geert_2

And no soap: that prevents bonding as well. Clean with whatever means you need, then with isopropyl alcohol, and then a couple of times with pure hand-warm tap water only. Rub and wipe dry with paper tissue, without touching the glass with your hands. Also kinks in the filament which prevent the filament from passing through the feeder, or the filament stuck under other windings on the spool which prevent it from unwinding, can cause underextrusion. I had the last one recently. So, never let the loose end go, never let it hang around, and never let it get under other windings.

The best is: make a few test models that have all typical features of your models. And then print them with various settings, to see the difference. Make them small enough so they don't waste endless amounts of time and material, but not too small, so the models still gets enough cooling (=without the hot nozzle sitting constant on top of the same place). Further: print slower and cooler than standard. This will give better corners. How slow and how cool depends on the model and material, and layer-thickness, so you have to try. For PLA, try something like 25...30mm/s and 200°C, and adjust from there. Unless this would go too slow and take too much time. Try all layer thicknesses and see which works best for your models. But 0.3mm goes 5x faster than 0.06mm. So, printing at 25mm/s and 0.06mm layers is 10x slower than 50mm/s and 0.3mm layers. Hence the need for relatively small test models. Personally, concerning colors, I would choose nice warm light-grey colors: stone, sand, rock, and similar natural colors, instead of white. White easily gets dirty. Or use the real colors the building will get, if available. (I never understood why architectural models are so often white?)

I kind of like the worn-out look of the first lady. Looks a bit like a corroded ancient statue that is in restoration, after being discovered in an old castle. Maybe you could keep it this way? Just fill and sand the splits, but no painting?

That overhang is very impressive, indeed. And what is the effect of this setting on the "curling up" tendency of steep overhangs? Where the edges bend upwards instead of sagging, causing the nozzle to bang hard into them, and causing vibrations and ugglyness too? And in extreme cases knocking the model off the bed? See these tests.

I was about to suggest this. 🙂 If they are totally separate, with a gap, and support is switched off, I see no reason at all why the base plate would be influenced by the X-character? Weird... 1. Have you tried printing them in one material, with only one nozzle? But with tiny gaps everywhere (even 0.01mm should do)? Just to see how it is sliced? 2. Have you tried putting both models into one STL-file? (Your models above are in separate files.) At 100% infill, on my single-nozzle UM2, the model below is sliced correctly in my older Cura. I quickly created it in DesignSpark Mechanical, and exported it as one STL-file. There is a tiny gap of 0.1mm between X and base. Base is 20mm x 20mm x 2mm. At less-than-100% infill, the area below the X is hollowed out according to the fill-percentage, but without outlines. STL-file: x_on_base1.stl

Also, when calculating the STL-triangles in MATLAB, make sure the coordinates of begin- and end-points of shared triangle-vectors are exactly the same, down to the last digit behind the decimal point. If these X-, Y- and Z-coordinates would be rounded off in your calculations, even to 20 digits behind the decimal point, it might create gaps in the STL-file, and then it would no longer be watertight. Let's say you have 3 triangles A, B, and C that share one corner. If this corner is (x, y, z): for triangle A=(10.00001, 20.00001, 30.00001), and for B=(10.00002, 20.00002, 30.00002), and for C=(9.99999, 19.99999, 29.99999), then this model would not be watertight, and would get messed up in slicers and CAD-editors, even though the gap of 0.00001 is not visible on-screen. (I think this is the problem that SketchUp has with 3D-printing too: it was originally designed for visual models only, like in computer games, but not for watertight printable models. Models look solid on-screen, but when you zoom in 10000x, you start seeing gaps.) Maybe you can add calculations to make sure that points that are very close together, do snap onto each other? Similar to the "snap" function in Powerpoint, Inkscape, and most 3D-CAD programs. Make this snap-radius adjustable, e.g.: "if points are less than 0.001mm apart, then snap them onto each other", but then in a math formula. So the three shared corner of triangles A, B and C in the example above, should all snap to (10.000, 20.000, 30.000).

