Contribute
Register

Building a G4 Cube in Australia

Status
Not open for further replies.
I think it looks just fine, Kiwi. Have you had a chance to stress test the i7 with the Cube closed up? Not sure if you're at that point yet or not.

It'd be interesting to see the temps.



Good Work!



Ersterhernd
 
I think it looks just fine, Kiwi. Have you had a chance to stress test the i7 with the Cube closed up? Not sure if you're at that point yet or not.

It'd be interesting to see the temps.

Good Work! Ersterhernd
So yes I have it all properly mounted, and working, just have post some pics. Temps are OK but could be better.

As an initial test have setup a fan/temp profile in bios, as would like to run it normally. Running case fan at 40-60%, CPU fan at 20-70%, with a max CPU temp of 75 degrees.

Running Prime95 (no overclocking) Temps rise slowly to 75 degrees then fans speed up to 100% for a while, then slow as temps reduce down, this pattern continues as the temps are maintained around 75 degrees.

I don't think this ideal, would like normal throttled fan rpm to be effective. We are in middle of winter, so summer could be a problem.

However I have some more work to do, which you will see the start of when can post some pics.

EDIT: I also did the same tests without the Max CPU temperature restriction. Temperatures would rise and settle around 83-85 degrees.
 
So here is where I am up-to with the cooling solution...

Chassis Fan Mounting


I mounted a case fan underneath the drive bays, to provide air intake from the bottom of the cube. I used a 92mm PWM fan which is only 14mm high.

I made up two brackets to hold the fan
IMG_1791.jpg


Here is the underside of the brackets with the fan attached, you can see the reason for the angled cutout.
IMG_1796.jpg


Then I glued two standoffs to the side of the case
IMG_1792.jpg


And in the middle glued some bolts (with heads chopped off) into some holes under the handle
IMG_1793.jpg


Here is what it looks like under the handle
IMG_HandleUnderside.jpg


Here is a picture showing the fan mounted
IMG_1800.jpg



Improvements

To improve cooling I have build a barrier to isolate the air intake to the bottom of the case. The theory is that it will reduce air being recirculated around the fan, so (in theory) ALL air sucked into the fan comes from outside the case (AND). No air can leave through the bottom of the case, it must be expelled from the top.

Of course this isn't going to be a 100% but should offer worthwhile improvement.

Here is the cutout made from thin flexible sheet plastic (Note plastic is white on reverse side)
IMG_1801.jpg


The cowling sits around the fan mounts in the bottom of the case
IMG_1802.jpg


With the fan screwed on top, it looks like this
IMG_1803.jpg


Then adding thick tape, to provide a good seal looks ugly, but does the job
IMG_1823.jpg



Did this make a difference, I didn't really take a baseline temperature before building it (my Bad). On the bench with the cube on it side, bringing my hand close to the bottom of the computer could definitely feel air being drawn into the grille, and mounted vertically you can feel airflow upto 30cm above the case.

This isn't the final state, still have some more work to do to channel the air internally.
 
Good Thermal Results - New Cooling Approach - Hugh Disappointment

Good Thermal Results

Today I did some major work on the Cube.

Here are the temps after running CPU Test Enormous test for 20-30 minutes. Note: These temps are about 10 degrees lower than anything I have tested previously for this build.

Results1.png


Geekbench

Results2.png


I don't know how to configure over-clocking but think could get some more MHZ out of this machine. Im summary I used Chameleon Wizard to setup smbios.plist for Mac Mini 6,2 then used ssdtPRGen to create the SSDT.AML file, no other changes except adjusting fan speeds.

New Cooling Approach

Basically I took a different approach than I have seen before. The approach is to treat the Cube like a laptop and specifically direct airflow through the computer in directed way.

Remember this design means I only need 2 x 92mm cooling fans, So means I have plenty of room (relative term) for optical drive, 3.5" HDD, and 2.5" SSD.

Here is how it works:

I - Air drawn in the bottom through case fan (see previous posts). With added air isolation blocking off holes at bottom of case to anything outer than the fan intake. This prevents air circulating around the fan, and to air escaping out the bottom, to be redrawn in by the fan.

II - A cowling covering the chasis fan directs airflow horizontally through the latching mechanism emerging at the bottom of the motherboard.

IMG_1853.jpg


III - This air cools the south bridge and other components on the motherboard

IMG_1855.jpg


IV - CPU Fan is located in-between the latching mechanism, close to the heatsink

IMG_1849.jpg


V - CPU heatsink has cowling that seals airflow from the heatsink to the Fan, meaning all air from the fan , must travel through the heatsink itself, no air can escape. This counteracts the problem where the fan is offset from the heatsink (see much earlier posts)

IMG_1841.jpg


VI - CPU Fan SUCKS air (NOT Blows) through the CPU heatsink. Air circulating around the motherboard is then drawn through the heatsink, exhausting towards the front of the latching mechanism.

IMG_1867.jpg


VII - The front of the latching mechanism is sealed, at the bottom by the cowling from #2 above, and at the front by the HDD drive that is mounted (almost) in the original location.

VIII - This leaves only one exit for heated air, and that is straight UP through the grille. As HDD is next to the CPU fan it is cooled (actually probably warmed) by air before it escapes.

During stress tests I can actually feel heated air on my face 50cm above the cube, I think this shows how effective this solution is at forcing air out of the top of the Cube.

I would love to have a pictorial diagram of this, time does not permit.

Hugh Disappointment

So after an enormous amount of work today, and some good temperature results. One huge disappointment. My Cube no longer fits inside the Can....…

The tests above were taken with the Cube protruding about a centermeter from the bottom.

IMG_1862.jpg


Here is the cuprit.

IMG_1871.jpg


To make room at the bottom I moved the HDD higher in the case, it is clearly too high.

:banghead:

As always welcome comments and feedback. Today has been a marathon...

Kiwi
 
I am greatly impressed with the concept and the execution. Consider a short skirt for the bottom of the can.

Good hacking,
neil

an after thought; " a mini-skirt"
 
I am greatly impressed with the concept and the execution. Consider a short skirt for the bottom of the can.

Good hacking, Neil

an after thought; " a mini-skirt"
Good thought, but I prefer the no-skirt approach. Simply put I lowered the height of the HDD.

It wasn't two hard in the end as the main HDD was held under tension against the Cubes internal frame via a backing plate. I had to push the cowling down a little bit, the weight of the HDD did the rest, splaying the thin cardboard where necessary.

Refinements, Questions and Power Switch

Thanks for your words of encouragement (it means something), however this was always an experiment, which I am not totally sure I have finished, so would appreciate feedback on my current dilemma.

Some facts: When I run the Cube out of the Can (on its side) I consistently get temperature of 5 degrees lower than when it is upright in the Can. Below are the specs of the fans I am using.

Case Fan - Noctua NF-A9x14 PWM Cooling Fan
29.5 CF/M - 2200 rpm - 20 dBA - 92x14mm

CPU Cooler - Xigmatek Praeton LD963
48 CF/M - 2800 rpm - 22 dBA - 92x15mm

The higher temperature inside the Can are disappointing, and feel like a failure of the design. I suspect that the case fan cannot supply enough fresh air, and when removed from the case it can draw in fresh air from its surroundings.

What to do?

  • Remove some of the plastic blocking air from being naturally drawn in through the bottom of the case.
  • Swap the fans over so the case fan creates a slight internal over pressure and can effectively pump air all around the heatsinks intake.
  • Buy a second "Xigmatek" and have matched fans.
  • I had even considered rotating the heatsink, so fins would be vertically aligned which should mean less resistance as the air is drawn up from the bottom of the case over the heatsink.
Its hard not to second guess yourself, so any feedback welcome…

On another subject

Power Switch.

The original Cube switch wont fit in its original location, and options of moving it seemed hard. The Edison solution seemed obvious but I wanted to try something different. So have ordered an IR transmitter and detector. This will sit on top of the HDD and shine up through the grille. The board I have ordered has adjustable sensitivity up to 20cm. The goal is to be able to turn the cube on by waving to it, i.e. passing your hand over the top of the grille.
 
Status of Build

So I have completed the basic build of the Cube, but waiting on parts to finish some remaining items. In the mean time it is now with it new owner until ready to finish.

When I complete the build I will post some final pics, including details of how my power switch goes.

Cooling Conclusions

This build was all about effective cooling of the 65w i7-3770s. Using two 92mm low profile fans. I have achieved temps (under load) in the low 70's, with fans running below 100%. My design also accommodates a full complement of Optical, 2.5" SSD, and 3.5" HDD

So Yes I think it was a success.

One of the critical successes (my opinion) was reversal of the CPU fan so it sucks air into a sealed cavity between the locking mechanism, so that air is forced up and exits out of the case through the top vent. At maximum RPM I can detect (using strips of tissue paper) air movement up to 1.5 meters above the case.

This means there is little re-circulation of heated air in the case. And for every litre of air expelled at the top there is an equal amount of fresh air being drawn in from the bottom.

Along the way I had a few issues, and in retrospect could have done things a little different, but at this point I don't want to retrace my steps, so offer some advice for anyone willing to extend my work.

  • Using this configuration it may be possible to not need a case fan at all. Just leaving the bottom unmodified, and sealing the cpu fan to the heatsink, and sealing around the locking mechanism.
  • If a case fan is needed I would use matched fans and to run the case fan at slightly higher rpm, this should ensure cpu fan has adequate supply of fresh air.
  • My initial design was to have the cooler rotated 90 degrees to have more alignment with the Fan. The cowling I created provides a seal and negates this misalignment problem. I think rotating it is still worth a try to see if there is a benefit, there are too many factors to really know which is better.
Finally I would recommend the Xigmatek - Praeton LD963 CPU Cooler in such a build.
 
Just a quick update on this build. Have had to re-order infrared led's so a bit of a delay while they arrive. Have fitted two extra USB's to bottom, similar to Rick's long term build, and one port on the top.
 
Good luck with the build! glad to see us aussies giving it a go :)
and love this case ill be doing similar when i finally find one
 
Good luck with the build! glad to see us aussies giving it a go :)
and love this case ill be doing similar when i finally find one
It took me a while to find my case, they are very hard to come by at a reasonable cost. I have been looking for a G4 iMac as well, nothing so far.

Kiwi
 
Status
Not open for further replies.
Back
Top