Scratch Build – In Progress BBJ-G

The problem is, I had that check time automatically box, checked. It just didn't. Never occurred to me that the time was off. The OS did not realize that I had switched MB, and had not scheduled a time check. At least no I have up to date chipset drivers, LAN drivers, GPU drivers, BIOS, brand new installation of Fusion 360, and so on. Just a heads-up: If login fails, or windows update fails, check to synchronize the clock. Back to build progress.

Another front corner is printed.

Image of bottom front corner part:


The crossbar in the front bottom, is also done.

Image of bottom front crossbar:


Some explanation is needed. The Backburner is a mod of the Sideburner. The Backburner uses the back part of this project, in both the front and the back. I decided to reuse the top and bottom parts, which means that the holes need to align. That means that the external holes need to align, and that the holes on the inside, those running perpendicular to the surface ones, are in the same place. This introduced quite a lot of challenges. It also maintain symmetry, which I love.

Image of the bottom, showing the symmetry of the holes:


Also, the front part is reused for an air cooled case. I simply wanted more room for taller GPUs, and CPU air coolers, and needed to add space for it. This is why the front part, in the bottom, is as thick as it is. This gave me another challenge. The screws. I decided to stick with 5mm-5mm-6mm screws, as I use them a lot. Reuse of parts, keeps the part count low. This result in the screws being sunk 10mm into the plastic. It really does not matter, as it is covered by a front cover.

Image of Screws being sunk into the plastic in the front:


I needed to spend quite a lot of time, to get the feet working. As I need quite a lot of clearing for the external cabling underneat the chassis, I need tall feet. That leaves me with an arm and torque issue, that I simply do not trust for a single screw joint. This is the reason why I use three screws, as to make it safe to tilt the case on its feet. This is a water cooling case, and that is what you do, when filling and draining the case. Since the placement of the holes, are at different distances from the edge of the case, what worked for the Sideburner, did not work at first for the Backburner. It was really fiddly to get this to align, and to keep the expression I was seeking.

Image of corner hole alignments in the front, red marking the distance to the chassis edge:


Image of corner hole alignments in the rear, red marking the distance to the chassis edge:


Please note that since I use the same corner in both the top and bottom, I need to counter sink the screws in the top. I thus fill inn the missing part in the bottom, with a washer.

Also, if you look closely, a lot of the screws have their threads in the crossbars, thus, the corner parts do not have threads on a lot of the holes. As some of the holes have covers on top of them, they are not countersunk. Since I have standardize on a design for these parts, using standard modules, everything falls into place nicely. It is not that complex, once comprehending the design.

Closeup image of bottom surface (print bed finish), and holes with threads and without thread.


The top-type surface, that is visible through the glass, is in it self not that smooth. But the pattern is calm and repetitive. It sure helps printing the crossbar in the same pattern, as everything just blend into this repetitive thing. It is not that bad in the end, and not something that stick out that much.

Closeup image of bottom front internal finish:


Image of progress this far, from the front:


Image of progress this far, from the side:


Image of progress this far, from the bottom:


Please note that this is designed for usability. It is designed to support standard components, as far as possible. Ease of maintenance is a core design goal. Form is secondary to function.

The assembly is rigid at this point, and surprisingly light weight. The new design for the screws and holes for the driver, really works out well. Things are coming along really great.

 

Hi, really like Your approach. Function shape the form ! Bauhaus and industrial movements !

Gteat


Was all the concept of the last build i done

 

Superb build! as well as all the rich info provided about your choices of design and material.

 

Thanks.

I am not into design history or even design philosophy, and have no formal education on the topic. But I did search a bit, and as I seriously design for symmetry, I guess I do not fall into this category? I spend a ton of effort and resources on form, just even more on function.

Thanks.

 

Next up were the feet. I design for access to the PSU from the bottom. I design for external cabling to exit the chassis in the bottom. I need some clearing under this thing.

Also, this box, when populated, is heavy. Like 20+ Kg. Tilting tall feet, and there is a ton of force on them feet. The taller the foot, the bigger the problem. As by force times arm, which is basic physics. This is why I use a slanted foot, with three screws. I also liked the feel of using them. I had some drawings with the box going all the way down, I just did not like them as much. There is a danger of ending up design with a cupboard feel to it, as feets like this, has been used for ages in furniture.

