Scratch Build – In Progress BBJ-G

Before figuring out how to solve the issues I was having, I decided to dismantle and get rid of the last Side Burner prototype. That beast is printed when I only had one printer. It took about two months to print.

Image of a real prototype:


That is a working desing, with six 480 rads. No heat stacking. 24 PWN channels, one for each fan. Three D5 pumps, each with a PWM channel. A ton of water temp sensors, about 10. RAM cooling actually works, been there, done that. Adding a ram block on the backplate of the rx5700xt, really works. And yes, I can carry this beast. It is really solid, but bloody heavy.

This time around, I had mounted the res in front, straight into the drain pump. That was a such a disaster. I realized that I needed a air inlet on top, as in the res on top, when draining the rads. This just took for ever. It also told me that I had messed up my current design, and needed to replace the res and the pumps. In retrospect, this was needed. And again, laborus.

The difference of the geometry of feet and corners are staggering. Sure, it is not perfect in the old prototype, but actually quite usable.

Image of geometry of corners in the old prototype:


In my current design, the corner parts padding the box vertically are 8mm thick, not 6mm. The foot is 47mm wide vs 37mm in the old design. That is really all the difference.

What both desings had in common, was that the side finish easily get blemished. I took note that the print bed facing parts just came out significantly better, and decided to revert to that, as I had used that in previous designs. Doing so, I hoped to solve the warping, and get a better finish. It worked out.

In the Side Burner prototype, I tried to embed the IO section on top, flat with the other parts. It did not turn out that nice. Warping is visible, more so than what is depicted here.

Image of top warping, old prototype:


This top warping is hidden, with the current design, with shallow plastics, resting on the top of the main structure.

Also, in the old design, cables were routed into the same channels. It is just spaghetti. Even the drain valve is hidden in this mess. It turned impossible to work with. This is why I currently separate the fan cables from the rest of the system, and have deviced a separate drain chaimber.

Image of sphagetti:


Here is an image of some of the parts that went into the last prototype:


That's a ton of work.

So, pump and res are now relocated to the left of the current design:


I need two pumps. I have tried to bleed this build with one pump, and it simply does not work. Backflow is an issue with this kind of pressure, so I recommend using a valve on the res, when filling and draining. I just leave it in place. Do note that the rubber inserts EKWB uses, lend themselves nicely for a 3D-printed design. And yes. I had a corner piece with extra holes in it, for holding the res. It had to go, meaning I needed to print one without the holes, and replace it in an nearly fully assembled box.

Since flow from res through the pumps down to the rad, follows gravity, water just flushes out now. When opening the res valve.

Also, not being a spaghetti fan, I went for a drain chaimber.

Image of drain chaimber from the top:


Image of drain chaimber from the bottom:




The drain port is at the lowest level both upright, and when the right side faces downward. With the res on the other end of the rads, gravity pulls the water out of the rads, one by one, when on the right side faces downward.

Closeup of placement of though port:


This new design has the added benefit of letting more air into the system. It also solved the issue of how to cover up different sized PSUs, and how to make a easily accessible spaghetti chamber.

Image of old PSU cover design:


Image of new PSU cover design:


A reprint of the front crossbars was in order. Also note that the hole in corners are gone, as I have reprinted two of front corner parts. That took 2 times 8 hours, and 2 times 14 hours. Just for them 4 parts.

The PSU is covered by to simple mesh covers, that are held into place with just two screws.

Image of right side PSU top cover:


Image of spaghetti kitchen:


Please note, that in the latter image, the power for both pumps are easily accessible and purposly on top. It makes filling a breeze. Also note, that there is a full EPS, PCI, and MB cable in that mess. As I simply use some cheap extensions. This is so bloody nice to work with. If removing the two screws on the MB raiser in the back of the latter image, the entire power cable routing is fully and easily available. There is no need to any zip-tie.

Please note, that I have redesigned the front cover on the MB tray riser in the latter image above. It is currently going through yet another revision, but this is really starting to shaping up.

 

People always struggle printing a motherboard tray. As did I. The issue for me, was that I wanted to print artifacts on the back of the MB tray, and thus could not print embedded standoffs.

