scratch build PowerCore

[Keir Graham]

Hi everyone,
First time on here for a scratch build :), I've done two previous build on Bit-tech years ago, and have come back for a third trying to be more adventurous than my previous build. One of my main constraints was that I wanted it small enough to easily take over to a friends, but at the same time I want it powerful enough for VR, and very unique design. This is the model of the design I came up with:

 

The first stage I took was to make sure I could obtain the heatsinks for the radiators I want to make, since this design is dependant on them. These were available at Bal-group.com, on this link: https://www.bal-group.com/heatsinks/27 , and they were happy to sell me only three:

Unfortunately they didn't have the exact size I needed (the other sizes they sell are just them cutting this size down for a fee), so I had to cut a section of each one with a slotting saw on the milling machine:

As the slotting saw doesn't leave a very good finish, I then used a face mill to skim a layer off to smooth it out:

As the base of the heatsinks are way to thick for what I need, I then again used the face mill to thin down of the base aluminium:

These are now ready for CNC to mill out the water and o-ring channels on the back:

Next up was testing the casting process, since this case requires such a difficult shape, the only way I'm going to be able to make it, is by casting aluminium into a mold, then milling, grinding and sanding afterwards. In order to test the process I've started off small, by casting the water junction blocks, that feed to the two coolant pipes on each side. I 3D printed out a mold, and poured in casting silicone (after spraying on wax mold release), to create a silicone end cap for a piece of stainless tube to act as a casting flask for a special plaster mix:

The hole in the centre holds a sprue, which then holds the parts you want to cast:


I then 3D Printed (In PLA) the parts and melted the wax sprues and the parts together:

I then also printed out a cap to hold the silicone tightly onto the bottom:

I then poured in the special casting plaster (Available Here: https://www.artisanfoundry.co.uk/product_info.php?products_id=98). This Plaster is a special blend for casting metal, don't use normal plaster. The Plaster itself sets in minutes, so you don't have long to mix and fill it.

Here's the result after I peeled off the silicone:


To remove any air bubbles in the plaster mix, I used a homemade vacuum chamber. Its a Instant Pot pressure cooker with a 14mm Acrylic top. I also used it to create a vacuum under the plaster mold while I poured the aluminium in:


To create the silicone ring on top that the mold sits on, I printed out a mold and pasted in some silicone sealant:

 

 

 

 

Edited September 20, 2022 by Keir Graham
Pics somehow got re-added at the end

[Keir Graham]

I didn't manage to get any pics of the casting process unfortunately due to the burn-out cycle taking 14hours and it had gone dark by the time I could pour in the metal and I was doing it outside alone in the dark, so getting pics and casting would have been tricky, so I'll just detail the process:

1. Create a plaster mould using the special plaster
2. Follow the plaster burnout cycle to remove all the moisture from the plaster and burnout the PLA. The burnout cycle I followed was the Jewelery one on this guide: https://docplayer.net/44984225-Sett...ent-for-jewellery-and-industrial-casting.html which is 14hours and up to 730c.
3. After the burnout cycle keep the mould hot by placing it in the oven on max 280c or so.
4. Melt the aluminium, make sure you melt more than you think you will need just in case.
5. Take the mould out the oven and place it on the silicone ring on top of the vacuum chamber.
6. Skim the dross off the top of the molten aluminium and switch on the vacuum pump
7. Pour the aluminium into the mould, give it 10-20 secs, then switch off the vacuum pump, and wait 10min or so for it to cool a bit.
8. Plunge the mould and castings into a bucket of water, this helps break up the plaster and free the casting.

There is also the option for degassing and using drossing flux, however I didn't opt to use them in this case. Gases only dissolve in the aluminium while its molten, and since I used the aluminium straight after it melted, and I used an Electic Kiln, and not a gas burner, I didn't feel it was worth while getting a bottle of Argon, as minimal gas dissolves in, in this case. For dross, since I was using a brand new crucible and decent ingot there was barely any dross to begin with, so I didn't bother using flux either.

