(In Win 901) Asteria II: Rearmoured

[jaffveyda]

Waste not want not
Since the actual shape isn't changing I thought I'd get the LEDs sorted anyway and give it a test.

Cool white 3528 LED strips on a 5mm thick PCB wrapped around an inner frame so the light shines outwards

Just as an experiment, I covered the underside of the light panel with silver foil to see if it'll help with internal reflection.

And then the LED ring is slotted inside the light panel

Note that all wires will be sleeved at a later date


So, let's give it a test!

Yay, ish kinda. The bottom dead spot isn't too much of an issue because the GPU covers that entire side of the motherboard so you won't even see it (shame), but the internal size of the light ring didn't divide to 50mm evenly enough to fill the gap. The corners are frustrating though. I know there are holes in the way so it was never going to be perfect, but I thought the LEDs had a much wider viewing angle than this to help fill the dead spots.

But then also look how the light doesn't fully penetrate the opal acrylic. Opal 050 has 37% light transmission and is perfect for lightboxes, but I guess that only applies if you're lighting underneath not directly into the sides.

So, time to look at other acrylics for my 3rd attempt...

• Opal 030 has 70% light transmission so that should help, but it's more frosted than a solid white
• Polar White Frost looks similar with a 60% transmission
• Crystal Clear Frost (oxymoron there, Perspex?) has 90% transmission, which is likely to be blinding given how damn bright these LEDs are

There are also a matching pair of panels like these for the storage covers which I haven't built the LED rings for, but could suffer the same issues. We'll see when I get to them, but if I'm changing the acrylic for the motherboard then I do want the storage to match when the lights are off as well as on.

Did you get normal ali sheets or bendable grade(it's so much easier to work with), none less great worksmanship. Shame i don't have space in my flat.

 

[LePhuronn]

The aluminium for the case work is 1050A grade at 1.5mm thick. It looks as close to In Win stock as I can get while still bend everything manually with a bench-mounted folder.

(P.S. don't suppose you could remove the quote from your post? Image quoting doesn't go down too well on these forums. Cheers)

 

[Cenedd]

Couple of ideas for you.
Bending: rather than drilling a line of holes, if you took a steel rule and a scriber (or some other hard metal point) you could probably score ali enough that it would help the bend - also keep it straight.
LEDs: If the LEDs are pointing outwards, the only light coming upwards through the acrylic will be that which is scattered internally. Since you have a reflective backing (I'm assuming the other side is more mirror'y than the backing) would it help if you angled the LEDs downwards slightly so that the light bounced off it more and then exited upwards?
As for getting them to fit the circumference and the corners, I can give you four ideas:
1. remake the piece but change the inner circumference so that it divides nicely into the length of the LED strips.
2. overlap the strip - being careful not to short out the terminals if it isn't an IP-rated (covered) type. You could do this in smaller sections to get into the corners...but it'd be fiddly.
3. change the size of LEDs on the strip. You can get smaller ones that are more densely packed. As long as they work with your supply (or you can change) you're ok.
4. drill loads of holes for individual LEDs and go quietly (or otherwise) insane wiring the delightful things up. This may be less onerous with single colour than with RGB LEDs! <twitch>

Gareth

 

[LePhuronn]

Cheers Cenedd.

The original test with the clear panel and sanding it down to get the frost effect worked quite nicely, but I couldn't get an even enough finish and had to recut it anyway, hence going Opal. And now I'm cutting a 3rd version, I'll rejig the internal dimensions to get an accurate length for the LED strips. I won't be changing the strips though as it took a while to find something with a 5mm wide PCB (everything else has 8mm, 10mm or even wider PCBs) because I'm space-limited

Still, nice to get a tangible test done so I can see where the issues and revisions need to be.

 

[LePhuronn]

Scoring is a good shout actually. I did originally try to slice in a bit with a hacksaw or file in with the edge of a square rasp, but didn't get very far. The alu is soft enough to get some gouging going on (as is clear by the state of the blanking plates ) so I'll get a cheapo craft knife on it or find a decent scribe.

I have backups of the front-side plates with the power button, but I think I'm out of the plain back-side ones. I'll see how they look when cleaned up and might redo.

 

[LePhuronn]

Drive covers

Yeah, these work nicely. Corners are still a touch dark because of the holes, but I'm experimenting with clear nylon screws since there is no load bearing on these covers, so hopefully having some clear material physically in the holes will transmit light back into the acrylic and light the corners up.

We'll see when the panels get threaded.

 

[Cenedd]

Now those are looking nice!

 

[LePhuronn]

Cheers dude, rather happy with those. I'll still recut them from whatever acrylic I go with for the motherboard rework just so things match when the lights are off.

Top layer is aluminium and will be brushed to match the Titan's backplate and the case exterior. Right now the logos are just cutouts, so I'm going to play with the glue gun and see if I can fill them in need some very delicate clean-up though to straighten some of the lines and remove spurs and spikes - unlike the rest of the metalwork I only have 1 set of these logo panels so I can't mess it up like I've done everything else!

 

[bobbarley]

Love how your build is progressing!

 

[LePhuronn]

Love how your build is progressing!

I better hurry up then, ain't I cheers dude, appreciate your interest and support.

 

[LePhuronn]

Greetings to all!

You remember me talking about that custom DDC top that would solve many pump placement problems? Well it kinda did and didn't work. Can it be done? Yes. Can I do it? Not so much, but this was exploration after all.

My big problem was actually cutting the acetal block to size and maintaining a perfect square. By the time I'd managed to straighten things up the actual block was a tiny bit too small in some dimensions. Not a deal breaker potentially so I continued the experiment nonetheless.

Let's get going!

When I made my first batch of light panels, I also cut some drill guides for the various holes and feeds for the pump top. Alas, no pictures of them in isolation, but this is the 3.3mm guide set for the volute side of the top in situ: 4 holes for M4 mounting threads and a pilot for the central feed.

The depth gauge on my pillar drill isn't amazing so I marked the required depth onto the drill bit with some masking tap.


I built the volute up in stages. First up is a 10mm drill to feed into the volute itself. Got my air compressor ready and waiting to disperse debris and offer a bit of cooling to the acetal.

Acetal is lovely to work with. You really only need woodworking tools, so nothing special, but you need to treat it with a bit of respect. Heat can build up very quickly and acetal goes stringy rather than producing chips, so you need to get that rubbish out of the way quickly so you don't gum up your drills, holes and make a mess. Best practice from what I've read is to have your drill speed rather low and then "peck" at the material as you feed. That way you don't get long strings building up and are easily removed. And the airflow also helps a bit with temperatures too.

Another reason I built this in stages is to maximise the efficiency of the Forstner bits I'm using. Forstner bits scrape away at material much like a spade bit; given the bit spins faster at the edges than the centre, I've found it's quite effective to remove the inner portions first, letting all the grunt apply to the outer parts of the bit.

So with 10mm removed first, the 20mm Forstner bit only has the outermost 10mm to scrape out. Turns out I didn't actually take a picture of the 20mm bit in place before milling.

That's how stringy things can get, and in this instance no amount of air helped disperse it. So lots of pecking and pulling the strings off to get this:

Now comes the big boy: slightly oversized 40mm bit (comes out at 41mm)

Even with only milling the outermost 20mm, no finesse with this pupper at all. The mess!

Did the job quite roughly too, so some clean-up of the edges required.

And this is where I stopped.

The M4 threads went in nicely first time but when I mounted this to the pump for a first fit I noticed that either a drill hole had slipped or I still hadn't squared everything up correctly, and the top was slightly out of alignment. I couldn't take any more material out to make it work, but the first half of the experiment was successful in that it's very easy to drill and mill a block of acetal.

I'll go back at some point and put some ports in just to test how much performance is lost with a circular volute, but in all honesty the thing that I need to address is getting square blocks. My jigsaw was pretty worthless cutting through 20mm, but a tabletop scroll or band saw might do a better job if I can guide, fence and cool properly. Or I could buy in pre-cut blocks. Ultimately getting the thing CNC milled is the best way to go, but at that point I might as well design a proper spiral volute and do this right.

For now it's plan B.

 

[LePhuronn]

Plan B is to modify my existing Aquacomputer pump top to add a new inlet port on the opposite side of the outlet to achieve my required crossflow design. The volute is already spiral, the internal milling and feeds are there, I just need to smash in a new port. This does also mean that both of the top ports need to be covered, so that's a new cover plate too.

Break out the acrylic guides and drill a 6mm inlet feed into the pump's collection chamber.

Lined up and ready for the 11.6mm drill for the port

Stringy mess, lovely black contrast to the white from earlier.

And thread that G1/4 (don't worry, I straightened the tap up after taking the picture )

Success! Inner part of an EK HDC 12mm in place

Because this top has a side inlet port, Aquacomputer milled out this intricate collection chamber in the upper half of the acetal block. All I did was pass a 6mm feed into that chamber and then offset the actual port down a few mm so the fitting will be as flush to the top of the pump top as possible.

Now close off those top ports with a plate since I don't have the space to use normal screw plugs.

I never actually intended to use this acrylic plate, otherwise I would've engraved the logo in the correct orientation but in practical terms the acrylic is only 1.5mm thick so it's actually bending in the middle rather than applying pressure to the o-rings. So since it'll probably leak, I guess we're on the stainless steel.

Rough shape cut from 1.2mm 304 shiny stainless (for no other reason the shiny sheet was £1 from eBay)

Shaping. Some of the corners would cut, others wouldn't so there was A LOT of rasping and filing down. And it's stainless steel so it's not exactly a quick job.

Drilling the M3 holes. My poor drill bits! Simple HSS bits and stainless steel don't play nicely together.

But after quite some time, aching arms and a sore back, I got there.

The stock Aquacover, my first version single port laser cut and the 2nd version no-port variant cut by hand. Crikey.


And after some tidying up and expanding the screw holes, we get this!

I've done a test fit and I have literally zero space behind the Titan given that the HDC fitting in the new port raises up 2mm above the top surface. But the important thing is this crossflow design removes all of my position restrictions and greatly cleans up my loop run in and out of the pump. That's a massive relief.

