geert_2

All plastics tend to creep under load, and deform permanently, especially PLA. Try printing a hook as cloth hanger, and after even a few days, you will see the deformation. Can be quite severe. So I am not sure if 3D-printing is the best for CNC-machine parts? Or maybe if you use glass-fiber or carbon-filled filaments? But I have no experience with them, so I can't give recommendations. I read that they tend to clog nozzles easily, and you need a hardened nozzle and feeder, as they are abrasive. Another option could be to print moulds in 3D, and then cast the parts with fiber-filled or stone-filled epoxies? These can be made very rigid. I know of industrial laser photofilm printers for the printing-industry that used a base of marble-filled epoxy, to give them the required stability and stiffness. That base was as hard and rigid as a block of marble itself. If you use sand as filler, it has to be pre-processed (washed with some chemicals) to make the epoxy stick better to it, making the cast model much stronger under tensile loads too. But I don't know the details of it.

Have you verified if your STL-file is solid and error-free? I could imagine that if there would be STL-errors, some things might print well in one orientation, but not in another? In the beginning I tried designing text in SketchUp, and import that in my designs, because my editor DesignSpark Mechanical did not have a text editor function. And at that time I did not know the work-around via the dimensioning-tool. But SketchUp-text had lots of gaps in its vectors, they did not connect. That caused text to import poorly, and some characters got lost, or would not convert from surface-models into a solid. Surfaces can not be 3D-printed, as they are infinitely thin. It looked a bit similar to what you have, except that I had that in the design phase, not in the slicing phase.

Yes, as GregValiant said: you can best adapt your model to the size it is going to be printed in. If you scale it down, everything below the nozzle-width falls away (unless you apply tricks). And other parts might fuse together because the gaps become too small. Or you might no longer be able to access certain areas with tools to remove supports, etc. So you have to adapt these in the design. This is a bit similar to logo-design in a graphics editor: you need to redesign your logo for each size it is going to be printed in. If not, when scaling-up, the white spaces will seem too large, and the logo will visually seem to fall apart in separate pieces. It will look too crude and dumb. When scaling down, the opposite occurs: the white spaces will seem too small, and everything will seem to fuse together, and details and fine lines get lost. It becomes an unrecognisable blur or mess. Both hurt visually. Thus: when scaling up, you need to reduce white spacings, and need to refine the design and add details. When scaling down, you need to increase gaps and thicken lines and features, and make the model cruder by removing details that are too small. You need to take the printing method and viewing distance into account when redesigning the logo. The same here in 3D-editing and printing, except that things fall apart or fuse together, and details are lost *physically*, not just visually.

Hi John, I think this is a good idea. And if it works for you, obviously, it works. If you would like, feel free to use my old warping test, shown in the photos above. I would be curious about the results, to see where the limits are? This is a small test that prints quickly: total model size = 50mm x 50mm; height=5mm; top-width of each arm = 10mm, bottom-width = 2mm. So, a tiny bottom area to bond to the glass, but a huge top area to produce warping forces, combined with steep overhangs peeling the model off the glass. And overhangs curling up. warptest11.stl

