DownloadWitch

[Jorrit Scharloo]

Hi everybody!

This is my first mod that I started producing early 2014; a small wooden computer case. Quiet by passive / active cooling combined in an efficient layout.

 

Naming

In my student days we had a very old PC (I think it was an old Pentium II 233 or something like that) in our dormitory that was called DownloadB****. My scratch build will be made of wood. So logically, taking into account the famous “witch logic†of Monty Pythons Holy Grail, the name of this build is therefore called DownloadWitch. This computer started only as a download computer, but now I use it more because of its low noise levels.

 

Starting points

Small, air-cooled, hot swappable. I’m trying to keep DownloadWitch cool with only the CPU- and PSU fan.

 

Components

• Motherboard: Asrock B75M-ITX

• RAM: 2x Transcend JM1333KLN-4G (8GB total)
• CPU: Celeron G540 met Scythe Kozuti 40mm Super-Low Profile
• Cooling: Scythe Slip Stream 120 Slim 1600RPM with fanguard (conrad 189506)
• PSU: Coolermaster Silent Pro M 500W

• 2x SATA Male to Female Adapter (dealextreme SKU 15885)

• 1x PCI-E PCI Express 16X Riser Card Extender Cable Ribbon (ebay toymodel55)
• 1x BitFenix Alchemy ATX 24-pin Extension 30 cm white (sponsored by my employer)
• 1x BitFenix Alchemy 4-pin ATX12V Extension 45 cm white (sponsored by my employer)

• 1x NZXT Front panel connections cable set white (sponsored by my employer)
• 2x StarTech Latching Round SATA Cable 60cm black (sponsored by my employer)

Arranged in DownloadWitch as seen below:

 

 

Tools

• Milling machine (milling time sponsored by Science Centre Delft)
• Laser cutter (laser time sponsored by Science Centre Delft)
• Handsaw, Drill
• Sanding paper, lots of sanding paper

Materials

Personally I like the contemporary combination of electronics and wood / bamboo; the contrast between these materials, when properly used, can make beautiful products. Furthermore I like the feeling / smell of wood. That’s why I want to use wood as my base material for DownloadWitch. Birch plywood is quite affordable and easy to buy at a local hardware store. The main construction of DownloadWitch will be milled on the CNC mill. The exteriour will be made of triplex, with a certain laserpattern to bend the wood.

 

Sponsors

[Jorrit Scharloo]

reserved

[Jorrit Scharloo]

Before I show the production of DownloadWitch, I would like to show the design process behind the scratch build. I designed the case with SketchUp and Alibre, alternating between the two programs.

After some sketches on paper, I modelled some preliminary models in SketchUp. In the below tree the main design evolved from “triangle†-> “triangle box†-> “box†-> “boxroundâ€.

 

 

 

Boxround was the starting point for the rest of the design: a semi-round box; two U-shaped bends interlocking with each other. I would like to investigate the possibilities to bend wood.

 

 

 

The preliminary orientation of the inner components helps with the natural air flow: cool air from below through the CPU cooler and drawn out from the case by the PSU.

 

 

Thanks to my sponsors!

[Jorrit Scharloo]

Focusing on the interior, first thought is where to place the hard disk drives?

The hard disks are connected via an hot swappable sata extender.

First thought was to place the hdd’s vertical as a sort of cartridge system.

This would make the case high and I’m not so fond to protrude the hdd’s through the top wall. Placing them sideways?

Or at the back…, better, but not in the air flow.

The ATX cabling can make the above corner, but it is not really nice, the opening at the corner.

Maybe make a big connector?

Somewhat higher case, hdd’s between motherboard and PSU.

Making room for the sata cables.

Too much pockets, making the side more simple.

 

Thanks to my sponsors!

[Jorrit Scharloo]

Received the cooler for measuring: see the bent plate?

After RMA I received a good one. Pablo is happy

The CPU position on this motherboard almost interferes with a PCI-e Add In card.

The PCI-extender will fix this problem.

Modelling the important parts of motherboard and cooler in Alibre. Fast render of model:

And placing this back into sketchup. Time to change the feet of the case. First try at integrating feet with the interior walls.

Integrating mobo tray as a cable guide.

Mobo walls have no function anymore, time to remove these.

Looking at the side

The PSU connector is too wide, time to reduce. The case is less deep. Reducing the length of the cable guide plates will make them relatively stiffer. In v25 the interior mobo walls are removed.

Wrapping the case exterior around the walls; the window must be small enough to have enough material left to wrap. Thought about lighting the side of the case, but decided not to.

