Omnifront.net‎ > ‎

Project Prism: A Computer Build Log

Here is a detailed build log of the latest incarnation of my personal computer: Prism. This is an ongoing project... that is to say... it is not done yet. Each time I make some additions to the computer, I plan on documenting and uploading pictures and such to the site. Pick a segment from the left to get started, or you can just browse the posts below, in reverse chronological order!


Also, a special thanks to my sponsors, for believing in this build and all that it can become!



Bitspower is sending a 'surprise' package of equipment, very excited to see what shows up!



Hughes is a client of mine for my day job, but they graciously supplied me with all the extra Aluminum I want, and access to pressure molds should I want them! Sweet!








Sleeving

posted Jun 5, 2014, 7:18 PM by Andrew Stock

It's been so long! Time flies when you're having fun I guess. 

Life has been busy, as always, but I've been fortunate enough to get pockets of time here and there to make some progress. I've piled up a few pictures, so it's time to share!

First, as indicated last time, I went ahead and purchased a fan to install on the back of my hard drive cage. It took quite a bit of research to find one that would fit considering the tight space, but I located a nice low profile 140mm fan that fit. I thought about a few ways to attach it, but decided to settle on simply tapping a few holes in the back of my drive cage, so that I could screw it flush against the surface. The area is completely invisible once the cage is in place, so I could have duct taped it  on for all it mattered, but this 'felt' nicer. 



Unfortunately, all the pictures of the cage assembled with the fan seem to have become corrupt, and my drive cage is now assembled and in use in the case... so no pictures for now... but the next time I get it apart I promise I'll share! 

Next, I decided to move on to some sleeving, at last! I started with the pumps, since they seemed pretty easy... just two sleeves each. Well... it wasn't as easy as I thought.  I did an initial sleeving with the shrink just running up to the pump... but it didn't look right, and you could see a bit of the colored wires where the sleeve met the pump. Time to take it apart!




The second attempt looked much better. The shrink is a bit crinkled up by the pump, but I'm not so sure there's anything to do about that... the hole in the pump housing is very tight, and practically crimps the wires by themselves, not to mention when they are shrinked. Not bad otherwise though, I'm pretty happy with it, for my first shot at sleeving. 




Next, I decided to make myself a custom USB to 4-pin motherboard header cable. This will run from my motherboard USB header down to my arduino hiding under the radiator housing, which will control my pumps/fans/etc. I started by disassembling an unused USB bracket, and cutting off the header portion of one of the cables.



Next, I took a regular old USB printer cable and cut off the "A" side of it, then stripped the wires:



A quick bit of soldering and a little electrical tape later:



I then proceeded to shrink over that part of the cable to tidy it up, reattached the USB header, and shrinked the head for good measure. 




I didn't think to sleeve this cable, especially since it will be completely hidden... but maybe I'll go back and sleeve it anyway, just for the heck of it, if I have some spare sleeve left over. I could also go as far as buying an empty USB header and some shielded USB cable, and cut myself a completely custom cable, but I digress... I have many other important things to get to first. 

Next, I decided to move on to sleeving some of my PSU cables. Many of these were nice enough looking... but why stop at nice enough? Surely I can do better. 

Appologies for the photos... they were taken late at night, under a desk lamp. 



I drew up some diagrams, to keep track of the pin locations. One thing I needed to do was "invert" the SATA power heads on this first cable, so that they were facing the right way as installed on my drive bay. The hard drives are inverted in the bay, so the power cables need to be inverted too! After figuring out what the wires needed to do, I measured and cut up one of the spare cables I had.



Next, I prepared my lengths of sleeving. Having no experience, I experimented a bit with the best way to get a nice looking end that isn't all bloated or frayed. What I found worked best was to hold the sleeve between my index finger and thumb where I wanted it to be cut, and use a very sharp pair of scissors to cut right next to my fingers.



Then, while still holding, I would use a lighter and heat up the end millimeter or two for about two seconds, enough to melt it just a little. Then, I would use my other hand and roll the slightly melted end between my index finger and thumb. This kept the end from fraying while keeping it about the same diameter as the rest of the un-melted sleeve.



After that, I'd push the cable through the sleeve, leaving about 5mm between the end of the sleeve and the beginning of the pin crimp:




Then, following the wonderful illustrative guides posted by Nils, I used some 15mm pre-cut heatshrink and placed it with the leading edge covering up to the head of the pin, and heated it to shrink it about 50% of the way there, then made sure it was exactly in the right place before shrinking it the rest of the way. At this point, if it was still in the right place, I would overheat it slightly, count to 5 (while lining up the pin with the plastic housing), and insert the pin into the housing... being careful not to get the leading edge of the still-warm shrink caught on the housing. That happened a lot. 



Not too bad!  Time for attempt #2!



Still lookin' good! Multiply that by about 50 more times... and that was my evening.  After completing the PSU end of the cable, I moved on to making the custom SATA "bridge". To hook it up to my drive cage, I needed to fit four SATA power connectors in relatively rapid succession. I took a while to do some research on various was that people decided to sleeve between sata connectors, but decided to go with shrink and sleeve. 




To sleeve this part, I first crimped on a SATA power head, then fit a 15mm shrink and a small piece of sleeve onto the cable. I locked the sleeve in place with the first piece of shrink, then slid it tight against the SATA head. 

 

After I had all five cables in place, I crimped the next head on:



I had some trouble initially getting these cables crimped on... I think my cable that came with my PSU had insulation that was slightly tougher or thicker than usual.  Anyway, I ended up using a small hex-headed screwdriver to force the cable down into the blades of each section, which seemed to work pretty well.



Here's halfway through the process of that cable:







Not terrible!  Some of the sleeve didn't end up stretching quite as much as the other pieces, but all in all considering this is my first go at a lot of this, I'm pretty happy with it. I'm not exactly sure how I want to approach v2 of the drive cage just yet, but I anticipate I'll have to redo this cable in order to draw power to some of the additional features, so this is certainly good enough for now. 