I solved this, by first making a M14 hole in the MB tray. The tray is 8mm thick. The treads are 4mm on the standoffs. Then I use a standard M3 srcrew, that runs through the M14 standoff. The m3 screw is fastended by a metal m3 nut. This is nothing that special, but that silly nut, goes nuts, all the time, and has some close misses dropping into my PSU, when it comes loose. That is why I use another m14 bolt, to lock the m3 nut into place, when the m3 nut is embedded into the m14 standoff. It is not that complicated, and it works really, really well. Do note that some of these holes are inside the external cable channel, and by this design, they do not protrude into that channel, and do not come loose, inside that channel.

Image of M14 standoff screwed into place, with inserted 20mm m3 screw:


Image of rear side of the same standoff, with the locknut beside it:


Image of fully mounted standoff, with secured m3 screw and nut:


That nut is secure in there. The locknut is fastened with a regular 8mm flat driver.

 

I also ran into drooping issues, when using supports. The brackets that holds the side covers into place, the glass parts, needed support to print. I gave it some margin, and printed an entire set of them, only to find that they all drooped and exceeded the 4mm thickness they need to adhere to:

Image of 3.7mm drooping to 4.7mm when using supports:


Image showing droop (the overhanging section):


Warping of the corners is mentioned above. Not only did I get blemishes on the side finish, but every part warped of the print bed.

Image of part pulling the magnet sheet of the printbed, when cooling, due to warping corners.


Printing these parts, I had two mid print crashes as well. Mid print a 8.5 hour print.

Screws can only be printed six at a time, at the most. Even then, prints often fail.

Image of failed print, due to layer shift;


There is a 4mm overhang to fasten glass or side panels. With the warp, there is about 2mm overhang left.

Image of ugly corner cover mounted:


Splitting the corner part in two, and printing it with only horizontal orientation, reduces the flaws greatly. This is the warping side of things, so warping is still clearly visible in the corners of the parts, if you look for it:

Image of nice corner covers:

 

Some images of the progress of some days ago:















I have installed a build into it, and gotten some RGB to work. I printed some mesh panels as well, and to be honest, mesh feels more sophisticated than untinted glass. Custom tinted glass will not come cheap. My gut feeling is to try to get some nice RGB fans, a nicer water block for the GPU, a rgb res pump combo, and a rgb pump. Not to mention, that I need a ton of more filament.

I have filled and drained the build a number of times now, and it such a breeze compared what is has been like with a regular case.

I forgot to mention, that I decided to replace the four main corner parts, as to remove unwanted holes, which was four times 21 hours print, and a ton of work replacing them corners. Once it was mounted, I realized that the one corner tops, had warped unexpectedly. If I ever fix it, heaven knows. The top IO section needs some love, but Fusion 360 is undergoing maintenance, so I simply cannot get that moving as of late.

Writing this on this build. Fans spin at about 330rpm, pumps at 800rpm. If max stressing a 3900x and a 5700xt at the same time, pumps reach about 1800rpm, fans 800rpm, and ambient is about 28c. Water temp is kept at max 40c. Water delta is about 2-3c before and after rads, at full stress test workload.

 

I didn't realize it was that big until the parts went in.
Side panels: No love for plexiglas? I think this would look great with red panels.

 

Red panels? I got a spool of red PETG. I might give it a go. It is a bitt dull, being all black.

The panels are like a window, the thing you look through to see the internals. Not sure how colored panels effects transparency of a mesh panel? Best thing is to try it out in real life, just play with it, get a feel for it, and take it from there. If the panels holds the color of the design, there will be no color for a glass version.

Also, colored PETG filament tend to be heavily saturated, as with little black in it. It quickly forces a high contrast expression, which is not my style. I do not design a clown suite, in which every color enhance each other, which is why a clown suite typically has the same saturation level for each color. I prefer to vary hue and saturation for each and every color, as it offers a way more relaxing and calm expression. A preference that leaves me scratching my head, trying to find colored PETG. Why this love for clown design by the filament makers?

