![\"\"](\"https:\/\/jimsghost.place\/wp-content\/uploads\/2021\/09\/2021-09-02_001.jpg\")
<\/a><\/a><\/a>When needed, any such group can be extracted from the overall model and examined individually, down to each of its elements. These elements, in turn, can themselves be groups of objects or standalone entities.<\/p>\n\n\n\n
This approach makes it easy to navigate through the 1,258 independent parts that make up the printer (assuming the object counter in the software isn\u2019t lying):<\/p>\n\n\n\n If you need to locate a specific piece in any printer assembly, it\u2019s as simple as navigating the object tree and clicking each time on the group containing that piece \u2014 no need to struggle with remembering the name of the file where the part is stored.<\/p>\n\n\n\n Let\u2019s take a closer look at a group of components I\u2019ve loosely labeled as the “printing assembly”:<\/p>\n\n\n\n Without any need for visual cues or annotations, it\u2019s immediately clear what the core elements of this group are: the filament spools mounted on holders at the rear of the printer, the filament feed motors for each spool located at the top of the printer, and the print head prominently positioned at the center foreground. While these main elements are not physically connected to one another, they all serve a unified purpose: to take a piece of plastic, deliver it to the heating element, melt it, extrude it onto the print bed, and cool the resulting output.<\/p>\n\n\n\n In the 3D model, every object is represented exactly as it exists in real life. Each part has precisely the number of screws and nuts it needs, exact virtual dimensions that match its physical counterparts, and is interconnected in a way that allows for realistic motion on the screen, just as it would in the real world. Everything is presented with meticulous accuracy \u2014 except for one thing: the filament feed tubes. These are represented in a more abstract manner:<\/p>\n\n\n\n As in, “They exist, and they start from here.”<\/p>\n\n\n\n I could have modeled them to scale, extending from the print head to the feed mechanisms across the entire printer. However, doing so would have made working with the model significantly more cumbersome. Constantly adjusting and examining the model to ensure there are no conflicts during motion \u2014 ensuring one component fits with another, and that some unexpected screw head doesn\u2019t collide with another part \u2014 becomes impractical. These types of conflicts are simple and quick to fix in the drawing stage but much harder to address on the physical device in the workshop. A small oversight in the model could lead to hours of extra manual labor. Consequently, the more I manipulate the model with the mouse, the less I have to manipulate it with a file on the workbench.<\/p>\n\n\n\n The filament feed tubes are among the few flexible components in the printer. Around 99% of the other parts are “rigid”\u2014they don\u2019t change shape no matter how you move them. Pull one rigid part with the mouse, and all the connected parts follow along seamlessly. Modern desktop computers can handle such calculations effortlessly, even on modest configurations. But these tubes are flexible. When they move, and as all objects connected to them also move, it\u2019s not just their relative coordinates that change\u2014their very shape is altered. The computer has to recalculate every polygon and vertex comprising the tube model, account for material resistance and deformation properties, check for surface intersections (both internally and with neighboring objects), and even factor in a simulated gravitational force that influences the direction and shape of each segment of the model.<\/p>\n\n\n\n These computations demand significantly more processing power. A quick mouse movement (even an accidental one) can trigger a chain of events that takes the computer seconds\u2014or even minutes\u2014to process before displaying the next frame. As a result, quick manipulations of the model become anything but quick and certainly far less comfortable. No matter how much one might strive to create a perfectly accurate project model, certain compromises must still be made for the sake of usability.<\/p>\n\n\n\n Whether filled with compromises or not, it was finally time to take all these components out of the computer screen \u2014 and place them on the workbench.<\/p>\n\n\n\n This process began with the print head block \u2014 as it is the most complex part, composed of elements where each plays a critical role in this circus:<\/p>\n\n\n\n The most crucial element here can be considered its base, which has a rather intricate shape:<\/p>\n\n\n\n The most challenging part about this base was that it had to be made perfectly on the first try \u2014 without any errors. First, I had only one remaining aluminum “brick” left from the batch I had purchased for the project. But that was just half the trouble \u2014 I could always buy more (albeit at the cost of time and money).