geert_2

To me this looks like overextrusion, maybe due to the nozzle being a little bit too close to the bed? But this is not necessarily a bad thing, because then it is squeezed well into the bed, thus giving a good bonding, and probably a nice underside too? At least, that is what I get on my UM2 if the nozzle is a bit too close. (I don't know your printer.) Personally I wouldn't care too much, if this is only on the first layer(s), and does not show up through the whole print.

Another thing that just comes to my mind: did you export it to STL as a solid, or as a bunch of separate surfaces? I don't know what the mathematical difference is, but in DesignSpark Mechanical there is such an option: "Connected Mesh" vs. "Simple Mesh". Maybe in FreeCAD too?

I have read that PLA gets harder and more brittle due to changes in crystal structure. Over time it changes from rather amorphe to rather crystaline. This makes it harder but more brittle. Placing it in an oven for some time, could sometimes revert the crystal structure back to amorphe, to some degree. However, it could also have the opposite effect, speeding up the crystallisation... The same mechanism is said to be at work in plastics that you need to put in boiling water, after which you can mould them by hand for some time. When cooling, they stay mouldable for quite long, before the crystal structure changes back to crystalline and they get hard again. Often the color changes from clear to opaque white too. So you need to put it in cold water to solidify it faster. That is how the manual explained it. Then there is hydrolysis, where water-absorption breaks molecules. I think this is a bit similar to UV-light breaking molecules down and making plastics brittle? (Although I am not sure how this would apply to PVA, since this is basically glue? Or maybe it is a reversible process?) I am not a chemist; so this is just what I read on the internet and in manuals. Anyway, in my experience, 3D-printed PLA snap-fit lockings that worked very well when freshly printed, do break when using them after a year: they have become too hard and too brittle, and they don't bend anymore. In addition, I found that in PLA-filament under continuous stress, micro-cracks do develop. This happens for example when you straighten a tightly wound piece of filament, and you keep it straightened (fixed) by clamps. Under a microscope these cracks are clearly visible. So I would not be surprised if some of these mechanisms could also be at work in PVA? But which ones, and in which proportions? I would welcome the views of plastic-engineers from big plastic manufacturers (Bayer, BASF, DuPont, Eastman,...) on this. Photo: micro-cracks growing in stretched PLA filament. This is colorFabb natural 2.85mm, but I have also seen it in other PLA-filaments. Photo taken through a microscope. Maybe this also happens in PVA? Try cutting off a piece, stretch it under a microscope, and see if you can see any cracks growing? Then try again at different temperatures and moisture-levels?

I haven't used FreeCAD for STL-export, so I can't comment on this specific issue. But in general my experience with complex shapes like this is that if you need to combine things (="union" function), make them *overlap* each other first. If the edges just touch each other, the software is more likely to have difficulties sorting out where the edges belong. Probably due to rounding errors? DesignSpark Mechanical usually gives an error-message that it can't do the combine, so it refuses, but it has never created invalid models. However, early versions of other commercial programs I tried, did sometimes create invalid models. And SketchUp nearly always creates invalid models, since it wasn't ment for 3D-printing anyway, but that is another topic... Did you make this screw by combining a rod and a spiral? If so, try again with a royal overlap?

This could be an interesting method indeed for mould making. Did you try to smooth the PVA with water to remove layer lines and artifacts? And if so, did that work well?

Yes, this is also a good idea, provided that the material can be glued well (PLA, ABS). Not for nylon or PP, or PE. In that case I would design alignment features into the model, so that both parts can easily be mated up. For example by cutting it along this blue line. Add a ~0.2mm tolerance gap (this depends on printer accuracy, material flow, temp,..., so you may have to experiment).