A few hours later:




And here it is installed:



And here's the underside, where it meets the PSU:



Actually very happy with how it came out! It's by no means perfect, but I'm satisfied for a first go as a complete sleeving 'novice'. 

And just because I see so many people do it, here's some sleeving gymnastics, to show how tight the stretch ended up being. 








With some balancing, it would stand up on end... but it was rather hard to get a picture of that considering how tall it was. 


Now, with that done, on to the next most challenging sleeving project... PCIe power cables! Why, might you ask, were these more challenging? Because of THIS:



Yep... "Y" cables were a part of it. 

I researched a bit online, but I couldn't really find a solution that looked 'right' to me, so I decided to go it on my own and just try a few things. Fortunately, I have a stretch of this cable that will be smashed between the motherboard and the motherboard tray... so it will be completely invisible... in case it comes out terrible.  I started by putting some single-wire sleeve over both of the cables, shrinking it in place at the end of the cable where they are crimped into one pin, then stretching it as tight as possible and clamping it in place with a locking wrench:




Next, I put a piece of sleeve up at the end of the wire (for later!), then I cut two equal lengths of sleeve and put them over the two separate wires leading away from the clamp, forming the "Y" halfway along the cable:



Next, I used my lighter to heat up the "Y" section, and mashed it together with my finger and thumb (in a glove... too hot!), to get it kinda held together. I read in some places that people used superglue to hold these sections together, but not having any, I went for melting. 



Next, I used that spare piece of sleeve that I put on earlier and slid it over the whole "Y" section, then shrank it in place:



Repeat that twice, along with several hours of regular sleeving:



Not bad, getting better with each attempt! The PCIe header on the right is my first go, the one on the left was my second. Noticeably better! 




I'd say it was a pretty good improvement from the "lazy" sleeving done by the manufacturer (on the left)!



Lastly, I made a few little odds and ends for connecting various headers inside the case. Sleeved for practice, why not!



And with all that done, I actually finally took my PC apart and got it back into my case! It's LITERALLY been years since it was reassembled! 

Still lots to do, nowhere near done! Until next time! 


Drive Cage, Complete!

posted Mar 15, 2014, 8:12 PM by Andrew Stock

Looking more and more like my life is allowing for monthly updates... sorry for the long delay.  I've worked little bits and pieces each weekend for a few hours at a time... so while I haven't shared in a while, at least I saved it up and have a large bit to share now!  Anyway, on to the fun...

Last we left the drive cage, I had cut and shaped most of the pieces, but they were just leaning together... nothing solid yet, and pretty rough. I decided the best approach would be to glue it together first, then use my dremel, sandpaper, etc to clean up some of the edges. Here it is with the inside wall glued in place, and the two supporting beams glued as well.



The two clear beams will serve not only to strengthen the inside wall (without them, only the edges would be glued to the "U", not very much surface area for the kind of weight I would be putting on it...) but they will also be the mounting points that will attach the drive cage to my case... more on that later.

With everything glued in place, I had some excess scrap to cut off the "U" (I left it intentionally, to give myself lots of room, as the measurements pre-bending were a little too hard to figure out on paper. ). I used some simple masking tape to mark the line I wanted to cut, then decided my jigsaw would be the best tool to cut off that edge. My table saw was a little too risky given all the work I put into the piece so far, and using a manual hand frame scroll saw would take ages. 






I left just a bit of an edge, as I didn't want to accidentally cut into my support beams. I then went back over this edge with my dremel and a sanding drum attachment to grind it down nice and flush. Next, I moved on to drilling some holes in the support beams to seat more press-fit screw housings into, using the trusty old drill press:




Initial fit test, all screwed in and looking just fine!



And from the back side:



Now, it's all screwed in, but it's still not much of a drive cage yet... time to get to work on planning the most important piece... the backplane!



First, I did quite a bit of measuring. Hours and hours of it, in fact. To get the drives to slide in nicely, I needed "sub millimeter" precision... the slightest bit to the left or right, and the drive wouldn't seat correctly. 

I ended up taking the approach of measuring it all out on a scrap piece, and giving the cut a few attempts, tweaking the measurements and re-cutting as I iterated through attempts. A little archaic, but it guaranteed a reliable/reproduce-able result in the end. Here is my first go at measuring out the cut points. In this first iteration, I was cutting out a block large enough to allow room for the connectors to stick through, along with all the pins, some of which you can see stick out of the PCB a bit, requiring room. 




After measuring it out a bit, I cut it out using the drill press to create some pilot holes, and the jigsaw to finish it off. The result...



Not so pretty. Fail.  The Jigsaw proved too squirrely for these small parts, and the large cutout was not the look I was going for... I wanted something a bit cleaner looking. At this point, I decided to drop the old metal ruler, and go for my high precision digital calipers, with 'actual' sub-millimeter measurements!



... alas, I'll spare you the multiple additional iterations of attempts I made before I finally got the measurements and process right. 

Here's the final backplane, all measured and marked, with pilot holes cut, ready to finish the openings for the SATA passthrough ports:



The biggest challenge in getting all this measured out was that the bays are not exactly perfectly spaced... one millimeter off here and there... which makes very little difference as I gave myself about 20mm of space between drives for slack... but makes a huge difference for the backplane, which all four drives need to seat against simultaneously. I measured and re-measured these lines more than 20 times... partly because I wanted them to be exact, and partly because at this point, I wasn't exactly sure how I wanted to proceed...

I could use the hand scroll saw, which would take hours, but I would likely end up with a fairly accurate cut. A power scroll saw would be perfect, but I don't have one of them yet unfortunately...

I could use my dremel with a cut off wheel... but the chance of messing it up would be pretty high, and the cut wouldn't be perpendicular to the surface of the acrylic, which would be pretty important for this high precision piece. 

I could use a hand router, but I didn't have any bits that would make this small of a cut. and I felt like it would be too overpowered/not precise enough at making straight lines. 