As far as I know, STL-files have no dimension units (mm, inch, meters,...). But if I remember well from older posts, Cura *assumes* that the dimensions are in mm? Is that correct? If so, maybe you can set that in MATLAB to render or export the model to a size of maybe 20...50mm? And then definitely run it through an STL-analysis and repair, just to be sure.

Yes, that is how I also understood it. That is what I meant by "moving the parts around on the glass": place the lowest model the furthest away from the home-position, and the highest the closest-by. And then hope that Cura is clever enough to print the furthest and lowest parts first, and then the nearby high parts. For low objects (<20mm) this seems to work in my older Cura version for my UM2-printers: at least it starts from the furthest away models. But this does not work for high models (100mm or so): then it prints them all at once. I have no idea about newer Cura versions and printers.

My old Cura version can display this model, although it appears extremely small, maybe 1mm^3. So I have to scale it up 20...50x to make it printable. However, this older Cura-version can not slice it: then it hangs up. Maybe there is something wrong with the STL-file? Maybe try saving it with different parameters? Or run it through an STL-analysis and repair program? But I have no experience with these, as I never needed them for myself.

For dimensioning threads, I think you could reuse the guidelines for plastic injection moulding. These threads generally use shallower angles than typical metal threads (M20, etc...), wider teeth, rounded corners, to improve the molten plastic flow, increase strength, and reduce stress-concentrations in corners under load. Indeed very similar to what you have done. If you use a standard thread, you can screw existing caps on it (e.g. recovered from PET bottles). Search for "injection moulding guidlines pdf", or: "part design guidelines for injection moulding pdf", "lanxess part and mold design guide", "injection moulded threads" (and then go to "show images"), and similar, but without the quotation marks. Most plastic manufacturers do have excellent guides on injection moulding.

This is indeed what I would expect to work, at least if you have 100% infill. If less than 100% infill, say 20%, then I could imagine that the slicer would generate a heavier structure below the letters, to support them. I guess this will have to be answered by the developers, or people with a dual-nozzle machine and a better understanding of the slicing-internals.

If the model is opaque, and the letters are raised, then maybe you could make the whole baseplate in one solid model/color? And only switch to the letters once you are above the baseplate? Thus without digging into the baseplate, rather like icing on top of a cake? If it has to be transparent, or if the letters are recessed into the baseplate, then of course this method would not work. (Note: I don't have dual-nozzle printers, so no experience, and just guessing.)

Can't you move them around on the build plate, and in that way influence the printing order? In older Cura versions you could print multiple models apart, as long as they would not collide with the print head and rods. So you couldn't put very much parts on the bed. I haven't tried this in newer versions. But especially for very small parts, printing model per model separately, instead of all together, might not always be a good idea: then the model does not get enough cooling, since the hot nozzle is constantly on top of it, radiating heat, and adding hot filament. So it has no time to cool and solidify. This is why I often print a dummy "cooling tower" next to my small models, so the nozzle moves away printing someting else, and the model has time to cool. (Just moving the nozzle away without printing is also not good, this interrupts the flow, and changes melt-temperature and viscosity, which shows up in the print. Flow and temp need to remain as constant as possible, especially for small models, in my experience.) The effect of heat on small models. The dummy towers minimise the effect, but do not eliminate it. (Printed at 0.1mm layers, 100% infill, don't remember temp and speed, ruler is in mm and cm.)

Thanks for the photos. It might be fillers in the black that decompose due to the acetone? Or due to UV-light, or age, moisture, or whatever else? Or the fillers that reduce layer bonding, and thus make it more prone to splitting? Also, I think manufacturer should mention in their specs for each filament: "Can be acetone-smoothed" or "Can be whatever-smoothed", and then the recommended product, time, method (brushing, wiping, dipping, vapour, whatever), and amount (thin, thick). I have just smoothed a handle (using your brush-on method), yellow PLA, and now I am going to put it in the lab-oven to see what happens. First at 60°C for half a day, and then we'll see.

