopic.php?p=556784#p556784
2. I am certain these cases are available in the US. Someone in the US has one of them. You can try emailing Matthieu Alirol at mylangear@gmail.com

It is hard to see in the current photos, but the GPU is far from starved, it, or they, get their own 120mm fan, it is overhead, but aimed at a 45% angle. Looks like you could put in a high end passive GPU or two and they would cool very nicely.

Can you describe suitable?
I'm guessing small, but what else?
Does the case have room for a tall CPU cooler?
5mm for mainboard standoffs, 3mm mainboard, 7mm cpu and socket, 165mm cooler -> 180 mm. plus mainboard tray, room behind the mainboard tray for cable management plus side panels. My case would be about 200mm wide minimum. Then I would think about adding a little to fit an 140mm exhaust fan...


I agree. Find out what needs to fit, then find a way to fit it. Maybe you'll find there is a case that's at least close enough. If you limit choices by designing a case first, your result might not be what you'd like.
What are suitable components?
How do you want to cool them (how quietly)?
What case would be good with these?

As for choosing quiet components, there is some truth to that, the component does have to dump it's heat into the air inside the case, but that's only part of the problem, the other part of the problem is keeping the air inside the case cool enough that it will readily absorb heat from the components, which is the side of the problem I'm working on now. I'm working under the assumption my system wide tdp will be 500w max and intake air will be under 30°C. I don't want any of my components over 70° C, the transition from die into heatsink is about 20°(base of heatsink is 20°cooler than core temperature) under load on my nehalem chip, may be higher w/ ivy, so for now I'll call that 25. How well the cooler dissipates to the air inside the case is more variable, and I don't have good data on heatsink base vs ambient, so I will split the remaining 25° gradient halfway between getting it out of the heatsink and getting it out of the case. Given the specific heat of air (im using 0.03 kJ/ K ft^3) the power (.5kJ/s) and the temperature difference (12.5K) we can calculate the absolute minimum usable CFM at 45, assuming perfect distribution of heat into the air (all exhausts equal temp, and all the air going into the case absorbs heat before it exits the case.) That's a lot lower than I thought it would be, I think either fan manufacturers are lying about their flow rates, I've made a mistake in my calculations, or I've got way more air throughput than I actually need.

In any case, at least I know I can turn the fans down low and still have enough flow to cool adequately. If there are any problems in cooling they can probably be solved with ducting.



Compatibility with high performance parts is a must, I'll be livestreaming sc2 at at least 720p, 1080p if I can find a new ISP(I know most viewers couldn't tell the difference without the option to select it, but they still complain), which would mean I can fully saturate an overclocked 4 core i7 in regular use(probably 3770k). GPU will be a 680 with aftermarket cooling, seems to be the best performance per watt. I have no need or desire for SLI or crossfire, that said there is currently room for it, and unless you can think of something better to do with the space I'll leave the option open for a big GPU cooler.

Quiet is also a must, but not necessarily while im fully stressing it, hence the fan controllers on front. This machine will be on 24/7 in my bedroom, so it needs to be quiet whenever I'm not using it, or only playing music/video.

Filters on ALL inlets.

It must also be vibration resistant and small, I'll be taking it to LANs. (hence the h100 instead of a large tower cooler, don't want one of those attached to my mobo when I'm bouncing along the road, it would take some pretty convincing drop tests for me to change my mind on that.)

Lastly, it must not look like Optimus Prime took a dump on my desk, while a minor consideration compared to the rest, I do not appreciate windowed cases, parts of the case that jut in or out for no reason(e.g. NZXT Vulcan) or pointless LEDs(I'll still buy them if the product is otherwise the best option, but there's a good chance I'll desolder them.)
Room, maybe, it would be simple to MAKE room for one if there isn't, but I don't currently have any motivation. Right now the whole case is 20 cm wide, 3.16cm of that is taken up by material thickness and cable management space. I'm using fan controllers designed for 3.5 inch bays so I'd have to make case wider rather than the cable management area smaller. I'd have to add 2.5 cm to the width to add a 140mm fan on back instead of 2 92mm fans. As those fans are probably going to be super low/off most of the time I think that can wait till revision 2.

I believe air circulation / heat transfer is far too complicated for the approach you are taking or the equations you are using. You need to engage in some trial and error.

Very true, I have very little hard data on getting the heat from the heatsink base into the air, though heatsink reviews are a good point of data. I've also seen a lot of situations where most of the air goes straight through without cooling a damn thing(lol lanboy air).

