![\"\"](\"https:\/\/jimsghost.place\/wp-content\/uploads\/2021\/09\/2021-09-06_002-1-640x329.jpg\")
<\/figure>\n\n\n\n![\"\"](\"https:\/\/jimsghost.place\/wp-content\/uploads\/2021\/09\/2021-09-06_003-640x329.jpg\")
<\/figure>\n\n\n\n![\"\"](\"https:\/\/jimsghost.place\/wp-content\/uploads\/2021\/09\/2021-09-06_001-1.jpg\")
<\/a><\/figure>\n<\/figure>\n\n\n\nWhen we first got our hands on serious laser engravers at work, I instantly fell in love with their control interfaces:<\/p>\n\n\n\n I\u2019ve always adored toggle switches and buttons in their classic form from the era before the mass extinction of technical designers in the late 20th century. Nowadays, you only really find them in industrial equipment\u2026<\/p>\n\n\n\n Naturally, about a year or so ago, during a visit to one of the local shops selling second-hand electronics junk, I couldn\u2019t resist saving a button I found sitting forlornly among a pile of modern plastic garbage. I didn\u2019t even know what I\u2019d use it for at the time, but I just couldn\u2019t leave it behind. It was a bit worn, of course, but industrial-grade components last an incredibly long time \u2014 this button will likely outlive all of us. Until now, it had been resting quietly in my stash, waiting for its moment in the spotlight. I pulled it out, cleaned it up, and washed it with alcohol \u2014 now it looks as good as new:<\/p>\n\n\n\n As you can see, I slightly bent the truth earlier when I said that all the components for this project were bought specifically for it, with none taken from my stash. This button, however, is the exception. It\u2019s just too good to pass up. I\u2019ve been waiting a very long time for it to find a truly worthy application\u2026<\/p>\n<\/blockquote>\n\n\n\n This button is designed to handle high-voltage circuits. It doesn\u2019t just break a single line \u2014 it cuts both at once. That is, it simultaneously disconnects both the live and neutral wires. And the best part? There\u2019s a neon lamp inside it. Literally. Not one of those cheap “LED backlit button discount 20% special offer” things. This means it doesn\u2019t require any additional wiring or separate power supply for the illumination. The backlight is powered directly by the same 110 volts from the outlet that the button itself controls.<\/p>\n\n\n\n In the printer, the button occupies its own housing on one of the front corners of the frame:<\/p>\n\n\n\n At the time of installation, the printer wasn\u2019t even close to functional \u2014 it couldn\u2019t even “meow” on its own. So, the housing parts for the button were printed using a Photon resin printer. Both the base and the cover:<\/p>\n\n\n\n The outlet block was made in a similar fashion:<\/p>\n\n\n\n This part is straightforward. It\u2019s just a standard IEC320 C14 “computer” outlet with a simple 15A glass fuse (6x30mm).<\/p>\n\n\n\n When power was finally connected to the printer, it was time to load it up with the rest of its components: power supplies, the controller, MOSFETs, and all the other essential elements.<\/p>\n\n\n\n The space designated for housing all the electronics was located under the printer’s workbed. There\u2019s plenty of room there:<\/p>\n\n\n\n Although, as you can see, while there\u2019s generally a lot of space, it wasn\u2019t possible to cram everything into one spot. I had to split the electronics into several groups:<\/p>\n\n\n\n Yes, this layout results in slightly longer wiring between some components. On the other hand, no matter how you arrange it, the total amount of wiring would remain the same. A few extra inches of length on some wires don\u2019t really make a difference, especially since most branches already measure in feet. What matters is that the entire setup is unified. It\u2019s a complete, self-contained device \u2014 “unplug it and carry it away,” with no external power supplies, detached units, or other add-ons.<\/p>\n\n\n\n The foundation for mounting all the internal electronics consists of two large aluminum panels integrated into the printer\u2019s frame:<\/p>\n\n\n\n These panels were marked and drilled in advance, before installation:<\/p>\n\n\n\n In other words, there was no “on-the-spot drilling” with sudden metal shavings flying over some circuit board. Thanks to standardization, this could all be prepared beforehand, even though some components were still en route in postal trucks at the time.