I have colorFabb glowfill, and indeed, it does not glow for very long. It is very bright for a *very short* time, and then quickly fades. Usefull brightness as a sign is a few minutes, maybe 5 at best. And then it is still barely visible, but unreadable, in pitch black darkness for indeed maybe 20 min. Just a blurred pale green shade. Useless as emergency exit sign, or as night-light in a room. In the old days they had radio-active glow-in-the-dark paints, used on clocks. These did glow *very* bright all night, for +10 years. This was really fascinating, and very usefull. But these paints are forbidden now, even though their radio-activity level was less than the normal background radio-activity in nature. Rocks, (natural) gypsum and river-water are radio-active too. Uranium, thorium, and radio-active argon are natural elements found in the earth crust. And much of the heat in the inner earth comes from natural radio-active decay of these elements. So the whole globe is a nuclear powerplant, sort of, and even a "melt-down". :-)

I believe the effect you see is overextrusion, due to the printer slowing down at the corners or ends, but the nozzle pressure is still up and takes some time to dissipate by leaking away. This gets more visible when printing lots of small segments. Printing slower and cooler helps, but does not eliminate the effect. But for me it is no problem. I haven't tried features like "coasting" (if I remember the name correctly, stopping to extrude just before a corner, so that pressure has time to drop). Maybe this might help? But if you overdo it, it might cause gaps at the start of the next segment. So try just a little bit.

I have tried it this way: - cut off both filament ends at 90°, - heat a knife in a flame, - align both filament ends in a holder (see pic), and push them down with your fingers, - insert the hot knife in-between both ends, and push both ends onto the hot knife (don't push your fingers onto the hot knife), - let these ends melt a little bit, - remove knife, and push both ends together, so they melt into each other, - keep pushing them down, and let them cool and solidify, - using a Dremel tool or similar, cut off the flanges, otherwise it won't go through the feeder and bowden tube. This is the worst part of the procedure. This works well, and it is usefull if you want to join different colors for artistic purposes. But for normal daily use, in my opinion, it isn't worth the time and effort. I keep left-over ends for doing cold pulls, or for plying them into hooks, or chains, or so.

Hoi cloakfiend, Have you ever tried *post-curing* an acetoned model? Thus: print, sand, acetone, wipe, dry, as usual, but then let it sit at an elevated temperature for maybe a day, or two days? The temp has to be above glass-transition temp, maybe 60-70°C, but below melting the model of course (may depend on the model what it can withstand). This could reduce residual stresses, so it might be less likely to split later on. And then primer, paint or plate as usual. If it is going to split anyway, it might already split during this process. I am not sure it is going to work, or whether there would be bad side-effects, but it might be worth trying on a test model? Another question: have you ever tried acetoning tough-PLA? This will have different additives than normal PLA or PLA/PHA, so it might behave differently? (I dont' have any tough-PLA, otherwise I would try it myself.)

Another factor is cooling: in small models it doesn't get enough cooling time. So the whole thing stays soft and sags. On sharp corners, the strand is pulled inwards like a rubber band. See the tests I did a few months ago. This is PET. Nozzle = 0.4mm. Layer-thickness from left to right (mm): 0.4, 0.3, 0.2, 0.1, 0.06. Top row: 50mm/s, bottom row: 10mm/s. You see the same rounding and not-enough-cooling effects in the thickest layers (left), although less than in your tests.

The screendumps below show why you should not use SketchUp. The vectors do not match up, they do not connect. So you do not get watertight solid models. SketchUp models are a mix of separate surfaces and half-solid parts. Especially if you then do boolean operations on these, it becomes a mess. Below this effect is shown in simple text, before it is extruded into 3D, but the same also happens in other SketchUp models. SketchUp was designed for visual representations only: of buildings, like in games, or for Google's original Earth and Maps views. Not for 3D-printing. If you want to use it for 3D-printing, you manually have to zoom in extremely, and manuall move and close the vectors (see the red arrows where the gaps are that need to be closed). Characters with openings (non-matching vectors) don't fill, don't extrude correctly into 3D, and give problems when printing. Idem for other models. A better solution is to use a program that is designed for modeling solids, for 3D-printing. I use DesignSpark Mechanical. Other people use Onshape, or Fusion360, or (student versions) of Solidworks, and lots of other packages. Search on Youtube for demo videos of 3D-editing programs, and then try one that appeals to you.