As for the equations, I know they're approximations, specific heat in particular varies with density and temperature, but it's probably within 10% of the conditions you'd see inside a PC.

In any case, that ideal number assumes everything that needs cooling can be kept at the same temperature, that you can evacuate exhaust gas without it ever mixing with intake gas, and that you can balance where air is directed precisely enough to get your exhaust temperatures to match.

Thinking a bit more, I think the biggest issues are that heatsinks don't evenly distribute their heat, in a tower heatsink for example, the bottom half of it can be 50 degrees and the top half 30, How much performance would you actually lose if you cut air throughput in half, but sent it through in 2 stages, first through the lukewarm half of the heatsink, then through the hot? can't be more than 5 degrees (assuming 25 ambient). Now what happens if you bring back the original flowrate through this new flow pattern?

The other big issue is that ducting the flow to where you need it reduces the flow rate considerably, an ommission by the manufacturers I guess, that flow rate isn't shown graphed against pressure.

This little discussion did give me a couple ideas for a volume efficient heatsinks and radiators, much harder than the current ones to manufacture though. :/

The omissions are not unintentional. When manufacturers are selling fans to engineers they provide the chart of you are talking about.

Even the engineers who do this for a living have to engage in trial and error testing. Nowadays when they are able to use supercomputers for specific applications... all that does is reduce the amount of testing required... but still does not eliminate it.

It's not plausible that your untrained logic, no matter how smart you are, is going to be up to the task. If it was that easy, all the cpu coolers would have the same performance. If you notice... even cpu coolers that look remarkably similar can have substantially different performance. It appears to be a lot more art and finesse than mere equations. And the equations that do capture what is happening need lots of data points and super computers to fully implement.

Also, take a look at:
viewtopic.php?p=557300#p557300

as far as heatsinks that look the same, exactly how the heatpipes are filled, and how the cooler is mounted(orientation matters)and amount of clearance between working fluid and die are important. I suspect even details like barometer and ambient humidity during manufacture can cause sample variance if the QC isn't super careful.

Yes. And on top of it, heat and air currents are just so darn hard to predict. That is why I would propose to you that no matter how well designed and thought out it is what you build... that you build it in a manner that provides you flexibility in testing it different ways. If you build something that has to be right the first time... you are almost dooming yourself to failure... or perhaps more accurately, diminished satisfaction.

I was planning on designing any ducting after I had it all put together and my GPU installed.

It seems like you got better answers here than on TL^^

Maybe you should post it on a trash forum, like EA UK, b.net, 4chan, just to get the entire spectrum of internet quality

Only you can answer this. All value is subjective. Of course, I selfishly want you to proceed and post pics!

You can always fab it out of something inexpensive for prototyping. Foam core or whatever.

A couple more images, flipped the back panel for finished side out, other small corrections and changes.


Also doing some sanity checking on airflow/ducting potential, if it's too loud or hot at start I'll try something similar to this:


Obviously I haven't chosen a GPU cooler yet, the above is intended to isolate GPU exhaust from the popular 2 slot coolers that vent into case, though I'll probably get a bigger cooler for GPU before I do ducting.

Is this going to be stamped from sheet metal or machined from solid stock? If machined, it's going to weigh a freaking ton. If stamped, it's going to take a lot more than just punching out flat pieces of sheet. There are a lot of very subtle three dimensional aspects even to something as simple as a solid side panel.

I don't like top mounted fans. I'd rather leave the top of the case solid and exhaust the air past the CPU and GPU heatsinks.

Unless you have a particular PSU in mind, I think I'd make the bottom fan grid rectangular, so that you don't have to worry about it aligning perfectly with the PSU's intake fan.

I wouldn't want an optical drive at ground-level unless this case always lived on top of a desk or the drive was very seldom used.

It's going to be CNC laser cut from 16 gauge stainless and TIG welded together. you can see the mounting hooks in the cut pattern, which will pull it flush.

I was planning on something in the earthwatts platinum series from antec for PSU, because they're fairly efficient, quiet, and share the same short form factor. Otherwise I'd have to find a short optical drive.

As for optical drive on bottom, it won't fit anywhere else unless I use smaller intakes and/or ditch the h100. It will indeed live on my desk and will only be used for installing OS. Maybe I should just delete it and install the OS with the dvd drive outside the case.