I thought I would add a few safety recommendations, as not everyone may be familiar with handling chemicals. (I have worked in the chemical industry with highly explosive and agressive chemicals, although very long ago.) When using chemicals like dichloromethane (=ethylene-chloride), ethylene-oxide, acetone, xylene, ether, and similar solvents, use good precautions. - Always use safety glasses. Not just any safety glasses, but the kind that look like diving goggles. So the spats can not fly parabolically behind it and still get in your eyes, as with normal glasses. - Always use good fume extraction, so you don't smell and don't inhale the chemicals. If unavailable, use it outdoors in the garden, or under an open carport. I have known 2 people getting cancer from inhaling xylene (another good solvent) regularly for prolongued amounts of time, disregarding our safety warnings, saying we were idiots. - Gloves might *not* be a good idea, contrary to popular belief. It depends on the glove-material. Most solvents go through latex and silicone gloves: silicone is watertight because it repells water, but it is definitely not oil-tight and not solvent-tight: they seep through quickly. That is why you need to impregnate a silicone mould with silicone oil first, before casting solvent-based two-component plastics. Then the solvents can not penetrate so deeply, extending mould-life. So, don't trust general-use gloves. If you get highly volatile liquids on your bare hand, such as ether, acetone, dichloromethane, ethylene oxide, xylene, and similar, they will evaporate immediately, feeling very cold. This may cause dryness and whitening of your skin. This is not good, but at least they don't penetrate deeply and don't do much internal damage, because they evaporate way too fast. However, if you wear gloves, and if the solvent would penetrate them (that is why they are *solvents*), then it can no longer evaporate and will keep penetrating your skin deeply for a very long time, maybe hours. This could do a lot of internal flesh damage and cause severe chemical burns. You may not feel these burns: some products like ethylene-oxide may take hours (sometimes up to 12h) before you become aware of the burns. But then it can not be stopped anymore and keeps going. And they may cause cancer on the long term too. I have known a guy who got ethylene oxide on his shoes, and even though it were "chemically resistant safety shoes", it seeped through and caused very severe burns, landing him in hospital for several weeks. So, if you want to wear gloves, be very sure that solvents can not go through. Don't use cheap general-use gloves. I always handle these products with bare hands, but I smooth only half of the 3D-printed part, up to a natural seam line. Then I let that dry, and do the other half. Or I hang the model on a wire, or on a screw, and dip it in the solvent. So I don't make skin-contact. If a drop spats on my hand by accident, it dries immediately. And I handle it in such a way that the fumes are extracted away from me (I have a fume-extraction cabinet in my lab). - Although dichloromethane doesn't seem to, most other solvents are highly flammable or explosive, and require only very little ignition-energy. Contrary to fuels like diesel or benzine/gasoil which require fairly high ignition-energy. Dropping a metal screw-driver can be enough to let some solvents explode (e.g. ethylene-oxide). Use any additional safety measures recommended by the manufacturers of the products. With such precautions, handling should be reasonably safe, and you can 3D-print smooth "hospital safe" models for medical use, without layer lines so they can be desinfected well. Smoothing is also very good for mould-making: it makes releasing the castings a lot easier.

I did tests with PET. In short: print as slow as you can, in as thin layers as you can, and as cold as you can. Due to printing slow in thin layers, the material will sit in the nozzle for a long time and start to burn/decompose, so you might need to print even below the recommended temp range. See my tests here: - upper row printed at 50mm/s, bottom row at 10mm/s - layer thickness from left to right (mm): 0.40, 0.30, 0.20, 0.10, 0.06 - dimensions of the test blocks: 10mm x 20mm x 10mm - watermark text is sitting halfway in the model A couple of years ago I made a whole thread on this, but I don't remember under which title... Maybe you can find it? Notice how the slowest printed blocks start to color brown due to sitting in the nozzle for a long time. Close-up of 0.06mm layer block: Close-up of 0.4mm layer block: Block as printed, and after polishing to remove most of the layer lines on the outside: Idem. Obviously, polishing the layer lines away, greatly helps improving clarity. Another watermark, thus hollow text sitting below the surface. Text is 3.5mm, leg-width is 0.5mm.

Do you mean cleaning the outside cone, or the inside, or both? After each print I immediately wipe the outside with a thick tissue, while it is still hot, so it is clean immediately. If material is burnt onto it, and I can't wipe it off, I use a brass M3 screw thread to carefully file off the ashes. Use brass or copper, never steel, because steel would damage the sensitive cone. And always be very carefull and gentle. Also, regularly applying teflon and/or PTFE to the outside helps reducing build-up of goo on the outside. I clean the inside with atomic pulls during each material change, to remove all old color, and any burned-in material. See my method here: https://www.uantwerpen.be/nl/personeel/geert-keteleer/manuals/