<\/p>\n\n\n\n Far more critical was the fact that the allocated personal-use hours on the company\u2019s CNC milling machine were about to run out. At home, there was no way I could carve out such a complex shape. I had literally two or three Saturday hours left to manufacture this final part on the company equipment. Immediately afterward, the machine \u2014 without a moment\u2019s rest \u2014 would be switched to production mode for company orders, running nonstop for the duration of the client contract. There was no telling how long I\u2019d have to wait for another “window” to slip in my personal project.<\/p>\n\n\n\n That\u2019s precisely why I wrote so much earlier about creating a complex yet precise model. Without this part, the entire project \u2014 yes, the entire 3D printer project \u2014 would lose its meaning. All the time spent so far would have been wasted. The project\u2019s not-insignificant budget would have gone up in smoke. That\u2019s why everything had to be taken into account, not just for this specific part but in terms of its interaction with all the other printer components \u2014 both those already completed and those yet to be made. Every single detail needed to be thoroughly tested and reviewed at least a hundred times in the model. There was no time for a second attempt.<\/p>\n\n\n\n I think it\u2019s obvious to everyone that I succeeded. Otherwise, this article series wouldn\u2019t exist. Nor would that grumbling printer with the odd name Marsh Turret \u2014 which is currently brooding over its latest protective panel for the electronics.<\/p>\n\n\n\n The base itself is attached to the X-axis carrier platform in a straightforward manner \u2014 with four bolts and nylon spacers:<\/p>\n\n\n\n What\u2019s far more interesting is everything that gets mounted onto the base:<\/p>\n\n\n\n At the center, there\u2019s something resembling a small syringe \u2014 this is the BLTouch. It\u2019s a probe sensor that allows the printer to “feel” the entire print bed surface before starting a print. With this data, it can determine not only the zero point for the Z-axis but also any potential unevenness across the bed, which can be accounted for and corrected during printing. Flanking the probe are the print heads, equipped with their respective radiators and heaters.<\/p>\n\n\n\n The intricate design of the mounting base must not only hold these three key components together but also allow for precise adjustments. This is critical! The nozzle tips of the print heads must be perfectly aligned at the same height. If not, the second nozzle might scrape off the material deposited by the first, or worse, it could dig into the printed layer and block the printer’s movement entirely. Either way, the print would be ruined.<\/p>\n\n\n\n The distance between the nozzles is equally important. In this case, it\u2019s exactly 50 mm. While it\u2019s possible to adjust this value in the printer\u2019s firmware, I prefer to avoid dealing with odd numbers like 49.34 mm. A clean, even number like 50 mm is both aesthetically pleasing and practical, and there\u2019s no reason not to set it precisely during the installation of the print heads.<\/p>\n\n\n\n The vertical distance from the BLTouch probe tip to the nozzle\u2019s end is another critical parameter. If this isn\u2019t set accurately, the printer will print either above or below the level “felt” by the probe during its operation. Again, this can be adjusted in the firmware, but such settings are meant for fine-tuning, such as compensating for minute discrepancies caused by nozzle heating, not for the primary calibration of the assembly.<\/p>\n\n\n\n These adjustments don\u2019t need to be made often \u2014 only during nozzle replacements or after particularly catastrophic incidents (like the time a cat decided to nap on the printer\u2019s warm bed mid-print \u2014 true story from the Black Widow, where the entire X-axis was knocked completely out of alignment). For such occasions, I\u2019ve created a special tool, which I\u2019ll discuss in the “Accessories” section of this article series. It significantly simplifies the calibration process and makes it nearly foolproof.<\/p>\n\n\n\n Next up\u2026 Bowden tubes for filament feeding.<\/p>\n\n\n\n These were something entirely new to me. Until now, I had never dealt with this method of delivering plastic to print heads, and I had to learn a lot about the pitfalls in this area. Thanks to the internet, I didn\u2019t have to discover all these pitfalls the hard way.