You then get a completed casting with incredible detail, you can clearly see the print lines:

 

 

The Kiln I used was this little one off eBay:

 

With a graphite Crucible for melting the aluminium:

 

Its also important to use an Aluminium alloy that is designed for casting, I bought an ingot of eBay (https://www.ebay.co.uk/itm/Aluminium-Ingot-BS1490-5KG-LME-Branded-Manufacturing-Alloy-Bullion/254330264143?ssPageName=STRK:MEBIDX:IT&_trksid=p2057872.m2749.l2649):

 

If your looking at getting into casting aluminium I would highly recommend VOG's channel on YouTube, its where I found a lot of this information: https://www.youtube.com/channel/UCkEYj8wtK3aEW8vSGhlB43g

I will also note my first attempt failed as I had filled the flask to high with plaster and couldn't get a decent vacuum seal, so on the 2nd attempt I didn't fill it as high, and I also added some vent sprues that came to the top, and also put in some sprues going up the sides from the bottom up, so the vacuum could get deeper into the plaster, I also didn't bother with the wax this time, and printed the sprues instead. This gave the results above.

Here's the 2nd attempt with the little vent sprues showing around the edges. I had to dig out the plaster a little to reveal them.

 

 

Now I had the completed castings, I then cut them free from the sprues:

 

I then drilled, tapped, milled and sanded to get a decent surface finish, and flat sides that the water cooling fittings can seal against:

 

And the results before finishing:

 

These are now ready for fine bead blasting to create an even, smooth finish, and then anodising.

I then worked on the casting for the outer shell of the case. I was attempting to cast it all in one in a very large plaster mould, with some help from SDTS Engineering: https://www.sdtsengineering.co.uk/

 

[Keir Graham]

Here's the 3d printed forms I made for the casting of some of the body panels, the 3D models were split into small enough pieces to print, and scaling them all by 1.8% to account for the shrinkage of the aluminium, I've used a bit of super glue to glue all the parts together then attach them to a sprue tree:

 

 

Sorry about the super vivid blue, my camera goes a little crazy with these! lol. The large pour bowl bit at the top, is to get some weight on the top to really push the molten alu into the mould, and as the alu shrinks it pulls in from the top, so you need a bit of a reservoir. These casting will also be vacuumed like I did with the small ones to really pull the alu into all the fine details.
These parts will be cast in the special plaster like the water junctions above, and the large panels are going to be cast using the ceramic shell investment casting process, due to there size it would take a huge amount of plaster and it aint cheap, so at least with the ceramic shell you only need a thin shell's worth of material.
Since PLA can expand and crack the ceramic shell as it heats up, I was going to use PolyCast filament for the large bits:
https://www.3djake.uk/polymaker/polycast-natural
This stuff is designed for this process so hopefully it will give good results. Its probably going to take another weeks worth of printing to get the main panels printed.

In the mean time for these smaller bits, I welded up some stainless boxes to cast these plaster parts in.  I kept blowing holes in the metal lol but got there in the end. I then hot glued the stainless to a wood base so I can easily remove it after pouring in the plaster:

 

 

I also added some print spool to the corners so that the vacuum can reach into all the plaster during the metal casting process:

 

Then it was plaster pour time. Once mixed I quickly vacuumed out all the air bubbles:

 

Here's the final results of the plaster casting. I've since put these in the oven at 130c for 16 hours to dry them out a bit, my oven has a steam vent so all the steam comes out the top rather than getting stuck in the oven.

 

Next up is the investment casting pieces. Since these parts have a big empty space in the middle, it would use way to much plaster, so I was going to have these done in the ceramic shell lost wax process as that only creates a thin shell over the whole part. The print spool I used was PolyCast Natural from here: https://www.3djake.uk/polymaker/polycast-natural . PS4 pad for scale :). After casting in aluminium, these will be welded together.

