Cooling the Cube

[rossi1959]

After modding the cube myself, and witnessing other builds on this site, I note that they all have a major limiting factor, well two really. Cooling and power reserves.

I think I have a fairly simple fix for the cooling that wouldn't look too garish and can be attached to the cube in a way that would not interfere with its mobility, and can easily be removed for servicing.

As I have a broken outer shell, I will do a mod on it with some scrap mesh from my G5 project, and install 2x 80 mm fans into the cube's lower air intake/cable access area, that will be powered by the motherboards fan header with a fan extension cable.
The fans will blow into the lower section of the case causing a positive pressure through the case pushing the hot air out through the chimney at a much more substantial rate.

I will be able to do the mod on my damaged case with the 2x 80mm fans without sentiment, but I would not be able to cut an undamaged cube, and can see that a pair of 60mm fans might be the limit for the standard opening.

A couple of these 50 mm fractal fans would fit nicely in the cubes intake area.
http://www.fractal-design.com/?view=product&prod=14


All thoughts and suggestions are very welcome.

Cheers.
Rossi.

 

[ersterhernd]

Hi Rossi, I've looked at your Cube build many times so am quite familiar with it. You've done a terrific job of implementing the internal power supply as you have. I did not have such a challenge, as both my DQ77KB builds use an external 160w 8.4A power supply with a Dell style round connector that plugs into the bottom of the cube through the I/O. I've actually ran the cubes on a 65W supply that I have here from my HP laptop, and had no issue whatsoever, even under Prime95 stress test. 160W is probably WAY overkill, but oh well.


The issue you mentioned about heat is current concern of mine as well, as I've stated in a different thread that I'm considering moving to an i7 processor in an upcoming Cube build. The max CPU core temps I've seen under the Prime95 test is around 71-73 degrees with the i3-3225. This is after running for several hours. I've installed 2 fans in my Cubes exactly the same way, as shown below. The intake fan offset from center slightly for maximum draw, and a smaller 60mm exhaust up top.

I have the intake fan throttled to 25% (500 rpm) via PWM, and the top fan throttled to 1600 rpm via a Noctua Speed Reduction wire installed in the case. If you put your hand about six inches from your face and blow on it slightly, that is exactly the feeling of airflow that comes out the top of my cubes. Its cool, too. When Prime95 stress test is running, the air blowing out the top actually feels like the heat coming out of a forced-air heat register in your house, MUCH warmer. Yes, it gets hot inside the Cube, thats for sure.

Your suggestion of the exterior mounted fans has merit, and could likely be done cleanly with some extra care in the installation. I don't think the aesthetics of the Cube would be negatively affected at all really, a good trade-off anyway considering the added static pressure inside the case to force air upwards. If there isn't room inside for an intake fan, then your suggestion of exterior mounting makes great sense, and based upon the results I've mentioned above, would likely be very effective in cooling the machine.

Maybe record some temp results before/after you do the mod, both at idle and under stress. It'd be interesting to see the difference.


Cheers,

Ersterhernd

 

[neilhart]

My 2 cents: I have assembled several versions and have run extensive tests trying to get the perfect G4 Cube cooling. You can always help the situation by using low heat generating designs.

It has been my observation that the air entry and air exit are by Apple design restricted. And with a full complement of cables typical of a user install, the intake area is even reduced.

The rear cable access cut out in the plastic case could be replicated to the other three sides. I have seen this mod done in the past and with a well polished out case, it is not objectionable IMHO.

Mounting a pusher fan inside the bottom plate is most logical. However mounting this pusher fan close the original mesh reduces the effectiveness greatly. There is the Cube that was done last year by a French Modder (search BitTech) where he used larger hexagonal mesh at the bottom plate. This material is much less restrictive (reference the mesh that I used on the Prodigy 'White One').

And at the top air exit, the Apple assembly, is way over engineered and again IMHO needs to be reduced to an nu-restrictive opening.

And the internal placement of passive items in the cube need to be well thought out with air flow in mind.

When I do my next G4 Cube hack, I plan to get deeper into cooling and use the Hex Mesh at the top and bottom and get the air moving.