I had a tiny router bit (a downward spiral bit) that came with my dremel, but I've never been able to get it to do a straight line correctly. Hmm...  If only I had some way of using the tiny dremel bit, with a big machine that will hold perfectly steady. Then I came up with a wacky idea... 



Put the dremel bit into my drill press, and use it as some kind of hybrid table router / scroll saw!  I cranked up the speed on my press to make sure that I was getting a very quick cut... the drill press wasn't made for this kind of "side to side" cutting, and my spiral bit has seen better days (not very sharp any more), so the extra speed helped make sure the lines were clean. 

Here it is after the first cut... works perfect! 



I went through and quickly finished the other four cuts, then broke out the hand file tools to clean up the edges and square everything out:



Perfect! Nice clean opening, just wide enough for the connections to fit through without exposing any of the underlying PCB. 

Here we are with all four drive bays populated and slotted in!



Again, perfect! SO glad it worked on the first try... would have been a lot of work lost if something got screwed up. 

After verifying everything roughly fit together, and using my filing tools for hours on end, I finally got to peel the backing off and assemble everything! I had to run down to the local hardware store and find some tiny washers and nuts that would fit the 6/32 MDPC screws I was using, then put it all together:



You can see in the closeup below, I had to put two tiny washers between the acrylic and the backplane PCB... this was due to those solder points sticking off the board, mostly. With these, it floated nicely in place. 




Here it is quickly put together, for a "quality / progress" check:






Getting closer! 

Next, I had to decide how I wanted to attach the backplane to the drive cage. My original thought was that I would glue it on... but I decided it would be way more useful if I could disassemble it when I needed to clean it or make modifications down the road. Well... disassembly means using screws, and the drive cage didn't have anywhere I could really seat my press-fit screw housings into... time to add a place!  




I used a few remaining scraps of my 1" thick clear acrylic to cut some wedges for along the inside top and bottom of the drive housing. These would allow me space to drill into, so that I could drop some press-fit screw housings into them and screw the backplane on. 

Here we are post gluing:




And with the screw housings drilled and pressed in:






Huzzaw! All screwed together!  







And back in the case, for a final look:









And finally, with the backing off of the drive plates:




That's almost it for v1 of the drive cage! That's right... v1. There will be more... later... but I think I am happy with the functional version for now... I just have one more step (adding a fan for airflow... it's in the mail right now...)

The flashy/added effect version will come in the future... hint... there's a reason the drive plates are made out of clear acrylic!  


Until next time! 

Drive Cage Creation

posted Feb 1, 2014, 11:07 AM by Andrew Stock   [ updated Feb 1, 2014, 11:09 AM ]

Hello everyone!

I've made some nice progress on the drive cage over the past few weeks, and got far enough that I figure it's time to share some of that progress! As you will recall from last time, I had cut out the side walls of the drive cage and cut the air vents out of them (albeit roughly.) First, I spent some time cleaning up those vents using a rat tail file, some sandpaper, and in some places my dremel with a sanding drum on it. I also completed the vents on the "D" shaped side of the enclosure. This just made the vents look a little more professional... like they were cut with a nice CNC instead of an 'ape with a saw'.  

Next, I moved on to the wooden frame that I was building earlier. I'll be using this frame as a "mold" of sorts when I start bending the drive enclosure into the "D" shape I've been talking about. In order to accomplish the angle, I used my table router with a 3/4" rounding bit. 



SO much easier than my original plan of shaping it by hand with a file/sandpaper. :P Nice, uniform, and a perfect fit to the original design:

 

Next, ever the cautious builder, I did a dry run of my planned process with some scrap acrylic first. My two big concerns were that A) The bend wouldn't happen uniformly, leaving me with a crease or corner and B) that the vent holes would deform as they are close to the area I am going to bend and they are the weakest part of the piece of acrylic. Fortunately, my heat gun came with a special head that seems to have been custom crafted for this purpose:

 

The little shield attached to the end of the gun in that picture directed hot air down while letting me protect the part of the acrylic that already had the holes in it... nice! I started by heating the area closest to the vents and applied no pressure downward... I just heated it until the edge sticking out started to sag on it's own a bit. I then moved the heat gun slightly downward, applied a little more pressure, down a little more, little more pressure, etc. The result:



Perfect! The bend seemed to take to the mold flawlessly, without too much deforming or anything. I think the real key here was excessive heat, allowing gravity to do some of the work, along with distributing the heat so that a large area was malleable rather than a small area that would tend to just bend. Either way, worked well.  I double checked it against the detailed diagram, just to make sure I had my measurements right:



Looks good! It's a little off, but that's due to this piece being a scrap, and relatively smaller than the actual piece will be. Now, for the larger piece, I made one small tweak. Since the surface I was heating was much wider, I wanted to make sure that I made a uniform 'bending point' so that the curve happened evenly. To do this, I figured I could clam a block on top of the acrylic piece I was about to bend, so that it was sandwiched between the mold and this block. This would also help ward off some of the heat from the delicate vent area, as I would be heating this piece much longer due to it's increased size:



I followed the same general process, heating an area, moving down, heating, moving, etc... but made one small tweak at the end. Once the piece was completely bent 90 degrees and more or less flush with the mold, I noticed that it was tending to try and bend back to it's original shape a little bit... more like an 85 degree bend.  To fix this, I clamped another board on to the side of the now-bent acrylic, then re-heated the whole bend for a bit... just to soften it up and let it reset into it's final form. I left the wood braces on for a good 15 minutes so that the piece was completely cool before moving it. 




With the heat gun part out of the way, time to move on to assembly! I cut some quick clear acrylic blocks that would serve as the 'mounting' point in my case, then set everything in place, just to verify.



Coming together! Next, I cut some acrylic strips that would be used as the mounting braces along the inside of the drive housing.



