geert_2

If you could somehow redesign the model so that there are no steep overhangs, it could work. Make smooth transistions, and no overhangs larger than 45° (from vertical). Might be worth trying?

Or a virtual gallery with nice big photos of the best models, plus a few tips and tricks on how to make them. Once people have seen these models with the infill shining through, it might well become a hype.

Technically, it is not easy to print little holes. When layed down, the molten sausages need something to grip too, otherwise they sag and go all places, out of control. Retracting and extruding again every 0.5mm also isn't going to work. Zigzagging the surface also doesn't work well, since corners are way too rounded. Apart from post processing like sanding or sandblasting, or so, the only method I know is setting the material flow to 30...70%, which will give a sponge look. But you have no control over the pattern: it will differ very much from material to material. See my flow tests on PLA here, which show the patterns (I made photos this week): https://community.ultimaker.com/topic/25522-photos-of-under-and-overextrusion-patterns/ Or see these photos, where another user has undesired underextrusion (probably ca. 40% flowrate), but his photos excellently demonstrate the sponge-effect: https://community.ultimaker.com/topic/25460-help-needed-on-layers-printed-with-random-patterns/ I don't know if it is possible to set only the flow rate of the outer surface layers to a lower value, but keep 100% for the inner surface layers and infill, for strength? (I don't have the latest versions of Cura yet.)

Before we had bought our UM2 printers, I had a couple of testmodels made on an industrial Objet 3D-printer of my collegues (60000 euro printer if I remember well). This is one that sprays liquid just like an inkjet printer, and then it is cured by an UV-lamp. These UV-cured models had a bit finer details than I can achieve with a 0.4mm nozzle on my UM2, but they are weak: very brittle, and they deform really quickly under continuous loads, even gentle loads. Today there are cheap machines on the market with far higher details, and also UV-cured resin, from 500 to 1000 euro on I think. But if you would consider such technology, first have a couple of tests printed to see how strong and durable it is. An alternative: make a silicone mould, and cast the keys in a touch polyurethane. This is probably going to be way stronger. Mould-making and casting is very educational too: it shows all the basics that occur in injection moulding too: undercuts, shrinking while curing, deformations, parting lines, unaccessible areas, draft, adding pigments,... So you can have the students do it. If you use fast curing silicones and PU, it can be done in a lesson.

I tend to stick with the wrong filament diameter, or wrong nozzle diameter settings somewhere. Or wrong flow percentage. Or something else along this line, which would affect the flowrate and cause this amount of underextrusion. It could be other settings that via-via affect flow too. Maybe hidden settings somewhere? So I would suggest you keep searching in that direction. If you compare your pattern to the one in my flow-rate tests, you will see that it is about 30...50% flow (which would *exactly* match a 1.75 vs 2.85mm mismatch. Keep the scale in mind: the big block on the left is 10mm wide, nozzle size is 0.4mm, in my tests. See here: https://community.ultimaker.com/topic/25522-photos-of-under-and-overextrusion-patterns/

Although I don't have a dual nozzle printer, I have read and seen on photos that PVA support material is very prone to stringing. So I doubt if you can totally avoid it. But you can sure try to minimise it. At least, be sure to keep the support filament very dry, and dry prior to printing if necessary, and preferably keep it in a closed box while printing. Make a box yourself or buy one of the available models (I don't remember the name, you will need to search the forum for this).

Could it be strings from the support material? On the place where there is a string, this room obviously can not be occupied by PLA anymore, thus leaving a gap in the PLA after dissolving the supports. I am not saying this is the cause, but it could be?

Completely off-topic: these are nice prints. They look like sort of ceramics or glass. I wouldn't think they were 3D-printed. Is that the gyroid infill shining through? And which materials did you use, looks like Polyalchemy Elixir?

De draden van de LEDs komen zeer dicht tegen de aandrijfwieltjes, vooral links achteraan. Zou het kunnen dat ze ergens contact maken, doorgesleten zijn en daar kortsluiting maken? Een voeding die stilaan defect begint te geraken en niet meer 100% stroom kan leveren, zou misschien ook een oorzaak kunnen zijn. Dan kan de spanning af en toe misschien even wegvallen.