Reducing the feet from 4 to 2 gives a less cluttered look.

Checked to see what effect a dark cable pocket gives, if the pocket bottoms are not nice (paint or something else). I’d rather not paint this. Only production will tell.

Next time: puzzling with hdd placement, pci-e extender, extract fan.

 

 

Thanks to my sponsors!

[Jorrit Scharloo]

 

Moved the pci-e extender from the right side of the case to a position below the PSU.

 

Therefore the sata extender is moved to the right side, this side still needs some more attention. I would like to see no visible fasteners on the outside of the case. The DIN 912 bolts from above picture can be used for magnets on the covering. But these bolts will be made of stainless steel, so extra magnet positions are needed.

 

Turning to the bottom of the case you can see that the I/O plate is too far from the wall, so I made the case smaller.

 

Some changes in the right side.

 

With this smaller case the hard drive is difficult to remove from the sata extender. Therefore a 'cardridge' system for these drives.

 

Turning to the left, the wide sata lane is changed to two small bands, since I will use round sata cables, approximately the same diameter as the sleeved ATX cables.

 

Some more puzzling with a fan and a sort of connector for the front void of the PSU.

 

Finally arrived at the final location of the extract fan: the right side.

 

Moved the hard drives back below the PSU. Hmm..., will block the PSU fan.

 

Back again? Better.

 

Started with the PSU void connector. A strain relief system is incorporated.

 

And a little begin for the hdd brackets.

 

A part of the bracket is fastened to the hard drive to guide this drive and to make a handle accessible from the outside.

 

Next time: final steps of the design process!

 

Thanks to my sponsors!

 

[Jorrit Scharloo]

No visible fasteners on the outside of the case means that the extract fan and grille will not be fixed on the outside plate. First read the following article: http://www.pugetsystems.com/labs/articles/Effects-of-Grill-Patterns-on-Fan-Performance-Noise-107/. Long story short: a standard fan grille works best regarding noise and air flow. Ordered a grille, measured it and modelled in 3D.

Fasteners visible below the cover. An extra fan plate is used to attach the extract fan. This fan plate also keeps the PCI-E riser cable folded and prevents the outside cover from bulging outwards.

Removing this fan plate shows the hard drives. Will need to route the fan cable.

The fan and grille are fixed by nuts between the flanges of the fan and DIN 7991 bolts.

Started drafting the parts in Alibre, looking for possible routes of the cables.

And a first testpattern to bend wood.

Forgot to join the lines, therefore the lasercutter cuts all the lines separate. Whoops!

Problem: the wood will not bend: 2 causes: wrong orientation of the wood fibres with repect to the pattern and wrong angle of the joined cuts. Wood bending by laser cuts is more by local torsion than bending of the wood.

Made a new bending pattern to test next time. Changed the hdd handle in Alibre.

The curved lines looks better.

Almost ready with the design, in the following exploded view all the wooden parts (22 unique items) are shown.

Time to generate NC code (my first time ever). After a few trials finally made compatible NC code for USBCNC (now EdingCNC). Next time first production!

 

Thanks to my sponsors!

 

[Jorrit Scharloo]

Laser time! The new pattern bends!

Assembled the fan plate.

 

My first time ever to use a CNC milling machine. Quite exciting to turn on the milling machine and to check whether my first generated production CNC code actually works!

 

Hopefully I use the correct feed and speed rates. I’ve been reading a lot on this material, even tried to make my own excel sheet to calculate correct feed- and speed rates, until I stumbled upon a trial of GWizard from CNCcookbook.com. This software is quite a handy tool for a beginner like me. Unfortunately I stumbled too late on this software before I generated my first CNC code for the first part. Didn’t even know where in the CAM software I could change the feedrate. Silly me So the first part, milled @ 1000 mm/min @ 10k rpm (minimum rate for the Kress motor) for a 2 flute 6mm mill, z-increment 3mm, takes a lot of time. Manually changing the speed to 300% in USBCNC is still too slow, with a tool change to 3mm it is more than 1 hour production time. But then again, the first rough part is there!

 

The base plate (stock) is made from 12mm Birch plywood. This part must be 9mm thick, so first the top side is milled with a 6mm mill. After that, the finer details are milled with a 3mm mill.

 

 

Unfortunately, the toolsetter is not properly integrated in USBCNC. Also not in the machine, it’s hanging like a weeping willow. See  photo below. So using an alternative way to zero the mill: take a piece of paper and put this between the mill and plate. Gently lower the z-axis till the paper is stuck between mill and plate. Zero enough!