I find, that the supply of of the shelf feet, are limited. They are particularly limited in their shapes, and the ones that are available are hard to work with. I have spent a lot of time working with how to solve this, and going all printed parts, makes the most sense, but turns in hard. Hard to get to fit an expression, hard to print, and hard to mount. For example, I learned the hard way, that long screws do not work for plastics, as the material do not transfer torque well for such a design.

As explained in an earlier post, the placement of the holes are different for all the feet. Since the front is 47mm, I need to stick with that all around, as to keep the feet somewhat similar. A foot is fastened with three short screws. The one in the corner is inserted in the big hole in the center of the foot, as the gauge of the threads for the screw holding the rubber sock, is m24, with 3mm thread.

Image of a rear corner part:


Since the slanting sides are 10mm wider than they need to be, in the rear, they cover some of the mesh.

Image of mounted rear foot, covering some of the mesh:


Closeup image of foot covering mesh:


Another issue I now suddenly had, was that the foot was protruding into the PSU access opening. It was a hard 90deg edge, that just felt like a pain work with. I tried going back to 37mm on the sides, but didn't like it. I ended up with a canfer, or a 45deg cut, to soften the protruding, and get rid of the stupid and irritating edge the 90deg resulted in. It is somewhat hidden from view. Do note that there is a lot of warping going on, on all parts of some thickness, particularly on the print bed side. In this case, we got two perpendicular faces, that both are print bed facing, on thick parts. It shows, if you look for it.

Image of edge canfer for the foot covering the PSU access opening:


Another image of the same:


To be able to insert the screws, through the threaded hole for the rubber part of the foot, I used m24 with 3mm thread. People need to realize how crazy that is, if using metal. As mentioned about torque equals force times arm, well, to tap in metal, at these dimensions, the tooling is somewhat huge. This build look a miniature build, as in tiny, using this tool.

Image of the insane tapper tool, tapping m24 hole in the foot:


The feet are designed to be flush with the final exterior, which is 6mm wide.

Image showing the 6mm frame overhang by the foot:


I have also printed the last cover for the rear mesh.

Image of mesh in the rear, with new central part mounted:


Image of cable channel, from the inside:


Image of progress this far, bottom view:


Image of progress this far, front view:


Image of progress this far, side view:


I am also in the prosess of getting more printers up and running. This is a lot of work.

Image of a Prusa being assembled:

 

Loving the progress

I just jumped on for a second printer too These things are a licence to print money i almost thought about a clone prusa 3 ! sooo cheap and soo good but ended up with a Ender v2 just for the upgrade path.

It must me starting to get some real weight to it by now? It's looking very solid

 

Thanks.

As for weight, it is probably closing in on 3Kg by now. As I am about to head into spool number 4, if I recall this right. I will update this with a table, down the road. That table will include the print time as well, and an estimate for print time with 0.4mm nozzle. Each foot took about 6 hours and 20 minutes. There are four of them, thus in pure print time alone, it took more than a full day to print just the feet.

As for solid, it just about solid enough. Once all the covers are on the bottom section, it is stiff in both x and y, but not z. (z as in height). There is some flex, if you push hard, to the bottom on the z axis, but nothing that I find concerning. As for the sections running top to bottom, they do flex, at this stage. About a 2-3mm, when force is applied. The solid rear, top, and bottom sections, are vital to counteract this. It is only 40% infill, that is, once you get past the perimeter, it is 60% air.

The only real concern I got, is the risk of a parallellogram effect, on the side glass. That might break the glass, when stress is applied to the case. I learned about this affect the hard way, once I tried to assemble a tall Ikea kitchen cupboard with a friend. We tried to raise the thing of the ground, before mounting the back plate. Since there was no real support to prevent a parallell shift, the 233cm cupboard simply collapsed, by first gliding into a parallellogram, and then the corners snatched on us. It actually fell on me. A 60*60*233cm cupboard. Just two clowns assembling a kitchen, or rather trying to assemble.