<\/p>\n\n\n\n And of course, the printer doesn\u2019t rest its considerable weight directly on these panels. It stands on sturdy legs (originally intended for “garage” shelving and rated to hold much more than the weight of the printer and its “considerable heft”):<\/p>\n\n\n\n So, the panels are essentially suspended beneath the printer’s frame. However, they do have small additional legs of their own, which can be seen in the area around the central crossbeam (the small black ones). These legs are more for reinforcing the U-shaped assembly of the Z-axis, which is mounted right in that section of the frame.<\/p>\n\n\n\n The power supplies for the printer were chosen with a significant safety margin. For example, Black Widow, also a 24-volt printer like Turret, only needed a single 600W power supply, which was more than sufficient.<\/p>\n\n\n\n However, Turret has two print heads, adding an immediate 40W to its power needs. Its heated bed is also more powerful and energy-hungry \u2014 220W compared to Widow’s 160W. Plus, Turret has an additional motor. It\u2019s not a huge load, of course, but an extra 1.5 amps is still an extra 1.5 amps\u2026<\/p>\n\n\n\n By my calculations, a single 600W power supply should have been enough for MARSH TURRET with a 15\u201320% power reserve \u2014 after all, not all components of the printer operate and consume power simultaneously. But, as we all know, everything is made in China. What calculations suggest will give you a 15\u201320% reserve might, in reality, be more like 5\u201310%. Or worse, nothing at all. And when that happens, the power supply might release all of its Magic Smoke and stop working entirely.<\/p>\n\n\n\n An 800W or higher power supply would have been a safer choice. Unfortunately, options of that capacity in the same size as a 600W supply either didn\u2019t inspire any confidence or had prices that shot straight into the Constellation of the Toad. So, I arrived at the solution of using two power supplies, splitting their roles, and ensuring a comfortable power reserve for each.<\/p>\n\n\n\n For example, the current setup for the heating elements is as follows: 220W (bed) + 40W (first nozzle) + 40W (second nozzle) = 300W out of 600W. This leaves a 50% power reserve for the heating supply, even when all heaters are running simultaneously. The situation is similar for the second power supply. I can\u2019t overstate how much this simple math improves a DIY enthusiast\u2019s peace of mind during long, multi-day print sessions.<\/p>\n<\/blockquote>\n\n\n\n As planned, both power supplies fit perfectly in their designated spots under the workbed:<\/p>\n\n\n\n On top of everything else, I really liked the design of these power supply cases. First, no random holes in the walls where debris, screwdrivers, dog drool, or stray fingers could fall through. Air is drawn in through a neat grill at the rear, just above the terminal block (which has an insulating cover), flows through the entire case, and is expelled by a fan at the front. As far as the case and exterior go, there\u2019s absolutely nothing to modify. They are self-contained and perfectly functional as is. Sure, the covers are made from basic thin sheet metal, but the load-bearing walls are solid aluminum.<\/p>\n\n\n\n To control the heaters, I used off-the-shelf modules based on MOSFETs:<\/p>\n\n\n\n My Black Widow printer had been running for years with its bed controlled by a module like this one:<\/p>\n\n\n\n I have no idea what specific MOSFET it uses \u2014 never cared to check \u2014 but these modules are fairly standard and are considered reliable workhorses for controlling heated beds in printers. They\u2019re often referred to simply as “MKS MOSFET,” likely because they were initially developed for Makerbase controller boards. However, nowadays, you can find them in almost any setup, even if the mainboard has nothing to do with MKS. It\u2019s a solid, time-tested option that I trusted for MARSH TURRET\u2019s heated bed.<\/p>\n\n\n\n For controlling the heating elements of the print heads, however, I considered using such a module overkill. Why would they need such a beast? More importantly, three of these bulky modules wouldn\u2019t fit in the designated area of the frame, unless stacked like a sandwich \u2014 and that\u2019s a terrible idea.<\/p>\n\n\n\n So, for the print head heaters, I opted for smaller, simpler modules:<\/p>\n\n\n\n These are supposedly also advertised as being suitable for heated beds. But I seriously doubt they could handle a bed of any reasonable size or power. Sure, they could manage it, but they\u2019d probably heat up as much as the bed itself. However, for the print head heaters, with their modest 40 watts, they\u2019re perfect.<\/p>\n\n\n\n In the end, all three heater control modules found their places:<\/p>\n\n\n\n Since these articles are written retroactively, it\u2019s worth noting that this entire thermal management system has already undergone extensive testing, inspection, multimeter measurements, thermometer checks, Dog sniffs, and Cat stares. It\u2019s been deemed safe, reliable, and fully operational.