geert_2

I am not sure what Cura is supposed to do if the gap is *totally zero*, or if the parts overlap. Maybe one of the developers can answer that? (I am not a developer, and not related to Ultimaker.) But Cura can't know the intentions of the designer, so I think it is best if it treats models exactly *as they are*, without making assumptions. Usually I leave very small gaps between 0.1mm and 0.5mm for supports or for different objects that need to be separated, and these of course shouldn't fuse: that would be a nightmare. :)

Are they connected in the CAD design files? In similar designs I made, Cura just follows the outlines of each layer, and thus automatically connects these (but I am using an older version, so I don't know about the most recent). If not connected in the CAD, then what we see here is the desired behaviour: this is what we use to make supports and other parts removable, like gears or blocks in Rubic-cubes.

I polished such a test piece, see the pics below. Both were printed at 0.06mm layer height, 10mm/s. The unpolished one (=same as from previous pics) was printed at 210°C, the new polished one at 200°C, thus far below the recommended range of 215...250°C. Hard to see on photo, but in real life the coolest printed one is less yellow. The difference in transparency is most obvious from the side, which looks dull in the unpolished model, and sparkling in the polished one. The top layer of the polished one is okay, but the bottom contains voids between the first couples of layers, which can't be polished out, unless I would remove all these layers. These voids are clearly visible from the top (the diagonal infill-lines), and they sort of destroy the transparency effect. So, this seems to be about how far I can get with standard PET. Unfortunately, on photo you can't appreciate the 3D-effect of the logo floating halfway the model, which makes it stand out compared to usual 3D-printed models. This concept is good for watermarking your models in STL-files you distribute. If your model contains enough of these watermarks, then plagiarists will have a hard time finding and removing them all. Selecting and deleting watermarks on the surface is one thing, deleting them from deep inside a model is another. In your CAD- and STL-files you can also model them microscopically small, so that they are almost invisible. These won't be sliced in Cura and don't show up in layer view (no need to, since they can't be printed anyway). But they are visible when zooming in a lot, in transparent or X-ray view. So, if you hide microscopic watermarks in corners, you can still easily claim ownership by zooming in in Cura.

Probably: "handle the real cause of the problem, rather than searching a way around it"? If the real cause would be for example a dirty and partially clogged nozzle, then the solution would be to clean that. Other possible causes might be incorrect feeder calibration (e-steps?), wrong feeder wheel tension and partial slipping, worn-out parts in the print head, too much friction somewhere, too low temperature, too high speed, or other things along this line... But I don't know your printers, so I can't really make educated guesses. A cheat is good as temporary solution, to get urgent jobs finished. But it is not a stable solution to rely on forever.

I did a couple of transparency tests on waterclear PET. See the pics. Test specs: - material: PET, colorless ("waterclear"), brand: ICE (Belgium). - model: block of 20mm x 10mm x 10mm, with a watermark logo included halfway at 5mm height, saying: "©️GEERT" (Caps-heigth = 3.5mm). - printing temp: 215°C, unless written otherwise. This is at the lower end of the specified range of 215...250°C for this material. - top row: all printed at 50mm/s, and 150mm/s travel speed (=the defaults for my UM2). - bottom row: all printed at 10mm/s, and 20mm/s travel speed, thus ultraslow. - layer height from left to right: 0.40mm, 0.30mm, 0.20mm, 0.10mm, 0.06mm. - nozzle: 0.4mm (standard nozzle of my UM2, non-plus). - infill: always 100%. - flowrate: always 110%, to really push the material into all corners and close the gaps. - exceptions to the specs: model at bottom-right (=10mm/s, 0.06mm): printed at 210°C to avoid overheating and discoloring. Models at top-row left (0.40mm and 0.30mm) printed at 225°C, to easier melt the material at that higher flowrate. - I have not post-processed any of these models (except cutting off the final "take-off string"), thus no grinding, no polishing. So the irregularities on the top and sides do distort the transparency and visibility of the logo. Results: - The prints are far from transparent, but in some cases the watermark can easily be seen. - The best results are at the lowest layer heights. - Slower speeds give clearer models. However, when printing very slow, the model tends to discolor and get yellow-brown. This seems a bit similar to the brown goo that sometimes accumulates under nozzles. - Transparency (or lack of) gets worse as layer heights increase. You can see there is something inside the "frosted glass", but you can't read the logo. - At the thickest layer heights, clarity improves again, but then a crystal- or prism-effect begins to occur, with sparkles and rainbow-colors, which drowns the watermark. - If the model would be only 2 or 3mm thick, it is still transparent enough and the watermark can always be seen, although it gets distorted in the thickest layer models. - When printing very slow at thicker layers, the model starts to bulge due to over-extrusion. - At 0.40mm layer-height and 50mm/s the model does not cool enough, stays soft and deforms.

