Fanless workbench-style PC: feasible?

New to PC silencing? Read & post your questions here. Dedicated to rosy_toes.

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Post Reply
wonderfield
Posts: 4
Joined: Sun Oct 02, 2011 1:42 pm

Fanless workbench-style PC: feasible?

Post by wonderfield » Mon Oct 03, 2011 2:23 pm

Hi there.

After seeing Lian Li's T60 workbench-style case and becoming pretty enamored of it, I've been tossing around the idea of building an ultra-quiet, moderately-powerful PC around it. My question here is this: given an unenclosed system, with little to no airflow, is it remotely possible to run a moderately-powerful PC? In other words, given a sufficiently large heatsink with sufficient surface area, is fan-generated airflow through said heatsink strictly necessary given an environment where dissipated heat isn't trapped inside a relatively small, enclosed space (i.e. a PC case)?

Specifically, I'm looking to pair a Z68 motherboard (Gigabyte's GA-Z68X-UD3H-B3) with a 65W i5-2405S and a Zotac GT 430 Zone (their passively-cooled GT 430, with a TDP of ~49W) and powering it with a Seasonic X-400 PSU. For storage, I'm looking at having a single SSD. Assuming I stick an extremely large heatsink on the processor — something with a fairly low fin density, I imagine, but otherwise very efficient — will heat transfer via convection be substantial enough to keep the CPU from overheating?

I suspect that at least some airflow would be required, but the optimum configuration here, in my mind, is to have no moving parts whatsoever. As someone who's built a number of loud, annoying, fan- and hard drive-laden PCs over the years, the idea of building a machine with no moving parts at all is kind of intriguing.

Thanks in advance for any responses.

HFat
Posts: 1753
Joined: Thu Jul 03, 2008 4:27 am
Location: Switzerland

Re: Fanless workbench-style PC: feasible?

Post by HFat » Tue Oct 04, 2011 4:43 am

It can be done at the cost of high temperatures... probably too high to be recommended unless you're using components that withstand heat particularly well. Running a PC in a workbench is also rather unsafe.
Here are a few more reasonable alternatives:

Use quiet fans. Especially if you want to use the PC for gaming, there's little point in having a totally silent PC unless you have very particular circumstances or you PC consumes substantially less power (in that case, going fanless would be cheap and relatively easy).

Or use a very open enclosure that allows convection, underclock a cheaper CPU and use a less powerful graphics card (or integrated graphics). That will protect your PC better than a workbench and you'll have lower load temperatures. For such a build, you'll want to drop the Seasonic PSU in favor of something less powerful and more effcient.

Or use a case designed to act as a big heatsink with fairly long heatpipes leading from the CPU to the case. You could use a slightly more powerful CPU and you might have to get rid of the PSU and the graphics card.
If you want more powerful integrated 3D graphics than Intel can offer, look into AMD's new offerings. Assuming such a case is available for their more powerful APUs, you could have a halfway-decent GPU and a decent CPU in one little package which can be cooled relatively easily with a few of heatpipes and a substantial metal case.

In any case, you can do better than that Z68 board which is pointlessly expensive for your application and will run hot, again without any tangible benefit.

wonderfield
Posts: 4
Joined: Sun Oct 02, 2011 1:42 pm

Re: Fanless workbench-style PC: feasible?

Post by wonderfield » Tue Oct 04, 2011 9:57 am

Thanks for the feedback. I should have noted that this project would be more aimed at exploring a particular idea (and having something unique) than getting a particularly good gaming machine for the money — this rig would be cycled in and out with my other PC, which is a moderately high-end machine, as needed to play new, higher-end games. I do intend to play games on this machine from time to time, but mostly older games (with vsync enabled so as to keep the GPU workload light), and when it's not gaming, it'll be running Visual Studio and various other development tools. Its actual typical workload should be fairly light. I've considered the idea of grabbing the fan mount rack for the case and installing a couple of Scythe fans, wired up for 5V, and having a simple front switch to turn them on/off if the workload becomes particularly heavy. For the most part I expect the machine will run without the aid of the fans.

The motherboard I've chosen is mostly for aesthetic reasons. An H67 board would work just as well, but I haven't found an H67 board that meets my aesthetic requirements. Price is not really a huge point of concern here anyway. I'd rather go overboard on certain components than to end up saving a few bucks and ending up with parts I'm not happy with.

As for the PSU, I did spend some time looking around last night for alternatives to the X-400, as the T60 case provides no option for fan installation below the motherboard tray, and stumbled across Kingwin's Stryker STR-500 fanless unit, which is an 80 Plus Platinum unit that has a large, top-mounted heatsink, unlike the X-400. Based on the specs, it should work well without a fan unless the ambient temperature becomes alarmingly high, and that's not something I expect will happen. It won't be particularly efficient at the loads it'll be subjected to, but if you have other suggestions for a better unit, I'd like to hear them.

