Acrylic Pyramid - the story, building and some testing - Overclock.net

dragosmp · 07-17-09

How it all began

Why an acrylic Pyramid? The idea actually occurred to me 2 years ago while talking to my lady about what could be innovated in regards to the fixed computing. After all, 90% of the users care more about the idea to have a good, silent and reliable computer to depend upon for different tasks they may imagine. Cuteness is an added bonus, which nowadays seems to actually be the differentiating factor.

This post will begin with a small introduction, then I'll show you the first version of the pyramid which didn't last long and then the real deal, the Pyramid v2.0: building, refining and testing.

Background info

Computer hardware reached a stage of maturity in which mostly any system is more or less stable (unless O/Cing or testing new settings in the BIOS) and even a cheap dual core CPU is plenty for the home users. So we end up with two factors, where apparently not enough progress has been made: silent computing and cuteness. Usually silent computing means low power consumption, so an added bonus could be the project might be environmental-friendly.

So my virtual rig is based on a decent dual core CPU, is reliable, but most of all it is silent and cute. The first idea was a sealed box with a tray on top where it’s supposed to have grass growing. The entire rig was supposed to look like a flowerpot, but underneath would be laying a high end media center PC – obviously it didn’t last long.

The Pyramid V1

Then it occurred to us the idea with a pyramid. It seemed like a good plan from two reasons: I’ve never seen one, and it seemed doable with reasonable low cost materials. This project, that I’ll call the first pyramid (or v1.0) was built on a steel frame for resistance and was covered with pine wood for sound-proofing. It used standard desktop computer components like an AMD µATX board, a single core CPU I had lying around, 3.5” desktop hard drive and a PSU that had all the cover removed, it was really only the PCB and the radiators left.

To cool this, I used a thermal-regulated 120mm fan that varied between 800RPM at startup and 2200RPM if the CPU radiator heated over 50°C. Needless to say, it didn’t work quite as planned; the cooling part of the pyramid was a severe issue that I didn’t imagine from the beginning. Since all the radiators were low cost stock heatsinks, they simply didn’t cope with the reduced airflow. Moreover the PSU (at the origins a 480W no-name) had an average load of about 80W, so its efficiency was horrendous (around 60%), so in actual I had something like 140W of heat with nowhere to go. Physics shows its nose where you least expect and the heat, following basic thermodynamics principles, went up in the cone of the pyramid creating a heat bag. The result was the pyramid could work for 2-3 hours before overhating; the result was a lack of stability, one of the primary objectives after all.

Here are a few pics with what happened, and how we managed to find a good use for the first pyramid case:

Yep, it became the home of the rabbit Nasuc and the guinea pig Rontz. Both are enjoying it a lot, especially around the edges where they can test their rodent teeth.

So what have we leaned? 2007 desktop parts weren’t power efficient enough for this adventure, and a better cooling system had to be deployed. Fast forward to 2009 then to the second Pyramid.

The Pyramid V2 (the actual build)

After the first try it took a while until I managed to find the time to build another Pyramid. The project began with a white sheet of paper with the same objectives: it’s supposed to be fast enough, stable, silent and cute.

I decided this would an acrylic build for two reasons: it’s easier to mold than steel (duh) and it’s even cuter. As for the computer components, I decided to go for a Mini-ITX motherboard Gigabyte GA-GC-330UD 17x17 cm and a low power CPU, the Intel ATOM 330. The question is: is the CPU and motherboard combo good enough for daily tasks? It’s supposed to replace my girl’s laptop (that was being used mostly @home) which was based on an AMD Turion x2 1.7GHz. The usage pattern: Firefox 70% of the time, Office 20% of the time, and movies the rest of 10%. An ATOM 330 can take care of this easily.

An alternative was offered by Nvidia’s 9300m chipset, also known as the Nvidia ION, which when paired with an ATOM 330 is capable of Full-HD 1080p playback. It’s a good option for a HTPC, but I’m not aiming for that; besides the added cost for the ION platform is huge (around triple) – so no ION for this pyramid. It might be included in the next version, who knows.

