Design thoughts: enclosure, Z-axis, and alignment.

I'm in the very early stages of planning to build a custom CoreXY printer, 400mm on each axis (big enough, but without needing to get custom heater or print bed parts), maybe with a little extra Y and Z. While I had thought of starting from an easy to source set of parts, I realized that I would end up spending much more time, and hundreds of additional dollars, to convert printer A into printer B, while still only having one working printer, and neither exactly being cheap. I have access to laser and water cutting, but not milling. While I have to pay market value, with no scrapyards within two or more hours of driving, used brand name rails seem to only be about double the cost of chiwins on eBay, so my current plan is to dump money into those, for linear motion, with a solid top plate (or two) for the motors and pulleys. V-slot would work, but those costs add up to being pretty similar to used rails, and have their own set of issues. No interest in printing large enough parts from ABS or others to justify a heated enclosure. My main goal is a combination of speed and precision across very wide ranges of print, up to about 350mm on each side, mainly with PLA and PETG.

What is the typical process of aligning the axes? Like, the diagonals measure OK on the enclosure, and X/Y motion frame, but now I have rails that I don't want to bind, grind down, nor cause the X rail to off-parallel. But, I don't have a machine with which to make an accurate jig. Would 3D printed guide brackets be good enough, following the other guide by feel, or a combo of the two?

I would like to build the frame with extrusions for easy assembly and mounting (2020?), strengthened by panels, with only the top and front being open. For the sake of visibility, I think it would be nice to use 1/4" polycarbonate sheet panels for the sides (no holes but for mounting screws), with aluminum on the front (that one a big enough opening to get in there and remove the bed), bottom, and maybe rear. Does anyone have any experience with that, or other thoughts? Acrylic can be kept looking clearer, and might be slightly better against frame twisting, but is a lot more fragile (also typically less flammable).

I'm frustratingly torn on the Z-axis. Linear rails and belts are how I'm doing it, for sure. However, there are a few styles that should be viable, and I'm wondering if there would be any real functional differences. First would be 3 rails, one front-left, one front-right, one middle-back, with a platform held between them: most expensive, and the most parts to design, measure, drill, assemble, and align. Another being the same, but with a floating leveling bed holder on each, rather than an additional platform: this one reduces parts and weight, and since the bed alone will be several pounds, I wonder if there is any real risk of anything bouncing around - this is definitely the one I like most right now, but I wonder if there are hidden issues (other than having to keep the printer upright during transport). Another yet being two rails down the side, with a platform between them (UMMD style): moderate amount of parts to deal with, but less aligning between them (compared with three rails on the frame), the advantages of cantilever beds without cantilevering; but I wonder about print artifacts, if I were to not have quite a rigid enough structure for the rails to be attached to.

Before going into buying mood I would recommend starting with SolidWorks or Fusion 360 assembly. This way you will not be buying parts 2 or 3 times and may not need to cut anything, but order precisely cut pieces.

I think 2020 t-slot is too small for a printer that size, even with PC side panels. The moving mass on the XY stage is at the top of the machine. Throwing it around at 100mm/sec or more will make the whole thing want to move around. 2020 is pretty flexible stuff, especially when the pieces get to be 500-600 mm long.

Alignment is most easily done by building the frame square to start and using it as a reference to line up the guide rails. But without a mill to make sure end cuts are square, you may have to design it so that the rails can be aligned regardless of the squareness of the frame. The XY stage should be relatively easy to align if you're using linear guides and have a flat surface, so you mostly have to worry about getting the Z axis rails parallel to each other and perpendicular to the XY plane. I built the frame square then used a metal spacer to align the Z axis guide rails to the frame. I have found that a digital protractor of this type provides a pretty accurate way to check alignment when measuring diagonals is difficult or impossible, and far more accurate than trying to use a bubble level.

You really only need two rails to guide the Z axis. Lifting the bed at three places is the most stable way to go and could be done easily with belts all driven from a common drive shaft. Put two of the belts close to the two linear guides and the third on the opposite side of the bed. You're going to need some method of preventing the bed from dropping when motor power is lost/cut. Since the bed assembly will be heavy, you'll want some gear/belt reduction of the drive to multiply motor torque. If you use a worm gear drive you'll get the lifting power you need and stop the bed from falling. I made a worm gear box that used a small, 40mm long NEMA-17 motor to lift the 3.5 kg bed assembly + 4kg load in my printer. Unfortunately the gears turned out to be poor quality and I didn't end up using it. The gears have to be very high quality to prevent repetitive Z axis artifacts in the prints.

I think that using the Z axis lifting brackets to lift the bed plate without a "frame" under the bed may be asking for trouble. The frame ensures that the leveling screws always stay in the same positions relative to each other. The frame also locks the lifting bracket's positions relative to each other. Without that frame, as the lifting brackets/screws move up and down, they may wander around a bit and that will directly translate to the bed doing the same. Without a frame, you would need three linear guides so there's a place to mount the third leveling screw. That said, I think it would be worth a try as an experiment to see if it works. If not, you can always add a support frame later. I transport my printer laying on its back and the bed plate stays put even though the only thing holding it in place is the springs (and the chamfered hole/slot in the plate).