I do a way more gentle pull than the traditional brutal cold pullings. Maybe this could help? - Manuall put print head in a front corner - Heat up nozzle (with bowden tube removed) - Push through some filament - Dial temp down to zero - Stop pushing filament, do a slight "manual retraction" of a few mm - Let cool down until nozzle is at room temp (25°C or less), so that the inner core of the filament also cools down well - If you have oil-free compressed air from a compressor, you can blow onto the nozzle to cool faster (but never use "air" from spray cans: this is often very explosive gas, and you don't want that on a hot nozzle) - When cold, gently try to rotate and wiggle the filament - Heat the nozzle to 70°C (for PLA, other materials need other temps) - While it is heating up, gently keep wiggling and rotating the filament until it comes out - Repeat as required. This method requires almost no force, but in my experience cleans as well as the original brutal cold pullings. The deeper cooling cycle to room temp is to totally solidify the inner core of filament. So, when heating up again, only the outside is molten, but the inner core stays strong and solid. The gentle pulling prevents damage to the nozzle, displacement of the coupler, and bending of the rods. It should also prevent breaking the filament. More info and pictures here (scroll down a bit): https://www.uantwerpen.be/nl/personeel/geert-keteleer/manuals/

Or resonance due to a particular model geometry? Or due to the nozzle banging into curled-up overhangs, or into a half-detached print, or so? Were the models still stuck to the glass? Or had they come off (maybe you can see if they came off before the jump, by analysing any spaghetti)? If you can get it to work again, try the same model again, but now stay around? Anyway, it seems like a good idea to glue or screw an edge to the tables. Or maybe put rubber antislip/antivibration feet or mats under it?

Maybe you could design it in the CAD file, like I often do for custom supports in hard-to-access areas? Then you can give it all the features you want. Question: how do you remove such a thick brim? Isn't that too hard to cut with a knife?

The biggest problem will be the layer lines and the little pockets of entrapped air. These will make the chocolate mechanically stuck, thus hard to remove. And they make the mould difficult to clean, and thus promote bacterial growth. So, if I had to do this for myself, I would select a material that can be post-processed very well, and that can be smoothed very well. And it should be temperature resistant up to at least 80°C. Probably I would try ABS, even though that gives of chemical fumes while printing. But when cold, it should be chemically stable. Lots of kitchen stuff are in ABS, like egg-cups, citrus presses. And ABS can be smoothed very well with acetone, into a really glossy finish without layer lines or openings at all. PLA won't be heat-resistant enough and is more difficult to smooth. Alternate solution: print the negative of the mould (thus make a mould of the mould), carefully smooth and post-process that, and then make a silicone mould to pour the chocolate into. Silicone is chemically inert, food-safe, heat-resistant, very flexible (easy to remove the chocolate), non-stick.

For people who do not have a mechanical engineering background, technical specifications such as hardness, tensile modulus, impact strength, etc..., are very hard to read. These numbers have no meaning because they don't relate to real-world materials for non-engineers. So, it would be good if all manufacturers could add a description *in plain language* of how the material feels, and how strong and flexible it is. This can best be done by comparing it to very well known standard materials like PLA, ABS (Lego bricks), PET (Coca-Cola plastic drink bottles), LDPE (lids of food boxes), PP (fridge boxes), car tires, rubber bands, human flesh, etc... A dummy example: Material ABC-123: - feels a bit waxy, similar to PP (polypropylene, fridge boxes) - glossy look, similar to PET (Coca Cola plastic drink bottles) - impact strength similar to ABS (Lego bricks, Playmobil) - flexibility similar to PP - good layer bonding similar to PLA and PET - low warping similar to PLA - mechanical post-processing: similar to ABS - glueing: cyano-acrylate, hotgun, most plastic glues - chemical post-processing: dissolved by strong solvents (e.g. acetone, thinners) - can easily be painted - chemical stability: can withstand mild acids and bases, is dissolved by solvents - heat resistance similar to ABS, starts to deform from 110°C, melts around 150°C, really liquid at 200°C - printing recommendations: nozzle 180...200°C, bed 80°C, speed 40mm/s, travel speed 120mm/s, fan 50%, layer height 0.1mm, good bonding to glass bed with product XYZ-456. - [then a couple of pictures of the filament, and of printed models, in an office environment] All this could go in the general description, so that we can estimate at a glance if the material might be suitable for us, without having to open all attachments with detailed specs. The pictures should be taken in an office environment, surrounded by common office stuff, so we can easier compare it to the real world. (Studio-lighting, althoug beautiful, often does deform the appearance and it makes it difficult to really interprete the pictures.) Could you consider this?