The components are listed here:
·Plexi case:
o1 sheet 4mm thick plexi 1m by 50cm – for the lateral triangles
o1 sheet 10mm thick plexi 1m by 50cm – bottom square
o1 sheet 1mm thick plexi 20cm by 10cm – hinges
ogummy soft plastic used for the legs
omagnetic hinges to stick the bottom to the upper pyramid (very Apple-esque, wouldn’t you say?)
oacrylic glue
osaw, drill, etc
·Computer components:
oGigabyte GA-GC-330UD motherboard with Intel ATOM 330 CPU – Ultra Durable 3 series
o2GB RAM DDR2 PC 6400 5-5-5-15 Samsung
oHard disk 2.5” 5400RPM Samsung HM320JI 320GB
oPico-ITX DC/DC 90W ATX 20 pin PSU 95% efficiency
oPower brick 120W from a Compaq R3000 laptop
·Cooling:
oHR-05 SLI used to cool the NB (the hottest chip @22W)
o2x Zalman ZM-NB4 Northbridge cooler used for the ATOM 330 (8W) and the ICH7 southbridge (5W).
oModified Evercool 80mm fan (800rpm @5v – 2300RPM @12V)

The actual building:

It all took around 2 full working days, but take this into account: I have never worked with plexi, all I knew was from Wikipedia and other places on the web. Two days is a week-end, it doesn’t seem all that much anyway.

Building the case

I began with the cutting and gluing the outer case parts. For the triangles and hinges I use the cutter method of cutting the plexi, as for the bottom square I used a fine saw. All the edges were sandpapered afterward to remove the debris.

In order to bend the thin 1mm plexi into hinges, I first cut 1cm by 2cm rectangles. Then I build from a steel sheet something like a mold which was bent at exactly the angle between the triangles. Finally I heated above a candle each rectangle and after it became soft I pressed it against the steel mold to get a hinge with the desired angle.

Important: you can also use a heat gun if you have one, but if using the fire method, remember not to remove the protective coating from the plexi before heating it – the smoke will gather as a black layer on the coating, and after you finished shaping the hinge you can simply remove it to leave a nice and shiny aspect. Additional black traits can be removed with some sand paper if needed.

In the same fashion I made the hinges to fix the fan, and the hard disk, both of which will be suspended from the sides of the pyramid. To fix the HDD hinges, in stead of measuring the distances between the hinges and the edge of the triangle face, I simply used a dummy HDD to have the hinges fixed onto at exactly the right place.

Gluing plexi isn’t easy, as you probably know. It’s a non-porous material, and all glues have problems adhering to its surface. There are chemical glues (some even suggest chloroform), but they weren’t used in this build. In stead I made a trick: using my multi function drill-gadget I scratched the surface of the plexi in the place I wanted to glue the parts. This way the surface became porous and the glue had no problem in keeping the things together.

I also placed 4 vent holes of trapezoidal shape near the top of each triangle; the cold air will enter through just on top of the fan. For the back plate ports I’ve cut a rectangle to be able to access most of the ports (safe for the LPT one).

Here’s how it worked out:

In the last picture you can see the magnetic hinges that have not yet been glued/ fixed. The white plastic part will be glued to the sides of the upper pyramid, while the metallic part will be fixed with screws to the bottom part. They are extremely powerful magnets; they can hold 2kg each with little problems. I’ve decided to put only 3, I might have put all the pack of 8, but there is no need for that.

Modifying the motherboard

This is home territory to me, so this went fast. The basic is on this motherboard there are 3 chips which consume (and thus dissipate heat) around 22W NB, 8W CPU and 5W SB. By this reasoning the NB got the HR-05 SLI, and the rest got the Zalman NB radiators. Good cooling means there is no need for a lot of airflow to keep these chips cooled - silence and points scored on the cute factor.

Here are some pics:

Testing

The first obstacle is that for some reason the Pico-ITX PSU hasn’t arrived yet. I found an old 300W PSU lying in the closet, so I went on to modifying it (aka reduce the number of wires in the ATX plug) to be able to power the pyramid. The thing is the 20 pin ATX connector has numerous black ground wires, red 5V wires and orange 3.3V wires. You would think they have some use, but for testing purposes I simply cut off all but one ground, +5V and 3.3V wires and connected together the ones from the ATX plug, so that the motherboard wouldn’t see any change. Did it work?