The only easy safe bet at the moment, is grey. Which is like the definition of a dull color. Color = 0%.

As for plexi, because of the situation we are in, sourcing tools for cutting plexi is my main challenge. There is one supplier offering cheap transparent plastics at 4mm, I just do not have experience cutting it, and it has a slight but ugly tint. I could use a power saw, but need the right blade for it, as not to melt the plastics. There are knives, not just sure what to use. Any suggestion for a quick and dirty way to cut 4mm acrylic would be highly appreciated. Just remember that resourcing tooling is a struggle at the moment.

Regular 4mm glass pre cut, is about 300NKr pr sheet. Cheap plastic is 139NKr, and I need to cut it myself. There is just a shipping cost of 600NKr pr order for the glass.

As for size, this is enormous for a printed case. It is 103 liter including feet, 89 liter excluding feet. It is 54mm wider than the absolute minimum width, due to ATX form factor, and the dimensions of the main pillars. It is 15mm taller than the absolute minimum in height given the rad height and the dimensions of the main pillars. Also, it is the absolute minimum in width, due to the rads in use. Dept is the width of a PSU (+2mm), 3 pillars (3*37mm), 10mm extra in the front to accompany taller GPUs, and 56mm from the rear of the MB tray to the fan. Them 56mm are like golden, when building in this thing.

I've also noticed, that the cleaner I make the design, the smaller the components feel. Time to get rid of some mess.

 

I have just updated the top IO section. The goal was to make the design cleaner and more calm internally, fix some usability issues, and improve external finish. Please keep in mind that I try my best to make a universal design, not a one-off. Supporting the most hardware, and sticking to standards, is thus vital.

The ATX case standard, is a standard for sheet metal production. In particular, the thickest walls I can make, is 1.2mm, or only 4 layers when printing at 0.3mm layer height. This might work for metal, but at these dimensions, plastic is flimsy, at best. Working with the standard, is like trying to put lipstick on a pig, and as we all know, it's still a pig. There are so many concerns getting this right, this post would need to be a book, to do it any justice. I rather just skip that, and show the parts, and offer some limited reasoning on the changes being done this time around. Keep it more blog like.

The internal side of things, of the IO section, is split into three parts.

Image of the new internal Sim cover:


Image of the new internal slot cover:


These two parts are split where the rear IO cutout meets the slots.

Image of joint between the slots and sim parts:


Image of front panel section, the third main internal section, slots into the sim section.

Image of the three main internal IO parts:


Cables are now properly managed, running from the front IO. In the latter image, cables run between the two holes marked red, the two to the right. The issue with the last design, was that it did not support enough cabling, and I did not like the look of it. Space is now plenty. The red circle to the right, highlights an area problematic for usability, but needed in some cases for structural sanity. In the old design, the top cover was fastened here, but that resulted in the outer slot cover could not be removed separately. Both the external sim and slot cover had to be removed, to remove the slot cover. By removing the fastener, but keeping the lip, the design is structural sane for the top sim cover, for the cover used in this design. And it can now be removed separately. The hole is needed for a cover exposing all the slots.

The internal IO parts, the slot and sim parts, are joined by two screws.

Image of slot and sim part joint:


In the previous design, the lip to right in the latter image, was a part of the top sim cover. It was there to align the outer sim cover, with the other parts. It turns out, in real life, that it was not needed. In the new design, it used to fill the gap between the front IO section and the two other parts. It now also function as a support for the added internal mesh, of the internal sim IO part. It is a bit thin in some places, which is just a result of it being a mod of previous designs, and this design not being made for this ground up. It works just fine, so I do not bother redesigning everything to make this a fatter section.

The issue with the old design, was the internal mess it created, which is not the expression I am seeking.

Image of the old IO section, internally:


The old IO section looks much like the usual mess in a PC case, with functional parts all over, plenty of surfaces, no lines or clear surfaces. A mess of functionality. It was actually rather clean at that, but just not my cub of cake.

The old slot section, was functionally sound, but distracted fromt he general expression I am seeking. Just a big sinkhole.

Image of the old slot section, internally:


In the end, the old design generated this big sinkhole focus point in the top.

Image of old IO top, internally:


In the revised design, surfaces are less distraction, and more flush with the internal top. I have kept the hole for the slot fastening, and kept the front panel section open internally. Keeping those two, makes a ton of sense, trying to working in this case.

Image of the new IO section, internally:


I probably should try to source black screws for the slots, and black m3 nuts. Other than that, this is probably as good as I get it. The design target is not to flash the design of the internal top IO, but for people not to notice it. It should be calm, make sense, and make sense while working with it. I actually have spent a ton of time on this IO section, which probably is the most problematic and ignored section of case reviews and case design. In the latter image, there is a full cover GPU block, a CPU and WRM block. It will be rather hard to device the tubing for that, any cleaner than this. Sort of a hint of why I built this, in the first place.

If removing the two outer IO covers, everything is easy to work with, and should now support a good usability experience.

Image of top, with outer IO covers removed:


Note that I have not made cabling tor the power switch or cmos switch yet, as these are on the back of the rear IO of this particular MB. I recess the CMOS switch, to make it impossible to press it accidentally without a tool. Them buttons on that rear IO are insane, just begging for an accidental push of the clear CMOS, at the wrong time, bricking the MB. I only have a 500mm USB 3.0 front panel cable, and it is too short. I have not made a USB C port on that front panel section, as I do not have such a cable. The point is, that the front panel part is small, and easy to replace and redesign. It is not structural stressed, suitable for a PLA print, if needed. This is the reason for it being a separate part. The front panel IO part is designed to be user redesign able, to suite future needs and personal custom taste. Like a big ass power switch, if people so desire.

The front panel IO top surface is recessed. This allows for parts to protrude a bit, and for the case to still be able to flip with the top down. This is why the front panel IO is recessed.

Image of front panel IO, recession marked in red:


There was another issue in the old design, and that was structural support for the last slot, as this is an eight slot design.

Image of old design, problematic are in red:


This is solved by adding a tiny bit to the second part of the internal slot section:

Close-up image of the added lip to hold the slot cover:


Since the holes for the devices to slot into, below the mother board, is actually below the printbed level of the main slot cover, I split this section into two parts.

Image showing a 2mm trim of the main internal slot section:


Image of the part that goes into that trim:


Image of the two internal slot parts and the internal sim cover fit together:


Despite all my best efforts, this is a still a pig with lipstick. The internal side is the nicer one.

Image of the inside of the internal slot IO section:


The external side of things are only so-so. Note that I used the top surface ironing feature for the slot covers. For this particular use-case, it made a ton of sense. It is worth noting that the uglier part is actually the GFX itself. This is expected. I am not in love with these slot structures, as they are messy. A pig is still a pig, even with lipstick on.

Image of slot cover:


Image of external finish of the slots section:


Since the IO is on the top of the case, routing the external cables needs a solution. Mine is that channel for external cables, running straight through the case. That channel is essential for any general design of this type, to work. It is a pain to design. In real life, cables are simply pushed up the channel from the bottom. The taller clearance below the case, allows for this to be done, from the front of the chassis. Even a CAT5 cable is easily pushed up this channel. No need to flip or move the case. The path from the IO ports to the channel, is by design short, at the top.

Image of top opening cable channel and top IO:


Moving on to the external IO covers, this used to be four parts. It used to be so, to allow for the smaller parts to be user redesign able. I never got it to work, and scrapped the idea. This allows for a cleaner top section, removes the need for four out of line screws, and removes some uneven joints on the top.

Image of old design of smaller parts:


The old design had this fastening point that needed a removal of the outer sim cover for the slots cover to be removed.

Image of old fastener, marked by red:


The old design was a two slot design. While most cards only have IO for one slot. Thus, I redesigned it for single slot. The old design needed some realigning of mesh patterns, after the redesign into two parts. Some mesh holes were removed to mirror the pattern across parts. This includes the front IO panel. The outer IO parts are split by case center, which is not where the internal IO sections are split, this is a carry over from previous designs.