I would also use brim, because steep overhangs also tend to cause huge warping forces, and in addition they tend to curl up. If you don't want a brim all around the model, design a custom one in CAD on the places where you need it. A custom brim gives you the option to make it thicker too, if needed, for example 2 or 3 layers. Try if your bonding method can hold an inverted prism down while printing. If yes, 99.9% of your other models will work too, because this is a hard test. This one is going to succeed well. This one might soon come off and produce "spaghetti", or it might barely survive. You see the warping due to the steep overhangs, and the curling-up of the edges, which makes the nozzle bang into them, and excert huge additional forces. Try such a small test model, to find the best bonding methods for your materials and models, and add custom brim as required to make it print well. Stay with the printer and watch closely what happens, so you can abort, and so you learn from the observation.

If you haven't done these already: Maybe take a gcode file that goes well at home, and try that on the printers at work? And take a file that goes wrong at work, and try that at home? Then you know whether it is in the gcode or not? Exchange spools of filament at home and at work, just for a testprint. Print both with the same gcode file. Then you know if it is in the filament? I was thinking about worn-out white teflon couplers, but from what you said, I think you already replaced them, so that can't be the problem anymore? Idem for worn-out feeder wheels, clogged nozzles, etc.

Because most of my models are small, or have fine details that I want to preserve. For example in the models above, the yellow bar has to slide smoothly in the red clamp, without too much tolerance. They are for medical use, hence the smoothing to remove layer lines and make desinfection easier. And I don't want to spend hours and hours of post-processing on filing and grinding. So, printing slower, and in thinner layers, removes most of the need of post-processing. Then I have less ringing, less thickening at corners (due to the head slowing down to take the corner), less stringing, less blobs, less unevenness due to layer lines, less risk of underextrusion, it is easier to remove the supports because they melt less together, etc.... Printing slow also greatly improves layer bonding: the next layer then has twice as much time to melt the underlying layer and bond well. The cooling is mostly achieved by printing two at the same time, plus a dummy tower for the top part, where there is very little material to print. So the dummy takes the nozzle away for some time. I would recommend you try several test pieces that are designed to show the effect on corners, overhangs, holes, rods, etc..., to find out what works best for you. Embed your typical model aspects in the test pieces. For PLA I normally print at 50mm/s, unless the models need to be very fine too. For PET, that is often just too fast with my models, my printer, my brand of PET; but this could be different for you of course. This drawing shows the dummy cooling tower to provide cooling when printing the tiny top area. It is a different model from the ones above, but the concept is the same. This dummy greatly improves quality of the top part. Below are testprints of a block of 10mm x 20mm x 10mm, transparent PET. Top row is printed at 50mm/s, bottom row at 10mm/s. Layer thickness is from left to right: 0.40mm, 0.30mm, 0.20mm, 0.10mm, and 0.06mm, printed with a 0.4mm nozzle. Watermark is sitting halfway in the model, as hollow text. Obviously, thinner layers, and printing slower, gives far less air inclusions in-between the extruded sausages, and thus better transparency. The brown discoloration is from sitting too long in the nozzle. Thus when printing slow, you have to print much cooler than otherwise, to prevent decomposition or burning.

I usually print PET at: nozzle 215...225°C, bed 90°C, speed 25...30mm/s, layers 0.06...0.20mm thick, and no cooling if there are no overhangs and bridges. If bridges/overhangs, then a little bit of fan is needed, but that reduces layer bonding. And yes, expect more stringing and worse bridging than with PLA. These are PET: the left one is as-printed, the right one is smoothed with dichloromethane. The thumbscrew is a standard nylon M4, with 16mm head.