 

This print spool is rather expensive so I decided to make the sprues and pour funnel out of wax I already had instead. First step is to melt some on very low heat. Instead of pouring imp using a stainless meat basting syringe:

 

And these nesting funnels:

 

At the same time I was also creating the sprues by pouring wax into a large diameter silicone tube. Once the wax sets you can just stretch the silicone over the end of a tap and force the wax out with the water pressure I also used a coin to plug the end while I pour in the wax:

 

And the results after melting the parts together:

 

I then printed some thin wall hollow sprue sections, put the wax parts above through the middle, then injected some wax to merge the wax and prints together:

[Keir Graham]

At this point I was still waiting on the getting the castings done at an investment casting company, and I hadn't found another alternative, so I decided to start making the custom CPU block, as this build is going to be so compact I need a waterblock that the tubes come off at 90 degrees, and I'll be cherry picking a motherboard that doesn't have anything in the way.
To start off I milled the edges of an acrylic block so I could get it into the lathe, and then rounded it off one side at a time:

 

I then bored out a circle so that the fins from the copper plate (From a corsair block) will it into one end:

 

Then I trimmed down the cylinder further to leave a lip on the bottom edge for the mounting bracket:

 

I then mounted it into the milling machine and milled out the channels for the coolant to flow into the fins:

 

Then using centering drills I made a pilot hole then drill through into the milled channels from the side:

 

Then I milled out a ledge and tapped the holes for the water cooling fittings:

 

I then used this rotary table to mill a 1.5mm channel for the O-ring:

 

I then drilled the ring of holes for the self tapping screws that hold the copper plate to the acrylic:

 

I flipped around the rotary table, and put the chuck back on it, then used it to mill out some notches for the mounting clamp to fit into:

 

And then finally after a lot of wet and dry sanding working my way up from 600 to 3000gritt, then a polish with Turtle Wax polish while it was spinning on the lathe here are the results:

 

The rads then arrived back from the CNC

 

I then spent around 2hrs tapping all these holes starting with a tapered tap, and then a plugging tap to get the thread right down to the bottom of the 6mm holes.

 

This part also arrived from the CNC, and I polished it like the waterblock with 600 up to 3000 grit wet and dry, then Turtle Wax polish. This is the pump top and it slots inside the next part, and seals with an o-ring around the bottom:

 

This next part is the cap for the CNC'd part. This made on the lathe starting from a block of acrylic like with the waterblock, then alot of polishing:

 

Here's some pics of how the parts fit together. The very outer edge and fitting wont be visible, and I was going to play around with various materials and lighting on the sides and bottom later:

This then sits on top of the DDC pump. Eagle eyed among you may have noticed the mistake I made, the outlet is on the wrong side for a DDC pump doh, so I ordered a DIYINHK DDC control board, which are fully controllable via PWM, and I can reverse the direction of the motor. This will then be used with a custom impeller which I was planning on making anyway.

Edited September 20, 2022 by Keir Graham

[Keir Graham]

This is the standard impeller with the top removed:

 

I've lathed off the blades so I can JB weld on a new set of blades that will be in the reverse direction, then I'll have to figure out how to balance it:

 

I then had a bit of set back when I discovered that aluminium castings don't anodise to an even finish, and you are restricted to only grey colours, sooo... I came up with a new plan to get the shell made from 4mm aluminium sheet instead, that's been bent on a CNC press brake. This unfortunately means I need to redesign the internals so that I can fasten everything with aluminium threaded studs welded onto the sheet instead of threaded lugs cast into the shell.

While I was sorting all out that with a local company I finished off the pump. I used JB weld original to attach the 3d Printed new impeller blades:

 

I then de-soldered the original circuit board

And soldered on the new PWM controllable board. Shows you how much the tech has come along, its basically all on a single chip now

 

Then I gave it a quick test, and it worked a treat. The pump will have a chrome around the edge, and a painted underside when its fully complete so you cant see all the way through

 

I then received the bent alu sheet that will base the walls of the case. I first milled down the ends so that they sit flush:

 

These two halves will be welded together like this:

 

I then trimmed down the top and bottom edges to make them perfectly straight:

 

Then milled a bevelled edge on the back edges which the back panel will slot in between:

 

Then milled out the notch in the bottom edge that gives a gap for the blower fan to draw air under the case. I will also draw down from the top as well.