Bonding to the glass is influenced a lot by the weather and by moisture also. So, how is your environment currently? Before I found the salt method a couple of years ago, I used to print on bare glass. On cold freezing winter days when the air was very dry and there was a lot of static electricity, I got good bonding. But on moist, rainy days, I got no bonding at all: the filament would curl up immediately. All this with standard Ultimaker PLA and colorFabb PLA/PHA, and with identical models and settings, and identical gcode-files. This huge variation was very frustrating. Also cleaning the glass played a role: when cleaning with window cleaners or dish washers, I got no bonding. When cleaning with pure warm tap water only, I got reasonable bonding (on cold days). We have so-called "hard" tap water here, very calcium-rich. The salt method seems to smooth things out: for me it gives a good bonding when the glass is hot (even on moist days), but no bonding at all when cold, so models come off by themself after cooling down. This is the big advantage for me: easy and fast to apply, and easy to remove: no mess, no need to remove the glass. I am still not sure why it works, from a chemical or physical viewpoint? I guess it has to do with modifying the surface-tension of the glass: higher surface-tension gives better bonding, lower tension (like after treatment with oil or soap) gives a worse bonding. But this is a guess. Maybe it also has to do with the chemical composition of PLA, if some parts of the molecules would have a polar charge and are attracted to salt, which is also polar (Na+ Cl-) ? Again, just guessing. On ABS the salt method does not work. On PET it does not improve bonding, but it seems to make removal easier after cooling down: here it looks like it breaks the bond between PET and glass? Also not sure why that is. It does not work as good as dilluted wood glue for very narrow but tall models, like lantern poles, or for models with huge overhangs that curl up, or for models that take very long to complete. It seems wood glue can absorb shocks and vibrations better than salt (not surprisingly). So it works best for low and wide models that can be completed in a few hours, like mine (see pic below). If you try it for Tough PLA, I am curious about the results and I would welcome feedback. Chances are it makes removal easier. But bonding during printing probably won't be as strong as dilluted wood glue or 3D-LAC.

I don't know about the procedure, so I won't comment on that. Have you tried loosening all frame-screws (not removing, just set loose) and realign everything, and then fix them again? If there would not be too much permanent deformations in the panels, this might work? (Maybe except for the top panel, which seems deformed.) The assembly manual should give info on how to align things, I guess? After that, I think you also need to inspect the rods carefully: when *manually* rotating them, do they move smoothly without wobbling, and without getting stuck? (Don't force anything if it doesn't move smoothly and easily.) Also, test if the electronics, motors, heaters and display still work? Manually heat up nozzle and bed, and "move material" to test the feeder? Good luck.

Maybe you need to make a distinction between "chemically melt" (dissolve) and thermally melt (high temp)? Most epoxies I have used are very exothermic reactions: they get extremely hot. So, these will melt a PLA mould for sure. I don't know about PU, but I believe it is also exothermic? So you need to search for slow-curing, low-exothermic versions (usually they do exist). Added benefit is that these shrink less than fast-curing But PU and epoxies may also bond very strongly to PLA, so you need a good separator: multiple layers of silicone or PTFE-separator spray, or other. And even then... Try it on small test pieces first. As far as I know, PU or epoxies will not chemically dissolve PLA. Silicone depends: there are sticky silicones like the sanitary versions. And there are non-stick versions for mould-making. None of those that I used caused real warm-up, so I guess these should be temp-safe. Also, silicones are chemically quite inert, so they should not dissolve anything. Another option would be to use the PLA to make a positive model, and then use silicone to make the mould from that PLA-model. And maybe make a hard shell in PLA too around the mould, to keep the silicone in place, so it does not deform (may happen if you use very flexible silicone). In Youtube, search for: mould making and casting. There are lots of good tutorials.