Again just my thoughts.
neil

 

[ersterhernd]

Hi Neil, all good points you make. It seems Apple was less concerned about heat in the Cube than we are on this forum. I've got 2 running side-by-side on my desk, one original and one of my i3 modded ones, and when holding my hand over the top vent the original is exhausting noticeably warmer air.

Haven't found a stress test for the PowerPC yet, but would interested in trying it out. Nor have I found any temp monitor software for the PPC, I don't know if the G4 even had temp sensors in it.


Ersterhernd

 

[rossi1959]

Thanks Guys,

All goods points, and with me using the Z77 ITX, I don't have the luxury of the dc to dc converter being built in to the motherboard, and have put the converter board in the centre cavity of the cube, leaving only enough room for an 80mm fan at the bottom of the case, and as Neil pointed out that the effectiveness of the fan to pull air is blunted somewhat, although it seemed to drop the temp a couple of degrees at first, but it was just holding back the inevitable temp rise because once it was stressed it rose to 61deg as it did before which leads me to wonder if the fan is just cavitating, and is still just relying on the convection to exhaust the static heat.

I am nearly finished the fan mount, and have deviated away from my original plan to use the leftover G5 case mesh, and have used the G5's rear fan mount instead, and after cutting the damaged section out you will see the damage was that extensive I couldn't square the cube up as good as I would have liked.





It doesn't look as weird as the pictures portray, it is only that the inner drum is in the background, and once turned around , it can hardly be seen.
I have just to get the fans and do some testing.

Cheers.
Brian.

 

[ersterhernd]

Rossi it will be interesting to see the results of this. Watching....


Ersterhernd

 

[MacTester57]

I only have a 80mm fan in the stock fan support. It blows the air thru the stock heatsink, which is connected with a custom made copper block to the i3-3225 CPU heat spreader. This works fine and is almost noiseless, but gets a bit loud while "pushing" the CPU with Prime95.

I think a top fan is unnecessary, because the intake fan produces an overpressure in the case and the only way out is the top grid. An important point is, that you use 4Pin PWM-Fans, that can be regulated through the BIOS sensor readings.

Mac Tester

 

[rossi1959]

I only have a 80mm fan in the stock fan support. It blows the air thru the stock heatsink, which is connected with a custom made copper block to the i3-3225 CPU heat spreader. This works fine and is almost noiseless, but gets a bit loud while "pushing" the CPU with Prime95.

I think a top fan is unnecessary, because the intake fan produces an overpressure in the case and the only way out is the top grid. An important point is, that you use 4Pin PWM-Fans, that can be regulated through the BIOS sensor readings.

Mac Tester

Hi MacTester, I am using an 80mm PWM fan in the bottom of the case, and have the 80mm PWM CPU fan, but the curious thing is, it is noisier under the Mac OS than it is under Win7, whilst testing the Intel HD 4000 in Win7 on an FPS game called stalker call of Pripiat," and on that note"( I am impressed how smooth it plays.) the fans stay quiet. The temps on mine show the heatsink is at 35deg whilst the cores are at 41deg, and I am curious in what your core and heatsink temps are. Cheers. Rossi.

 

[minihack]

Hi Rossi,

Looking forward to seeing how this develops and what you come up with.

I know there are all sorts of different comments already and I don't disagree with any of them, having done a few different Cubes myself.

I think though there is an infinite scope for inventiveness here (even on such a small platform) and will read with interest and see how your solution works out!

 

[rossi1959]

OK Folks,

I thought this mod would cool the cube, and it has, but to a degree I hadn't expected.
It exceed's my expectations by a long shot and my only complaint is noisy fans.(92mm Noctua's would solve this)
The fans I used are old 775 CPU cooler one's I found in my junk boxes, and they pressurise the cube very well indeed, with the air coming out of the cube's chimney at a rate as if the fans were pushing from just behind the chimney vent.
As you can see in the screenshot below, I only ran the stress test for 35min as I could not see any sense letting it go on any further, because the temps were stable and were not deviating from the one's displayed for the length of the test.
This test really proves to me that the most effective way to cool the cube is by forced extraction, (but lot's of it) the air coming through the cube is not even warm, and actually feels cool, and I can only put this down to the displacement of the tepid air in the cube, by the massive amount of cool air pushed in by the 2x 90mm fans.