 

Only the straight bit at the top is finished, the sloping ends I will file down later:

 

Next up was milling out the pattern in the side of both halves for the coolant pipe section:

 

I got a bit carried away and forgot to leave space for the bevel at each end grr, so I had to just go with a rounded edge design instead. I needed to recast the water junction blocks anyway, as they were designed to fit onto a casted wall, not one with welded on threaded studs.

 

The diagonal bits will be filed down later:

 

I then started work on the jig that will hold the parts in the correct position so I can mill the 3 corners:

 

I gave it a quick test to see if it was going to work, it was a bit rough, but doable.

 

The welding made the jig not quite flat anymore, so I milled it back down flat:

 

I then re-clamped the first part down onto it, and began to mill out the large cut-out section:

 

The middle sections that will be welded together later also fit on the jig:

 

I then used a narrow handheld belt sander to sand down all the diagonal bits to give this result:

 

Since I've gone this route of making this outer shell using belt sheet aluminium, I then needed to completely re-model the interior of the computer, to mount everything via threaded spot welded studs instead of cast lugs. So I decided to learn Fusion 360 and remodel it on there. I also slightly tweaked the outer design too. In order to get the model to match these parts, I used a scanner to get an image of the profile, then used that as a canvas to model from. I then 3D Printed the profile to check it matched, and made minor alterations. I then sent this shell to be welded.
They did a great job of joining the two halves together. With the amount of time I spent making this part I didn't want to risk my amateur welding lol. They cut a thin slice off another scrap piece of this same alu for use as a filler rod, so that the anodising will be an even colour.

 

Once Grinded/sanded it looks like its all once piece now:

 

The big block of alu arrived too for the back panel so next up was to manually mill out the shape of the panel (to save some CNC cost). I gave the top and one edge a quick skim over with the face mill and end mill:

 

Then figured out how to mount the thing as the milling table is barely wide enough, while also giving room to mill out:

 

Also got this new toy 12.5 mill roughing end mill:

 

Cuts through like butter, so much quicker than normal end mills, cant believe I hadn't not used one before:

 

I then cut out the sections for the PSU and Motherboard IO, and switched back to a normal 2 flute end mill for getting a smooth edge:

[Keir Graham]

I then flipped over the block and tilted the miller head the other way to mill out the opposite side:

 

 

This took about 16 hours to get this far!, as I was getting more and more nervous about messing up as I went along, so the work got slower and slower, as I was triple checking each cut. Here's the results before I cut the sides off and trim down the bottom so it actually fits:

 

I then milled off the sides and end of the back panel:

 

Then milled the hole for the power button:

 

I then sanded the panel down with an orbital sander but the forgot to take a pic of it before the next step.
I rented a capacitor discharge Spot Welder from www.taylor-studwelding.com:

 

I couldn't get in to all the areas I needed threaded studs with the big gun, so I also rented this mini gun that goes into the end of the big one:

 

I tried a few, then found the best technique was to press a blunt center punch into the alu to make a very shallow dent, that the little pip on the bottom of the stud just about sits into, then give it a quick sand:

 

I also spent hours marking out all these spots to put the studs onto, then welded them all on:

 

You get a fair amount of splatter, these are M3 threaded studs:

 

Next up I started making the new water junction blocks for the outer coolant pipe section. I'm going to have to scrap the cast ones I did, since I found castings don't anodise very well, and colour wise your stuck with grey.

 

New vise too

 

I then used the belt sander to curve the edges, I still need to finish it by hand sanding to match the chamfer:

During this work, I've come to the conclusion that those DIYINHK DDC pwm coolant pump boards are crap, this thing is sooo loud now. I thought I had messed up somehow but after reading online it seems that these boards are just noisy af. Sooo I'm gonna have to get another acrylic piece CNC'd and use the standard control board I think.