<\/p>\n\n\n\n It was immediately clear that these tubes should enter the print heads as vertically as possible and that their ends needed to be perfectly flush to ensure there\u2019s no gap between the tube and the nozzle:<\/p>\n\n\n\n What was less obvious (though somewhat expected) was the fact that the fittings meant to hold the slippery PTFE tubing in place aren\u2019t all created equal:<\/p>\n\n\n\n That\u2019s putting it mildly\u2026 The overwhelming majority of fittings that come bundled with print heads are nothing short of useless junk! One way or another, the tube wiggles, twists, bends, and generally behaves like a worm on a frying pan. A standard cheap Chinese fitting usually lacks the “backbone” to keep the tube firmly in place and prevent it from misbehaving. Inevitably, a gap forms between the tube\u2019s cut and the nozzle, where molten plastic builds up. The extruder thinks it has pushed out a certain amount, but in reality, some of that material doesn\u2019t come out of the nozzle \u2014 it simply fills the gap, squishing around like oatmeal in Sir Henry\u2019s boot. This leads to gaps or over-extrusions in the print.<\/p>\n\n\n\n Thankfully, I didn\u2019t have to experience this disaster myself. After seeing plenty of examples from others, researching the topic, and heeding advice, I invested in a high-quality set of fittings recommended by “leading printer experts.” I made the right choice! These fittings feel noticeably sturdier than the ones that came with the print heads. The “collars” you press to release the tube click distinctly, like good buttons on an old-school mechanical keyboard. Meanwhile, with the stock fittings, you can barely tell if you\u2019ve pressed the miserable thing or not\u2026<\/p>\n\n\n\n But I didn\u2019t stop there. If you look at the photo above, you\u2019ll notice that before the tube even reaches the fitting, it passes through another clamp (the gray one). With this additional mount, the tube has absolutely no chance to wiggle in its seating. When the screws on this secondary clamp are tightened, all possible lateral forces are removed from the fitting on the print head. This way, the fitting handles only pulling forces, while the additional mounts absorb any side-to-side stresses. Together, they hold the tube so firmly that it has no chance of going rogue.<\/p>\n\n\n\n Somewhere around this stage, the puppet finally turned into a real boy. For minimal 3D printing capabilities, all it took was temporarily attaching a couple of fans \u2014 on a wing and a prayer \u2014 to cool the radiators of the print heads:<\/p>\n\n\n\n That\u2019s it! The printer was already capable of doing its job. It could even attempt printing something super simple\u2026<\/p>\n\n\n\n At that point, since the printer still didn\u2019t have its own spool mounts for filament, the “feeding mother” turned out to be the frame of the old Black Widow standing next to it:<\/p>\n\n\n\n This gave my friend \u2014 who was closely following the project \u2014 plenty of material for snarky comments, like tabloid-style headlines: “The corpse of the old Widow feeds the new marsh!” and so on.<\/p>\n\n\n\n The system for feeding filament from the spool to the print head was completed back during the assembly of the Z-axis. While it doesn\u2019t logically belong to the Z-axis group, it made more sense to install it at that stage for easier assembly.<\/p>\n\n\n\n For the filament drive, I opted for more advanced dual-gear extruder blocks with a shared gear coupling between the two shafts:<\/p>\n\n\n\n By comparison, the old Black Widow only had a single drive shaft. It pulled the filament by pressing it against a polished passive roller on the opposite side. Despite the Widow\u2019s extruder motor being equipped with a reduction gear, the filament still couldn\u2019t be fed through the nozzle with much force. The shaft would simply slip against the filament, failing to push it into the “combustion chamber.” To compensate, I had to heat the plastic significantly above its normal working temperature to increase its flowability.<\/p>\n\n\n\n For example, I had to overheat PLA all the way to 235\u00b0C \u2014 which is utter nonsense, considering its typical working temperature is 190\u2013210\u00b0C. Honestly, 235\u00b0C is even too much for ABS. At that temperature, PLA was practically on the verge of disintegrating.<\/p>\n<\/blockquote>\n\n\n\n Somehow, I managed to adapt back then, and it worked\u2026 But for the Marsh Turret, I decided not to adapt to anything. Instead, I adapted the printer itself for proper operation from the start.<\/p>\n\n\n\n On top of the dual gears, I installed full-sized NEMA 17 motors for filament feeding instead of the smaller “cut-down” versions (like the 17HS12-1204S and similar) commonly used. These are the same motors that power the entire movement system of the printer.<\/p>\n\n\n\n Yes, it\u2019s overkill. Yes, sometimes it feels like the extruder could push filament through the nozzle without even melting it. But I\u2019d much rather deal with that than the chronic under-extrusion I constantly fought against with the Widow.