I would think long and hard about building a large machine without the ability to print ABS. If I were printing large objects that represent a substantial investment of time and material, I'd want them to be around for a while. PLA is not a material you can count on. It gets brittle when it absorbs moisture from the air and melts in a hot car. You can't use it around anything that gets warm, let alone hot. PETG is marginally better.

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

Quote
piper3d
Before going into buying mood I would recommend starting with SolidWorks or Fusion 360 assembly. This way you will not be buying parts 2 or 3 times and may not need to cut anything, but order precisely cut pieces.

I intend to do just that, for much of it, and will likely have to order precision-cut extrusion (most likely Misumi, unless someone at my makerspace is willing to pitch in, or we can make use of of the local tech school's gadgetry for milling), but that's a ways off. I haven't touched any real CAD in a very long time, so I expect that will eat up a good bit of my free time, as I get other details worked out. I will likely buy some things before that's done, but that would be rails, pulleys, etc., for which I would want to get actual dimensions to use in the rest of it.

Edited 1 time(s). Last edit at 07/15/2019 03:11PM by yellowedplastic.

Quote
the_digital_dentist
I think 2020 t-slot is too small for a printer that size, even with PC side panels. The moving mass on the XY stage is at the top of the machine. Throwing it around at 100mm/sec or more will make the whole thing want to move around. 2020 is pretty flexible stuff, especially when the pieces get to be 500-600 mm long.

I was thinking that, but it would be nice to use such ubiquitous material. And, that when I just sketch things out, the necessary width and height for a given print volume just keep adding up .

Quote
the_digital_dentist
You really only need two rails to guide the Z axis. Lifting the bed at three places is the most stable way to go and could be done easily with belts all driven from a common drive shaft. Put two of the belts close to the two linear guides and the third on the opposite side of the bed. You're going to need some method of preventing the bed from dropping when motor power is lost/cut. Since the bed assembly will be heavy, you'll want some gear/belt reduction of the drive to multiply motor torque. If you use a worm gear drive you'll get the lifting power you need and stop the bed from falling. I made a worm gear box that used a small, 40mm long NEMA-17 motor to lift the 3.5 kg bed assembly + 4kg load in my printer. Unfortunately the gears turned out to be poor quality and I didn't end up using it. The gears have to be very high quality to prevent repetitive Z axis artifacts in the prints.

The specific driving I will probably just experiment with. A simple idea, which I've seen in some very large printers, are counter weights, which would be trivial to implement, combined with gearing down by belts (9:2 is easy enough, just browsing RobotDigg). It would be easy to try, and easy to move away from if it doesn't work well enough. With the actual supports, rails, and vertical belts arranged well, changing the Z motor section of it shouldn't be much trouble.

Quote
the_digital_dentist
I would think long and hard about building a large machine without the ability to print ABS. If I were printing large objects that represent a substantial investment of time and material, I'd want them to be around for a while. PLA is not a material you can count on. It gets brittle when it absorbs moisture from the air and melts in a hot car. You can't use it around anything that gets warm, let alone hot. PETG is marginally better.

That's a good point. I'll compromise from the start, then, and design it to be enclosed, even if I don't initially put a door or top on it.

In a screw or worm gear belt drive, the weight of the bed assembly keeps the screws/nuts/gears loaded in the same direction at all times so there's little possibility of backlash. If you counterweight the Z axis, be sure to leave the drive mechanism partially loaded so that the counterweight won't cause backlash. If the bed assembly weighs 5 kg, use a 4 kg counterweight, for example.

Printing ABS isn't just about enclosing the machine. When you heat things up they expand. You have to allow for that or you may have problems, especially in a corexy mechanism made using linear guides. Linear guides are great bearings but they aren't tolerant of misalignment, even the relatively small misalignment that can result from thermal expansion.

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

Quote
the_digital_dentist
Printing ABS isn't just about enclosing the machine. When you heat things up they expand. You have to allow for that or you may have problems, especially in a corexy mechanism made using linear guides. Linear guides are great bearings but they aren't tolerant of misalignment, even the relatively small misalignment that can result from thermal expansion.

That's why I originally was thinking to just eschew ABS printing for specialized variants of lower-temp plastics. Heat-wise, I would be fine with the right filaments, but I think you were right about long-term moisture. Smaller PETG prints I have are worse than cheap eye glasses. That will mean either mimicking your X axis, or making a sturdier than normal rod-based X axis, that can itself be more forgiving. I've seen that done to good effect on these here intertubes, but I have also noticed that such machines making good prints always use printed bearings, tuned to their specific rods.