For this sort of things in PLA and PET (I haven't used nylon yet), I usually take 0.3 to 0.5mm gaps. As gr5 says above the bottom layer sometimes does fuse a little bit, due to that layer being squeezed well into the glass and spreading out. But it is easy to cut. Further there are occasional strings, also easy to break. I would suggest you design a couple of small test models with variations in gaps, and print them at different speeds and temps in the range you usually use. The gap in the keyring here is 0.5mm. For size-estimations: the ribs on the pink support block are 0.5mm wide, with 1mm gaps. The pink block itself is 5mm wide. Vertical gaps between the pink support block and the overhanging yellow part are 0.3mm: they fuse slightly, but can be wiggled out. The cyan part is ca. 50mm long and 25mm wide. Gives you an idea.

Yes indeed: Z-Corp, that was it back then, thanks. The toothbrushes and gears printed in stone that I saw didn't feel right, but brick- and concrete buildings did. :-)

If the above things would not work, and you have a screw extractor for removing broken screws, maybe that could be an option too? But I would suggest you watch a couple of tutorials on screw extraction first, on Youtube. Or maybe a wood screw might also work? First drill a small hole in the center of the filament, at *very* low speed so the filament does not melt. And then insert the wood screw and pull?

If you want to print this in full color, then you need (someone with) a gypsum full color 3D-printer. But I don't remember the brand name of these printers. They use gypsum powder that is glued together, and sprayed with inkjet colors. So you can have all colors you want, although a bit dull. The gypsum gives the walls a stone-like texture and feeling, and a stone-like weight of course. After all, gypsum is a sort of stone. So these printers were often used for architecture. I have seen several buildings and other models in this material. It is not good for technical parts (too hard, too brittle, too rough texture), but it is excellent for models of buildings indeed. Maybe companies like Protolabs, Materialise, Melotte and Shapeways can do it? I don't know their current range of printers?

Ah okay, thanks for the clarification. I haven't seen or felt Tough PLA in life yet, so I was just guessing that it would be a bit more flexible. So then I would definitely give cyanoacrylate a try, but still on a test piece first. I have good success with both Locktite standard cyanoacrylates, and the versions for plastics.

For *standard* PLA I have good results with cyanoacrylate glue. Never had any problems after hundreds of glued models. But I never used tough PLA. Maybe you could try it on a few small test pieces? Don't waste a nice model on tests, just design and print a couple of small test plates. I don't know how well cyanoacrylate will handle flexing models, but if the area that is bonded does not flex, or not too much, I think chances are good.

Mine typically last reasonably long, several hundreds of hours. But if you print at hotter temps or higher speeds (=more pressure), they might wear out faster. If you have a microscope or good magnifying glass, you could look into it from the bottom and see if the inside is deformed? Or maybe it was a bad batch?

SketchUp is not suitable for 3D-printing: it does not produce "solid" models, but a sort of "empty cardboard models" with gaps in the seams. Like paper models you glue together. SketchUp was only ment for visual representations of buildings, not for 3D-printing. It will drive you nuts, and you will waste huge amounts of time. I would recommend you switch to another free program, such as DesignSpark Mechanical (for geometric models), or Fusion360. Or even Blender, if you need organic models, although Blender has a very high learning curve. There are lots of good demos and manuals in Youtube. Watch a couple and see which interface and workflow appeals to you, and try that. It will cost you some time to learn, but you will soon earn that back.

If your printer runs out of filament, and if it does not have an "end of filament" sensor and switch off automatically, then the remaining filament in the nozzle may burn and clog the nozzle. Because it will be sitting there at hot temperature for maybe hours. Normally Cura lists how many meters you need. Based on this, and on the weight of the spool (minus the empty spool), you could calculate how much is still left on the spool.

Do you put that cut-off piece in a drybox too while printing? Or if left hanging in the open air, doesn't it cause problems in long prints (like more than a day)?

Apart from all printing problems: does that model need to survive mechanical loads, after the wood stick is inserted? If yes, you would need to make the walls much thicker and stronger. Now they will break at the slightest bending of the sticks, due to the large lever. If purely decorative, it might be okay of course. And indeed, as said above, I would print this with the backside on the glass, and the opening on top. Then there is no top layer, thus no problem. For best layer adhesion, print slow and rather hot. If printed in PLA, layer adhesion should not be a problem (I never had any troubles). If printed in PET, switch off cooling fans.