A minimum layer time may not be enough: if the object is very small, the nozzle may keep sitting on top of the object, and radiate heat, so it can not cool down and solidify. Setting the nozzle outside of the print is also not good, because then it leaks, causing defects on the outer walls. And this interrupts the flow, causing different material temperatures and viscosities, which also shows up. The flow also has to be as constant as possible. So that is why printing a dummy block next to the real model is usefull. Further, in my experience it is best to keep the printing area per layer constant for small objects. Sudden changes from a large surface to a small one, also show up. So, for small objects that need more cooling time, I often make the dummy block less or more into the inverse of the real model. Then the total printing time per layer is constant, and while the nozzle is printing the dummy, the real model can cool. I recommend adding a flange to the base of the dummy, to make sure it sticks well and is not knocked over. This first pic shows the effect of not enough cooling (dimensions mm and cm). The pictures below show dummy blocks: the theoretical concept, and a real example.

In DesignSpark Mechanical there are settings for STL-quality: coarse, medium, fine, custom. Coarse gives the same as you show here. Fine is okay for 99% of prints. With "custom" even better results can be got, but then the amount of triangles and thus the file size of the STL go out of the roof. Search for a setting like "STL export options" or something similar.

Do you mean the ribs on top of the supports? I usually design them on a 0.5mm grid, so they are usually 0.5mm wide (a bit wider than my 0.4mm nozzle), and 1.0mm separated from each other. But this is not critical, and occasionally they may differ in my designs. The vertical gap between the support-ribs and the bottom of the real model usually is 0.2mm to 0.3mm. The idea is to get the supports as close as possible to the real model for best accuracy, but without fusing them. So, dimensions may depend on the material you use, printing temp and speed, amound of fan (more fan = more cooling = less fusing). I would suggest that you make several small test pieces with slightly different dimensions, and then select the one that works best for your situation. It may require some trial and error...

Yes, this is a good point to consider. My prints usually require 1m to 3m of filament. So I rarely have more than 1.5m left over. But if you have half a spool, but just not enough for the next large print, then indeed it makes sense.

This is probably because the bonding is not so good? Can you see it wrinkle already while printing these layers? I have noticed that in PLA from the ICE-brand the outer edge also curled up a little bit, although otherwise it printed very well. So there seems to be quite a difference in bonding from PLA brand to brand. For Ultimaker and colorFabb PLA bonding was better for me.

In my tests I wiped off the white residu with a paper tissue a few minutes after the acetone was dry, but the PLA still flexible. But I have to be very carefull how I handle the model, otherwise my fingernails get imprinted in it. If I pick it up too soon, when just dry, then my fingerprints also get into it. So, this is a great way to immortalise your fingerprints in PLA. :-) There is a spot where the model is hard enough to handle but not too hard, where wiping off works best for me. But I have 100% filled models, so the inside stays hard. I am not sure how it would work with your hollow models with thin walls. It is the left model in this pic. The middle one was heat-gun treated (which doesn't work: trapped bubbles expand due to the heat and explode into craters), and the right one was untouched, if I remember well. For reference: the plates are 10mm wide, and text caps height is ca. 5mm. This is colorFabb Dutch Orange, but it also works with red and yellow.

Another option would be to just print everything, and then compare them to the design on the computer-screen and puzzle... :-) I would not do this for the main buildings for clients, but if it is for the surrounding decorative buildings ("filler buildings"), which often don't matter too much, this could be done.

I once tried melting PLA together with this tool. It worked fairly straight forward: cut both filament ends in a 90° angle, put them in the guides, and keep them in place with two fingers. Heat a knife in a bunsen burner, and insert it in-between both ends. A soldering iron should also work. Slide the ends towards the knife or soldering iron, let them melt, remove knife/iron, and push molten ends together. Keep a while until cool. This goes easier than describing it. But then I had to cut off the inevitable flanges and smoothen these out with a cutter knife or Dremel-like tool. It worked well, but it takes so much time that it isn't worth the effort. But this technique could be usefull if you want a vase or toy with colors that change over and over, on a single nozzle printer. So, now I use these ends for atomic pulls, like kmanstudios and several other people here. If you have kids, you could soften these pieces, and make toys out of it, or wrist bands, etc.

Thanks.