Yes it did, while I’m a bit confuse about the +3.3Vrail that now is @4V… but for testing purposes it worked. I’ve held the system in the BIOS for around 10 minutes (similar to full charge) and the CPU temp stabilized at 36°C and the fan at 1300RPM, where it’s quite silent.

Now I’m waiting for the DC/DC Pico-ITX PSU to arrive, it should arrive next week (hopefully). It’s been a great experience, now I’m waiting for your comments, suggestions for improvements, questions or whatever!

D.

______________________________________________
My original build log – Pyramid V1 2007-2008

Daney · 07-17-09

Great dude!

I've been waiting for you to make a thread about it, +rep.

What was the reason that made you change the heatsinks on the NB, SB and CPU? I thought atom build hardware produced very little heat.

dragosmp · 07-17-09

Yes, they do produce very little heat, but even that heat must be dissipated somehow. The NB had this tiny fan spinning at over 2000RPM, so that had to go. The CPU might have had a good enough heatsink, while the SB had one that was just very very small.

Since my only fan is a 80mm low RPM, the airflow is much diluted on the lower part of the pyramid and things could overheat. Besides, the HR-05 SLI was gathering dust since I sold my sk 939 DFI Nforce 4 Ultra mobo, and the 2 Zalman heatsinks were 6€

ModMinded · 07-18-09

Cool, you should link to this in the case mods forum as well!

dragosmp · 07-18-09

Cheers mate, I'll see what I can do to move the thread to the case modd section, looks like there's the right place for it.

*EDIT: We're here to stay

dragosmp · 08-15-09

…and the Pico-ITX PSU comes into action

After a month of waiting (I’m not kidding) my Pico-ITX PSU finally came. The specs and some pics are below:

120W at the ATX plug and around 93% efficiency for any charge
between 12V and 24V DC input power

The thing is it doesn’t have a 4pin ATX 2.0 plug, but I happened to have one in a box which can be connected to a MOLEX plug. Here is how the pyramid looks with the PSU and all the wiring mounted:

It’s crowded and the cable management absolutely sucks; in my defense I’ll say that it’s for pure testing purpose: to see if it BOOTs, if it loads Windows and if it doesn’t outright burst into flames. Later I’ll cut’n’mod all the wiring (including the SATA cable) and make them as sleek as possible.

So here’s the first Windows install, an USB Windows 7 Ultimate x64 RC. All went well after I managed to make the motherboard actually “see” the HDD, I had to set the HDD mode to “Enhanced” – nothing else worked. BIOS pics galore:

The last pic shows some overclocking options. I didn’t really have time to test, I had about 2 hours to do all I wrote in this post, but while clicking to see if it actually works, I tried to disable “indexing”. The CPUs spiked @80% and stayed there for about 5 minutes, while the CPU went at around 40°C (BIOS read after restart).

CoreTemp reported something like 55°-60° - the time to restart the CPUs may have cooled a bit, but 20°C difference is maybe too much? If you happen to know some monitoring software that supports the ATOM 330 properly, please leave a note. Till then, I've just placed a -16°C offset in CoreTemp and now the BIOS and the utility seem to be around the same level.

Catch you later.

dragosmp · 08-15-09

Now I'm actually typing on it the pyramid, it's in testing phase.

Earlier today I managed to sort out the cable management and even to run 3DMark 2001.

It didn't brake any record, come to think of for the 3D standpoint it may actually be weaker than my first real OCer rig: a Duron 750 and a GeForce 2MX 400 with 32MB of Vram. The pyramid consumes probably less then a tenth than what was my high end rig at that time.

reustache1 · 08-15-09

nu mai bine pui iepurele sa sufle in piramida sa o raceasca?!

p.s: am vorbit codat pentru 95% din cei de aci!

)

Beat · 08-16-09

Pretty sick idea man! I like it a lot.

Very sweet indeed.

Dylan · 08-18-09

Looking good.

Subb'd

		Overclock.net - Overclocking.net > Specialty Builds > Silent Computing
Acrylic Pyramid - the story, building and some testing

08-15-09	#8 (permalink)
reustache1 New to Overclock.net Join Date: Aug 2009 Posts: 1 Rep: 0 Unique Rep: 0 Trader Rating: 0	nu mai bine pui iepurele sa sufle in piramida sa o raceasca?! p.s: am vorbit codat pentru 95% din cei de aci! )

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)