Maybe try this method for overhangs? In CAD design a custom support structure into the model, and optimise it for ease of removal. In this example, all bridge pilars and plates are 1mm thick. Ribs on top of the supports are 0.5mm wide, with 1mm gaps. Gap between ribs and underside is 0.2 or 0.3mm (don't remember exactly, I use both). Horizontal connection strands are 0.5mm wide and 0.2mm high (=2 layers of 0.1mm). Long overhang is 35mm or so. Print cool and slow. In my designs, this is *very easy* to remove with a sharp cutter knife or scalpel. The concept is based on an idea of another user (smartavionics if I remember well?) who was developing alternate supports for use in Cura. Maybe you could get this done with more advanced settings in Cura too, but I am not using the latest version, so I can't tell about that.

I would suggest you disconnect the bowden tube at both ends, and then: - Check the tube is not mounted the wrong way around: the wider opening should be at the feeder. - Manually heat the nozzle to print temp, and manually insert a piece of filament straight into the nozzle, and extrude: does this go smooth, without too much resistance? To make sure the nozzle is clean, and temp is correct. - Manually push some filament through the bowden tube, while still disconnected at both ends: does that go smooth without too much resistance? - Then connect the tube again at the nozzle, and feed manually from the back through the tube, and extrude (heat nozzle). Does this go smooth? - With the bowden tube still disconnected at the feeder, feeding a piece of filament, and pull on it to see if it easily slips? For example if the internals of the feeder would not apply enough pressure anymore due to wear? - Some more tests along this line? The idea is to *separately* test all parts of the traject. Probably you have done these already, but maybe you overlooked something? Or maybe something has deteriorated recently? I found that it is often a combination of things that causes underextrusion in my UM2: - Most important: too tightly wound spools, near the end, especially of old PLA that has become very hard: this causes extreme resistance against unwinding, acting like a strong spring, and it causes very high friction in both the bowden tube and in the teflon coupler and nozzle. My solution here is to manually unwind a bit of filament and wind it in the opposite direction around a skater wheel (diameter 7cm). Then let it roll up on the spool again, but now it has a bending radius of ca. 30...50cm, the same as the bowden tube: this causes no more friction. This alone prevents 95% of underextrusion problems for me. - Deformed teflon coupler after a few 100 hours of use. - Kinks or sharp bends in the filament, so it has difficulty passing through the feeder and tube. - Rare, but has happened: dust picked up by the filament due to static charge, and pulled into the nozzle, causing partial clogs.

Definitely *avoid* SketchUp: it will cause nothing but problems, because its models are not watertight: they are no solids, but sort of "cardboard models" with gaps in the seams. I use DesignSpark Mechanical, a freeware limited version of the commercial SpaceClaim. For my technical models this is good enough. On Youtube you also find good tutorials for this. Go to the DesignSparkRS channel and scroll down to 3 and 4 years ago. It is easy to learn too. Not suitable for organic models or photorealistic rendering. It depends a lot on your personal taste and requirements. I would recommend you take a look at demos of several programs, and choose the one you like most.

Normally you can select a point (=endpoint of line) and move that. If you press the space bar after selecting, when in the move tool, you can enter numeric values. But the best is not to work on lines in sketch mode, but rather on 3D-planes and edges in 3D-mode. Contrary to most other 3D-CAD systems, DesignSpark Mechanical (and it source SpaceClaim) are designed for "push-pull modeling" of solids. You can keep editing existing shapes: it won't break the tree-structure because there is no tree... If you haven't done so, I would highly recommend watching the tutorials on Youtube. Search for "DesignSpark Mechanical", and on the DesignSparkRS channel, in the videos scroll down to 3 and 4 years ago: there are a lot of excellent tutorials. (Later tutorials of 1 year ago are not so good, so take the old ones.)

Are there steep overhangs that curled up, and where the nozzle bumped into? Causing it to skip steps? Or maybe a loose pulley could also gradually cause this? First it might cause an X-shift due to sitting loose. And then the edges would curl up, or the deformed part would otherwise get in the way, and the nozzle bumping into it might cause and Y-shift then, due to skipping steps? Just guessing, but maybe worth checking? Or drivers overheating, and temporarily shutting down? But this seems less likely, at least here in winter. :-)

I don't have an UM3, so I can't comment on the print cores. But maybe you could also check the SD-card? It could be a damaged card too, or dirt in the connections. Or a damaged file, incorrectly sliced or written (e.g. empty)? Or check connection if via USB?