I glued those to the inside of both of the faces of the drive cage using some of my IPS Weld-On #3... which was just as much of a hassle as I remembered from putting together my reservoir. This stuff is thinner than water, so it has some weird properties like refusing to stay in a syringe, then with the slightest pressure shooting all over the place. Needless to say, I screwed up during applying one of the eight pieces, and marred a bunch of the face of the drive vents. Doh!  Oh well, more of an excuse to use some of the Novus polish stuff I bought. I sanded down the marred area with 1000 grit sandpaper to get rid of the bigger ridges, then 2500, then #3 Novus, then #2, then #1... and done. Back to it's original shine, like nothing ever happened. It only took me an extra hour or two. 



With that out of the way, I reassembled things just to check progress:



Still lining up well! On to the next piece, the actual drive 'plates':



I cut out a bunch of clear acrylic squares, then used some spare drive plates and dead hard drives to sketch out where all the mounting holes would need to go. 



After finding all the locations, it was back to the drill press to drill all these holes. I started out with using the "right" drill bit that was tight fitting to the M3 and 6/32 screw sizes that would be mounting the drives... but found that the slight variances in placing my holes due to cutting this all manually with a drill press instead of a CNC or anything lead to stuff not quite lining up right... so I backed off and drilled the holes a tiny bit larger, to give myself some wiggle room when mounting the drives. 

There we are, drive plates all cut and ready for polish!






Next on the agenda, gluing the drive cage together, and then cutting and placing the 'backplane' area, where all the little SATA adapters will be seated. I have no doubt that will prove to be a lengthy exercise with a lot of measurement, screw ups, re-measurement, and eventual success.  Looking forward to it... until next time! 

Drive Cage Beginnings

posted Jan 6, 2014, 3:40 PM by Andrew Stock

It's been woefully long since my last update.  Family and holidays will do that, I suppose. I've made a few tiny bits of progress that I will share, anyway! Hoping for a free day sometime soon so that I can keep moving on this... otherwise it's going to go on for 2 years instead of just 1!  


Anyway, I've pressed forward with designing and preparing for the drive cage that I eluded to in my previous post. I did quite a bit of thinking on just how I was going to go about doing this, and I think I finally have a design and process now!

I started out with pencil and paper drawing a few concepts and getting the measurements right. You can see the different phases of the design on the sheet on the left. Once I settled on one that worked / I liked, I did a "final draft" to-scale on paper to get the measurements completely solid (on the right):



Next... I started with wood. Yes, wood! 



I cut some blocks out of some spare wood that I had laying around, and glued them together in a "U" shape. 



I plan on sanding down the corners of this wooden U to make the basic shape of the drive enclosure. This will ultimately be a "bending mold" that I will use when shaping my acrylic housing.

Next, I drew out my drive cage pattern in detail on the paper backing of the acrylic:




Once it was all drawn out to my satisfaction, I used a jigsaw to cut out the pieces. I could have used a table saw, but the jigsaw felt much more accurate... and with these tight fitting pieces, every half-millimeter counts! Once cut out, I put them on the drill press to get some pilot holes created for more jigsaw work:



After that, I took some blocks of wood and clamped it down to my table saw, then used the jigsaw to cut out the four ovals. These ovals will ultimately become "air vents" between each hard drive, through which exhaust air will pass into the body of the case, for expulsion by the rear case exhaust:



First a rough cut... the jigsaw couldn't quite make the turn necessary to complete the whole cut:



Next, I used my dremmel with the router attachment to round out the corners:



One side done!




And that's about as far as I've gotten.  Lots more to do on this (obviously), but hopefully I'll get a little more time soon!

Until next time.. Happy Holidays!

Reservoir Complete!

posted Nov 10, 2013, 10:05 PM by Andrew Stock

Hello Everyone!

It's been a bit of a delay since my last update... had a few weddings to attend followed by a delightful Windows 8.1 upgrade corrupting my storage drive ... so things have been busy. Fortunately, I'm back in business now, and have make a few bits of progress on the reservoir!

In all previous photos, I just had the reservoir and housing 'sitting' on the midplate in the drive bay area, not really attached to anything as shown in this picture:



I played with a few ideas on how best to attach the reservoir, but figured the easiest method would be to have some supports attached to the sides and use the slots in the drive bay with some screws to mount it in place. First, I measured a few times and cut myself some acrylic blocks with the appropriate counter-angle to my reservoir:






There we are, nice and snug. I got the acrylic bars about where they were going to be, then marked off some spots to drill holes into them:



Tossed them in the trusty drill press / vise:



After the holes were drilled, I hammered in some of those little "press fit" screw housings so that I had some metal-threaded 'taps' into the bars, then screwed them into place on the drive cage:







A top-down view of the housing/supports as they will be once glued together:



Next, I had to 'actually' glue my housing together. As you can see from the above picture, up until this point it was in two pieces: the sides, then the top/tube/bottom. I glued them together using some of my IPS Weld-on #3 so that they would maintain a nice 'neat' seam. 

After that, I sanded the face of the reservoir housing so that all the edges were even and such. I left myself some slack knowing that I would need to even up the edges as getting them perfectly exact was near impossible. As you can see below, the face sticks out just a bit pre-sanding:



The easiest way I found to get a nice even flat face was to tape a sheet of sandpaper onto a board and rub the face of the reservoir housing back and forth over it:




After the face was nice and flush, I powered up the hot glue gun and ran a few beads of hot glue on the 'angled' face of each supporting beam while it was still attached to the drive cage, then seated the completed reservoir and housing against it, making sure the face was flush with the case. After that, I ran a few beads of hot glue around the support beams to make sure they held well... this will get a bit heavy when it's full of water, and I don't want it deciding to give up the ghost under the weight after everything is all together. 




The glue job is a bit ugly, but fortunately once my case is completely assembled it's impossible to see this area. In fact, I'll be hiding cables and such here, so I really don't mind if it's a little ugly in the name of stability. 

A view of the mounted reservoir from the sides:



Nice and clean looking! 

With the reservoir all put together, I now finally have all of the 'hardware' for my cooling loop in place! Huzzaw!



Now, I just have to do the tubing and the 'functional' part of the water loop is complete! Obviously, I won't do that until almost last... so much more to do before that! 

Anyway... that killed about two weekends getting the reservoir housing done... looking at the pictures back to back it seems like about 15 minutes of work.  That's how these things go, I've come to find... the small stuff takes the longest! 

Anyway... on to the next big piece of my case. I'm in the planning and design phases of it now... can you guess what it is? 




Until next time! 

Radiator Cable Routing

posted Oct 19, 2013, 4:19 PM by Andrew Stock   [ updated Oct 19, 2013, 4:27 PM ]

Well, my hand is feeling quite a bit better at this point, and the work I was doing that caused the original injury is now complete! It doesn't look like much work now that it's all together below... but it took quite a few hours to get this all done and right. 

First, to make sure there were no further injuries, I went out and bought the tool I should have been using all along... a drill press vise!



Now that acrylic bar isn't going ANYWHERE.  The drilling position for repeated presses is a bit more precise, too, which is always nice! 
The other real trick to getting clean holes this large without shattering the acrylic was to slow down how quickly I was pressing down on the bit substantially, especially as I was about to breach the other side of the bar. I was going about a millimeter every second or so before... I dialed that back to about a millimeter every 3-4 seconds... more of a 'shaving' action than a 'drilling' action at that rate... but it worked perfectly.

I did a bit more planning on how I wanted to route and separate cables down under the radiators, then went to town drilling holes on my spacer bars. I ended up deciding on three paths due to the eventual number of cables I would be sending between cells... one for power, one for sensors, and one for data/uplink. I'll be powering and controlling 8 fans from this little area, along with the sensors for water temperature and such... so lots of wires coming and going. 

In addition, I needed a square hole that could fit a USB type B cable for powering and controlling the Arduino, so I got to work filing one of the holes into a square:



Perfect! 



The cable will route through the end spacer of my radiator housing and directly into the Arduino, like so. The Arduino will be mounted to a thin sheet of acrylic... or perhaps in some kind of housing that will be attached via velcro to the underside of the housing... I haven't decided yet. 



Here we are with most of the routing in place:



The center cell in the picture will be where a small 'power distribution' breakout board will be located. I'll have two standard 4-pin molex power lines from the PSU come down and power that board, which in turn will send power to all the fans and pumps and whatnot. That distribution board is still a work in progress... but should be together for next time! Cya then! 

Reservoir and Radiator Modifications

posted Sep 21, 2013, 5:03 PM by Andrew Stock

Now that I had the basic reassembly completed, time to start preparing some of the components I've assembled for installation into the case!

First, I started with rebuilding my reservoir that was 'disassembled' by my cat a while back.  I decided to go ahead and cut a new top/base, since the original one was a bit marred by the glue and probably wouldn't seal well. I also relocated the holes to be more central, since I'm not putting the light bar in the center of the reservoir (for this iteration, anyway... ).




Next, I used my old tapping rig that I made a few posts back to tap the two holes for G1/4 fittings to screw directly into the base:




I'm sure there's actually a tool meant for this kind of thing... but this works well enough.  I did change up my process slightly to ensure a cleaner/sharper cut. I would now set the tap up like below, then shoot a small amount of cutting fluid into the hole/on top of the bit before starting. This just lubricated the blades on the bit a little more, making it cut easier. 



With that done, time to glue the top and bottom back onto the tube! I changed my process here a little bit as well, after the learning experience from gluing it the last time. I started by putting the wooden guide triangle you can see in the top right of the below picture around the tube, but about 5mm away from the end. This way, I could use it to center on the acrylic triangle using my fingers, but it wouldn't pull the glue out of the seam with capillary action like it did last time. 



I also glued the unit "upside down" this time, putting the tube on the table and running a small line of glue on the top of the tube, then pressing the triangle onto the top of the tube until it held. I would then flip the whole thing over and shoot glue around the outside edge of the tube where it meets the acrylic triangle, allowing capillary action to pull the glue into the seam if there were any gaps. I still wish the seal would look cleaner, but this weld-on #4 is pretty watery and hard to manage right. I could use a different glue, like Weld-on #16, but I've heard it has a white appearance when it dries, which would be even worse. 



Once the "base" of the reservoir was glued to the tube, I used the reservoir housing piece as a guide for gluing the top of the reservoir onto the tube... I couldn't use the wooden guide, as it would be permanently stuck on the tube afterwards, sandwiched between the bottom and top. Glad I noticed that before I mindlessly went ahead and did it. 

After that, I filled it up with water and plugged all the holes to let it sit for a few days and make sure it doesn't leak. I checked back after a few days, no water spots or leaks... dry as a bone. Just to be sure, I did a quick cheap "pressure test" by sealing a fitting onto one of the openings and blowing into a tube... just to make sure it didn't force any water out. I know there are pressure testing rigs that will do this too... but my method is cheap, and just as effective.  



As I mentioned, I didn't really like the look of the seems... but I am hesitant to re-do it again in vain hopes for a better result... instead, I am thinking of a good looking way to mask it. I though about perhaps some kind of black vinyl design or something around the top and bottom of the tube, which would hide the seam area and ports, which are clearly visible. To try and fix it, I quickly threw on some electrical tape I had, just to see what it looks like:



Before and after, with the prism lit up:


 


 Hmm... not bad... it definitely does a good job of hiding the ports and seams, but it can be better. I'll have to think. Perhaps if the vinyl 'mask' on the top and bottom is some kind of design or something...


Anyway, on to other tasks for now! There were a few modifications to the radiator housing that I had been planning in my head for a while now. I wanted to make it so the cables from all the fans were hidden, or at least better organized. 



To do that, I disassembled the housing and marked off where I wanted to cut a few holes, using a wax pencil:



I clamped it down to my drill press, dabbed on some cutting oil, and put a few holes in it where marked:







Reassembled it, and fed the cables through...




Much nicer! 



I also considered rotating the fans once, so that the sleeve shows a little bit (since will have that nice MDPC sleeving on it!):



Not sure which I will go with... leaning towards the second one... especially since the fourth fan at the back looks more similar to the other three, whereas in the top picture you can see the cable for the fourth fan has to 'backtrack' a bit to go to its hole. This second way, they are more uniform. 

Next, time to plan out where to put all the cabling!



Hooking up 8 fans makes quite a spaghetti mess of cables.  I had some cheap-o 4-way fan splitters that I picked up, but I wasn't sure if I wanted to use them or not. They end up leaving a lot of extra length of cable to deal with, and they are ugly (though they would be hidden anyway...). Meh, maybe I won't use them. Too messy. Time to think of something better...



There are these nice spaces that I left myself underneath the reservoir where I can hide power cabling for the fans and pumps, as well as an arduino controller to manage the fan speeds and such based on water temperature. My initial thought was just to cram all the cabling down there, but looking at it upside down, I thought of a better way to do it. I quickly cut out some acrylic squares to use for prototyping:



Hmm... much neater this way! All I have to do is cut some holes through the clear acrylic blocks for the fan cables to pass through, and I can make a simple PCB that will fit in the bottom area to which all the fans will connect. Then, I just have one cable going from this "daughter board" over to the arduino "main board" in the cell next to it! 

First, time to cut some pass-through holes for the fan cables. I planned on putting one hole through the center of each block, which should be more than enough to fit all the cables:



Now... being the cautious person I am, I pulled out one of the scrap acrylic bars that I cut too short, and used to to try out cutting the holes. These are large diameter holes through rather thick acrylic... and I have had this particular acrylic shatter on me before for no good reason... so I wanted to practice a bit and make sure I didn't ruin any of these acrylic cross-bars by breaking them. 

I set up one of the scrap bars, clamped it down to the drill press, and ran a few holes through it. The first hole went fine until the last little bit, where it caught the edge and cracked the surface. Blah. I scooted the bar over, made another hole... same thing. Ugh! One more time, with more oil and a little slower... this time, the bar completely exploded, jammed on the drill press, and chopped up my hand in the process.



Stupid bar!  Alas, I had to call it quits there for a while... being down one hand with an already finicky piece of acrylic could only spell disaster from here... so I stopped for the weekend to nurse my wounds and think about how best to make these holes. 

Until next time! 

Putting It Back Together

posted Sep 7, 2013, 12:28 PM by Andrew Stock

Hello everyone! 

Picking up where I left off last time, I was starting to reassemble my case. Last time, it was just the clamshell that was really put together... only a few tabs fitting into slots, and a few screws to put the wheels on. Now, it's time for some rivets.  Sorry in advance for the terrible photo quality. It was 110F outside, so far too hot to work in the garage... thus I had to work inside, where the light is not so good. 


Here's an initial shot with the rivets just set in place while I tried to remember how this whole thing went back together... man, more complicated than I remember. Fortunately, I took a lot of pictures before I disassembled it... which helped greatly in remembering which pieces went together in what order. Like a giant 3D puzzle. 






It took quite a bit of doing, but I finally got the power button all reassembled and attached to the midplate again...



You can see the midplate is just a little wavy from the welding/warping... but fortunately, this area will be completely covered by the reservoir, so it will not be noticeable. 

Next, I started riveting using a trick I picked up a while ago while reading forums on the proper way to use these black rivets without scraping all the paint off. I put two layers of masking tape over the "nut" on the front of the rivet gun, so that when it presses against the face of the rivet it has a bit of cushion:



I replaced this tape after every rivet to make sure that each one went perfect. Maybe a bit overboard, but it worked perfectly. 



Looking good!



Got the motherboard plate reattached to the midplate and case clamshell. These were a pain in the butt to do... upside down, with not enough room to put the rivet gun head flat against the surface. Luckily, it came out good. 

Next, I started putting some of the Bitspower hardware back into the case. In the midplate, I decided to use these fill ports as a sort of "pass through" port, rather than routing a tube through a blank hole. I thought it looked a little more "professional" that way... plus, it gave me an excuse to liven up the otherwise plain midplate area with some nice shiny chrome hardware. 



When I cut the holes for these originally, I made them a very tight fit to make sure there was no vibration or anything. Well... it turns out I made them just a BIT too tight... with the primer/paint/matte coat on them, they no longer fit.  Fortunately, easy enough to fix with some P400 sandpaper. I just sanded down the inside ring till I was back to bare aluminum, and the plugs fit right in.

After a long day of riveting, this is all I have to show for it:



Doesn't look like much, huh? 

Here we are, somewhat reassembled!







I'm not super happy about the visibility of the weld spots on the underside of the midplate... but it's pretty difficult to see them from any angle other than this one, shot from underneath the case while it's up on a table. So I'm not too worried about it... for now. Maybe for V2 I'll look at rebuilding my own midplate... or maybe just do a scratch build while I am at it. 

You can also see one of my wifes stuffed animals in the reflection... a flamingo, wearing a sombrero, named Mother Clucker. I didn't realize I caught his reflection in this shot until posting it... but thought it was funny enough to leave. 







And here's the prototype radiator just set in there for effect. I still haven't cut the new bottom/top portions of it... I am just using the old ones for sizing for now. 

Up next time, re-doing the reservoir with a new bottom/top, and doing some cable routing on the radiator housing... and maybe sleeve the pumps?  Till next time! 










Painting, done!

posted Aug 9, 2013, 11:44 AM by Andrew Stock

Hello Everybody!

Man... it's been a hell of a month. This is quite literally my first day off... in fact, my first 8 hours in a row off... since I got back from vacation in the middle of last month!  Work has been absolutely insane. But, I finally have a little time to myself... and where else would I spend it, but with updating everyone on my progress! 

As I eluded to last time, I was going to work quite a bit more on giving my remaining exterior pieces the mirror finish they deserve. These pieces were absolutely brutal to complete without screwing up. Took a very long time, especially the parts with perforation. I'll just jump right into the pictures and let you enjoy the outcome... not many words necessary! 









*phew* Nice to have all that done! 

Once all the exterior "clam shell" pieces were completed (the parts that make up the frame of the case) I went on to doing the doors. If you will recall, these doors had quite a bit of custom Dremel work done on them, making them an extremely difficult piece. Couple that with the fact that I wanted them to be absolutely flawless since they will be the 'star of the show' so to speak, and you end up with a heck of a lot of care and effort going into getting the finish just right.

I'll go through the exact process I followed, just in case anyone wants to duplicate my steps. Sharing knowledge about the build experience is what this log is all about, anyway! 


To give you an idea of where I start from, here's the piece after drying for a few weeks without any sanding or modification done:



You can see it's a literal "sea" of orange peel... everything is blurry like you are looking through security glass. It's probably a combination of my painting skills (or lack thereof! ) and my paint gun that produce this... I'm not really sure if you can get a paint job without this level of orange peel... but fortunately a bit of effort will clear it all up!

I start out with P1000 sandpaper from 3M. I pick up mine in the paint section of my local automotive shop... but I am sure you can find it elsewhere. They have a nice "kit" of sandpaper that has 1000, 1500, 2000, and 2500 all together for pretty cheap. I also have some rubber sanding blocks that have little flaps with spikes in them, so that you can tuck the sandpaper around the block and lock it in place. That makes sanding flat surfaces SO much easier than trying to evenly distribute the weight of your hand or anything like that. 

I also get a painters tray with a couple of cups of water, put a single dot of common dish soap in it, and stir it a little to distribute the soap. This helps the sanding block glide a bit better by reducing the surface tension of the water between the block and the surface of the material. I then throw all my sandpaper in there and let it soak for at least a good 15 minutes before using it, so that there isn't any dry paper left. 

Finally, I have a large bag of color-coded microfiber rags that I use so that I know what rags I used on each step of the process, and I can avoid using the wrong  rags with the wrong grit of sandpaper/polish, thus producing larger scratches than intended while cleaning off the surface to check that orange peel is gone. 

Now that all the setup is out of the way, I start by sanding in one direction (we'll say left to right) across the entire piece as much as possible. Sometimes I have to switch directions on detail pieces or corners, but the more you can stick to one direction, the easier it is later on. Every 20 or so passes on a spot, I will stop sanding, flush the area with water and dry it with a rag, and then check to see if all the orange peel is gone. This helps me ensure that I go no deeper through the clear coat than I have to, which will help make sure I don't go all the way through into the color coat by accident, since I am sanding this so many times!

Here's one of the case doors after the first pass with 1000. This part definitely takes the longest because you have to sand the entire surface until all of the orange peel is completely gone... leaving a hazy (but scratched up) clean surface. 





You can see, if you look very very closely in that slightly more white section in the center of the picture, there is a tiny extra deep pock-mark of orange peel. All that has to be gone, or it will be extremely obvious in the final finish! This first pass is also the most dangerous pass, because P1000 sandpaper has a lot of removal action... so if you grind on one place for too long, you'll go right through your clear coat and be back at square one! I've done it a few times... it's painful.  

You have to be extremely careful along edges and corners while using P1000 also. I would do no more than 10-15 passes when sanding "over the edge", as you remove way more material from edges than you do from surfaces. I use my fingers along the edges I am sanding to create a sort of "bumper" that will keep the sanding block from going off the edge of the surface, which keeps it from digging in too deep and removing too much material. The problem is, orange peel likes to hide at the edges... so I usually have to sand "along" the edge even if that isn't the direction I am sanding... and once I'm satisfied that I've sanded down the previous layer enough, I go over it a few more times in the correct direction to get everything uniform. 

After all the orange peel is gone, you switch to the next level of sandpaper (1500) and switch directions (up/down now). This pass also takes quite a bit of work, because P1000 leaves fairly deep grooves in your clear coat, requiring many passes to remove it all. You can see at the center/bottom edge of this shot where I haven't quite gotten rid of all the "side to side" scratches. Gotta get every single one!



These little groves along the edge of the case wall were a disaster waiting to happen, too... you can kinda see that the clear coat and paint is slightly risen compared to the rest of the surface... which is very easy to grind through by accident! Not to mention, every little ridge is like it's own corner... so the edge of this case panel was extremely hazardous to sand. I honestly expected to screw it up... but I got lucky!  Here we are, done with 1500. You can see the shine starting to show through!




Now on to 2000 grit, changing directions yet again! As the grit gets higher, it gets easier and less passes are required. I'd say that 1000 took as long as 1500, 2000, and 2500 combined. All you have to do is sand until the scratches from the previous direction are no longer visible, and then you are on to the next grit!




And finally... 2500 grit. For this level of grit, I switched from a rubber sanding block to a sponge. The reason I switched, was that I felt like it was a bit softer and less likely to cause deep grooves from trapped material or whatnot. I would flush off the piece much more often at this level, and try not to push down at all... just let the weight of the sponge do all the work, and use many many many light passes. 



Now that all the sanding is done... on to the rubbing compound. This stuff is almost like a soupy toothpaste, that has tiny bits of grit inside it. I would always use a particular color of microfiber cloth that I didn't use for anything else, because the grit gets caught in the cloth and you don't want to accidentally use it on a piece that you are polishing instead. I would lay a ribbon of compound onto a small area I was going to work on, maybe 6 inches x 6 inches, and use the cloth with my hand in it to "buff" in the opposite direction of the 2500 grit fairly hard... maybe 50 or so passes over each area, giving a fairly strong downward force to really work out the 2500 scratches. I didn't need to use a lot of rubbing compound because I went to the 2500 level with sandpaper. Previously , I was stopping at 2000, and it took a lot of rubbing compound. Once I had gone over an area with rubbing compound, I would use a cloth of the same color to wipe off the area and get rid of the extra residue. 



Looks pretty much done, right? Nope! Still have polishing to do! Unfortunately, the difference wasn't easily capturable by my camera... it's a very subtle nuance. But it's worth doing the polish layer afterward!

To polish the piece, I would use yet another color of cloth, another similar ribbon of polishing compound, and go in a swirling motion instead of an up/down or left/right. This makes it very obvious if there are any scratches left from either direction that need to be worked on. Sometimes, if I uncovered a rough patch, I would go back to rubbing compound and work on it a bit more to get rid of any residual scratches... especially on corners, where I was more careful with sanding. After polishing an area, I would get another cloth of the same color as my polishing cloth to clear/clean the area and validate that no scratches were visible anymore. 



Done! 



Here's a quick before/after I took. The piece on the left is after clear coat without being modified at all. The piece on the right is after it's first polish pass letting it "rest" for a week. You can see a couple of scratches visible on the right piece, particularly in the bright areas. I noticed that if I let the piece sit for a few days/weeks after polishing, some "ghost scratches" would start to come back. I intend to go over the pieces one more time with rubbing compound and polishing compound to stamp those out. My guess is that the compound would get in the scratches making them disappear... but once the compound dried or got wiped off, it became visible again. I'm betting another pass will get rid of those! 




Anyway... now that all the months of painting are coming to a close... time to get back to the bench for the real fun stuff... reassembly! 

I started off by cutting some more custom-length M3 screws with my metric wire strippers. I needed about 8 of these to reassemble the bottom of the case and casters. The original screws wouldn't work anymore, as I added the bottom plate and acrylic piece (for a future feature!) to the stack, making them too short. I measured out the needed length, then cut each screw to be the same. 



Here we are, screwing the casters back on to the bottom plate:



Good! We're back on wheels again! 



Next, I put the other half of the clam shell back on, and assembled the drain plug through the whole business:



Alas, that's as far as I got! 

Hopefully some more time soon, as work starts to slow down again! Can't wait to get everything re-assembled so I can start to sleeve all my components! See you next time! 

A Mirror Finish

posted Jul 7, 2013, 6:17 PM by Andrew Stock

Hello Everyone! 

Last time we left off... I made a few "minor" mistakes in trying to achieve my mirror finish on my exterior pieces, cutting all the way through my base coat in one or two places. Whoops. 

Anyway, I went on to repair the pieces by sanding down a few layers of the clear coat, and in the mistake areas all the way back to the primer. 



After that, I laid down a new base coat, followed by a heavy re-coat of clear. Hopefully this will give me a bit more of a buffer of clear that I can cut through to get rid of all the orange peel. Now, I just had to wait a week for the clear to dry again. 






... and of course, I cleaned out my spray gun again. I picked up some different cleaner (lacquer thinner instead of mineral spirits, works much better...) and a few extra tools like a solvent-proof spray bottle to help with cleaning.




One week later (this weekend!) I finally got to make another shot at wet-sanding the exterior pieces. I started off with the bottom plate of my case... which will actually not be visible... but I figured it's a good piece to test on. 

First, I filled a painting tray (one of those plastic things you use with a paint roller to paint a house) with water and a dab of soap, and put all of my sandpaper in to soak for a few minutes. Once soaked, I started with 1000 grit as I had a lot of orange peel to cut through. This part took the longest, I had to be very careful not to cut through edges, but had to smooth out every single little dot. After about an hour of working on the piece with 1000 grit, I ended up with a finish like this: 



It took a bit of experimenting, but the process I worked out that more or less "guaranteed" that I didn't cut through the edges was to use my fingers along the edges like a "bumper" of sorts... like so:



The other thing I did was make sure that I never pushed the sanding block over an edge that wasn't parallel to the direction I was sanding. So this is fine:



And this, is not!




Next, I switched to 1500 grit and changed directions. I went over the entire piece until I could no longer see the side to side marks from the previous sanding job with 1000 grit, meaning I had leveled out all the deep cuts. 



No more cuts visible from the 1000 grit going side to side means I'm ready to move on:



Next, I went over the whole piece again in 2000 grit, once again switching directions. You can see that I am starting to get a good reflection going at this point... but we're nowhere near done yet!



Once all the lines from 1500 were gone and only 2000 grit lines were left, it was time to move on to 2500 grit. I also switched from a rubber sanding block to a sponge with the sandpaper wrapped around. This helped cut down on the depth of the lines... but as a result took many more passes to get rid of all the 2000 grit lines:






Looking even closer! Now, I moved on to some rubbing compound meant for renewing your finish on your car:



The rubbing compound has a bit more grit to it than the polishing compound, and does a better job of removing the scratches from the 2500 grit with less material required. I switched directions while using the rubbing compound one more time, going back and forth until no lines were visible from the 2500 grit. I used a microfiber cloth that I picked up at the automotive store I bought the rubbing compound and such from. I also used 'color coded' cloths so that I didn't mix up the cloth I used for rubbing compound from the one I used for polishing. I used one orange cloth to apply the the rubbing compound, and another clean orange one to remove the compound. I then used one blue to apply the polishing compound, and another blue one to remove the compound. The result:



Perfect!   



Hello there! 

(It looks a little hazy, but that's because my shutter length / ISO wasn't quite right on my camera or something... it was hard to capture this picture... )

I then went on to give the same exact treatment to the top of my case:











Hooray! All that work is finally paying off. All in all, these two pieces took me about 6-7 hours. It's quite an extensive process, but most of what took so long was dealing with the complicated angles in the top of my case and such... it was very slow work trying to clean up all the edges in the design without cutting through. If you have a plain surface, it should go much faster!

Now, I just have to give the same treatment to the two case doors, the other half of the case shell, and the PSU bracket. I'm hoping I'll get them all done next weekend, or maybe sometime into the week after! I have some time off, and I know what I'll be doing with it. 

Until next time! 


1-10 of 59