Image of old IO design. Red lines show holes removed in new design, and the difference of split of parts:


Image of new external SIM IO cover:


The external SIM IO cover is fastened to the internal SIO IO by two screws. This helps combating warping defects. A bit of a struggle fastening them screws.

Image of IO parts, holes to join inner and external IO covers marked in red:


In the latter image, the I had forgotten to canfer the edge to the left in the picture. I had to redesign the part, only adding that canfer, and reprint the outer SIM IO cover. That is like a 5 hour print.

Image of outer SIM IO cover in assembled (the one lacking canfer)


In the latter image, the mesh patterns aligns fort the patterns of the corner parts, the areas behind the IO section, and the pattern of the front panel part, align with the external SIM cover part. The joining of the parts, now results in the screws are now in sync pattern, front to back. The end result being less messy, more calm and structured. The latter image shows what is accessible by removing just the outer slot cover.

Adding the outer slot IO cover, and I finally get the expression I am seeking. It literally enclose the pig inside a box.

Image of fully assembled top IO:


The move to single slot for the GPU, fully functinally supports most GPUs, and for those that need more opening, that can be designed easily. I really like this. It is super tight, nice to work with, and I think it is a nice design to a messy situation. The ression of the IO, also hides the contacts, exposing only the cables. I really like the balance this strickes. Note the short distance the the external cable channel.

Image of GPU IO:


Rounding up this post, the main challenge designing this, is the maximum print size of parts, and the puzzle it forces sections like this into. That said, the warping and drooping is not solved by bigger prints, just enhanced.

 

This is now getting somewhere. The vision I had for this case, is starting to show. This is now getting fun.

First of, is fixing one of two issues with the PSU and MB structure. The front cover for the MB cable channel, I found it too short. Sure it support full EATX, or whatever to call it, but for regular ATX, If felt too short to me.

Image of old MB front cable cover:


I made the lower part taller, and extedned the opening at the top.

Image of lower MB cable cover:


I made some finer adjustments to the upper part, and canfer the edges.

Image of upper MB cable cover:


Image of the two MB cable cover parts together:


Image of new MB cable cover assembled:


For some cheap extensions and otherwise just stock cables, this is not that bad at all. Swapping MB would probably generate a similar result as well.

There is still some slight droop of the left PSU cover, one that I plan to fig by simply adding some tape to the top of the PSU. The droop is there, as to allow some tolerance for PSU variance in height, and by experience, there is a need for such a tolerance.

I have made a new cover for the bottom left cable channel, to support the pump and thermal sensor cables. There need to be a monitoring of water temp after the pump, as the pump is max rated for 50C. There is an inline sensor mounted before the last bend.

Image of cabling and tubing of the rear pump:


Note that for this build, the fitting use in the latter image, is on the super complicated side, for this build. This is as bad as it gets. For this build.

Cabling for the front pump and RGB is routed into the PSU cover. It is a bit longer, a bit more visible. In the end, not what most people would notice that much.

Image of cabling for the front pump and RGB:


Stil to be fixed, a cover for the RGB strip, and extending the RGB cable to run appropriately to the MB.

Image of horrible RGB tape hack for the res:


Image of both pumps and res assembled:


Please do note how water flows downward with gravity, and the super complex tube runs in this build.

The simplicity of this design, well, the apparent simplicity, is starting to pay off. That pump cabling in the front, not the main feature really. The simplicity of them tubes runs, just exactly what I have been targeting. No need to explain how water flows, or tube runs for this one. That's not what this design is made to flash off. It is meant to be the underlying expression, facilitating main components show off.

Front view image, no panels:


To be able to plug in the power cord, there is a need for an angled connector.

Image of angled connector on power cord:


Image showing tight fit for power cord:


Image of PSU chamber without power cord:


Image of PSU chamber with power cord mounted:


All external cables exits the case at the bottom. Cables are messy, but the nature of things.

Image of totally unorganized external cable mess of fully working computer:


Some might object to the cabling being on the top, calling for the beauty of cable spaghetti at the rear, dangling in mid air. I hocked up what I usually do, minus the LAN cable. It gives me this mess:

Image of top fully populated with cabling:


Adding a few more cables really do not alter much.