<\/p>\n\n\n\n Why didn\u2019t I power the print head heaters directly from the controller, as is commonly done? After all, the controller has decent (relatively speaking) MOSFETs of its own. The heated bed, sure \u2014 everyone uses an external MOSFET for that. But I haven\u2019t seen many setups that also route print head heaters through external MOSFETs. However, as I mentioned earlier, I was determined from the start to fully separate the power systems for the heaters and the control electronics. Completely. No mixing. Like flies and cutlets. Yin and Yang. Avada and Kedavra. Two power supplies, each powering only its designated group of devices, with no overlap or interconnection. And that\u2019s why.<\/p>\n<\/blockquote>\n\n\n\n Strangely enough, the group of control modules is the most boring part of the entire printer setup:<\/p>\n\n\n\n And that\u2019s it. No need to invent or come up with anything. Everything\u2019s already been invented and designed. All you have to do is put it in place and connect the wires.<\/p>\n\n\n\n About three years ago, I burned out the Widow\u2019s original controller. Essentially, it was the same standard MKS-base Mark-something-or-other from one of the earlier revisions, slightly modified by the manufacturer to suit that specific printer. It was simpler and faster to replace it with a “standard” one. By that time, “standard” controllers had become much more advanced and capable. Over the years, the new controller proved to be incredibly reliable, bug-free, and required absolutely no attention.<\/p>\n\n\n\n Back then, upgrading Widow\u2019s controller was a quick, 30-minute plug-and-play operation. I removed the dead controller, installed the new one, plugged in the connectors, flashed the firmware \u2014 done. I even used Widow\u2019s stock firmware, which was essentially a preconfigured version of Marlin. I didn\u2019t have to change a single line of code in the configuration. The only issue was that the mounting holes on the new controller didn\u2019t align with the original mounts. Since I was too lazy to adjust them, the new controller has been hanging off Widow all these years, attached by one screw and a bit of hot glue\u2026<\/p>\n<\/blockquote>\n\n\n\n After scouring the internet to see “what\u2019s new since then,” I was disappointed to find that there\u2019s essentially nothing revolutionary. Sure, they\u2019ve been playing with form factors, experimenting with motor drivers, and introducing 32\/64-bit processors that are questionably relevant for this field. But fundamentally, nothing groundbreaking has appeared. It\u2019s safe to say that the technology for simple Cartesian printers has reached a stable, finalized form. Over the past couple of years, it\u2019s mostly been about adding gimmicks like “now with Bluetooth and banana flavor!”<\/p>\n\n\n\n So, in this case, I chose not to follow the typical DIY enthusiast path of “something new, untested, and exciting.” Instead, I went the boring route of an industrial equipment designer: “what\u2019s tried, tested, and works well.” I found the exact same controller model from the same manufacturer that I had installed on Widow to replace its original:<\/p>\n\n\n\n However, I had to configure the firmware for it from scratch. By now, a solid second version of Marlin had been released. And, of course, the new printer\u2019s setup was quite different. So, I had to dive deep into the configuration file and meticulously adjust almost everything in it.<\/p>\n\n\n\n Now, onto TL-smoothers \u2014 what they are and why they\u2019re necessary.<\/p>\n\n\n\n Bipolar stepper motors, which move the entire printer, are notoriously noisy. Personally, I couldn\u2019t care less about the noise itself. However, that loud whining is merely a symptom of a bigger issue. The motors don\u2019t just produce audible noise \u2014 they “scream” electronically, too. The interference they generate in the circuit is enough to make you want to lie down and give up. Currents jump back and forth, inducing, reflecting, and ricocheting off each other \u2014 chaos! This affects print quality. The parasitic microvibrations (the ones you can hear) cause visible artifacts on the printed surfaces \u2014 “waves,” “ghosting,” “reflections,” and other well-documented defects.<\/p>\n\n\n\n A TL-smoother, placed between the motor and its driver on the controller, is designed to suppress much of this noise and interference. This not only quiets the motor\u2019s operation but also significantly reduces the aforementioned print defects. At its core, it\u2019s an extremely simple filter assembly made up of a bunch of diodes:<\/p>\n\n\n\n You could easily put one together yourself with whatever components you have lying around<\/a>. It would work just as well. But as a ready-made module, it\u2019s more convenient and compact:<\/p>\n\n\n\n The cost of such a module is negligible \u2014 around six to nine dollars for half a dozen of them. However, it\u2019s highly recommended to add a heatsink on top of the SMD diodes because they tend to heat up quite a bit during operation.<\/p>\n\n\n\n Whether TL-smoothers are needed in a printer is a purely personal choice. Many keyboards have been smashed in debates among DIY enthusiasts over this. On one hand, the motor driver on the controller is supposed to handle all this “nonsense” on its own. If it doesn\u2019t, it\u2019s a lousy driver. And I agree with that. However, from personal experience, even with a good “quiet” driver, having this module in the circuit undeniably improves print quality. I\u2019ve had the chance to test this both with Widow\u2019s original controller \u2014 equipped with drivers so terrible it\u2019s unclear why they were even there \u2014 and with the new controller, which had decent drivers. In both cases, having the TL-smoother in the circuit made a difference you could both hear and see.<\/p>\n\n\n\n For me personally, I\u2019ve decided that TL-smoothers are a must-have in my printer, no matter what controller or motor drivers I use.<\/p>\n\n\n\n And finally, the last component in this group \u2014 the Dual Stepper Motor Module:<\/p>\n\n\n\n It\u2019s a trivial piece, honestly. Nothing you couldn\u2019t rig up with a soldering iron and some wire. In fact, calling it a “module” feels like a stretch. It\u2019s really just a convenience tool for wiring.<\/p>\n\n\n\n The Z-axis on my printer is powered by two motors working in sync. However, the controller only has one driver and a single output for the Z-axis motor. There are two ways to connect two motors to one driver: wiring their windings either in parallel or in series.<\/p>\n\n\n\n Both options are viable, but each comes with its own caveats. With parallel wiring, you maintain a lower (half of the nominal) but consistent torque at any speed. With series wiring, both motors achieve maximum torque at start and stop, but the torque drops significantly as speed increases.<\/p>\n\n\n\n For a 3D printer, where “better to be steady than strong” is often the guiding principle, parallel wiring tends to be the better choice. If one motor can handle the entire axis on its own, two should manage it just fine, even at half power each. On the other hand, this is the Z-axis we\u2019re talking about, which moves far less frequently than the other axes, and its travel distances are measured in millimeters or fractions thereof. For the Z-axis, the ability to deliver strong bursts of torque at start and stop points becomes more relevant, making series wiring seem more appropriate. But then again\u2026 the Z-axis typically uses a leadscrew, whose properties make questions about “start and stop bursts” more theoretical than practical. If the motor drivers on the controller can supply enough power for two motors, why impose limits on Z-axis speed and travel length?<\/p>\n\n\n\n Ultimately, I decided to go with parallel wiring, as it\u2019s the simplest solution. Thankfully, there\u2019s a handy module available for contact routing. If it turns out that parallel wiring doesn\u2019t provide enough power and the driver can\u2019t deliver additional current, I\u2019ll rewire everything in series.<\/p>\n\n\n\n Unfortunately, I couldn\u2019t find any pre-made modules for series wiring, so I\u2019d have to make one myself if needed. Not a particularly complex project \u2014 just a board with three connectors and a few traces\u2026<\/p>\n<\/blockquote>\n\n\n\n As it turned out, everything worked perfectly in the simple parallel configuration. I didn\u2019t even need to crank up the driver\u2019s output power (though, for safety\u2019s sake, I increased it by 30% above what\u2019s needed for a single motor). At the same time, the driver doesn\u2019t heat up any more than it does when running a single motor (as mentioned earlier, the Z-axis doesn\u2019t operate often or for long). Still, I added heatsinks to all the drivers and set up active cooling for them.<\/p>\n\n\n\n The printer has already been tested in a relatively “long run” (continuous printing for 3.5 days). No overheating, no missed steps, no anomalies of any kind. The Z-axis moves up and down with a speed I\u2019ve never seen before, not even on the Widow. For now, everything stays as it is \u2014 the parallel wiring setup works flawlessly.