There is a reason for me loving these covers, preventing parallell shifts. The rads in the rear, play a similar role. This is by design. Just like a back plate of a cupboard is essential for structural integrity, the covers for the cable channel plays the exact same role. This is why it is backed into the corners. If not doing so, the entire structure is depending on the corner rods, and the corners, which simply would need to be stronger than these are.

 

Happy update. The printer part finally left the UK, after being stuck at the airport like forever. Hopefully I will get a few more printers up and running, making some real progress.

 

My Prusa simply hates elastic filament. I have to print without any retraction, and slow as a lazy snail.

The filament I use, is hard rubber, and pretty slippery on many types of surfaces. I simply need a different extruder, one is in the mail. For now, these will have to do. Due to the lack of any retraction, I need to print one part at a time. I have just printed four, one for each foot. The priority is to get the assembly on its feet, as not to scratch the bottom while working on the build.

Image of foot with rubber sock:


Next out, is the frame middle corner parts. This is one case, in which functions trumps form. Printing this in one piece, makes it significantly stronger. It also introduced two different print finishes, as one is print-bed-finish and the other is side-finish. The lines of the side finish, runs perpendicular, to the lines of the top and bottom corner parts. The only trade-off as for form, is the surface finish. Given how much stronger the part is, being printed as one, this is a fair trade. It is visible, but it one of the issues when printing things like this chassis.

Image of middel corner part:


Image of middel corner part, print-bed-finish side:


Image of middel corner part, side-finish side:


Image of assembled middel corner part:


Image of progress this far, front view:


Print time for this part was 7 hours and 36 minutes.

 

M24 with a standard screwdriver head. You really should make a much wider screwdriver slot and make a wider tool to turn those. -Or make the classic coin slot keyway.
I've noticed some twisted screw slots already.

 

I sort of lost you on this one.

Most of the screws show no wear at all, even after the abuse I have given them.

The m24/3 is far less demanding to tighten than a m14/2, as applied for this build. Far less torque is required to fasten a plastic screw, than a regular metal one. They do not come loose, as for instance the m5 of slotted aluminum do.

My current design, is just about wide enough for the driver bit. The driver bit, needs to be forced into the screw head. Also, the opening is slightly slanted, as is the surface of the flat bit. This result in a bigger contact surface. Screws are designed for a specific drive bit, at 8mm wide. The fit is super tight.

Image of the cross section of a 20-5-5-6 nut:


Since the screws are never embedded, they can universally be scaled for the x-y plane. I scale them to 98%, and keep the z scaling at 100%. This leaves me with just about right tolerances, for this build.

I tap the embedded nuts, ensuring they are up to spec. I scale the cross section down to 98% to get the right tolerance. I need to use one, and only one, driver bit, as the hole in the screw head is design for just this bit. The fit is super tight for the bit, as it needs to be, as not to deform the screw head.

That being said, any plastic screw can be deformed by using excessive force. I am rather surprised how much force that is, given that this is a key weakness of the material used. Plastic will wear quicker than metal.

I have no idea how to design a universal coin slot, that would be anywhere near the design I am currently using. The fit will end up loose. The contact surface will thus be tiny, increasing the pressure on that smaller surface, resulting in deformed plastic.

 

It has been a long day, so I need to make it short.

I thought I elaborate a bit more, on the screws, and how this version has worked thus far.

If nuts are embedded into the plastics, I want the screw to fail, and not the other parts. Screws are far easier to replace. If using a metal screws, stripping plastic nuts, is a real possibility. At current, I am not that far off starting to strip threads. I have yet to strip any threads. The only issue, is the head of the nut. I have never ever had issues with the screw threads or the nut thread failing on me, not as a result of use or excessive force.

Image of what to expect, within a few mounts and unmounts:


At one point, I tried to force the screws, trying to see if I could force them flush with the surface. The examples below, are all of abuse, not of accidental mishaps.Note that the nuts are fully functional, even after this abuse. No biggie to loosen the screw.

Image of what to expect, when you are stupidly seriously forcing it, and try to stop, when you feel something went wrong:


Image of stupidity, when you deliberately force the nut to deform:


Note that in the latter image, it seems like the driver is not sticking that deep, so it seems I got even further adjustments to make. The opening is nice and tight, both in x and y.

