With this lecture, we are shifting into the mechanical system to take a look at the motion mechanical subsystem that governs how a 3D printer physically executes the instructions. For how and where to deposit material delivered to it by the electronics control board. And as discussed in earlier lectures, the motion mechanical subsystem moves everything that needs to move. This function is obvious from the name of the subsystem, but I'd like to remind you that the three primary systems of a 3D printer: control, mechanical and extrusion, operate independently. So it is key that the activity of the motion mechanical subsystem syncs up to the extrusion system. It isn't just enough to move everything where it needs to move. You also need the tool head to arrive when it needs to arrive so that the machine deposits material in the correct places. It isn't enough of a challenge that these different systems run in parallel to the same clock. It is also necessary to plan 100% of the instructions for the motion mechanical subsystem before the print begins the job instruction file. From the moment the machine triggers the homing signals, the entire sequence of instructions will be executed. And there will not be an opportunity to check back in on how the paths and speeds are working out until after the part is complete. For thousands and thousands of lines of instructions, the machine will follow the plan you set at the beginning of this process in the control software when you slice the model to produce the job file. And neither you nor the machine itself can check into monitor if the instructions are being carried out successfully nor intervene. In fact, the only mode of intervening that is currently available to an operator is to pause or a border print. Pausing might work to move something out of the way such as a stray bit of plastic or cutting away a ridge that has worked up onto the path of the tool head. But then your only option is to resume the job instructions where you left off or to abort the project after all. And once the job instructions are truly stopped, the homing references and number of steps and micro steps advanced by the steppers are no longer confidently known by the machine. You will have to start the part over again. Sure there are advanced techniques to resume a print at the current layer. But remember that in order to do this, you need to confirm what layer you are on precisely. Which can be difficult when the layers number in the hundreds or thousands or anywhere from point 0.04 millimeters, 2.0-3 millimeters thick. And even then, you are typically restarting the instructions on that layer, meaning it's time to trust the job instructions again to complete your job without intervention. What you can't do is tell the printer say you are missing a bit on the side there and the side of this panel is a bit ugly. You're extruding that material here and your extruding too much material there. And a critical feature is missing here. Those kind of instructions are well out of scope for the present day technology. I'm not saying this to encourage fear and despair in the tools we have today. You'll find that the tools we have today are pretty great. And the desktop printers of today, which have been advancing since 2010 tend to complete jobs more often than they botch them. I'm providing this context to point out the level of precision and uninterrupted consistent function that is required for 3D printer hardware to work well and deliver excellent results, from the virtual to the physical. Bring those two considerations, parallel actions from system to system and lack of feedback and intervention in the printing process into the same engineering challenge. And you quickly realize that this is a bit like preparing for a space mission where the operation of a device is so far away. You have to trust that the machine you designed will function as intended without a human operator able to control it in real time. Let's review what we discussed about the coordinate system in the context of these job files and then advance the concept further. The coordinate system established for a machine identifies the homing position and how the machine understands movement in the three dimensions. It governs how a job file is executed. Imagine for a second that you are designing a machine in a virtual universe to autonomously navigate a virtual maze. Only your machine has no sensors and no brain making this more difficult than a smarter robot but not impossible. So what do you do? You analyze the maze, make your measurements and calculate a path the machine needs to follow in order to navigate the maze without crashing through a wall. You test your machine to ensure that its sense of where it is and how it must move to follow its instructions are spot on and that sorted. You activate your robot and head to lunch confident that when you get back the machine will be where you expected it in the virtual maze. However, while the metaphor for what you need to accomplish with a 3D printer is pretty spot-on in some ways, a 3D printer isn't in a virtual world. It's in a physical one. And while your slicer does produce a maze solution for how to fabricate your part, your path instructions and your execution are not precisely the same. A 3D printer drives around physical objects with real weights and is subject to friction. What's more? Your motor is to drive this behavior aren't on the tool head. They are elsewhere mounted on the frame of the machine. You have to use belts, cables, gears, or sliders to drive them from a distance, a motion train that itself will contribute to tiny imprecisions and friction. If your virtual machine solver follow the same path in the real world without considering momentum, weight and friction, those tiny issues would stack up and your machine was slam into the maze wall. When you design a machine for the real physical world, you need to prepare for the complications of objects moving through the real world, dealing with the momentum and friction and stack up of tiny imprecisions. You need to identify the risks, find ways to compensate and constrain the in precision motion lash, etc, for each element that makes it difficult to follow a sequence of precise plans you deliver with each new part of the maze instructions. Then and only then are you confident that your machine will follow with instructions to correctly go from homing to the completion of the part. Some of this work happens in firmware with adjustments for motion planning and similar such as acceleration strategies that ramp up and taper off. Speed along each movement axis so that the momentum introduced by the weight of the tool head doesn't cause the extruder to miss its mark when it's squirting out a very precise payload of molten plastic. A lot of time and energy for improving machine behavior goes into both firmware development and introducing compensations into the job instructions themselves. And the rest of this work is put into the mechanical design of the system and the choices of what kind of motion mechanical components are selected for your machine design. Cables versus timing belts, sliders versus rods Z+XY versus Z+Y machine designs, machine kinematics. Let's start off with degrees of freedom and offer insight into one of the most common confusions first time 3D printer audience's experience. Most desktop 3D printers are three degrees of freedom kinematic chains. But you'll notice that even while the machine designs may be arranged in terms of ZX+Y or XY+Z, most of the strategies end up separating the XY from the Z as far as path executions when running print jobs. These machines print entire layers before moving