More than 50°C is too much for PLA, it will warp. It is best to stay below 45°C. While ethylene oxide may sterilise well, I am not sure if it is a good idea to use it. It is extremely explosive and requires very little ignition energy. It can explode in concentrations between 3% and 100%, contrary to most other gasses which only explode between 5% and 15% or so. The smallest spark is enough, e.g. from dropping a steel screwdriver. Almost all production plants in the world have suffered severe explosions. The plant I have worked in, also exploded a year after I left in 1986, and that shock was felt like an earthquake over 30km away, even though it was only one destillation column that had exploded. And even though the column was designed to "take off like a rocket", instead of exploding. According to witnesses it did take off indeed, but then still exploded at 500m high in the air... When in storage, ethylene oxide may spontanously decompose, or polymerise, both of which are very exothermic and often lead to explosion. Thermal runaway can occur from a bit above 50°C, if I remember well. Shaking it (e.g. during transport), or any contact with almost any other material may also lead to exothermic reactions and explosion, if in sufficient quantities. Usually, contact with other materials lowers the thresholds, making it more dangerous. Further it is chemically very agressive towards human tissue: it penetrates the skin deeply, without you noticing it. And then it starts burning and eating away the flesh, without any way to stop the reaction. Probably that is why they use it to sterilise? Spilling is not a problem if you spill a drop on your bare hand: it feels ice-cold and evaporates immediately, before it has time to penetrate. But if you wear gloves or clothes, it can not evaporate, and then it penetrates the skin. It goes through chemically resistant safety shoes, if you step in a puddle. We had several people severely wounded by these chemical burns. Know the risks before you start playing with this. If using ethylene oxide, all equipment should be equipped with good flame arresters, and be explosion-safe. And all bottles should have good safety valves, even small glass bottles. Maybe you could look into things like "isopropylalcohol + water" instead? They are less dangerous. I don't know anything about sterrad.

Yes, I was going to say it looked too big indeed... :-) But in my model, the connection strands are not plates, but just sort of tiny "hairs": they are 0.5mm wide (=a bit more than nozzle-width), 0.2mm high (=2 layers of 0.1mm), and 1mm long (=the distance of the gap). Before you print the whole box, try the concept on a small test piece first, to find the best dimensions for your materials and printer. The concept is clearer in this pic, where only the outer supports are shown. For reference: the upside-down stairs and the table plates are all 1mm. For small models, these tiny strands provides enough strength, and they are easy to cut.

Wow, if all that white is salt, then indeed you used way more than I usually do. I prefer a thin layer that is almost invisible: after applying the salt, usually my glass looks a bit "dusty" like a drinking glass that has been unused for a couple of years. For me, the first pic below (orange testmodel) has already too much salt; while the last one (with label "the salt method") has the optimal amount. Hardly visible when looking down vertically on it, and a bit misty when looking horizontally at it. But of course, it is best if you try different methods and choose what works best for you. Different materials, printers, and environments (temp, moisture) may all have an influence. For other materials than PLA, I recommend that you carefully watch the print, and don't leave it alone, until you know how well it sticks, or not. Or do a testprint like in the first pic, with inverted prisms: due to the very small bottom area to stick to the glass, and the very large overhangs creating huge warping forces, this is a good bonding test. Also, the overhangs tend to curl up, and the nozzle tends to bang brutally into these curled-up edges, making this a very hard test.

Most silicones I have seen can withstand 200°C, and up to 260°C for a very short while. But molten nylon is likely to be too viscous at this temperature: more like thick sirup instead of water. You would have to inject it at very high pressures and speeds (which would deform or damage the silicone). It will also be very difficult to get entrapped air out. And indeed, it will cool quickly upon touching the mould walls, making it difficult to fill the mould. You would need liquids that chemically cure, like ahoeben said. But then still make sure they are not too exotherm, as this may damage the mould, or the epoxy/PU may catch fire. Also, saturate the silicone mould with silicone oil prior to casting. Otherwise the vapors of the epoxy will penetrate the mould and make it fail brittle and fail very soon, after a few castings. Silicone is water-tight, but not oiltight and not solvent-tight. Although I haven't used it, I read that PU exists in tough flexible versions similar to nylon (they use that to protect car undersides and trunks). Edit: there is a reason why plastic injection moulding machines require heavy metal moulds, and clamping forces up to some tonnes, depending on the model...

If you design custom support structures, maybe you could add sort of dovetail things in it? (I hope this is the correct English word.) Like they used to do in old woodworking. See this: the bottom red symbolises the custom support, the middle white is the PVA, and the top red is the real model.