A couple of years ago I did a lot of tests with small text: caps height 3.5mm, character width 2mm, leg width 0.5mm. Recessed text, thus engraved, always came out worst. This is because the nozzle can not get into tiny openings in characters like: N, M, K, H, G, B, 8, etc... So they get completely messed-up and the openings (engravings) get closed. You are better off making raised text. While it still looks a bit Flintstone-like, it is easier to read. If it has to be recessed somewhat to prevent it from being damaged easily, consider recessing a plate surrounding the text, but raising the text itself. See the quick concept pic below. When printing transparent materials, you could also consider making a watermark text. Thus hollow text inside the model. It will still get deformed, but for a watermark that is less critical than for surface text. Watermarks are expected to be somewhat abstract and deformed. I also made a characterset optimised for 3D-printing. See here (and then scroll down a lot). Be sure to copy any files of interest, as I am not sure I can keep them up due to changes in policy and website-engine: https://www.uantwerpen.be/nl/personeel/geert-keteleer/manuals/ Pictures: Concept of raised text in a recessed surrounding plate: Watermark text concept: both as positive text (solid characters inside a hollow) and negative text (hollow characters): Raised text, 3.5mm caps height: Watermark text inside a 10mm x 20mm x 10mm block. This is printed very slow and cool, in thin layers, otherwise the text would not be visible. Left: as printed, right: after polishing the block: More watermark text. Nozzle is 0.4mm, and thus printing-lines are 0.4mm apart. Text leg-width is 0.5mm. Mini-text, caps height 2.5mm, character-width 1.5mm, and leg-width 0.5mm. This is the smallest you can get with a 0.4mm nozzle: A bit bigger text, raised in a recessed area. Caps height is ca. 7mm: Watermark text, as hollows totally inside the model. Top model is smoothed in post-processing to reduce layer lines, bottom model is as-printed:

This can also be used for bonding PLA and PET. Or bonding can be an undesired side-effect if you are not careful enough. But I don't know if the bond will hold forever, how strong it is compared to the material itself, or if it is susceptible to shocks (like cyano-acrylates: they can be knocked apart) or to heat-decomposition. Here a few pics of the bonding. Also note that the smoothing removes layer lines, and seals underextrusion if not too bad, but due to the high-gloss and reflections, it might make the surface look rougher than the dull raw print before, in some cases (see the sphericons). When rolling downhill, these sphericons roll straight down first, then take a sharp corner sideways and go back upwards a little bit, and then go down again, repeating this zigzag cycle.

Thanks for the feedback. I only see it now, must have overlooked it before. With that amount of shrinking, it could be good for artwork, toys like model railroad stuff, and other models that don't need exact dimensions. But less suitable for rulers or tight-fitting technical parts. If you would have real models that you can show, I would also appreciate that. In green state, after sintering, and then after further post-processing (tumbling, sanding,...). If possible, also in close-up? Would be good to see the differences in material structure.

For me, the salt method works very well for long, low, and 100% filled models like rulers, that take 1 to 3 hours to print. Thus for my typical models. I use it all the time, and almost never use any other method. It gives excellent bonding for PLA (may not work for any other materials), when the glass is 60°C. But no bonding at all below 25°C, so models come off by themself after cooling down. No need to remove the glass from the printer. I just drop a few drops of salt water on the glass, wipe it, and ready to go. That easy of use is the most attractive for me. However, narrow but tall models like poles and towers, are more likely to come off. So, for tower-like models I would recommend a brim if you want to use the salt method. I guess because the salt has no flexibility to absorb shocks, contrary to most glues? Or use another bonding method instead. No method is perfect, they all have their good and their weak points. Also, the bonding seems to reduce when the printing takes a long time, a whole day or so. In that case too I would recommend using another method. Not sure why that is, since I still don't know the why of this salt method, I can only guess. But I rarely print such big models, thus no issue for me. Also the infill-percentage has a huge influence on bonding: a 15mm cube, 100% filled, sticks so hard you can lift the whole printer. You would have to hammer it off. But if that same cube is printed *unfilled*, with only one 0.4mm thick sidewall, then it tends to come off: it is sort of peeled off from the sides. I found this out when using such dummy cubes as cooling towers next to fine real models. So, for such empty dummies (or empty near the bottom), I design a brim in CAD now: then it stick rock-solid as before. See the pic below. Overhangs exert a lot of warping force, so they are also more difficult to keep down, regardless of what method you use. If you want a hard bonding test, try printing an inverted prism like this. In that way you can compare various methods. My typical models: long and low, ruler-like, 100% filled. The built-in watermark ruler is in mm and cm. Custom brim designed in CAD around the hollow (at its bottom) dummy tower, and around the feet of the custom-supports. Not needed nor desired for the other parts. Inverted prism as warping test: this one might complete, or might come off and produce spaghetti, thus it is not safe to walk away from the printer now. This is a very hard test: very small bonding area, huge overhangs and huge warping forces, and edges curling up so the nozzle hits them very hard. Even within one material (PLA), and one bonding method (salt method here), there can be a lot of variation due to brands, colors, fillers, and even moisture in the air. Another warping test: this one warped, but completed without falling off.