Anyway I finished off the bevelled edges on the new water junction blocks with a small file:

 

After tapping the holes, these were ready and waiting for the final surface finishing at the end of the build:

 

I had to drill a hole all the way through from the side to connect all the holes, then I'll make some plugs to screw into the ends to seal off this side hole. The tapping was done straight after the drilling with the tap in the milling machine, and the gears in neutral so I could manually turn it, so as to get the tap dead centre.

 

I then spent most of that weekend milling out the bar mounts for each heatsink. Two of the mounts will also be the mounts for the back panel as well:

 

There is tons more to come, but its getting late now, so I'll post the rest tomorrow

[Timpelay]

Wow mate this is incredible. I see the hours fly away.

Impressive to see you work with your tools.

[Keir Graham]

9 hours ago, Timpelay said:

Wow mate this is incredible. I see the hours fly away.

Impressive to see you work with your tools.

Thanks Timpelay, yeah it took a ton of hours, and I haven't even included all the little bits here and there of failed experiments, and design changes.

 

After a ton of small adjustments here and there, I finally managed to get the rads to fit perfectly. I wanted to get the mounts to sit perfectly as I'm planning on having a bit of thermal paste between the rads and the case, to essentially use the rads and the entire case itself as surface area to get rid of the heat:

 

I've also started playing around with different fans. I tried a PS3 blower fan first, and man those things are powerful on full, way over the top of what this needs I think, so I took the blower fan of a GTX 970 and cut the fins off:

 

I then 3d printed a 140mm blower fan blades peice:

I found a company in China for getting all the main CNC bits done: https://www.rapiddirect.com/ they gave quotes very fast, and responded to every question within an hour (during their work hours), and they gave a rapid turn around and delivery to the UK. All the following parts came to USD 898.50, which given the complexity I thought was pretty good 

Before I sent off the files for CNC I needed to make sure the 3D model would exactly match the shell, given it was made by bending aluminium sheet, and its complex shape, to the eye it looks perfect, but there a slight variations which would mean the CNC'd tops wouldn't fit flush with the edge all the way around. My first idea was to use a robot arm and trace around the perimeter of the shell:

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

but unfortunately for some reason it didn't work out that accurate, even though when the path was re-traced on the arm to looked perfect. So I resorted to getting an A3 scan of the end of the shell, and using that as a canvas to draw on in Fusion360, I then 3D printed out a template from the model, then adjusted the 3D model based on how it fit, and repeated, over and over and over until it matched perfectly:

 

 

I then sent the model to RapidDirect and they have sent back these impressive quality parts:

 

 

The raw finish of the CNC is very smooth too, so there wasn't much sanding to do. The tabs on the inside I'll need to mill down a little though, as I forgot to account for the raised stub at the base of the threaded studs, so they didn't quite fit on the case perfectly at the moment.

I also had the pump housing re-done as well, so I can use a normal DDC, because I made the original one in a mirror image:

 

I was amazed how smooth they got this downward slope, with barely any rounded edges at the bottom:

So thats all ready for polishing

 

I also had an aluminium impeller made as well for the pump, rather than the 3D Printed one I had before:

 

This took me longer to figure out than I care to admit lol, but eventually managed to clamp the top panels in a way that I could trim down the mounting tabs a bit:

 

All the bits arrived from the laser cutters too

 

Next up was a hellish amount of polishing, this must have taken me about 18 hours all together, as I had to start off with 80 grit sand paper to get out the deep dents in the aluminium where the bends were, and a few deep scratches. I started off with an orbital sander and discs from 80 - 800 grit, then swapped over to foam sanding discs with WD-40, then finally a polishing disc for a polishing grinder and the brown metal polishing compound.

 

For the tricky to reach spots I used a multitool with the sanding head

 

Then I stuck some foam to the blade attachment (Didnt work without the foam from the pic above), and stuck some wet and dry to get the areas I couldnt fit the normal sanding head in.

