geert_2

I didn't know I had a photo of the above effect: expansion of the molten filament after exiting the nozzle. But I just stumbled upon it again, so here it is. You can clearly see how the sausage expands immediately after exiting the nozzle, during the first mm. In the zoomed-out image the sideways expansion is from ca. 46 pixels to ca. 51 pixels, thus about 10%, very rough estimation. Nozzle opening is 0.40mm, so that would be from 0.40mm to ca. 0.44mm.

That vertical banding is the "ringing" gr5 referred to: the mechanical oscillations of the print head after taking a sharp corner at high speed, due to its weight, play in the bearings, and the rubber drive belts. So, printing slower reduces this effect. You could also try the opposite: printed a lot faster, and see how the effect changes. For the best settings, you need to ask gr5, as I never experimented with these settings. You can also see the effect in the red print below, due to the light-reflections which exagerrate it. For size-reference: the nylon screw is an M4, and filament is 2.85mm. Printed on an UM2 with factory default settings for this, and 0.4mm nozzle. Instead of printing whole boats, you could also test these effects on small cubes of 15mm, and then change settings (speed, flow, temp) on the fly via the printer knob, to test the effect. That would cost you less time and less filament.

No, it's based on my experience with my own printers. The teflon couplers tended to wear out quite fast if the printer was running on higher temperatures (above 220...230°C). And that resulted in irregular prints and underextrusion. From the outside it can be difficult to see if the coupler is worn out, because that is only at the inside. But there can be other causes of irregularities: irregular friction in the feeding traject also has a lot of influence on these printers. For example, an irregularly wound spool, kinks in the filament, or a spool nearly empty (=bending radius too tight). And also too much play on the Z-axis screw, or dirt. Or a worn out nozzle (=the conic top being grinded off by abrasive filament, or the inner opening got too wide). Further, overhangs often tend to curl up and cause very ugly walls, especially in thin layers. This is far less if you print at 0.2mm or 0.3mm. For me, usually printing slow and cool improves quality, as gr5 also said. This reduces "ringing" and expansion at sharp corners, reduces strings and hairs, reduces the risk of underextrusion. Also, sudden changes in "layer printing time" do often show up as horizontal lines in the side-walls. Maybe this is what you get when it starts printing the deck of the boat? It seems around that height? But in general, although not perfect, your last white boat doesn't look too bad to me.

There is a way to mimick this effect, but it requires changing the CAD model. On top of your model, create these hexagons, and leave a very tiny gap in-between them, for example 0.001mm. This gap will fuse while printed, but the slicer will see it as separate polygons, and slice them accordingly. So it will outline each hexagon separately. Then use multiple outlines for wall thickness. But the problem is that the nozzle has to go from the center to the edge after completing each hexagon, and this may create ugly lines or defects.

Quite some time ago this has been discussed on this forum, with lots of photos. But I don't know the topic name. Maybe you can find it back? They did this by first printing one layer ("the symbol"), in a thin layer height (e.g. 0.1mm). And then, using a thicker layer height (e.g. 0.3mm), they printed the next part over the first, without removing that model from the build plate. But I don't remember the details. Some people got really nice results.

If it is an old standard UM2, chances are that the white teflon coupler on the nozzle is also worn out. If you do a cold pull ("atomic pull"), and it has a huge blob at the seam between nozzle and teflon coupler, it is worn out and needs replacement. See the blob in the white cone below, which shouldn't be there at all. It should look straight like the orange cone at the bottom. Obviously, I did these atomic pulls while changing filament color, so they have both orange and white. Yours should have only one color, and no black dirt. If it has black dirt, do multiple cold pulls until the black is removed. I don't know how familiar you are with nozzle cleaning and cold pulls or "atomic pulls"? If not, first search on the Ultimaker site for more info. Next, you could try the alternative and softer method I use on my machines: this doesn't require hard pulling, but only very gentle handling, but it also works well for me. See the manual here (and then scroll down a bit until you find "atomic pull"): https://www.uantwerpen.be/nl/personeel/geert-keteleer/manuals/ Also, while on that same page, I would recommend you download and print a couple of these horseshoe clips: they are much easier to handle than the standard clips. These are in PET, but I have also used them in PLA.