the Z. Unlike you might expect with certain multi axis CNC Mills and similar that might execute three axis moves and Mills and maybe even add in a lathe component or similar. Note that we aren't yet talking about the mechanical arrangement of the machine design. We are still just talking about what happens when a print occurs. There are a few reasons for this and this list essentially defines desktop 3D printing mechanical design. First, job preparation tools slicer software. The way most 3D slicer tools work is to divide the physical object into a tall stack of 2D problems. And while there is a bit of chicken and egg problem here, this arrangement is a perfect match for how FFF printing instructions function at this time. There is an additional computational consideration. It is easier to solve the problem of sorting what order to create outlines and infill on a series of 2D images. Tough as it is than to consider all of the possible routes for moving in all three dimensions. Second, tool head and obstacle avoidance, also tool insertion. There are certain types of features that it is difficult to produce with CNC because you can't lower the tool head into a pocket that doesn't accommodate the size of the tool head. Well, when you are building up layer by layer, you don't even have to consider tool insertion complexities. Any part of the build volume on that plane can be reached by the nozzle every time eliminating that complication from consideration. You solve every single 2D slice one at a time. Third, open-loop control style. Freer movement in all three axis comes with risk for losing position, slipping a step and facing more difficult to dampen adjustments in velocity and direction as you move in all ways. Next, part count on a budget. A lot of these concerns mentioned before could be addressed differently with a heftier parts budget, especially with onboard motion solvers and control systems. But keeping the costs affordable for the desktop 3D printer is a great deal of the point. If the costs per unit approached industrial equipment, the costs to own and use the equipment would rise as well. And when iterating a part adds the quality of fussing over a PO approval process, the value of this technology to improving design processes and problem solving through very inexpensive iteration is diminished aside from the best funded projects. Rigidity on a budget. To best bring together the physical and virtual world, you need rigid frame and nimble systems operating within these fixed positions and relationships. While numbers of 3D printer frames are impressively locked down, compare the rigidity requirements for hefty cutting tools that are fighting extreme rotational forces as well as moving around heavy spindles and similar. By reducing the rigidity requirements for a 3D printer down to just what can keep the physical world closely match to the virtual model, dealing with the forces a plane at a time, eliminating the real tough nonlinear loads subjected to a CNC tool head. You can keep the cost of the machine significantly down as well. Precision on a budget, fewer stack tolerances. With precision, you need to think not just of the cost of the parts but of the liability of incrementing the stacks tolerances. Desktop 3D printers tend to stay mechanically simple because more systems and components would bring with them additional requirements to counteract those additional imprecisions. Efficiency coordinated movement, travel movement, and finally, efficiency. A 3D printer can direct the tool head in practically any direction within a layer at the drop of a hat. Because movement in each direction doesn't trigger ramp ups and downs on tool pressure and other forces. So the machine can more easily switch between coordinated and travel movement with less concern over context beyond what is on a single layer at a time. Components of the motion mechanical subsystem. We will break the elements of emotion mechanical subsystem down into a few separate areas organized by their critical role: mechanical arrangement and mechanical distribution. Mechanical arrangement, how the structure and position of the physical components of the motion mechanical subsystem themselves function to execute instructions to deliver material to where it needs to go. Mechanical distribution, how motor actuation is translated into what is needed to operate the machine. Two machines that appear to have very similar mechanical arrangements might have differences in how the mechanism itself is driven as in the case between a standard Cartesian machine and a coreXY. My challenge here is the speak to the entire class of desktop fused filament.a 3D printers rather than approaching this from the perspective of a single vendor. This can be tricky as there are a number of little changes that might come down to preferences and critical trade-off decisions for the machine developer. Classic machine design trade-offs such as rails sliders lead screws versus belts and pulleys versus rack and pinion that lead to a large range of potential components in this system. Mechanical Arrangement. What is mechanical Arrangement? Okay, so now that we have discussed the coordinate system, the underlying mathematical landscape that governs how a printer interprets the world. It is time to talk about mechanical arrangement, how the structure and position of the physical components of the motion mechanical subsystem functions execute instructions to deliver material where it needs to go. Earlier, I defined mechanical arrangement as how the structure and position of the physical components of the motion mechanical subsystem themselves function to execute instructions to deliver material to where it needs to go. If the coordinate system is how the machines understand 3D space, the arrangement is a physical placements themselves what the humans who take the machine apart would see. There are quite a few potential arrangements and you can see the reprap family tree for a hint at the wide diversity of strategies currently in practice. Cartesian Arrangement, there's the X head and YZ bed, XZ head and Y bed, XY Head and Z bed and you typically have a rectanguloid build area. Delta Arrangement, working with in a Cartesian plane, but actuated via carriages moving up and down three towers triangulated to position the tool head anywhere within a cylindrical building envelope. Polar arrangement, a rotating bed plus X and Z axis. Escala Arrangement, an unarticulated arm executes XY moves without complex drivetrains. Mechanical distribution. Mechanical distribution, how motor actuation is translated into what is needed to operate the machine. Two machines that appear to have very similar mechanical arrangements might have differences in how the mechanism itself is driven, as in the case between a standard Cartesian machine and a core XY. While Cartesian and Delta motion systems are by far the most popular choices these days for desktop 3D printer, there are a number of other robots tiles in use. From Scarah to Polar to Extruder heads mounted on five six and even seven axis robotic arms. There are a number of impressive and rather odd ball approaches that produce excellent printed parts, one interesting consideration when looking at these more exotic machine design styles. What additional considerations are necessary when processing digital objects into instructions for motion control? I shared details about a few of my favorite exotic printer designs in the resources with the course one. Want to check out in that list, the reprap wiki family tree that charts the development of a vast taxonomy of printers clustered in terms of decisions around motion mechanical systems.
