[Protoyping] Introducing the Mac Pro Jr. - X99 E5-2699v3 64Gb AMD R9 Nano

[phunguss]

I have a x99 motherboard with Xeon e5-2696v3 (18 core) that I wanted to Hackintosh into a G4 Cube case, along with an AMD R9 Nano 4gb. I have had the motherboard/cpu for three years, just getting around to this now as life always gets in the way. Not enough cubic space (350ci or 5.7liters) for water cooling without the radiator being external, so after the new Mac Pro (super cheese grater) was announced, I decided the components would fit into a tower case better, hence the concept of the Mac Pro Jr. (610ci or 10liters). The Jr. is 63% the size of the original G5/Intel Mac Pro tower (the new mac pro is relatively the same size as the original). It will stand about 12.5 inches tall and 5.25 wide.

I have all the components running on the bench in a water cooled configuration with a 120x120mm radiator, but to give a little more cooling capacity and more quiet operation, I will double the radiator to 120x240mm and make it push/pull.

I built a cardboard prototype to test component size and have started on the aluminum case. Hopefully this weekend I will get some of the components bent and either spot welded with TIG or attempt brazing.

(Not the actual motherboard, another project just for sizing)

Conceptually, I have to figure out how to mount everything before I start bonding metal, as I only want to build one case, not five.

Comments and suggestions welcome.

 

[bixmix]

Oh man if you get this figured out I will have to try and attempt to recreate it. A good SFF version of the original cheese grater tower is my holy grail.

 

[phunguss]

I purchased a couple sheets of perforated aluminum to try to get close to the hole diameter of the original Mac Pro Tower. The smaller size has 40% holes and the larger 50% holes (more air flow). On the 5x12 inch face/back, that gives me about 22-27 square inches of free air flow.

I used a 1-1/4 inch dowel and placed the panels in my bench vise. Make sure everything is square, and bend it over to 90 degrees.

Very nice roll.

 

[phunguss]

So I ordered some flux and brazing material as I only have a stick welder. My MAP torch did not get hot enough or I didn't wait long enough... so I went old school. All screws.

Time for the outer shell! Estimated size 24-5/8 x 11-3/4. Again, mostly hand tools and the occasional power tool, this time being a jigsaw. After my sloppy corners, a little hand filing smoothed them out.

Then I made my own roller bender with a 1.5 dowel and a 1 inch box tubing scrap. I did a few test bends and gave it a go.

It looks groovy, so slip fit it on, no adhesives yet, thinking of putting 12 screws on each side to hold it in place, since I am not welding it. I still have to make mount holes for internal components and do the front/rear cutouts.

Compare to the cardboard aided draft (CAD).

Baaaaaby Mac, Mac, Mac!

 

[phunguss]

The front panel ports are: two USB3.0, audio in/out, and illuminated power button. I could not attach it to the perforated aluminum face well, so I decided to use the aluminum angle and mount it to the frame on the left side. The holes are sloppy, and I will eventually redo this part. Once mounted to the frame, I cut the mesh panel to fit around it. I mounted the fans on the radiator and just placed the radiator inside for the photo, it is not mounted yet.

I completed the internal mounts for the motherboard, so I could find where the IO panel cutout needed to be. I also mounted the power supply to find where that cutout needed to be. I have to build a mount for the video card and place the PCI slots on the rear cover.

Final shell assembly. I cut top/bottom panels and attached with 4-40 torx screws. The screws are under the handles, so not too visible near the seams where mesh meets solid panels. The outer shell is connected with 12 4-40 torx screws on each side. The component access panel is attached with 4 screws. Not as easy as the original MacPro latch, but size is a limit.

Finishing. I think I will sand all the deep scratches and then sandblast it. Then either anodize it or cover with clear matte paint.

 

[bixmix]

This looks so good. Cannot overstate that. What do you think the final cost is going to be after all is said and done for all the materials?

 

[phunguss]

This looks so good. Cannot overstate that. What do you think the final cost is going to be after all is said and done for all the materials?

The Case: The cost is probably around $100 for the raw materials. I purchased a 30x48" sheet and an 8ft section of angle, a couple flat stocks from the local hardware store, and ordered the perforated online in 12x12" sheets, all for around $200. And I have enough materials to make a 2nd one. All the screws were free, as I salvaged them from hard drives, but new would be minimal cost.

The Computer: Mobo was 250, CPU was 900, RAM was 420, Misc water cooling items near 200, GPU+waterblock 500. So under $2400 for parts.

 

[phunguss]

The PCI-slot cage took some work. Several ideas, and changes, and once I built it, things changed. So this is the final back plate:

I ended up building it wider than I imagined (fat fingers need to be able to unplug DisplayPort latches), and it almost didn't fit on the rear panel. I wasn't sure how I was going to attach it to the rear panel, until I drew this design and decided to bend the rear panel. This gave the rear mesh some rigidity from all the access holes, and allowed my PCI cage to be mounted to the side frame. It also increased the strength along the I/O panel.

Bent most of it in the bench vice with a hammer and piece of dowel to whack it.

Testing the video card faceplate...

Cut the back hole and slot the folds.

Bend the tabs.

Add some pop rivets and screws.

The final back:

The last piece to build is a the radiator mount that doubles as the PCIE riser card support.

 

[phunguss]

Case complete, installation and testing of computer components:

First up, the radiator mount. I made the case size based on the radiator size, so no real mount needed. There is barely a quarter inch on any dimension, and the front panel wires and fan power on one side push it to the limits on the other. So, just a couple notches in the frame and bend them over a bid to work as a holding mechanism. The PCIE riser support will hold the rest of the radiator in place.

The bottom has both sides tabbed, but the top only has the front side tabbed. The radiator is placed in location, then held in on the rear by the PCIE support bracket. This bracket hold the PCIE riser and has a spot for a bolt on 2.5 drive on the bottom.

After making this bracket and installing the push/pull radiator (with 12mm thick fans on both sides + 3mm offset for fan scraping), it appears my PCIE riser will not be able to plug in the USB on the rear (left of picture). I have other risers, ones that I have modified for other cube projects, so I used one of those and it fits fine.

I then added all the soft tubing, filled it up, and powered it on! It is tight, I don't think it would have fit in the original cube, even with the radiator external.

Some final burn-in and testing before I close up the case and call the project complete!

 

[bixmix]

Hey I guess my last post got deleted. But I wanted to ask if you'd provide some of the details of the material you used? How thick was the aluminum (both the perforated and not)? From the pics it looks like 0.0625" thickness? I think the original Mac Pro has 0.125" holes so I was going to try for that or .25". Where did you source the aluminum? Do you have specific measurements for the cutouts and all that? The dowel rod bending is very clever. I am about to try to recreate something like this and want to nail down the details before I start buying materials.

Also thinking about anodizing at the end. I have 0 experience with that and will prob have a local service give it a go. Do you think I need to get it sandblasted prior to doing that? I am guessing from some research that type 2 anodizing is the way to go...