 

 

This ‘paper zeroing’ is not very accurate. A fraction in the offset of the z-direction means that the 6mm and 3mm paths are distinguishable. But very well, this is acceptable in this part. In the picture above the part is already sanded, the part came quite rough out of the milling machine. Never had blisters on my fingers from sanding before, now I do !

 

 

As you can see, two M3 screwed inserts (RAMPA) are already assembled is this part.

 

Till next time!

 

Thanks to my sponsors!

 

[Jorrit Scharloo]

 

Milling Session 2

DW-301D: MOTHERBOARD TRAY

Making my second milled part. Made from 12mm Birch plywood stock. Somewhere in the milling program:

 

The wood splinters at the top side of the stock while using the 6mm mill. The 3mm mill doesn’t do this as much (see the small upright wood fibers in the inner corner). Fortunately, only the stock is ‘damaged’ and not the real part.

 

 

DW-101D: HDD BRACKET and DW-201D: HDD MOUNTPLATE

From 18mm stock plate.

 

The parts come out quite rough (again), so first rough sanding with the sanding machine. Sanding the HDD mountplate:

 

Sanding the HDD bracket.

 

The rest I'll sand at home by hand.

 

At home, my first test is whether the HDD bracket fits in the ventilation plate. Unfortunately, the parts have a too small tolerance. So more sanding by hand.

 

Temporarily fastened a donor hard disk on the mount plate. It fits!

 

This joint assembly also fits!

 

Removed the donor hard disk to further sand the parts.

 

Details of the mountplate before sanding.

 

After sanding with P800. The layers in the wood are more pronounced.

 

Sanding the HDD bracket. Used a small piece of rolled up P800 sand paper to sand the holes.

 

Before and after sanding. The 0,5mm height difference in the bracket is quite splintered. Next time I will avoid such small z-increments.

 

The color difference and the saturation between sata extender and the wooden part is great. I don’t think I’ll be oiling or waxing the wood, because then the difference will be less.

 

 

 

Thanks to my sponsors!

[Jorrit Scharloo]

 

Milling Session 3

 

Time for a new production batch. From left to right: front plate, cable connectors around the PSU bracket and the back plate.

 

 

 

DW-005D: PSU BRACKET

 

First milled part is the PSU bracket. Unfortunately, there's some minor damage on the part. Might it be possible to mitigate these damages with painter's tape? Let's see. Left pocket for placement of the two hdd brackets. The right pocket is for the PSU.

 

DW-006-7D: CABLECONNECTORS

 

Added some painter's tape on the stock material. Hopefully this works better.

 

The painter's tape helped a little. But still some damage on the parts. There's still long wooden fibres at the cutting lines.

 

 

Also the top surface of the part is damaged, fortunately these surfaces are at the inside of the cable connector assembly (not visible when assembled).

 

I’ve sanded the cable connectors and assembled the wooden dowels. The screwed insert gave some problems, the plywood delaminated when screwing the insert in. So filled the loose layers with wood glue and allowed the glue to harden in clamps.

 

DW-002D: BACK PLATE

 

Tried one more time to put painter’s tape on the stock. Tried to change feed and speeds, to lessen the problem of splintered wood at the top surface. The painter’s tape wants to get stuck in the mill, so not a good solution.

 

 

 

But still damage occurs, especially on parts perpendicular to the wood grain of the top surface. Need to find another solution for this problem.

 

 

 

After some hand sanding.

 

 

Initially I wanted a single slot case. But in the night before my milling day I thought a double slot PCI-E case will be more flexible future-wise. I looked in the data sheet about minimal IO openings of the PCI-E standard, but these dimensions are only for the contact area. When checking the part with a real HDMI and VGA connector, the housing of the connector does not fit in the slots. So this part needs a little redesign.

 

DW-001D: FRONT PLATE

 

I also milled the front plate of the case. No big damages here. The hole is for the power switch. Top and back side of the front plate.

 

 

When looking online for other workarounds for the damaged milled top surfaces I finally found a better solution, a negative spiral mill! This will reduce the chance of a damaged top surface. The disadvantage is that the bottom surface may be damaged earlier. But the stock plates are tightly pressed to a temporary plate, so this chance might not be big (I hope). The difference between a positive and negative spiral mill is the direction of the cutting faces. The wooden chips will not go up, but down when milling in the stock with a negative spiral mill. Ordered a 6mm negative spiral mill (2 flutes) and waiting for delivery.

 

Till next time!

 

 

Thanks to my sponsors!