Image of bit inserted into screw:


Image of bit inserted into screw:


Image of bit inserted into screw:


I use a bit, as I need one shape for all tools. That be a short driver, a slim long one, or a wrench. A wrench is needed for tight places, to achieve enough torque.

Image of bit, adapted to wrench:


I know from experience, that if I use a bit that is 1mm narrower, it eats plastics. This is the first build I use this design, after a ton of failed designs. This actually works, but only with this particular bit.

As for progress. I have printed 3 middel corner parts without extra holes, and one with holes to mount the reservoir.

Image of corner part with holes for res, top view:


Corner part, side view:


Progress this far.

 

*Two posts later.* OK! It's just a suggestion for a client version. In my experience, they never use the right tool, and always use brute force.

 

As often, You're often right (I said often to not say always)

 

Anyone using this build, has to use a specific bit, if not, they will ruin the case. I guess that has to be seriously stated, like a million times, if these ever get sold to clients.

Another screw update.

As I mentioned yesterday, it seemed like the hole for the driver bit, was too narrow, and the bit did not hit the bottom of the hole. This is getting crazy. I had adjusted the scaling from 98,5% to 98%, without compensating for it. The driver only got 3mm in. By adding 0,098mm in width, for both x and y, for the hole only, fixed it. That's crazy accurate, as the diameter of the nozzle is 0.6mm or more that six times of that adjustment. That is how close the tolerance is at the moment, less than 0.1mm, for the bit in use.

The good part, is that I really struggle to deform the screw head now, using a driver. I tried to abuse a screw, using all the force I could. The result below, is a worst case scenario, one of absolute abuse.

First image of nut deformation, when applying stupidity of force on the screw with a driver:


Second image:


I also get a serious grip and torque given the driver used.

Image of driver, bit, and deformed screw:


That is a bloody good result. After all, this is the main weakness of this material. Unless you are Brutus, these screws now takes a serious beating. Also, I added .098mm in the longer direction, as not needing to force the bit into the hole. That probably weakened the joint of screw and bit, but the risk of scratching the screw trying to get the bit in the hole, is like gone now.

I finished another rear corner part. The top right rear corner, is the same as the lower left rear corner part. Progress is slow, as parts for more printers are still in the mail, progressing like a snail, due to the current stress on shipping services. Do note how the different surfaces reflect light differently, and as I have tried to explain in detail, there is no remedy for this, as this is the nature of 3D-printing.

Image of progress thus far:


That it for now. I only need to redesign like 15 different screws now, but I am actually proud of this result. The way I see it, I actually just made plastic a real viable option, even for more heavy usage. Screws will now actually take a beating, and their lifespan is now great. The way I see it, this is good enough. Also, I will share the source code for making these, as well as what I have done. The reason being, that if a client are to trust a build like this, he really should be able to print new screws down the road. Given prints need to be within 0.1mm accuracy, sharing the source is the only reasonable path forward. Just one more thing, on the to-do-list.

 

With a larger build plate and a good amount of supports you could print that whole thing in one shot ! looking amazing so far

I just brought my third printer ! here comes the print farm

 

Thanks. I took a huge risk making this, as it takes a ton of time and resources. Really appreciate it, if people actually likes this, otherwise it has been such a gargantuan waste of time.

As for bigger printer, the ones that are sanely prized, all have moving beds. The increased momentum, is a real issue, at that bigger size. Also, the risk of a crash will increase, and when it happens, it will ruin the print. That said, with a general frame like this, who would not try to fit a 3D-printer inside of it? The only challenge is designing the heatbed, the rest is plug and play by now.

Also, this build could be printed without hardly using any support at all, in a few prints. It is specifically designed to support that, if I ever get to designing my own printer.

 

I have been bussy altering some features of the design, like the nuts. I have decided that the orientation of the PSU is unsafe. Not of the risk of water leaks, but since I simply have too many near misses, in which screws and nuts nearly falls into the PSU. Rotating it, opens up opportunities, and needed a redesign. Just generating stl and gcode files of all these parts, takes like forever.

I also have had a few misses. I forgot to include a tolerance of the tap facing the glass on some parts. Eight hours later, printing a part, it is trash. I need for it to be 14 layers, or 4.2mm. It was at 3.9mm, which sort of is too narrow for 4mm glass.