As for workbench-style machines being unsafe, that won't be a problem. I've built watercooled systems that were far more unsafe :)

dancingsnails
Posts: 21
Joined: Thu Jan 07, 2010 12:05 am
Location: San Francisco

Re: Fanless workbench-style PC: feasible?

Post by dancingsnails » Tue Oct 04, 2011 10:47 am

Also it may not be completely silent. Just because there are no moving parts doesn't mean there won't be noise from vibrating inductors and such. Without any case they may be audible - and not a good kind of audible either.

quest_for_silence
Posts: 5275
Joined: Wed Jun 13, 2007 10:12 am
Location: ITALY

Re: Fanless workbench-style PC: feasible?

Post by quest_for_silence » Tue Oct 04, 2011 11:00 pm

wonderfield wrote:will heat transfer via convection be substantial enough to keep the CPU from overheating?

It's feasible, not optimal, and it strictly depends of ambient temperature and load (I would rather a 2390T than a 2405S).
I have a CM Bench (another stupid purchase) where I test my new components before putting into enclosures (sometimes even up to two months): if ambient temp is above 26-28°C and I run FurMark/Prime95, even my Scythe Orochi can't stabilize the CPU temp (not to mention less massive GPU heatsinks).

Broadly speaking, I don't recommend using in a completely fanless system a GPU with TDP in excess of 40W.

quest_for_silence
Posts: 5275
Joined: Wed Jun 13, 2007 10:12 am
Location: ITALY

Re: Fanless workbench-style PC: feasible?

Post by quest_for_silence » Wed Oct 05, 2011 12:36 am

HFat wrote:For such a build, you'll want to drop the Seasonic PSU in favor of something less powerful and more effcient.

Just a side note: what do you have in your mind?

boost
Posts: 661
Joined: Wed Jul 12, 2006 10:29 am
Location: de_DE

Re: Fanless workbench-style PC: feasible?

Post by boost » Wed Oct 05, 2011 1:14 am

If you're looking at a completely new system, check out the Nofen sets. They're standard size towers and should be dead quiet when used with an SSD but fairly expensive.
I would rather use this set or build a similar system with a large CPU heatsink, i.e. Scythe Orochi, Thermalright HR-01 or Silentmaxx BigBlock, and a fairly open case than use a bench table.
A good, very low RPM <500 RPM fan fan should be inaudible from one or two feet away and will improve temperatures a lot.
A GTX430 is not a big improvement on the integrated graphics, for multimedia or light gaming a socket FM1 system with an AMD A-3800 would be very cost efficient. You might need a slow fan for this CPU.

quest_for_silence
Posts: 5275
Joined: Wed Jun 13, 2007 10:12 am
Location: ITALY

Re: Fanless workbench-style PC: feasible?

Post by quest_for_silence » Wed Oct 05, 2011 2:06 am

boost wrote:multimedia or light gaming a socket FM1 system with an AMD A-3800 would be very cost efficient.
But it's not energy-efficient, when properly taxed, up to that I think the A8-3850 with a discrete graphics such as 5570 should account perhaps better for a few less watts (and the IGP is maybe on par with a 6670, power-wise).

boost
Posts: 661
Joined: Wed Jul 12, 2006 10:29 am
Location: de_DE

Re: Fanless workbench-style PC: feasible?

Post by boost » Wed Oct 05, 2011 4:53 am

quest_for_silence wrote:But it's not energy-efficient, when properly taxed, up to that I think the A8-3850 with a discrete graphics such as 5570 should account perhaps better for a few less watts (and the IGP is maybe on par with a 6670, power-wise).
I would argue that the most common use case by far for a PC is idling in Windows. In that scenario the A8-3850 is a bit worse than the i3-2100. The article doesn't compare the 65W version A8-3800 to a i5-2500. Load power consumption is worse by quite a bit, but how often do you "properly tax" your system. The power consumption isn't that much worse that it requires a bigger PSU or even a heatsink. Maybe the fan will spin up a bit more, but only under full load, but that's it.. Which brings me to my second point: total cost of ownership. AMD CPUs and mainboards are cheaper than their Intel counterparts. The money you save on materials will buy a lot of electrical energy, if and when it actually uses more and it will only under full load.

quest_for_silence
Posts: 5275
Joined: Wed Jun 13, 2007 10:12 am
Location: ITALY

Re: Fanless workbench-style PC: feasible?