Also try using mechanically interlocking features. Since this is highly recommended for injection moulding in overmoulding, according to the manuals, I think it also applies for 3D-printing. At least, it shouldn't hurt. Think of a dovetail system, or hooks, or chains that interlock. Think of these concepts, but then all melted and squeezed into one solid block. Have a look at how it is done in tooth brushes: most of them also have overmoulding with interlocking features.

If you are courageous and have some electronics knowledge, and you want to rule-out the bed-heater influence, you could try to fix a LED-lamp onto the heating-connector. And then babysit the printer to see if the bed-heating LED and print-lines occur at the same moment? (Use a diode + LED + suitable current-limiting resistor, all insulated, not a bare LED alone). But here you risk making short circuits and burning out electronics, so you must know what you are doing. While babysitting a print, also watch if there are any feeding issues: bends in the filament cause increased resistance in the bowden tube and nozzle, or the filament momentarily stuck under other windings, etc. There could also be dirt in the Z-bearings, not only the Z-screw. These are a sort of chain-ball bearings (don't know the correct name), which can get stuck if dirty, or if they are lubricated and the oil gets sticky (some types should not be lubricated, but I am not sure which are used here). Or maybe place a heavy weight on the bed, left and right of the Z-screw, to compress any Z-play, and minimise any irregular friction caused by dirt, compared to this heavy load? And see if it gets worse or better? Not sure if this is a good idea though... Risks overloading the Z-drivers. These things are getting more and more risky, so use them with caution, and only if you have a technical/electronics background. Obviously, I am running out of ideas here... :-)

If it is at *random* heights, or at a fixed height not related to model-features, I would think of a Z-axis issue, such as dirt (thoroughly clean the screw), or play in the nut or driver. But I am not familiar with diagnosing and solving that. If it is always at the exact same height as big changes in print-area, or changes in infill, or changes in wall-thickness, then I would think that there is still something there to improve (thicker walls?), even on bigger models. I have seen models where the infill-pattern shined through the walls, causing indents. A bit similar to the structural ribs shining through the hull of big container ships. Or, if random: maybe a non-constant filament diameter? Or the filament temporary getting stuck, resulting in a lesser flow? Or minor changes in filament color, which show up as darker or lighter lines? I do have a yellow filament that has this: first I thought this was a printer-issue, but upon closer inspection, it was the filament that had those color changes; probably the pigment not mixed well enough. Long ago there also has been a discussion about the bed-heater drawing a lot of current, which could shift the ground-level (zero-voltage) of the electronics, which in turn could shift the perceived temperature from the nozzle temperature sensor, which then could cause the printer to "adjust" nozzle temperature, resulting in a different viscosity of the melt and different nozzle pressure, which might show up in the print. But I have no idea if this could be the case here or not; I don't know the electronics. Could be lots of things, hard to diagnose...

Do you mean that the 8mm rod on which the pulley is mounted, could be shifting (or be shifted) axially, due to too much axial play, or an incorrect plastic spacer? Or one of the pulleys being mounted a bit too far away from the printer-housing? That could also explain the phenomena.

For PET, I get good layer adhesion by printing much slower and cooler, and *without fan*. But then the bridging and overhangs suffer (but most of my models need no bridging and no overhangs). The "printing cooler" is to prevent the filament from burning and decomposing in the nozzle, due to the much longer transit times. I don't know if this would also work for PVC? Anyway, watch out for chlorine gasses, if it would decompose. Should be easy to smell, like in a swimming pool where they overdid it.