For keeping filament dry, I use these bags with desiccant (left=package, right=actual bag), and keep all in a big sealed food box. The blue dot turns pink when moist. These bags can be dried in a microwave, or conventional oven at low temp. They can be found in car-accessory shops as they are used for drying car interiors. For reference: the ruler is in cm. This is good to keep filament dry. But to actually dry already wet filament, it may not be enough: you may need elevated temperatures to first "shake the watermolecules loose" from the plastic, before this bag can absorb them.

Make a small and simple test piece and try to print that with default values: 40...50mm/s, 200...210°C, 0.1...0.2mm layers. Anyway, 240°C is way too hot for PLA: this could easily make the PLA decompose and block the nozzle. So you would first need to check that the nozzle is clean. Then also check if the teflon coupler is not worn-out, if the little fan at the back of the nozzle is working well, if there are no kinks in the filament, if the spool can unwind freely, etc. But definitely start by cleaning the nozzle until you can easily push filament through manually. Search for the how-to guidelines for nozzle-cleaning on the Ultimaker-website (I don't know the link), or try my old manual with a slightly different approach here (scroll down a bit): https://www.uantwerpen.be/nl/personeel/geert-keteleer/manuals/ On the Ultimaker website, and sites of Ultimaker distributors, there are also a lot of other good manuals on underextrusion (=no filament coming out, or not enough), and on checking parts. But I don't have the links, so you will need to google a bit.

The little "insect antennas" are when the nozzle leaks a little bit while traveling through the air. Upon arriving at the next wall, that droplet is deposited on the side of the wall. On the next layer, the drop is deposited on the previous drop, since that is now what the nozzle encounters first after flying through the air. And so on..., causing upwards tilted "insect antennas". I haven't printed with nylon yet, so I can't give any recommendations. For PLA, usually printing slow and cool helps, but I don't know how that would work for nylon: it might negatively affect layer-bonding. Increasing retraction also might have other side-effects. I would first try the slow and cool approach on a test-piece, where the "slow" is to prevent excessive pressure in the nozzle, and the "cool" is to make the melt less liquid and thus less likely to leak.

The "0.1mm" should be "1.0mm", to bridge the gap of 1mm between walls and support. Also see the STL-file above, and the picture, where most other major dimensions are 1.0mm also (the width of all plates, and the stair-cases). This is what works for me, but of course, you may want to adapt it to your situation. Different materials, models, and printing circumstances may require different parameters. Before printing big models, I would recommend you design a few small test models with lots of variations, and print these with your typical setup, so you can see what works best for you. For bonding PLA, I use standard Loctite cyanoacrylate glue, after sanding the surface a little bit.

Wasn't that print head the same as the UM2 (non-plus!)? Or have there been different UM2Go versions? In the first case (=like UM2 non-plus), I think you don't need any spacers, because this was with a spring instead of fixed-length metal piece.

It could also be that the little fan on the back of the nozzle is not working. This would cause the filament to heat up too much, and soften before the nozzle, so it gets too thick and gets stuck. Or that there is a kink in the filament, so it can not unwind from the spool, or not pass through the feeder. Or a totally worn-out white teflon coupler (especially on UM2 non-plus). Or any other obstruction along the whole traject. PS: normally you can just drag and drop photos into your post while writing. It should work, I do this all the time.

I have this phenomenon very occasionally in one of my two UM2s, not in the other. There is no clear indication why, although it is most often when playing around with the control-knob a lot. If I don't touch the knob, it doesn't happen. And it gets worse in very dry weather. So I don't know if this is a firmware-bug, electromagnetic spike issue (heater or steppers switching on or off, causing spikes), electrostatic discharge issue (like sparks in freezing weather), communication-issue between knob/display/mainboard, mainboard hardware, or a combination... The randomity makes me think it might be some EMC-issue interfering with the communication between mainboard, knob and display. But this is pure guessing...

Enable the standard supports as Smithy says, or design your own custom supports into the model. Sometimes custom supports may be desirable in special circumstances: for example to make the support stiffer so you can more easily grab and pull it out with pliers, or to make it extend so you can grab it, or to make special holes in it to insert hooks. Usually I prefer to design custom supports into my models, so I have full control. Except for large and easy to reach areas, where the standard supports do a very good job. For example, the red and orange blocks below here are custom supports, to prevent the bridges of the yellow part from sagging too much, since the blue spoon has to slide through that yellow part. This part is too small to get in with a knife, so I need the supports to extend, so I can grab them. The supports also need a little bit of extra brim for stability. For reference: text caps-height is 3.5mm, text legs are 0.5mm. So, all this is very small. Tha same one plus a few other types of supports I tested or used through the years: Another one, if you don't want the support to go all the way down and damage the part below it. Same, now also on the center bridge: the bottom of the support-bridge will sag and produce some "spaghetti", but the real bridge will be reasonably okay after some cleaning up. The above one printed. Keep in mind the small dimensions: the "plates" are 1mm thick, the text is 3.5mm caps-height. Nozzle-diameter is 0.4mm. And another one that could be usefull for springs or cylinders. The red support needs to be cut out later.

Are these diverters also printed in nylon? Or in another material (they don't look like nylon on the photo)? Could you keep us updated on how well they survive outdoors, and if there is any UV-degradation over time? We haven't seen very much outdoor applications on this forum.

This option exists in older Cura-versions, with the "minimum layer time" and "cool head lift" options. I don't know if it still exists in new versions? I tried this, but it did not work very well for me. The problem is that if the nozzle is simply moved away while waiting, the filament gets hotter and more liquid in the nozzle. So it starts leaking more, and the more liquid plastic flows differently upon starting the next layer. Further, the added heat during this waiting period, also has to be removed in the cooling cycle, so it does not really help very much. Any differences in layer printing time, or in molten filament temperature, show up as horizontal lines and deformations on very small objects. In my experience, you need a very constand flow rate, constant temperature, and thus constant layer area, for best results. But the nozzle needs to be away from the object long enough, so it has time to cool down. So, we need to print multiple models at the same time, or print a "dummy cooling tower" next to the real model. Ideally, this dummy should have the opposite layer-area as the real model, so the total area per layer is constant for the whole model. This is especially true for very small models, with 100% infill, and thus a lot of stored heat which has to be removed. When models get bigger, this is no longer required, as each layer gets enough cooling time anyway. See these pictures I made a few months ago:

Originally, I also thought so. But I got inconsistent results when measuring these extruded strands: they could be anywhere between 0.40 and 0.60mm. So, something was off, but I didn't know what. Later I saw a Youtube-video of "nozzle-developers" (I don't remember from which company), and they said this very common assumption was not true. When the plastic comes out of the nozzle, it is still molten. Molten strands of plastic usually have a strong tendency to contract. This is because during extrusion their very long molecule chains are stretched. These chains are then under a very high stress. When that stress falls away (=as it exits the extruder), the molecules tend to relax to their previous shape. So the strand gets shorter and wider. At least, as long as the plastic is still molten. This effect is similar to when you heat most plastics in a flame or with a hot air gun. Especially for thin rods that have been extruded or injection-moulded under high pressure. The width of the rod expands, but the length contracts. I haven't discoverd this, I am just echoing their findings. :-) After watching that video, I tried this on my UM2 printers. I manually fed material through a very hot nozzle (with bowden tube removed). When pushing hard, I could get sausages of up to 0.8mm to 1.0mm, out of this 0.4mm nozzle. When extruding manually at very low temperatures, barely above melting temperature, and at very low speeds/pressures, the expansion is far less, and I usually got to ca. 0.45mm. So you can not use the extruded sausage-diameter as an accurate reference of the nozzle diameter, only as a crude estimation. A better way to measure the inner diameter might be to use copper wire strands, clean the tip (often deformed after cutting it), and then try which ones fit into the nozzle. And then measure the width of these strands with calipers. I have grinded down a soft steel injection needle, and use that. I made it for the purpose of cleaning, but I also used it for measuring. The needle tip of 0.39mm could easily get into the nozzle opening, but a bit further where the needle was still its original 0.41mm, it could not get into the nozzle anymore. So, given the +-0.01mm accuracy of my calipers, and my imperfect sanding of the needle, nozzle opening was quite accurate. Today, a couple of years further, the same needle easily goes up into the nozzle all the way, and it still has some play at 0.41mm, so the nozzle has worn out a bit. If you mathematically want to calculate volume, when feeding electronically via the feeder, you also have to account for the partial slipping of the feeder wheel. The feeder bites into the filament and deforms it. At low speeds and low pressures, these are nice square indents. But at higher speeds and pressures (=more mechanical resistance), the feeder has to push much harder, and instead of squares, these indents become stretched, long diamonds. You can see this deformation under a microscope. So, the effective amount of fed-through material is less than calculated, due to this "partial slipping". To compensate, you need a higher flow-rate. Due to the huge amount of other variables involved, I wouldn't know how to accurately predict or calculate all this. So I use the good old and proven, "trial and error", and use that experience as stable starting point. Any method that gives good results, is a valid method. :-) What I often do for new materials (but you could also do it for new nozzles), is remove the bowden tube and manually feed some material through the nozzle. Then I adjust temperature on the fly, and watch and feel how it melts and flows. This gives me some feeling for the material (or nozzle). Then I print a couple of test pieces, starting from the default values, and on the fly I adjust speed, temp, flowrate up and down, to get an idea of the edges up till where it still works fine, and where it starts to go wrong.

You could do this kind of tests on small test pieces, of only a few mm high. In these tests, provide as much different features as realistic for your typical models: different line-widths, different holes, different extensions (poles), overhangs, curves and roundings, bridges,... If you use bolts and nuts, also model the required hex-recessions into these tests. Idem for dovetail- or snap-fit mechanisms, etc...

In my experience: - If bed temp is too low: the model is very stiff, and the model-edges generally do not lift much, but the whole model may suddenly pop-off the glass, due to insufficient bonding. - If bed temp is too high: the bottom of the model stays too soft, so the edges may lift due to shrinkage forces acting upon the higher layers, and the whole model may gradually peel off the glass. - Walls too thin: walls are pulled inwards (similar to the effect you see), and the models tends to lift at the edges and gradually come off the glass. It is like a cardboard box that you fold inwards. - Not enough infill: walls may be unstable and sag if printed rather hot, and if the bed is rather hot. - Models with high infill (70...100%) rarely deform, althoug they may lift edges if the bed is too hot. - Very small models need an extra "dummy cooling block" next to them, so the print head is moved away for some time, so this small part gets time to cool and solidify. I needed to find a balance where the bed temp is high enough to make the model bond well, but not so high that it deforms, sags, or peels off. The optimal bed temperatures differ from material to material: for PLA it is 60°C, for PET it is 90°C in my system. But this could be different for your printer and materials of course. Thin-walled objects with sharp corners and no infill, need a brim. This could be the standard brim, or a custom designed brim in CAD, depending on the model (sometimes only one little area needs a brim). Otherwise I get the effect you see. Models with high infill, generally need no brim for me. I am not sure that these are the effects you see in your models, but it could be. Have a look at this: the dummy cube (green, top-right) needs a brim because it is hollow at the bottom. Otherwise the walls tend to bend inwards and it tends to come off the glass. The supports (pink and orange) also need a brim because they are very small, only a few millimeters, and they have long overhangs which tend to curl up, making the nozzle bang into these curls and knock the part off. The rest prints perfectly fine without brim. All have 100% infill (except the hollow text and ruler, and the dummy cube). For reference: text caps-heigt = 3.5mm, and its legs are 0.5mm. The green dummy block is to provide enough cooling time for the top-section of the yellow part, otherwise these top layers do not solidify. This prints well in both PET and PLA.

This looks like severe underextrusion, which could be caused by a lot of factors: too low temperature, partially clogged nozzle, worn-out teflon couplers, incorrectly mounted bowden tube, dirty feeder, slipping feeder due to wrong tension, too high speed, wrong filament diameter or setting, too much friction in the whole feeding traject, spool wound too tight (often when almost empty), material decomposing in the nozzle, way too moist, and so on. Somewhere on this forum there is a good list of causes (but I don't remember the name exactly), or google for underextrusion causes on the internet. I made tests some time ago, and yours looks like an extrusion-rate of maybe 50%? (Pictures are a bit out of focus due to taken on a dark winterday, but you get the idea).

The cones can not solidify because the hot nozzle (200°C) is continuously on top of it, and it keeps radiating heat, so the model can't get below 50°C. The "dummy cooling tower" allows the nozzle to be busy for some time, far away from the object, so then it has time to cool down. This greatly reduces this overheating deformation, but it does not totally eliminate it. In very small objects you might still run into it. This is another picture showing the effect: they are 20mm high. The ones printed standing (left, printed in different temperatures and speeds) are deformed due to insufficient cooling. The ones printed laying on their back (right) with several together, are okay. Try various approaches, and various temps and speeds, with and without "dummy cooling tower", so you see the difference.

Even if you could get it to print, you are likely to run into cooling issues, like this below. Unless you would print multiple parts next to each other, or with a dummy cooling tower next to it (=the square blocks in this picture). An option might be to make the walls 0.5mm thick, so they print well on a 0.4mm nozzle, even after converting to STL. And then manually drill out the hole with a separate drill chuck. But it won't have much strenght. I don't have a 0.25mm nozzle, so I can't comment on that. Maybe you might want to search for a totally different solution too, if the purpose allows that. Or go for SLA if it doesn't need any strength (these light-curing resins might become very brittle and might deform under loads).

In my experience, normal PLA gets *much* harder and stiffer after a year, which is a disadvantage if parts need to flex a little bit (like the carabiners above). While PET keeps its original flexibility. How does tough PLA behave in this aspect in the long term, in your experience?

I recommend that you experiment with various concepts on small test models, to see what works for your application. In this spool holder for my UM2 printers, I use several different mechanisms: - Two snap-fits for the main holder (cyan part), where it fits into the printer, just like the original. This snap-fit has long travel-ways, and the openings in the printer have a lot of dimensional tolerance, so it is not likely to break. - The pins (red) are clamped and kept in place by other parts. - The rotating head and bearing are mounted on the main house with a big inox M8 bolt and self-locking nut (mediumblue and darkblue). - Both halves of the bearing-housing (yellow and spring-green) and the bearing itself (purple) are secured with nylon M4 bolts and nuts. - The bearing (purple) is a standard 608 bearing (as in skater wheels). My designs often are a compromise between ease of printing, ease of assembly, functionality (it has to work), and strength and stability. For printing, I want a big flat bottom plate sitting on the glass, and an orientation that does not require too much overhangs and bridges (as I have single-nozzle printers). If supports are needed, they should be in an area that does not damage the model (like in the main cyan housing: supports are required and would be on the inside, and it would be printed with the wide opening down). The anti-unwind clamp shown below also uses two M4 nylon nuts and bolts to clamp things, and it can very easily be printed on its back. One bolt clamps the filament, the other clamps the whole thing on the rim of the spool. By itself, the red models snap-fits and freely slides around the rim of this spool.

If you are on Windows, I would suggest you try DesignSpark Mechanical for editing: this is also free (just requires registration) and is much easier and more stable than FreeCAD. There are a lot of good tutorials on Youtube. Have a look at them and see if you like its concept. Or try any of the free online-programs (I have no experience with them). Then for mounting, in my experience: For multiple-part objects, I often mount them with nylon screws and nuts. Provide the holes and hex-indents for the nut and screw in the model, so they sit recessed and are easy to assemble and disassemble. Press-fitting parts requires a lot of testing to get it right, and PLA (and lots of other 3D-printing materials) tend to creep under mechanical loads, so they deform over time. So the fitting may become loose. Snap-lockings work well for PET. For PLA they work when inserting, but after a year the PLA gets harder, and then the snap-locks break apart when you try to unlock them. 3D-printing screws and threads is hard, and tapping them is even harder, especially in PLA: this melts almost immediately. Glueing with cyanoacrylate works pretty well for PLA. Depending on the circumstances, this can be a good option. One of many possibilities of a clamping bolt and nut (this one has to slide and be adjustable):