Post by quest_for_silence » Wed Oct 05, 2011 5:08 am

boost wrote:but how often do you "properly tax" your system.
Maybe you've missed this point: he's talking about the feasibility of a totally fanless open air benchtop (and not of a generic quiet peecee).
With reference to that, forty or fifty watts of more heat do matter: current Llano generation is inefficient, it is a fact and not an opinon.

boost
Posts: 661
Joined: Wed Jul 12, 2006 10:29 am
Location: de_DE

Re: Fanless workbench-style PC: feasible?

Post by boost » Wed Oct 05, 2011 7:38 am

quest_for_silence wrote:With reference to that, forty or fifty watts of more heat do matter: current Llano generation is inefficient, it is a fact and not an opinon.
It includes a nice enough GPU for multimedia and light gaming. The integration level is compellingly efficient. Not to mention the price makes it very cost efficient. I said is was hot under full load, it is not power efficient. I don't know why I'm taking heat from you, though. IMHO it is just an opinion, a very narrow minded one.
quest_for_silence wrote:Maybe you've missed this point: he's talking about the feasibility of a totally fanless open air benchtop (and not of a generic quiet peecee).
Most reviews of CPU heatsinks use an open testbench. Here are some reviews of the Silentmaxx BigBlock on a testbench: Dexgo, Teschke and Planet3D (all Googlenglish). The consensus is that a 65W CPU can be passively cooled even on a testbench. The Scythe Orochi should yield similar results. The A8-3800 which I suggested is a 65W CPU. The Intel i5-2500K, despite being label as a 95W TDP piece, effectively is 65W, too.

wonderfield
Posts: 4
Joined: Sun Oct 02, 2011 1:42 pm

Re: Fanless workbench-style PC: feasible?

Post by wonderfield » Wed Oct 05, 2011 8:33 am

quest_for_silence wrote:I have a CM Bench (another stupid purchase) where I test my new components before putting into enclosures (sometimes even up to two months): if ambient temp is above 26-28°C and I run FurMark/Prime95, even my Scythe Orochi can't stabilize the CPU temp (not to mention less massive GPU heatsinks).
Just the kind of answer I was looking for. Thanks.

I imagine what I'll do then is install the fan rack with a pair of Slip Streams @ 5V or 7V, running constantly, and mount a good PWM fan to whatever CPU heatsink I end up with, having SpeedFan (or perhaps just the BIOS) kick the CPU fan in whenever it reaches a certain temperature threshold. My only concern would be producing significant-enough (and adequately directed) airflow to satisfy the cooling requirements of the GPU, RAM and motherboard components, but that's something I'll just have to roll the dice on if I'm going to stick with the bench case. If it ends up not panning out, I can explore other options.
boost wrote:A GTX430 is not a big improvement on the integrated graphics, for multimedia or light gaming a socket FM1 system with an AMD A-3800 would be very cost efficient. You might need a slow fan for this CPU.
The GT 430 isn't a great performer, no. It is, however, marginally better than the GT2 graphics in the i5, supports 120 Hz refresh rates from its DVI outputs (which the GT2 doesn't, regrettably) and it's — and this is really just a personal preference thing here — not an AMD graphics product, which is something I find incredibly pleasing (as a current HD 5850 owner).

Asus does produce a passively-cooled GTS 450, which is quite a bit faster than the GT 430, but it's currently unavailable in the U.S. and would probably be completely unusable in my situation with its 105W TDP.

I don't have any particular issue with AMD's CPUs, though I do for the most part prefer Intel's offerings right now.

quest_for_silence
Posts: 5275
Joined: Wed Jun 13, 2007 10:12 am
Location: ITALY

Re: Fanless workbench-style PC: feasible?

Post by quest_for_silence » Wed Oct 05, 2011 11:24 am

boost wrote:The integration level is compellingly efficient.

There are more than some evidence that it isn't exactly so (Anandtech, SPCR, Lost Circuits, Tech Report, et c.).
Just for example, the A8 *with a discrete graphics* is more power-efficient than a Sandy Bridge, according to SPCR.
boost wrote:I don't know why I'm taking heat from you, though. IMHO it is just an opinion, a very narrow minded one.

I don't understand what you're saying (writing).
boost wrote:The Scythe Orochi should yield similar results. The A8-3800 which I suggested is a 65W CPU. The Intel i5-2500K, despite being label as a 95W TDP piece, effectively is 65W, too.

Apart that I own the Orochi, so I know its capabilities, but the scheme of things remains equally not favourable to Llano, even with the A8-3800: six dollar less and much slower (2.4GHz vs 2.9GHz), you trade in some economic efficiency for some energy savings, but still about 30W more for the passive CPU heatsink. For almost the same power difference, you may use a discrete HD6570 or HD6670, with their own heatsink and much more horsepower than the APU, while with a GT430 (just slightly better than the A8) perhaps you may save some watts over them (the same goes for the HD5570/5670).

Post Reply