Image of part with lack of tolerance for the glass, I need to cut of a layer on the protruding edge:


Also, this very part had a near crash, in which a blob of filament built up on the nozzle, and fell on the part during printing. I tried to cut off the protruding part, but it didn't turn out to be a winner.

Image of result of a near crash:


As can clearly be seen in the image above, once a near crash like this occurs, it messes up the nearby printing as well.

I finally got some printer parts form E3D, and started mounting a couple more printers. It takes like forever, particularly since I got the Prusa kit, and since I am replacing the heatbreak. I have been constantly printing, none the less. The main corners and crossbars of the top section is printed, as well as a ton of screws. I did all the tapping at once this time, and it took a few hours alone.

Image of progress thus far, top front view:


Image of progress, front view:


Image of progress, side view:


Please note in the latter image, that the covers for the cable channel are not done, caused by the one print that failed. I like the calm look of the second or internal mesh, as it generates a more uniform plane or surface. The internal mesh, has no real function, and is purely a formative addition.

The crossbar for the MB tray, is recessed 4mm. This is done, as the backside of the mother board tray, is one surfaces of the channel for the external cables. The MB tray is 8mm thick, this to achieve enough stiffness, and to accommodate a locking screw for the inserted m3 nut. More on that later.

Closeup image of section one half of the MB tray is fastened to, 4mm recessed:


Wider view image of same MB crossbar:


That is all for now.

 

I hope not !!!

Just past our 400th print on 5 printers So i'm hoping that one day i can print this up overnight

 

This build is a bit challenging. I simply have had to address a ton of issues, and it has been quite a messy journey. I have simply not had time to blog, as I simply need to get this done. I have taken some pictures along the journey, and will rather try to share them in retrospect.

Lets start with the feet. They were modeled just fine, but warped. They simply looked ugly, particularly neighboring other problematic warping parts.

Image of warping foot:


There is a lot going on here. The main issue being that I had printed near identical feet before, without this horrendous result. It seems like the adding of 10mm in width pushed this filament over some edge. There is supposed to be no gaps at all in the image above. It just looked dreadful.

I had noticed that the top surface of prints tend not to warp, and decided to give it a go. Please note that printing one foot, takes about 7.5 hours. I have never heard about this trick to fight warping, I just went for it, as it was all I had.

Image of foot being printed with support, as in up side down for anti-support-mindset:


As expected, the bottom of the foot now warps in the corners, but it hardly shows in normal usage, and the foot still works as intended by desing.

Image of foot with everything fixed (like a million hours later):


The difference is quite dramatically improved. Note the slight canfer at the foot outer edges. That is, this is a water cooled build, and when filling draining, or maintining this beast, the box can be tilted in any direction, resting on the rubber "socks". By cutting off some of the corners, the hard plastic never hits the table.

Image of foot fully dressed with socks:


At this point I thought I was done, but no. I swapped the position of the pump, and made an accessible drain port. That was similar to the one hole for the PSU. This meant that I needed to print another foot, with a canfer to get rid of some hard edges. So, in short, there are four feet, but I have printed 9. They take about 7.5 hours each. That is excluding the rubber part and the screws.

I have struggled for ever with the settings for elastics filament on my Prusa. Turns out, they use 15% too much filament, for their stock settings. I just didn't think they could be that dumb. Never occurred to me. After getting extrusion right, that is the right amount of filament, I dried the filament, and it printed somewhat nice. The printer still jammed, every once in a while, and it was snail slow and laborus. I actually spent an entire day focusing on printing these rubber bastards, ending up with just 3. In one day.

It turns out, that this filament should be dried at 100C and not at 65C, and that it is a requirement for every print, every time, for any success. I had to look up drying guidance of the filament manufacturer, to find the issue. (as in reading the instruction, hidden in a table on how to dry their filament, that I needed to search actively for) I also got a Hemera, but looking at the quirks for installing it on the Prusa, that to is quite labor intensive. I have not gotten to it yet. I need to print at way to high a temp, and that makes the filament cook, resulting in small bubbles in the end print. It is not to bad, but at the current speed, printing these "socks" takes more time than printing an entire case.

Close-up of corner with socks:


Please note that the image above is of a top corner. This box can be flipped on any side.