<\/p>\n\n\n\n The Display.<\/p>\n\n\n\n The display is positioned separately from the rest of the control electronics, on the front part of the printer\u2019s frame:<\/p>\n\n\n\n Years of experience have shown that when using a Raspberry Pi with OctoPrint as part of a printer setup, the role of the printer\u2019s onboard display becomes almost negligible. It\u2019s always easier, more convenient, and faster to control and configure printing from a large computer screen with a mouse using an intuitive interface, rather than scrolling endlessly through a million text menus on a tiny printer display. For the same reason, I see no point in having a graphical touchscreen onboard. Even though it\u2019s a bit more advanced, it\u2019s still small, and its interface is never as convenient as an app on a proper computer. Not to mention, such screens cost twice as much and come with so many quirks that configuring the controller firmware to support them could drive you mad.<\/p>\n\n\n\n That said, the printer still needs an onboard display with basic controls, at least as a backup system. If the network goes down or some other issue severs the printer\u2019s connection to the outside world, the onboard Raspberry Pi with OctoPrint becomes useless. And, let\u2019s be honest, having a dedicated display on the printer simply looks cool.<\/p>\n\n\n\n This “coolness” was my guiding principle. Touchscreen color graphical displays don\u2019t fit my highly unconventional sense of aesthetics. They\u2019re instantly disqualified without discussion. That leaves the standard “pixel” LCDs. However, most of them are just plain ugly, especially those horrid blue-backlit rectangles that are considered “standard.”<\/p>\n\n\n\n Fortunately, persistence paid off, and I stumbled upon a very charming display from the Mini12864 <\/strong>series:<\/p>\n\n\n\n It\u2019s smaller than the typical displays for printers of this type, but that\u2019s perfect for a backup control element. The rotary encoder is on the side, not underneath the screen like on “standard” displays (which used to drive me nuts!). Plus, it has RGB backlighting for both the screen and the rotary encoder, allowing for color customization to suit any twisted taste\u2026 or so I thought.<\/p>\n\n\n\n In reality \u2014 yes and no. Well, yes\u2026 but also no. More precisely, yes, it\u2019s RGB backlighting, and you can configure it. But no, because you might not want to. If you connect this specific display (or a compatible model) and enable it in Marlin firmware, the default behavior is to have the display\u2019s backlighting change color according to the printer\u2019s current operation. For example:<\/p>\n\n\n\n And honestly, I liked this much more than a static, manually chosen color “to suit my taste.” I decided not to change anything in the configuration. I left all the backlighting settings as they were in the firmware. This kind of dynamic indication is far more intuitive than any text could ever be. A quick glance at the display is enough to understand what the printer is currently doing and why it hasn\u2019t started printing yet.<\/p>\n\n\n\n A sleek, unobtrusive display that has been officially recognized as a fantastic choice for my MARSH TURRET\u2026<\/p>\n\n\n\n The display has also been certified by the Cat. Unlike the Dog, the Cat\u2019s color perception is perfectly fine. He sat and stared at it for several minutes, watching the colors shift. At one point, he even pawed at the rotary encoder. He left with a very satisfied look on his face.<\/p>\n<\/blockquote>\n\n\n\n There\u2019s one more set of components I forgot to mention in the previous articles of this series. While these components belong to the electronics, they are still an integral part of the printer\u2019s axes rather than the overall system. So, I should have written about them earlier. But here we are\u2026<\/p>\n\n\n\n Endstops, or limit switches \u2014 I\u2019ll call them “endstops” for simplicity:<\/p>\n\n\n\n Printers need these to determine the zero coordinate, the reference point from which all printing will be based. For printers of this type, it\u2019s customary to set the origin point here, in this corner of the bed:<\/p>\n\n\n\n The printer itself doesn\u2019t care where the origin is. Humans, however, decided it should be here.<\/p>\n\n\n\n You may encounter printers where the origin is at the center of the bed or the far left corner, but this is rare. Such setups are usually dictated by technical features of the machine, where there\u2019s no other viable option.<\/p>\n<\/blockquote>\n\n\n\n The printer doesn\u2019t mind where it starts counting coordinates, but it must know exactly where to begin. It can figure this out in one of two ways: either a human tells it, or it tries to find an extreme position by itself and assumes that as the origin.<\/p>\n\n\n\n Humans \u2014 lazy meatbags that we are \u2014 don\u2019t enjoy manually setting the printer to its zero position every time. We\u2019d rather let the poor robot do that job for us. And that\u2019s where endstops come in. The robot moves one axis until it hits an endstop. When the endstop is triggered, the robot considers that point as the zero position for that axis. Then it moves another axis until it triggers its endstop\u2026 and so on for all axes.<\/p>\n\n\n\n Endstops themselves come in different types. They can be mechanical and contact-based, or “sensor-based” and non-contact. My Widow had the latter. They were absolutely dreadful! If you accidentally nudged them even slightly \u2014 just a hair\u2019s width \u2014 they\u2019d stop working. Worse still, if a magnetized screwdriver happened to be lying nearby, the sensor would fail to detect proximity and wouldn\u2019t trigger. I\u2019ve lost count of how many teeth were stripped off the belts because of that nonsense. This time, I decided: no sensors, absolutely not. Old-school, reliable “lever microswitches” \u2014 as the forefathers intended. And industrial-grade ones at that, built to last! Tough enough to chop firewood with or to use on space shuttles.<\/p>\n\n\n\n V-166-1C25:<\/p>\n\n\n\n I got the idea from one of our workplace CNC mills during a maintenance session. These exact switches were used there. I\u2019m not sure if they\u2019re standard for all parts of the machine, but they\u2019re definitely installed on the tool carousel.<\/p>\n\n\n\n For the Y-axis, installing the endstop was the simplest task. An adjustable aluminum plate was mounted on the frame, and the endstop was attached to it. It sits in place, waiting for the bearing block of the bed to bump into it:<\/p>\n\n\n\n In normal circumstances, the endstop is adjusted once and left as-is permanently. For this reason, I didn\u2019t bother with any complicated adjustment mechanisms.<\/p>\n\n\n\n The X-axis endstop, however, turned out to be a bit of a mess. The initial mounting design I came up with ended up being completely useless. I don\u2019t even have the model for it anymore, just a couple of photos:<\/p>\n\n\n\n What I was thinking when I made that “ear,” I\u2019ll never know. Securing something like that to a polished, well-oiled steel rod is obviously not a great idea. It would rotate, sure, but worse \u2014 it had a tendency to slide backward whenever the carriage bumped into it. Inevitably, I tightened the screw too hard, and the mount simply cracked\u2026 which isn\u2019t surprising given that it was made of photopolymer resin.<\/p>\n\n\n\n Fortunately, by that point, MARSH TURRET had already started attempting its very first prints. I temporarily taped the remains of the first mount in place and modeled a new one \u2014 more robust and reliable.<\/p>\n\n\n\n I\u2019m not entirely sure anymore, but this might actually have been the very first functional part that Turret ever printed:<\/p>\n\n\n\n All the usual test prints like boats and other random stuff came after this mount. The print quality at that time was absolutely atrocious, but it worked nonetheless!<\/p>\n\n\n\n This mount solved all the problems. Unlike resin, the plastic is more flexible and doesn\u2019t crack as easily \u2014 I can tighten the screws as much as I like. Now, the endstop stays firmly in place and doesn\u2019t budge.<\/p>\n\n\n\n The Z-axis endstop was even more interesting. It also needed to be mounted on a rod, so I printed a mount for it using photopolymer resin:<\/p>\n\n\n\n However, the rod for the Z-axis is much thicker, and the mount itself is bulkier. Overall, this version sits in place quite securely.<\/p>\n\n\n\n What\u2019s interesting here is that this endstop isn\u2019t even necessary! That\u2019s because of BLTouch:<\/p>\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
<\/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<\/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\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
<\/a><\/figure>\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\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\n\n\n\n
\n\n\n\n<\/a><\/figure>\n\n\n\n\n
<\/a><\/figure>\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<\/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\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\n<\/a><\/figure>\n\n\n\n<\/a><\/figure>\n\n\n\n\n
<\/a><\/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<\/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<\/figure>\n\n\n\n