Some time ago I made a series of tests to see what different percentages of underextrusion and overextrusion would look like. Here are some photos. The percentage of the flow-rate is written on the model, thus "30" means a flow rate of 30%. So when experiencing underextrusion or overextrusion you can guestimate the amount by comparing it to these tests. Best is if you first run these tests on your own machine on a moment when it is good, with your own materials. So you can compare later on. There may be differences between machines, and materials. Sharpness of the photos is not optimal due to not enough light, and thus a big lens opening and not enough depth. But you get the idea. Next summer when there is a lot of daylight again, I will make better photos. Material is colorFabb PLA/PHA, color "warm red". This filament is slightly translucent, which reduces contrast. I should have taken a color that was totally opaque... The "big" block on the left is 10mm wide, nozzle 0.4mm. Flow-rates of 100 and 110% look best. Below there are gaps and sort of sponge-patterns. Above lines start to get too thick near corners. At 150% I stopped the test prematurely due to too much overextrusion (didn't want to do damage). File: flowtest2.stl

Forgot to say: the original poster is obviously familiar with epoxies, but for other readers who are not: epoxy-curing is a very exothermic reaction: it generates a lot of heat. I have had epoxies going far above 100°C when curing, melting the plastic cups in which they were mixed, and once even melting a thin lead container. In extreme cases the part could cause fire or explode due to internal stresses (fortunately I haven't had that). So, temp while curing is likely to go above the 50°C limit of PLA, and the mould may deform or even completely melt. This could also happen when impregnating larger volumes of the "sponge PLA" I mentioned above. Silicone usually is quite heat-resistant for a short while (~200°C or more), so that is less of a problem. Use a slow-curing low-exothermic epoxy, or apply in very small quantities, in contact with PLA. Also, slower curing versions tend to shrink less, which might also be an issue in some cases: shrinking could cause gaps in the material, or deformations, just like shrinking in injection moulding.

Why not simply apply a gasket-type silicone, before mating the parts? Like in automotive applications? That would fill all surface-defects like layer lines too. If you want a removable gasket, use a non-stick silicone. Otherwise use a sticky one. Note that most silicones are water-tight because they repell water. But they are not water-vapour tight. And they are not oil-tight: solvents, oils and liquid parafines leak straight through it.

If you want a sort of sponge look, then after the print starts, manually reduce material flow to 30, 40 or 50%. Try this on a small test cube first. Of course, this "sponge" won't be very strong, but it will have the general shape of the model, except for fine details like text which are totally lost of course. But I don't think it will get very strong after impregnating with epoxy. The bonding between the epoxy and the PLA (or whatever else filament you use) will be critical. Normally in particle-filled epoxies, the particles need to be specially treated to make the epoxy bond well to them. So you would first need to treat and chemically activate the PLA-surface. It might work with a special "plastic activator" spray? But then the procedure would become complex: print, post process, activate, let dry, impregnate a little bit with epoxy (not too much, otherwise it leaks away), apply next epoxy layer, post process,... Then I think printing a mould and fill that with epoxy might be better. Definitely use multiple layers of a good release spray or release liquid. But even then releasing might be a problem. Or print a master model and make a silicone mould of it, and fill that mould with epoxy. You could also print a "shell" around the silicone mould for more stability. If well designed, you can put the model into the shell, and then pour silicone into it. So the shell acts as a container for the silicone too. Let cure, open shell, cut silicone, remove master, and your mould is ready. Use plenty of silicone oil as release spray to saturate the silicone, and prolong mould life. Such a mould can be re-used several times (depending on shape, up to 50x), which saves a lot of time in the end. It's messy, but if you use epoxy anyway... Flexible silicone mould release much easier than hard plastic or plaster moulds. That is why I would go with a hard shell as container, but pour silicone into it for the actual mould. The shell is only for stability, and for containing the silicone. Make sure you can easily open the shell. Probably you are familiar with mould making. If not, search on Youtube for mould making and casting. There are lots of excellent tutorials, both with thin liquid silicone and thick silicone paste (brush-on types).