<\/p>\n\n\n\n Of course, the journey didn\u2019t end there. A few temporary fans and feeding filament from the Widow\u2019s “corpse” were all well and good, and I even allowed myself almost a week off to tinker with the Marlin firmware. At long last, I could work with a printer that physically existed!<\/p>\n\n\n\n However, the print head assembly was still far from complete. Beyond fixing the “temporary fan setup” into something more permanent, the print head also needed a proper layer cooling system and some lighting.<\/p>\n\n\n\n For this, I designed a custom attachment for the print head assembly. To ensure ease of maintenance in the future, this part was made fully removable and self-contained:<\/p>\n\n\n\n This module incorporates no less than four fans: two standard fans for cooling the print head radiators and two “blower” fans for cooling the printed layers.<\/p>\n\n\n\n Naturally, the base for this construction wasn\u2019t made from aluminum using a CNC machine. Given the internal geometry of the air ducts, that would have been far too labor-intensive.<\/p>\n\n\n\n Instead, all the parts for the fan modules were produced on my photopolymer Photon printer.<\/p>\n\n\n\n While that worked out fine, the Photon\u2019s resin vat limits the maximum print size, which isn\u2019t very large. In theory, I could have squeezed the housing into two parts, but I decided to play it safe and split the structure into four sections, which could be printed in pairs:<\/p>\n\n\n\n After printing, the parts were glued together to form a single unit:<\/p>\n\n\n\n Painting, assembly, and installation followed:<\/p>\n\n\n\n I reused the same clever trick for the print head radiator fans that I had previously used on the Widow.<\/p>\n\n\n\n Under normal circumstances \u2014 when the heating block is functioning properly, equipped with a suitable heat break tube, and printing occurs within a stable temperature range \u2014 the radiators of the print head barely warm up. They feel just slightly warm to the touch and don\u2019t require intense active cooling. A gentle airflow is more than enough to prevent heat buildup.<\/p>\n\n\n\n If the radiators become so hot that you can\u2019t touch them and need constant, intense cooling, something is seriously wrong with the system. The problem isn\u2019t the airflow \u2014 the issue lies within the print head itself. Such overheating of the radiators is abnormal and must be fixed.<\/p>\n<\/blockquote>\n\n\n\n Typically, the cooling fan is mounted to blow air onto the radiator. At first glance, this seems logical. However, in practice, the airflow from the fan breaks against the radiator and assembly, scattering in all directions. Most of it ends up blowing directly onto the print bed, simply because there are fewer obstacles in that direction.<\/p>\n\n\n\n When printing with plastics like PLA, this isn\u2019t an issue. Such materials benefit from cooling during printing and often require additional cooling systems. But things are very different when printing with ABS or similar plastics. Forced cooling during printing is a disaster for these materials. They shrink unevenly, causing the print to crack, delaminate, or detach from the bed, turning the print into a frustrating mess. For such plastics, it\u2019s crucial to isolate the printing area from drafts entirely, allowing the part to cool gradually and uniformly. Clearly, a fan blasting air directly onto the print bed doesn\u2019t contribute to a stable temperature environment.<\/p>\n\n\n\n To minimize this unwanted airflow, I “invented” a brilliant trick \u2014 I flipped the fan around. Instead of blowing air onto the print head, it now pulls air away from it. The radiator still cools effectively as air is drawn through it, but this way, the airflow from the fan becomes much easier to control and direct away from the print bed.<\/p>\n\n\n\n<\/a><\/figure>\n\n\n\n<\/a><\/figure>\n<\/figure>\n\n\n\n
\n\n\n\n<\/a><\/figure>\n\n\n\n<\/a><\/figure>\n<\/figure>\n\n\n\n<\/a><\/figure>\n\n\n\n
\n\n\n\n<\/a><\/figure>\n\n\n\n<\/a><\/figure>\n\n\n\n<\/a><\/figure>\n<\/figure>\n\n\n\n
\n\n\n\n<\/a><\/figure>\n\n\n\n<\/a><\/figure>\n\n\n\n
\n\n\n\n<\/a><\/figure>\n\n\n\n<\/a><\/figure>\n\n\n\n<\/a><\/figure>\n\n\n\n<\/a><\/figure>\n<\/figure>\n\n\n\n
\n\n\n\n<\/a><\/figure>\n\n\n\n<\/a><\/figure>\n\n\n\n<\/a><\/figure>\n\n\n\n
\n\n\n\n<\/a><\/figure>\n\n\n\n\n
\n\n\n\n<\/a><\/figure>\n\n\n\n<\/a><\/figure>\n<\/figure>\n\n\n\n<\/a><\/figure>\n\n\n\n<\/a><\/figure>\n<\/figure>\n\n\n\n<\/a><\/figure>\n\n\n\n<\/a><\/figure>\n\n\n\n<\/a><\/figure>\n\n\n\n<\/a><\/figure>\n<\/figure>\n\n\n\n<\/a><\/figure>\n\n\n\n<\/a><\/figure>\n\n\n\n<\/a><\/figure>\n\n\n\n<\/a><\/figure>\n<\/figure>\n\n\n\n\n