Top mounted fans though, I hate side mounted fans even more, and there's not enough room for a radiator on back.

There are somethings you overlook until the problem slaps you in the face yes.
I'll go off on a bit of a tangent here, on my most recent build (nearly 2 years old by now) I used a P183, but one detail I missed was the positioning of the intakes in the p183, and most importantly the external 3.5 inch bay. Unfortunately I neglected to overlay an image of the motherboard and GPU over the case when selecting components, and I didn't notice that the first x16 slot being in slot #3 was a problem until I got it into the case, and found that the GPU could only really draw air from that 3.5 inch bay. Wasn't something I could fix without modifying/switching case or swapping mobo. I ended up modifying case to improve airflow to the GPU.

I wish cases like this were available. You could move the ODD to the top, but that might require the case to be taller. You could also use an external ODD or install from USB stick.

Agreed.

I have optical drives in two of my systems. One is a HTPC, and needs it. The other is a gaming system and I haven't used the optical drive since installing the OS. I can't remember the last game I bought physically - Steam took care of that.

But both are negative pressure cases, and the gaps around the drive bezels are just place dust gets sucked into the cases.

You could always get a slim USB drive for the rare occasion you need optical.

Apparently the big manufacturers see enthusiast mATX a futile market.

LIAN LI V351 W 10.98; H 10.31; D 14.69; (ITX? & MATX)
BGEARS BG01206 W 11.20; H 9.00; D 14.70; (ONLY MATX)
APEVIA X-QPACK II W 11.25; H 9.00; D 14.50; (3 WINDOWS) (ONLY MATX)
APEVIA X-QPACK W 11.2; H 9.00; D 13.80; (3 WINDOWS) (ONLY MATX)
LIAN LI PC-Q07 W 7.60; H 11.02; D 8.19; (ONLY ITX)
LIAN LI PC-Q08 W 8.94; H 10.70; D 13.58; (ONLY ITX)
ARK MINI ITX PI-01 W 8.00; H 7.00; D 12.75;
SUGO SG05 W 8.74; H 6.93; D 10.87; (ONLY ITX)
SUGO SG06 W 8.66; H 6.97; D 11.26; (ONLY ITX)
SUGO SG07 W 8.74; H 7.48; D 13.78; (ONLY ITX)
SUGO SG02F W 10.6; H 8.30; D 15.50; (ONLY MATX)
SUGO SG03B-F W 7.87; H 14.17; D 12.28; (ONLY MATX)
APEX MI-100 W 8.70; H; 5.10; D 11.80; (ONLY ITX)
APEX MI-008 W 8.70; H; 5.10; D 11.80; (ONLY ITX)
LIAN LI PC-Q09 W 7.85; H 3.86; D 7.85; (ONLY ITX)
ANTEC NSK1380 W 10.60; H 7.90; D 13.20; (ONLY MATX)
ANTEC NSK 3480 W 7.87; H 13.78; D 13.98; (ONLY MATX)
ANTEC NSK 3300 W 7.80; H 13.80; D 13.80; (ONLY MATX)
ANTEC NSK 4400 W 7.80; H 16.50; D 16.60; (ATX & MATX)
ANTEC NSK 4480 W 7.80; H 16.50; D 16.60; (MATX & ATX)
LIAN LI PC A03 W 7.40; H 14.49; D 17.60; (ITX & MATX)
LIAN LI A05 W 8.27; H 15.00; D 19.29; (MATX & ATX)
LIAN LI PC-A06F W 7.36; H 14.76; D 19.29; (MATX & ATX)
ANTEC SOLO W 8.10; H 17.00; D 18.50; (ITX, MATX & ATX)
CENTURION 590 W 8.27; H 17.32; D 20.08; (MATX & ATX)
ANTEC ELITE 341 W 7.28; H 14.37; D 15.75; (ONLY MATX)
SHUTTLE SG41J1 W 8.46; H 7.48; D 12.79;
LAN-GEAR.COM W 11.4" X H 8.6" X D 14.6" (ONLY MATX)

Okay, so maybe it's just corsair: http://www.overclock.net/t/1220311/so-c ... t_16555249

Still, if that's the competition, I should sell these.

I think you should start by assembling a model with Styrofoam or cardboard or something. Then posting the pictures here.

But maybe before doing that determine what the target components are. You should make the case as small as possible to take those components. The biggest decision is what CPU cooler you are going to use. I would suggest something along of the lines of the Prolimatech Megahalen / Thermalright True format.