I think it is best to test all parts of the whole feeder traject *separately* first, to determine *where* the problem is, before changing parts. I would suggest: disconnect the bowden tube at both ends. Remove the filament from the printer, but leave the spool on the printer. Then step by step: - Try to manually unwind filament from the spool. Does it go smooth, almost without resistance? Sometimes filament windings may get stuck under each other. - Feed a piece of filament manually into the bowden tube (with tube disconnected at both ends). Does it go smooth, almost without resistance? If there would be kinks in the filament, or a too thight bending radius, this might cause a huge amount of friction. - Heat up the nozzle to the correct temp for that material. Manually feed filament into it (=directly into the nozzle, without bowden tube). Do you get a thick nice flow, without having to apply too much pressure? Keep pushing for some time, a few minutes. Does it keep going well. - Using the "Move Material" function of the printer, feed a piece of filament through the feeder (=without bowden tube, only the feeder). Does that go well? Try to pull on the filament, or try to keep the filament from moving, while you use the "Move Material" function, and see if it slips too soon? If I remember well, user gr5 said that the feeder should be able to lift the whole printer, whithout the feeder slipping? Try something along this line. Somewhere something in this traject must be out of spec. Or it could be a combination, e.g. partially blocked nozzle, worn out coupler (its color looks okay in the photo, but it is hard to see if there are dents on the inside), worn out bowden tube, worn out feeder (e.g. if used for abrasive materials), kinks in filament, filament mangled on spool or a too tight bending radius near the end of the spool. So the sum could be too much.

Hard to see in the photos. In addition to underextrusion, filament that is too wet might also cause a rough surface. If you make a test cube of 15mm x 15mm x 15mm, 100% filled, same temp and speed, does that come out well? If the cause would be a bed leveling problem, it should be smoothed out after max. 1mm height.

Is the bowden tube installed correctly (=wider opening at feeder)? If the feeder wheel would be okay (=not covered in plastic, and teeth not grinded away), and the nozzle is okay, and the bowden-tube is okay, maybe it could be the sum of friction in the whole feeding traject that is too high. Then you need to examine each part of the traject for friction or blockings. Also, I vaguely remember someone else where something was broken in the feeder, causing issues.

Now I am going to guess... Could the underextrusion *in the cube* be a temperature issue? This appears near a drastic change in printed surface per layer. If the "Minimum time per layer" setting would kick in, flow would be reduced, heat demand would be less (since hot filament stays longer in the nozzle), and thus the temperature would be reduced. If this reduction would overshoot to the point where the filament would not melt enough, it might explain this. Another guess: is it always the same time into a print (thus not at the same layer, but time) when this happens, at a given printing speed in mm/s? Let's say always after 40 minutes at 50mm/s? In such a case, it could be a feeding problem. After a print is completed, if filament is sitting still too long, then it might gradually get flattened in the feeder due to constant pressure of the feeder wheel on one spot. Thus causing a slow gradual deformation. This might cause wide flat in the filament, which has trouble getting into the nozzle. This flat would have to travel all the way through the bowden tube to get to the nozzle, so at a given flow rate, it should always take the same time. To test for this, at the start of a new print, could you mark the filament with a black alcohol marker pen, just at the entrance of the feeder? And then watch if the problem always occurs when that marked spot reaches the nozzle? If not, then you can exclude this as a cause. If yes, you could further investigate. The Z-stepper drivers (the chips, not the motor itself) getting too hot could also be an issue. I had that occasionally in the beginning with an UM2. But it showed up rather randomly, and its phenomena were different: the Z-plate would just drop 5mm. It didn't look like this. After reducing motor current from 1200mA to 900mA or so (I don't exactly remember), it never happened again. But as said, all these are guesses...

You don't mention which printer you have. On my Ultimaker 2 printers I have added a wire to wipe the nozzle after priming. It is spring steel wire, like used in dental bridges. This prevents the priming blob from getting dragged around, and molten filament on the outside of the nozzle is wiped away too. Due to the spring-effect, it does not hurt the nozzle. In the beginning I manually pulled it straight down and kept it aside with a pincette ("tweezers" in English?). See the pictures below. It is based on an idea of another user (I don't remember his name), who called it "filcatch", but I made it in steel instead of 3D-printed. I am not sure if this is your problem, or if this method could be applied to your printer too. Use common sense if you would try. Depending on the condition of the nozzle, and on filament type, during printing some filament may accumulate on the nozzle too. Printing slow and cool reduces the effect, but it is hard to totally eliminate. In PLA it usually is no problem, but in my PET it happens more often. And then gradually that residu gets brown, sags, and causes a big brown blob on the prints...