[Keir Graham]

And here's the result, its still needs a final polish to remove some of the streaks but its almost there:

 

Then onto finishing the fan. The edges are 3D printed with soft fabric glued onto the sides:

 

Here is it in the bottom of the case:

 

I've finished the polishing of the case shell and top by this point, however the apodizers are concerned about being able to get an even black tint across both the sheet, and milled parts, soooooo, my next idea was to get it clear lacquered instead, with a "candy" tint to keep the mirror effect but have a black tint. I found this lacquer online: https://www.cerakote.co.uk/shop/cerakote-coatings/performance-clears/clear-aluminium-mc-5100/ this 
Which seems to be super tough, sticks to polished aluminium, and a company near me sprays it, but they hadn't tinted it before though, so I was waiting for them to test tinting it, before they give me a price.

Anyway onto the build....

I managed to find an arbour for polishing wheels that fits my grinder, so this made things alot easier to finish the polishing:

 

Next up was sanding and polishing the blades for the vents in the top and bottom in each corner:

 

I then realised all the hex's I had laser cut all had a little bit sticking out so I had to file them all off, took ages:

 

I then sticky taped them to a peice of wood, then sanded and polished them:

 

I then glued them all to a 1mm polycarbonate sheet, along with the rest of the panel and using a 3d printed holder to get them centred. Each of these will be cut off this sheet later. The battery was to press the hex part out of the 3d print holder:

 

That was then set aside, while I modded some G1/4 plugs:

 

It passed the pressure test

 

I also found these sweet fittings from Barrow, that have RGB LEDs built into the compression fitting, which Ill be using with these water function blocks:

 

I then polished and hand bent the 1mm alu panels that go behind the tubes on the outside of the case:

 

For the backlight, I thought about using a Tv backlight, but it didn't have the light spreading ability I was looking for unfortunately:

 

So I've had to resort to Surface Pro LCD back light sheets:

 

I wore gloves for this as the sheets mark very easily:

 

I glued, and taped the edges onto the back of the panel, I'll probably stick some fabric or something on the back as well to cover it all:

 

This is how it will look, for this test its just a small RGB strip on one side, when its complete it will be a strip on both sides, and it will be the length of the whole side:

 

I had the water distro block for the GPU 3D Printed via HP Multi Jet Fusion, so its water/air tight:

 

Edited September 21, 2022 by Keir Graham

[Keir Graham]

I then tapped the holes, and milled flat the surfaces. then sanded it all:

 

 

Then started investigating surface finishes for the main body. I tried for a quite a while to find a company within 150 miles that would black chrome plate aluminium, as a one off but I had to give up the end. I then tried various black chrome paints, and tints but I wasn't happy with any that I found. So its pains me to say it, but after all that polishing I ended up with just going for a super glossy polished black car paint. Not what I originally wanted however I think its still turned out very nice.

I also got back the rest of the inner parts which had been ceramic coated with Cerakote. Sorry about the fuzzy pictures of these black ones, my phone camera didn't like it.

 

I then chopped up more of those Surface Pro back lights to sticky tape to the back of these panels, however annoyingly I forgot to take a picture, but you get the idea from the previous light panels I did, I then stuck LED strips down the edges:

 

then added the heatsinks:

 

Next up I glued in the magnets that will keep the top panel vents in place, so they can be easily removed for cleaning (there will be filter foam on the inner side soon):

 

and finally we get some computer hardware in this thing:
Specs are Zotac 3070 Twin Edge with waterblock, 16gb ram, Ryzen 7 5800x, 1tb NVME SSD under the heatsink and a 500gb one on the back, 4tb HDD,
Gigabyte B550 I AORUS PRO AX motherboard, 600W Corsair SF600 SFX PSU

 

Tape on the PSU as I almost dropped a bolt in it lol. The HDD is mounted using a rubber pad on the screw side, and an O ring on the other so the vibration is dampened.

 

JB welded, and super glued some small HDMI/Display port extenders to the GPU IO plate:

 

Installed here, and also connected up the Cooler Master Master Accessories RGB Fan Controller, which will run all the RGB LED strips

 

More wiring for more RGB I also changed the 3d printed brackets that hold the RGB panel in behind the coolant pipes to some bracket/strip combos that I could attach the RGB strips to: