Announcement

Collapse
No announcement yet.

DFrame FCE

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • DFrame FCE

    I've owned an original, gen 1, In Win DFrame for a number of years now. This chassis was never designed to allow for the incorporation of a custom water cooling system:

    1. The three fan brackets, which run along the bottom of the chassis, are awkwardly spaced, and thus, prevent a proper mounting of a radiator. This design also prevents fans from being placed under the radiator in a functional orientation (`2" gap in between) and with a top mounted layout, space becomes an issue.

    2. The brackets are also supported by extremely short screws, penetrating the motherboard tray about 3/16" which can cause problems in regards to weight capacity. Ultimately, running a bottom rad, with 3 fans and potentially mounting a reservoir, proved to be too great a risk for me to take...

    3. The motherboard tray design doesn't allow for the mounting of a reservoir, nor does the top fan bracket. It was, again, too risky to use the fan bracket as a mounting platform, especially for something as important as a reservoir. The motherboard tray has, imo, the most arbitrary screw hole layout, of any system I've ever worked on. In fact, while I took dozens of measurements, there came a point where I couldn't tell whether In Win designed the case using imperial or metric systems. When I show a picture of one ot the templates for the mounting system I designed, you'll see what I'm referring to in more detail.

    These are the primary issues I had to overcome in modifying the DFrame in order to accommodate a custom cooling solution. There are also less obtrusive issues which I'll cover as we go through this build log, as is the case with any custom build, tbh.


  • #2
    So I sat back and stared at the DFrame, for a seemingly endless period for time, one question after another, with little to no success in finding a solution to my problem. Whilst looking through the rear glass panel, it dawned on me that I could replace the glass with an aluminium panel, and mount fans and radiators to it. Fortunately, the spacing between the mb tray and the glass panel, would come to 68mm, plenty of space to house a 30mm thick rad. However, the fans, complimenting the rads, would have to be mounted to the exterior to allow for proper, if any airflow, and so I would also need a fan cover plate. With this idea in mind, I disassembled the rear glass panel, traced the perimeter on a piece of bristol board, and started drawing out what would ultimately become the radiator support panel.

    I really took my time planning these rear panels, both in terms of aesthetics, and functionality. The lengths of the curves along the edge of the glass. Where to place and how far to separate the rads to accommodate the front panel i/o shield. Where do I run the fan power cables and what controller can I fit beside the rads. I came up with the design, shown below.

    Comment


    • #3
      The radiator passthrough cutout was taken directly from a pair of HWlabs 360gts' I had picked up for this project. Slots were included to allow for some freedom. The dual rectangular cutouts were positioned to allow for a smooth transition of fan cables (pwm+rgb+drbg combo cables) to the interior, where they could be wrapped in a circular pattern using 4 cable management clips before being attached to the controller. The hole spacing for the clips was designed to allow for up to, what's referred to, as an 8 cable clip. Now, while the rectangular cutout dimensions allow 3 sets of fan combo cables to pass through, the theme of this build is industrial, with very limited RGB. So I opted to purchase be quiet fans.

      When it came to fan selection, I wanted something that could minimize vibration transfer, and the silent wings 3 fans utilize a rubber ring along the fan perimeter. This ring sits less than 0.25 mm lower on each side when compared to the mounting posts' rubber washers (total post+washer = 25.7mm vs. 25.3mm). When sandwiching the fans between the radsupport and fancover panels, it actually creates a seal with the fan frame, as a result of the post+washers, nearly indiscernible difference in overall height.

      The fancover panel would take a great deal of design time, as it would be the primary aesthetic focus of the rear panel. I created a number of iterations, one's that included hex patterns (restricting airflow), or completely open (dust accumulation and overall protection). With that said, the final design is pictured below. The centre of the fan cutout is 42mm in diameter, in order to cover the label and soldered connections of the fans themselves. I went with a hexagon pattern to attach the centre circle to the cutout's perimeter. Each fully exposed hexagon is perfectly symmetrical...a task which required the use of manual inputs for vertex coordinates. I actually set the grid to 0.001mm at one point, but to no avail. It probably didn't help that the fan cutout perimeter was made to 117mm to match the centre perimeter of the circular fan frame itself (be quiet SW3 fans are ~119mm in diameter). The overall design is a good balance of aesthetic appeal, while sacrificing very little airflow.

      I had to take serious consideration when selecting the screw length in reference to putting the entire assembly together. The system works as such: radiator, 0.5mm (contracted) gasket, 4.78mm aluminium panel, 25.3mm fan (contracted), 3.25mm aluminium cover. I wanted the screws to penetrate about 4mm into the rads (HWlabs 360gts have protective fins beneath the M4 screw holes, set to 5mm, but they can still be bent if you apply too much force). so the total length is ~37.75mm. I added 2mm worth of washers (rubber+stainless) to make up the difference and then used 40mm screws to fasten everything together. Important side note, the fasteners are all stainless steel. This helps to prevent corrosion.

      Comment


      • #4
        This leads me to the shortcomings in the design. A 5mm thick aluminium (4.78mm to be precise) sheet was required to replace the glass if I was to use the original sand blasted aluminium mounting supports. With the radiator mounted on one side of the panel, and the fans on the other, necessary for spacing between the mb tray and the rad, there is no longer direct contact between the fans and the rads (5mm gap). Also, be quiet fans are round on one side (exhaust) and octagonal on the other, instead of square, so there would be some loss in efficiency for that reason as well. Fan removal could be easier, at least for the center mounted fans.

        Weight! I was very concerned about attaching such a heavy panel to the rear of the chassis. The mb tray is 1/2" thick but the entire system is aluminum. I test mounted the panels in an empty chassis, fans and rads included, and was pleasantly surprised that, even at a 30 degree tilt, the system didn't tip over. Moving along...

        Comment


        • #5
          The second set of custom panels were designed for internal mounting of fans and reservoirs. I needed to replace the front fan bracket with something much more substantial, if I hoped to mount a "floating" reservoir. I decided to use 1/2" aluminium bar stock as a mounting point for a mating panel, which would hold the reservoir, and subsequently fans, as I added those as well.

          The bar stock was cut using a horizontal band saw to 155mm, and then shaved down to 153.5mm using a hand file. I've provided a figure for illustration purposes (notice the ridiculous hole spacing...141.5mm, IN WIN!!!) I machined out the holes at work using a 3.4mm parabolic spiral flute drill bit. The parabolic shape really helps to remove chips. This is an HSS (High Speed Steel) standard finish bit. Do not use black oxide, or cobalt drill bits when machining aluminium (TiN is also alright). They are made for much harder materials. I set the rotation speed to 3600 rpm (half the recommended speed for aluminium given all factors involved) since the mill redlines at 4200 rpm. Each hole was machined independently, as the milling table moves with a terrible taper, so I had to drill a hole, remove it from the vice, then drill the next hole...so very time consuming. The hole distances had to be precise, (+-0.125mm). You can create a panel with a single axis slot, but further freedom along a perpendicular axis will create a loose fit, and for my intentions, I wanted this to be as secure a connection as possible. A9 cutting fluid was used. This is a very important step. When working with aluminium, failure to use appropriate lubrication can result in chips fusing/melting to the freshly cut/drilled surface. I've watched a saw almost tear aluminium, rather than cut it, creating a terrible surface finish. It took roughly 15 minutes to machine each hole (8*M4 + 4*6-32), while liberally pouring A9 fluid...very slow feed rate for a flawless finish. I actually ended up offsetting the holes in relation to the companion panel by 1mm (center = 7.25mm vs 1/4"). I was concerned that the "floating panel" would vibrate against the mb tray if I made it symmetrical (see photo) Once the bar was complete, I took it home, and manually tapped 8 holes out (also used A9 during the process). And thus I have my completed mounting bar.

          Comment


          • #6
            I purchased an SC resonance for this build, which includes their 140mm universal mounting hardware. I set out to design a panel around the 140mm mounting hardware, so that the edge of the bracket would not overhang the panel I was making. I actually made a paper cutout of the fans to see if they would fit in the chassis, and sure enough...I had 12mm of space remaining between the top and bottom. Interesting that the chassis could, so accurately, accommodate 3*120mm fans in the front. With that I set to work and drew up a CAD model, this time with hollow fan cutouts. While this is an industrial theme build with minimal RGB, I left the cutouts hollow in case I ever wanted to switch to something like a corsair ql, or lian li uni fan. I left enough space for the fan cables to run parallel, between the fan and 1/2" bar stock.


            I ran into a problem, however, when I mounted a single aluminium panel to the bar stock, there was still considerable flex at the bottom, despite being 3.25mm thick. I suppose I took too much material away in using hollow fan cutouts, thus compromising the panels integrity...I won't go into an explanation of Young's Modulus (stiffness) vs material strength. So I had to add a second, identical panel, in order to drastically improve the rigidity. Now there's almost no flex, and no audible vibrations are transferred to the mb tray, even when running the pump at 100%.

            Comment


            • #7
              Paint. This was an enormous undertaking. I ended up stripping the original paint, entirely by hand. The problem with using power tools with abrasive wheels and disks, is that they can cause pitting in the aluminum as it heats up. This pitting is difficult to hide, even with a thicker coverage of paint. So for ~100 hours, I removed the paint using soft sanding bricks and sponges...240 grit brick to start, followed by 320 grit. Once I began to see exposed aluminum, I switched a 400 grit thin sanding sponge to really get into the crevices and welds, before finishing with an 800 grit standard piece of sandpaper. The last step left a rough enough surface for a primer to bite its teeth into.

              After completing the sanding portion, I washed all the panels in degreaser, and it's at this point you have to be extremely careful not to touch the panels with your bare skin...make sure the room is free from contaminants. Nothing worse than having to re sand a panel because loose hair got caught under the paint, or you brushed it with your elbow, so the surface finish looks inconsistent.

              Paint was applied using a micro spray gun attached to a dual piston compressor. I would highly recommend this setup for painting panels if you don't have the space, or need to paint on the go. A typical airbrush compressor reserve air tank has a capacity of 2L, with the largest alternatives possessing a 3.5L tank. You'll blow through that reserve in a matter of seconds while painting large surface area panels using broad spectrum stokes. Microspray guns also allow for a spray diameter of up to ~10" which really helps covering larger surface areas quickly. Buy a shop compressor+large capacity air tank if you have the space (they're loud), but you'll get an extremely consistent flow of paint, free from blowouts.

              I used a primer/sealer (2 medium 1 light coat) to start, colour keyed (black in this case). It's very important to start with a proper primer for the material you're working with. After the primer dried, I used a black charcoal metallic paint to get that metallic shine (2 medium, 1 light) followed by 2 light coats of pearlescent black. This gives the surface that sparkle. Finally 2 medium and 1 light top coats, gloss clear. The top coat is actually for outdoor use, so it's fairly resistant to fading.

              For the radsupport panel, and the internalfansupport panels, I went for a much more basic paint job, utilizing only a black primer, with a matte black clear coat to top it off. This gives the panels a much more subdued look when compared to the chassis. All in all, it was 120 hours prep for 2 hours of actual paint (not dry) time...worth it!!!

              Comment


              • #8
                More photos...

                Comment


                • #9
                  Custom Cables. All of the sleeving and wires are from MDPC-X. They make my favourite sleeving, and the colour combinations are extremely vast. Everything has been sleeved* down to the front panel i/o wires. Sleeving the fans worked better than expected, I was able to run all four wires down a single piece of small xtc sleeving without any real issue. I always colour code the wires with sharpies so I can attach them to the correct ports on the pwm connectors.

                  I use the heatshrinkless method for atx power cables, although I've never quite seen anyone use my methodology. It's more time consuming, but you have a much higher probability of getting a clean melt on the sleeve. I actually use a heat gun, set to 800 degrees F, and slowly rotate the sleeve and heat shrink around the outer rim of the heat gun, overhanging the element by about 2mm.

                  I use molex pins to crimp 2 wires together, when making split wires. This really creates a solid bond, although I still solder them together. Each cable has been made to length (roughly). I had originally planned an arch for the 24 pin, however, I couldn't get the cables to bunch together near the end, thus leaving a heavily exposed mess of cabling when viewing from the front (your primary perspective). So I had to disassemble the 24 pin run, and remove all but one of the cable combs. This allowed for a much better, much more hidden, flow of wires when viewed from the front. Disappointing? certainly. Although I'd rather have it look its best from the front, then from the back, which you'll unlikely see for more than a few hours/year.

                  I went with a combination of liquid carbon and white carbon. The liquid carbon was the closest colour I could find that had the reflectiveness of the aluminum, and the black intertwined to match the chassis. Nils now sells aluminium sleeving, however....hahaha. The white carbon was used to give a little more contrast to the build, and I really wanted the cabling to stand out, given there's so few places to hide it. The white also works well with the shiny silver bitspower+black darkside adaptor fitting combination.

                  Comment


                  • #10
                    Tubing. I've used mayhems 16mm OD glass tubing in this build. Some may criticise the excessive use of fittings, however, I really like the appearance of parallel tubes and fittings. and the shiny silver bitspower+matte black angled adaptors, really compliments the industrial theme. I always pipe up my systems with PETG beforehand, so that I can get an incredibly accurate length measurement for each tube segment.

                    I've always used a dremel and a diamond cutting wheel to cut glass tubes. I much prefer the solid body bit against its EZ lock counterpart, although the latter is superior for sanding down the jagged edges (it's got some pretty significant flex). Before I start cutting, I draw 4 lines to mark my cut, and then wrap a piece of masking tape around the tube. I then leave a 10mm gap, and proceed to mark the edges of the next tube. I'll grab a scrap piece of styrofoam at work, and drill out a 1/2" hole, entirely through the foam, and a second 1/2" hole, drilled half way through the foam, for sanding purposes. I then place the tube into the through hole, and place the foam into a vice. I've found this method to be the best for preventing vibrations, and accidental destruction of the tube, as a result of over tightening a vice. Because the hole in the foam is drilled at 1/2" (12.7mm), the 16mm OD tube fits quite snugly into the foam.

                    I like to go quite slowly around the OD of the tube, very gradually creating a break. I do this approximately 5mm away from my measurement line (hence the 10mm gap between tubes). At this point the tube is separated. I then take the freshly cut segment and put it into the half drilled hole, where I cut another 2-3mm away (roughly the thickness of the diamond wheel itself when in action). At this point, it's time to grind down the last ~2mm, and this requires extraordinary patience. I constantly spray the glass with water as I grind away (although I do this throughout the cutting process as well...just not as vigorously, I'll say). Eventually I reach my measurement line, at which point, if I've taken my time, will create a very smooth, very flat perpendicular surface. Following this, I sand the edges with 60-120-240 grit sandpaper. Wet, with a splash of soap. This helps to keep the ends of the tubes clean as they are rounded off.

                    Comment


                    • #11
                      Component selection. The cooling components found in this system were chosen based on their performance reviews. Consequently, many of them are highly restrictive in comparison to the more well established brands. HWlabs rads are widely known to be highly restrictive, and each 360gts creates a pressure drop of approximately 0.9psi @ 1.0gpm, making them twice as restrictive as their, say, ek counterparts. Couple this with a pair of bitspower QDCs, and the back end of this system could produce a total pressure drop of 3.3psi @ 1.0gpm. Now I'm using the newest renditions of the bitspower QDCs, which appear to be much better than their predecessors in terms of flow optimization. Speaking of optimization, the optimus cpu block is the first I've used from the company, and it is exceptionally well constructed, however, I believe the pressure drop lies somewhere in the region of 1.5 psi, twice that of an ek velocity block. Finally the heatkiller gpu block is more restrictive than a typical ek counterpart, but only by a small margin (around 0.20psi). Originally, I believed that I wouldn't be able to hit a solid flow rate on a single d5 next, but I was proven incorrect once I got everything up and running.

                      Comment


                      • #12
                        Component List:

                        Aorus x570 extreme motherboard (full board cover plates are important when running in an open air chassis. Dust accumulation is far more difficult to clean on traditional board types)
                        Ryzen R9 3900x
                        Radeon 5700 xt (swap to a 3080 when available)
                        G.Skillz ddr4 3600 cl16 4*8
                        1tb WD black SN750 m.2 ssd
                        1tb Samsung 860 evo ssd
                        9 * Be Quiet Silent Wings 3 fans


                        Cooling Components:
                        Singularity Computers Resonance
                        Aquacomputer D5 Next pump
                        Heatkiller IV gpu block (black + nickel)
                        OptimusPC cpu block (acrylic + nickel + anodised black mounting hardware)
                        2 * Hardware Labs 360 gts radiators

                        Fittings:
                        Bitspower Shining Silver
                        Enhance Multi Link * 14
                        14mm rotary extension * 2
                        Soft Tube Compression * 4
                        Male QDCs * 4 (two for fill and drain ports...work remarkably well)
                        Female QDC * 2 + 1(1 additional for fill/drain connection)
                        Hot Air Exhaust * 1

                        Darkside Black
                        90 rotary adaptor * 8
                        45 rotary adaptor * 2
                        15mm extension * 2
                        20mm extension * 1

                        EK Black
                        Dual female adaptor * 5

                        Comment


                        • #13
                          Final Thoughts. I'm quite happy with the way everything turned out. I set out to create a rad box that could be mounted to the DFrame chassis, and the results speak for themselves. Coolant temperatures and subsequently, airflow are fantastic.

                          This build log turned into a bit of a tutorial at times, but I wanted any newcomer to understand the puts and takes when it comes to custom fabrication, and some of the, seemingly insignificant, details which have to be taken into consideration. I had actually planned to include many more details, but I'm already past 4000 words, hahaha, so if there are any questions, or any design elements require clarification (ie. what did you use to vertically mount the gpu? A. cablemod vertical mounting bracket. It leaves 30mm of space...and so on).

                          I'd like to send a shoutout to Daz for hosting this modding competition. I had no idea it existed until more recently, and the work my fellow competitors do is absolutely exceptional. Very well done all around.

                          Anyways, I've included a smattering of photos at the end here...I need a better camera than one found on a 4 year old cell phone.

                          Happy new year to all!!!

                          Comment


                          • #14

                            Comment


                            • #15

                              Comment

                              Working...
                              X