But why not consider building a case around a NH-C14 or a NH-L12 without a top fan. You could pack a lot of power in a very small volume while doing a good job of cooling the motherboard components. I would build short shroud around the top of the cooler. That would permit the CPU cooler fan to act as the case fan. You could then use two slow 120mm to exhaust air from the case.

Maybe you have a baffle that permits the video card to have its own air stream? Take a closer look at the Lan-Gear cases. They bsically sort of do that. The video card gets it own separate case fan.

If you are going to build it to hold traditional hard drives, consider using Noise Magic No Vibes III:
viewtopic.php?p=539359#p539359

I also think the case fan blowing air into the case and the cpu cooler can should be one in the same. And that even in the event of the loss of one fan, that there is enough redundancy to keep the case from frying itself.

Don't forget you need to design for the future not for the past.

I thought the fan above the GPUs in the lan gear case was an exhaust, where do they get the intake? If that is an intake, why is it right above the GPU IO/Exhaust?

System will be built in a month or two, so components might change.
CPU will be 3570k or 3770k, OC to however much I need to live encode 4 mbit 1080p in FMLE on 2 cores(plus 2 HT cores if it's an i7). Probably around 4.5 to 5 ghz. Not set in stone, if more information indicates SC2:HotS will be able to utilize more than 2 cores, I'll probably go with a 6 core CPU instead.
GPU will be a 680 w/ non reference cooler.
Monitor will be 120hz 1080p. I'm targeting a minimum framerate of 60fps.

PSU will be earthwatts platinum 550.
For CPU cooler I was planning on a Corsair H100. As I said earlier, i'll be taking this to LANs, and don't really like the idea of hitting a pothole with a big tower cooler on the mobo. I don't know that the smaller air coolers are enough to handle that OC.

The NH-C14 has more than enough horsepower to handle overclocking. See one of its better reviews here:
http://www.frostytech.com/articleview.c ... 562&page=5
Where it compares favorably to both the Corsair Hydro H50 and the CoolIT Domino ALC water cooling systems... which will not cool your motherboard or your video card anywhere near as well as the downdraft NH-C14. This would be especially so if you built a low profile case that gave the NH-C14 direct access to cool external air. I would expect that air to be perhaps 15C cooler than the internal air inside your case. It will go first to cool your CPU and then forced by the bottom fan to cool you motherboard and then your video card. It would almost be like strapping on a turbo charger.

If you take a look at its CPU cooling performance handling the 150 watt load, it seems likely it will handle the overclocking even with just the lower fan... reducing its profile even further. Designing a case that gives it direct access to cool external air will substantially increase its cooling capacity even more.

And it is implausible that a Corsair would be quieter. The Corsair H50 tested 10 db louder than the NH-C14 with both fans on high. The Corsair H100 is likely to be even louder than the Corsair H50, don't you think?

The NH-C14 is a low profile non-tower cooler with a special strut intended to assure users that it is safe when jarred.

I would be no less cautious about traveling around and jarring a water cooling system.

Is chipset cooling or case temperature really going to be a problem? Take a closer look at that airflow sanity check image I posted(the one in color).I don't think so. The h50 with fans on high isn't relevant at all. I'll be using the h100 with the fans on a controller, and possibly replace the fans for something else. According to the reviews I've seen, the h100 performs very well with fans on low, and the performance scales through 200w+ overclocks.It's not like I'm going out of my way to jar the case; if you're driving around some vibration is unavoidable. The difference is that most of the mass is attached directly to the case rather than the motherboard. That strut adds rigidity between the fins and the mounting bracket, whatever it's intention, it doesn't remove any load at all from the motherboard. Low profile or tower, it's still 1kg hanging off the mobo.

I am not able to fully figure out your diagrams. But to the extent I do understand them it seemed to me that you could perhaps accomplish more with less... that is house the same CPU & GPU power, with
less case (both in size and weight)...
less noise...
and less cost.

And I still harbor the believe that it is unlikely that either Corsair will approach the quietness of the Noctua. At the end of the day you have pump noise and also the high impedance radiators that you have to push air through. If you actually have 200 watts of heat perhaps you have to put up with that.

But I can't believe that, even overclocked, the Ivy Bridge will ever even approach saturating the Noctua's heat capacity dispersion capability.

Just my opinion.

What kind of wattage do you expect your Ivy bridge chip to generate with and without overclocking?
What are the exact dimensions of the proposed case?