I read on the datasheets that Aquasys dissolves at 70°C-80°C: does that mean that this is the minimum temperature to make it dissolve, or is that just the optimal for the fastest results? How does it perform in cold or hand-warm water, thus below 45°C, for PLA? It would be good if you could some feedback on your experiences.

If it are rods, then it should be the *outer rotating rods*, those with the belts. Because it is their wobbly rotation that causes the pattern (I think). Try measuring and calculating if the circumference of the pulley wheel and belt matches the distance between the spots? Visually guestimating, it could. Or you could make it more easy: using a marker, mark a spot on the pulley and on the belt, rotate the pulley, and each time it comes along the belt, then mark the belt as well on that same matching spot. After a few rotations, your belt should have marks with the exact same distance as the pattern. If that is the cause. I am not saying it can't be the inner rods which are carrying the head, but that seems *very* unlikely, since they do not rotate. So they would have to have a sine-wave or zigzag deformation all along their whole length, and that seems unlikely. But when printing, after 2 or 3 layers, you will most likely not see anything anymore of this deformation. It will probably be less noticable than the layer-lines and print-lines. So the question is whether it is worth worrying about? My experience with disassembling and repairing complex equipment (like cameras and stuff) is that it sometimes ends worse than it was before. If it was my printer, I wouldn't want to fall into the trap of "keep repairing it until it is broken", a risk that is always there with any fine and complex equipment.

Hope you can read a bit of English. I can read German, but not write it. I use a standard thin injection needle, of which I cut the sharp tip off (that would damage the brass nozzle), and rounded the edges with sanding paper. It was a little bit too wide, 0.41mm, and did not go into the nozzle opening. So I had to grind the sides also until they were 0.39mm ... 0.40mm. And then very gently poke from the bottom upwards into the nozzle, and gently push and scrape. Do not use side-forces, so you do not deform the nozzle-opening. However, I never had a total blocking, only reduced flow. Not sure if it will work for you. I read that other people use acupuncture needles, but I could not find them here. To scrape debris from the side-walls of the nozzle, I use an M3-thread with rounded end, so it does not damage the nozzle. But that only works if the nozzle is already free, to clean the sides; it can not remove plugs. Then I do very gentle scraping, using it like a file. Always very gentle.

I don't know Fusion 360, nor your design and requirements, so it's hard to give good advice. But in general, if you have special support requirements, you can always design them in the CAD-model itself. Then you have full control. I often do this because my models are tiny, and the supports are hard to reach: there is no room to get in with a knife or plier. See the example below: these are self-hanging supports, and easy to remove. The slot where they are hanging in, is only 5mm wide. If printed in PET, I need the supports, otherwise it wouldn't bridge. If printed in PLA, I could bridge this gap without supports, but I want tighter tolerances, since a ruler has to slide in this opening later on. The ribs on top of the support allow a smaller gap between support and model, without them fusing together: usually the gap is between 0.2 and 0.4mm for my designs. The supports extend a bit longer than the area they need to support: then I can knock them down from above with a screw driver. The bonding of the supports to the walls is based on stringing of the filament: it works, and makes removal really easy. Similarly, you can invent all sorts of support methods, and optimise them for your specific purpose.