With a case like this, a desktop USB hub makes sense.

Image of messy desktop with USB hub:


Image of a desktop without any cable management at all:


In the latter image, the focus in the case, will be the parts that are highlighted. The noctua fans is an obvious candidate for some RGB ones.

Image of front view with side mesh only:


Image of side view with side mesh:


Image of right side view, with mesh:


Image of desktop with full mesh:


A word on that mesh. The mesh area is almost four times the max size of the print bed of my 3D-printer. The mesh pattern is super tailored to allow transparency, and to compensate for elephant foot. Every print layer is tailored for the mesh pattern. As in adding a canfer on the back to each and every hole in the mesh, to increase transparency at an angle, yet keep some structural integrity. Not to forget, the case can be flipped on the sides, with the mesh in place. By design.

Close-up image of front with mesh:


Image of left side, bright ambient light, full mesh cover:


Close-up image, front, side mesh only:


Image of desktop, all covers, no mesh:


Image of PSU section, no mesh:


Image of rads and over complicated tube runs:


Close-up image of rad tube runs:


Image of right side view, without mesh covers:


In the latter image, only one tube run is hidden from view. Most exposed cables are in that image.

Draining this build, is first done upright, like in the image above. It drains the bulk of water, but far from the lot. As is always the case, for any water cooled build. The valve for the res is almost the highest point in the loop, the drain valve is the lowest point.

Image of chassis flipped on right side for draining:


Flipping for draining, the res will now be empty. The res valve will be the higher point in the system, the drain valve will again be the lowest point. Gravity forces water to flow sequentially through the rads and the GPU and CPU block. As far as this is possible. The res valve functions the same way in both orientations. It lets air into system, allowing flow to the drain valve. If it wasn't clear by now, this The Core Feature of this design.

This case is built for water cooling. That is, to be filled, bled, and drained. Most of the water of the rads, is actually drained. The air is easily bled, read, the use of two pumps in serial.

Personally, I favor the tint effect of the mesh, dimming the internals, highlighting the RGB by a night effect. Leaving me in a bit of a problem. Where do I get affordable slightly tinted glass? Custom cut at that?

Also, when using RGB, the all black works. Focus is drawn to the lightshow. Any tint on the mesh covers, and that will tint the color of that light show.

Replacing the pump top and pump-res combo with something current from EKWB would be awesome. I got a 5700xt ROG, and the water block for that, is RGB plexi, and would kick ass in this build. Adding some Corsair RGB fans on them rads, would also spice things up at bit. And some Corsair RGB RAM. Unfortunately, Corsair do not offer RGB pump tops, at least not last time I checked. If I can find the finances for it, I will give it a go.

 

This build might appear dead, and it sort of is.

Some geometry issues arose, due to warping, and I did some redesigns of the corner covers to combat the issues. This lead me to cut the number of nuts, and I really loved the result. Long story short, I decided to go for a 3 layer construction incorporating gliding covers, to hide almost all screws. Also, I continued working on the mesh, as I simply did not quite like it. It struck me that I was not exploiting the strengths of the plastics, and started experimenting with a wire-like mesh instead. I finally had gained modeling proficiency to achieve such a design. That lead me to drop the mesh technique used in thus far in this build, to drop it -completely. In the end, the use of plastic screws and embedded nuts, a 3 layer design dedicated to even out geometric inaccuracies, and the use of a wire like mesh, all of which I seem to be pioneering, means that almost every part of the old design is redesigned. To save time, I rather did the exploring and development of the techniques, on my air cooled design, as it is smaller and easier to work with.

Also, I really want to get my signature design out the door, and rather implemented the new techniques into that design first. When i return to the Back Burner jr. design, I will need to redo it ground up, as everything is changed a bit. The design is pretty much finished, but modeling it is a pure pain and labor intensive.

I rather walk through the new design and technique, in a blogg making the first Side Burner implementing it. Not sure if there is a any use continuing on this blogg after that one. Time will tell.

 

You can drop it here and add a link in the first post to the rebuild point.