If not yet done, maybe you could open the old good and new bad gcode files and compare them line per line? If possible for the same object? If old gcode files of the same model print well, but new don't, then the underextrusion has to be a slicer or slicer-settings issue, and not an STL and not printer-hardware problem. This apart from the bed-leveling, which is something separate. In my UM2-printers I set it a bit closer than the required 0.1mm, for optimal results. And I adjust on the fly while printing the skirt around the object, if necessary. But this could be very different in other printer brands (I don't know yours). Probably you have checked and double checked it, but could it be that the printer-definition in Cura has changed? So it thinks it is printing to another model, and it is using wrong defaults?

If you wouldn't mind babysitting the printer, you could make it all in one design (=one part in one STL-file). And then manually pause the printer at the appropriate height, change filament, and continue? If you would print a dummy block next to the real print, you can hit the pause button while it is on the dummy block at the right height, so it doesn't stop halfway a layer in your real model. Not practical, but it might work. There are people who do similar things to insert metal (or other) inlays halfway a print.

Could it be that the nozzle cooling fan is not working correctly? This might at least explain underextrusion after a short while. Also a too high required material flow - more than the printer can deliver - would explain that. The first layers are usually printed at lower speeds and sometimes higher temps. Nozzle jams, feeder jams, spool jams, too much friction somewhere, too low temp, too high speed, too thick layers, too thick layer lines,... But it wouldn't of course explain erratic nozzle-movements. If the file-code would be garbled, it seems most unlikely that the general shape would still be correct. I think. And if only the infill-generator would mess-up, then the walls should still print fine. So I can't set aside the idea it has to do with the 1.75mm vs. 2.85mm filament: or an incorrect setting, or a rare bug triggered, or a "stuck value"? What happens if you *manually* feed material through the nozzle while hot, thus bypassing the whole feeder traject (if possible on your printer)? Good, easy flow or not? I also think designing and printing a small test cube (e.g.: 10mm x 10mm x 10mm, 100% filled) might help sorting things out. First print at the default or at very safe settings (e.g.: 210°C and 30mm/s for PLA). With brim. Then print at your current settings. Also, make a close-up film if you can, so we can see the nozzle movements and extrusion. Due to uneven extrusion, or uneven deposited material, could it look like the nozzle is moving erratically without actually doing so? Just guessing... But since we don't know, we need to make educated guesses.

I think that cyanoacrylate glue should work well. It does exist in larger ("industrial") bottles too for bigger surfaces. At least it works very well for glueing PLA to PLA in my models. Might be worth a try?

Some time ago I ran flow-rate tests on a test model, ranging from 30% flow rate, 50%, 70%, ... etc..., 100%, ... up to 150%. (But I don't have photos of it yet.) This looks like a flow rate in-between 30% and 50% in my tests, although closer to 30%, maybe 35...40%. As mentioned above, a setting of 1.75mm filament when using 2.85mm filament could explain this: that should result in a real flow rate of 37%. As would a way too high speed, or a way too low temp for the amount of material you try to extrude. Assuming that the hardware is okay.

I once hit hard on a PLA model with thin wall and 20% infill, and both the wall and infill got severely crushed and dented at the place of impact. The damage looked weird and very "un-plastic like", but rather like a crashed car. So I am not sure PLA infill is going to protect, unless the soft shell would be very thick and distribute the loads widely enough, and the infill would be nearly solid and very thick too. So I would rather print the whole thing in nylon if it has to be tough, or in thick solid PLA if it has to be hard. Another option could be to print the shell, and then fill the inside with polyurethane. PU exists in foam, in rubberlike solids (after curing), and in very hard solid versions. I think a very tough and hard PU like used in skater wheels might be a good choice. Or print a mould in PLA, carefully seal it with a non-stick layer, and cast it in PU. Then you have a very tough model in a single material. This might go a lot faster and cost less, once you have the setup right, if you need multuple models. A lot of artists and sculpters use this method in various forms. If it has to be light-weight, they tend to fill it with PU-foam. Or sometimes they work the other way round: first model the thing in PU foam, and cut and remove foam as desired. And then they seal the whole model with a hard and solid PU layer. So, a combination of 3D-printing for a detailed model, and casting for strength and speed, might also work well. Then you have the best of both.

Looks like that gold-painted girl in the James Bond film Goldfinger... To get to the bottom of things: if this is recycled PLA, couldn't it be that there is dirt in it which causes the clogs? Or particles from other sorts of PLA with a higher melting temp, or ABS or PET particles, or so? Or are the glitter-particles way too big? Ultimaker Pearl and colorFabb gold (which looks more brown than gold to me) also have glitters in it, but they are so small it never caused any issues for me.