This top will do for now, however I've hooked up with Lucas from Mp5works who's milled me some aluminium pieces (update coming) so there will be a conversation coming about a 2nd attempt at the proper bespoke top. If that comes to anything I'll certainly let you know

Thanks for reading as always, hopefully a bit more to come soon!

 

[Cenedd]

Looking good. Probably far too late as you seem to have a working plan B...but could be useful for future stuff.

Squaring blocks using only a vise: https://youtu.be/-lgMcDOkeg8

For stainless, you want to treat yourself to some highly expensive cobalt drill bits. They handle the heat better so they don't dull. Water is a perfectly good coolant to keep the heat down - a plant mister or a water pistol would do. These 3mm bits are a whole £3 for a pair including delivery: Excellent quality and they're stub length (rather than the usual jobber length) so they don't flex as much - less prone to wander. Whole range of sizes available too: store

 

[LePhuronn]

That video is awesome Cenedd. I wish I had the kit to pull that off. My problem was my mitre block is a little loose so all supposed 90° angles were out just enough so none of the hand saw cuts were sufficiently accurate.

I knew HSS bits weren't really up to the task of drilling stainless, but I don't use my 2.8mm bits often so I saw them as disposal rather than investing in some cobalt tips. Nice to see a cheap set should I need them in the future!

 

[NoustieChris]

Top notch workmanship on this dude.
Can't wait to see the finished article

 

[LePhuronn]

...Can't wait to see the finished article

Nor me, its been going on for years! My Skylake is looking pretty long in the tooth now

Many thanks.

 

[Cenedd]

Those cobalt bits work very nicely in acrylic too - they're sharp. And they do an 11.8mm bit for your G1/4 tap. I tried the original imperial size spec'd for G1/4 (28/64ths or about 11.51mm) and I think the end result was a little bit tight - fine on some fittings but not on others.
I know last time I tried I got away with a 35mm Forstner hole from nothing in 10mm acrylic but my bits are Bormax with the sharp saw teeth down the side. Slow speed; few hundred rpm if I remember correctly - too fast and it melts. Your biggun looks like it's got carbide teeth which are hard and tolerate lots of heat but often not very sharp.
You can do it with a rotary table - but they're pricey - or a trepanning tool if you can find/make one.

Cut and shape of the top plate looks good though

 

[LePhuronn]

Hi there, a little bit more for you before I, no doubt, disappear for months again

A replacement PSU mount and version 3 reservoir mount, beautifully milled in the same 1.5mm aluminium by Mp5works. Many thanks, Lucas! The focus for this post is the res mount.

It's been commented on a few times "I can't wait to see how you're mounting the res" given I've said I wanted something unobtrusive, almost invisible to give a floating appearance. For my money all the res mounts I've seen are a bit unwieldy and no finesse to them. Hopefully this will achieve the look I want.

So, folded up and mounted to the Aqualis ECO 100:

Simple, out of the way. In what can only be described as a miracle, my intended millimetre precision has actually worked for once!

The plate matched perfectly to the chamfer of the acetal base, so no unsightly overhang, and the vertical portion just close enough to give clearance without a massive gulf. Those chunky M4 screws will be replaced with slimmer button heads for the final build.

So, in situ then:

Positioning is about 90% accurate here. I need to be careful because I don't have much room below the res and the Titan, and I can't push it up too far otherwise the 90 degree fitting on top fouls the roof of the case chamber. Dead centre in the chamber though, and will be coming away from the PSU area by 4 or 5mm.

I'd say that was pretty damn invisible

Since this is only 1.5mm aluminium, it's not going to hold up to much abuse and weight as-is. That's borosilicate glass for a start, and then add in 100ml of water too. You can see it sags a little already with the unsupported test placement. The mount itself will have a bit more support from being screwed into a M3 standoff a bit further back (that small hole underneath the vertical portion), and the res itself will be supported at the top.

The idea is that acrylic ring will be screwed into the roof of the case chamber with a chunky M4 bolt and hold the res top vertically in place, along these lines:

Should help stop the res from tipping sideways along the mount's fold line. After some proper mounting I'll test it with water and give the case a wiggle to see what happens. I already have a dual-screw version drawn up in case I need more support, but will have to make sure it can't be seen (single screw version will be obscured by the 90 degree fitting).

Failing that I'll just use and abuse the 2nd G1/4 port in the res top and screw the damn thing directly into the roof

So that's it from me for a while. I'm currently solving a conundrum with my storage plate assembly to try and thin it down a bit; with everything stacked up it's currently about 3mm too tall, so I have some refinements to do. That will be helped by how far I have to trim the DDC heatsink's fins, which in turn dictates how much I can space out the glass side panels from the main body.

We're getting there slowly, thanks for being here!

 

[Cenedd]

Nice. <face>I love it when a plan comes together!</face>
A quick spray from a satin black rattle can and that aluminium circle supporting the res will show up even less....not that it clashes with the aluminium in the rest of the build.

 

[LePhuronn]

Bear in mind it's all a little misleading right now because everything will be black powder coated to get that stock look back. But yes, once the case is black and the res mount is black and the acrylic ring is black it should all nicely vanish from sight