To repair: trying to bend the rods the other way round, doesn't seem like a good idea to me: chances are that it gets far worse. So I wouldn't do that. I would say: now that you know the most probable cause: if you can work around this by using a slightly thicker first layer, then try to live with it? To rule out the glass for sure as a cause, what you could try is: mark and measure the exact centers of the "dots" you see, from the edge of the glass. Then rotate the glass 180°, and do the test again. If the cause is the glass, then the pattern should have rotated also. If the rods, it should not have rotated. To prevent further damage, never lift the printer by the rods, so don't grab into the top opening, but grab from the bottom. And use very smooth and soft atomic pulls to clean the nozzle. Have a look at my method here (and then scroll down a little bit): https://www.uantwerpen.be/nl/personeel/geert-keteleer/manuals/ Note: do copy any of my manuals and models that you find usefull, as I am not sure I can keep them up here on our website, since our website-engine, layout, functionality and rules are changing here (e.g. in the future we can no longer upload zip-files).

Glad you are back. I already wondered what happened. Yes, I understand the frustration of losing abilities. But the good thing is that concerning art, you are still way above average. Most people would not be able to produce that at all, no matter how much time they got. Like I would never be able to compose a song, no matter how long I got.

I don't know your filament, nor your printer and print-bed, so it is very hard to give advice on what bonding methods and preparations you could use. But the chemicals you are using, are toxic and cause cancer, especially xylene. Some of my previous collegues got cancer after using xylene carelessly for years, against all our safety-advice. So use them carefully in well ventilated rooms, and don't touch them. Preferably look for non-toxic stuff, or less toxic. I have glass beds, and most of the time I never clean them: I just wipe them with salt water, and go. For my PLA prints this works well. If the glass is greasy, I clean it with isopropyl alcohol first, and then with pure tap water. (No soap, no detergents, no window cleaner, no thinner,...: they often reduce bonding.) But that might not work for you. I leave the heated bed on, because otherwise the PLA has no bonding at all. Maybe you could first print this in PLA until you get that right? And only then move to more difficult materials? There are a lot of variables and design-features that might need fine-tuning: to get your overhangs smooth instead of rough, to get the gap between support and print right for easy removal, to get the bonding right, and to fight the warping.

Off-topic, how to get pictures right: There appear to be some bugs in the image-insertion module in the forum. Inserting and moving pictures around by dragging and dropping, does not seem to work well. To get pictures right, I need to do this: - Ffirst drag and drop them from Windows Explorer into the grey area at the bottom. Not in the message you are writing, but just below that window. Then they will be uploaded, but not yet included in the message. -Now position your cursor on the right place in your text, where you want the picture to appear. Preferably type a few Returns to make new lines and room. - And then click on the Plus-icon on the picture you want to insert. Now it comes on the right place. If you made a mistake, delete the picture from the message, and insert again by clicking on the plus-icon.

It could be the glass: glass is never totally flat, it has bumps and pits. If you look through a car window, especially on older cars, and in curved areas of the window, you will see the lens-deformations caused by this unflatness. But it could also be the rods not being totally straight. If they would only be 0.025mm curved, almost nothing, that would show up as a 0.05mm difference in layer height, in a sort of sine-wave. Thus a full layer in fine settings. If both rods have it, you could get a pattern like yours. Maybe you can check if the repeating frequency is one full turn of the rods? Bent rods could be caused by lifting the printer by the rods, instead of by the frame. Or by doing hard atomic pulls (which is why I use a much smoother method without pulling hard). Or it could have been in the rods from the beginning, if their extrusion was not perfect, or if they have been handled roughly during transport (e.g. if it were 3m or 5m long rods loaded on trucks by a crane).

Your first layer should probably look more like these: