geert_2

While there could still be underextrusion (hard to see), that alone would not cause the rounded corners, I think. Too much cooling would rather cause no layer bonding, and thus splitting layers. But the shape would more accurately follow the nozzle-path. Yes, putting a desktop fan in front of the printer at its lowest speed, and maybe 1m away, is worth trying. That would give a nice gentle flow.

I made a quick drawing to illustrate how I understood it ("a picture says more than..."): 🙂 It does not really matter what the diameter of the nozzle is: it just has to be able to fill the gap with molten plastic. And that gap is: line-width x layer-height x line-length (=red bar in the drawing). Although of course a 1.0mm nozzle can't draw a line of 0.4mm wide, and neither can a 0.1mm nozzle. So nozzle-diameter and line-width should be about the same, both about 0.4mm, to get decent results. The extruder has to push that volume of filament through the nozzle. In calculating how far to move the filament, it will have to consider the filament diameter 2.85mm, travel distance of the filament (e-steps?), and maybe losses due to "partial slipping" in the feeder. I think...

I have done line-heigths of 0.4mm with a 0.4mm nozzle (=the only nozzle-size on my UM2), just for experimenting, and I started to see a bit similar deformations. Corners got rounded more than at lower heights, and it didn't cool well and deformed. From left to right: layer-height = 0.4mm, 0.3mm, 0.2mm, 0.1mm, 0.06mm. Top row = 50mm/s, bottom row = 10mm/s if I remember well. Each block is 20mm x 10mm x 10mm. This one is printed at 0.4mm layer height, 0.4mm nozzle: The next one is 0.06mm. (Also, you see it starts to discolor due to sitting too long in the nozzle at elevated temperatures, even though I printed it below the recommended temp range. Material is PET.)

Two top layers is not enough to fill the roof smoothly. You probably need 4. Further, on such a small object, I think it is not getting enough cooling: these thick sausages don't cool down and stay molten for too long, so they get dragged along and get deformed. Plus there is the hot nozzle sitting on top of this, radiating heat and preventing solidifying. You will need to print multiple models at the same time, and bigger models. For a 1mm or 1.2mm nozzle, it will be hard to print details finer than ca. 1.5mm. This is a guess, but it won't be far off, I think. :-)

This took me some time to understand, but as I see it now, it is the *printed* sausage that counts, not the nozzle. The extruded amount has to fill the printed sausage. So, ideally if there would be no air trapped between the sausages, and at 100% infill, it would print perfectly *rectangular bars* with dimensions: line-width x line-height x traveled distance. I guess that is where this calculation comes from? The nozzle-opening limits accuracy: too wide, and it can't print fine lines; too narrow, and it can't extrude enough material. Is this correct?

Since today there is no link to the forum ("Community") anymore on the Ultimaker website. So, for people who don't have the sublink "https://community.ultimaker.com", from which the forum is still reachable, it seems to have completely disappeared. Probably just a glitch that has been overlooked? Could you have a look at that and repair the link? Thanks. PS: my browser is a standard Pale Moon, which is a Firefox split-off that has kept the traditional menu-structure and status bar, like in the good old days before Firefox went evil à la Google. Pale Moon is a very standard browser that follows the HTML- and CSS-specs as closely as possible, thus any standard compliant website should work. So I don't think this has anything to do with my browser? My operating system is Windows 7, but that shouldn't make any difference for websites either.

The overshoot on the corners is most likely mechanical: oscillations and vibrations because the printer head can not immediately stop and change direction. Plus it will extrude a bit more material per mm while the head is slowing down, compared to printing straight lines. This effect can be reduced by printing slower, and by changing some settings, although I don't know which ones (never experimented with that). I have no idea about the shift. Can you see that shift in the layer preview in the slicer, when zooming in a lot? I guess if you move that transition from straight to rounded corners up and down, the shift also moves? If you look close while printing, does the printer change direction at that point, from clockwise to counter-clockwise, or so? In which case it couls also indicate a mechanical problem such as too much play?

In future posts, it is best if you place the pictures directly in your message: you can drag and drop image-files (JPG, PNG) from the Windows Explorer directly into this message-board. Not everyone is willing to follow unknown links. I believe your intensions are good, but we all get too many spam mails with links to virusses, and fishing mails, so people tend to become cautious. That said, I am not using the same Cura version, and I always design my custom supports in CAD too, so I can't answer this question.

Controleer ook eens of de printkop nog soepel beweegt, door hem manueel over en weer te schuiven (met de printer uit natuurlijk)? Dit voor het geval één van de lagers geblokkeerd zou zitten door vuil, of er iets tussen de riemen zou zitten.

Did you clean and oil the X and Y rods, so they move very smoothly? More friction could also lead to more noise. On one of my UM2, I once had a belt rubbing against the flange of a pulley, but that made a squeeking sound. After I carefully wiped the edge (only the edge!) of the belt with chemically inert silicone grease, the noise went away. Don't use petroleum oils or grease on rubber, this may damage it. Only use inert lubricants. And don't get it on the teeth: it could make them slip.

This is severe underextrusion, caused by not enough material flow. But there can be a lot of reasons: blocked nozzle due to burnt residu, dirt particles in the nozzle, kinks in the filament which prevent it from moving well through the feeding traject, spool not willing to unwind for whatever reason, printing too fast, too cold, and probably many many others. User gr5 has a good video and info on this, search for it. I guess the flow in your model is only ca. 30 to 50% of what it should be, once the problem occurs. See the photos of the tests I did a couple of months ago (flow-rate in percent, in black marker on the model). This model is quite small: if I remember well, the left block was 10mm wide and 5mm high. Made on an UM2 (non-plus).

If it is only in the back corner, and it was not earlier, maybe you could have a look into these: 1) Is there something stuck under the glass, so it bends upwards? Sand, dust, a burr around one of the holes in the aluminum plate, or so...? 2) Or did the glass itself deform over time? Try rotating it 180° clockwise and see if and in which corner the problem remains? 3) In the left back corner, the filament has to make a very tight bending to get through the bowden tube. This may cause a lot of additional friction in the bowden tube. But it may also cause extra friction in the nozzle itself, since the filament enters the nozzle under a different angle, compared to printing in the right-front area (at least on Ultimakers, but this could be different on other brands). When printing hard filaments like PLA, this is an issue on old UM2 printers (non-plus). I don't know if it is also in other models. Disconnect the bowden tube from both ends, and try feeding filament through it *manually*, with the head in various positions (front-left, back-left, center, back-right, front-right), and feel if there is any difference in friction? 4) If it is also on other areas: maybe a bad spool of filament, or different brand? Kinks in the filament? Or any other changes in settings (speed, temp)? Or the spool is nearly empty, which adds a lot of "anti-unwinding resistance" to the feeding traject?

Probably not the answer you are looking for, but an option you always have is to design any supports directly in your CAD model, and switch the auto-generation off. Then you can customise it to your specific needs and design-in all stability and other features you need. For example in these tiny models, I can not get in with a knife to cut the supports loose, so I need them (=orange and red parts) to extend from the model to grip them with pliers. And due to the relatively large overhang and tiny ground-surface of the supports, I wanted to give them some custom brim for more stability. The top layer of the supports have ribs, so they don't glue to the real model, to make removal easier (for my single nozzle printers). These are things that standard supports can hardly do, since they can't know the intentions of the designer, the purpose of the model, or the mechanical post-processing tools the designer has available.

For an immediate solution, yes. For a long term solution, I think it might be worth looking further into the idea of drying on the fly, but then over a much longer path, and with a means of evacuating the released moisture. And with a means to cool the filament down again after dry-heating, before it enters the feeder. If PLA would enter the feeder at 50 or 60°C, it would get totally deformed by the feeder wheel pressure, and would cause grinding and/or blocking in the bowden tube. So, after the heating and drying cycle, there needs to be a cooling cycle too (without moisture entering again). It might not be easy to put all that in a small housing with no friction, and to make it still work after a week-end of standing still. But I still like the idea, especially in compact spaces where a separate dry-box might not fit well. :-)

Looks like a leaking nozzle, maybe not tightened enough? Or maybe there was some debris on the thread when it was tightened? I think they need to be tightened when warm, so any filament-remains on the thread would be molten. If fastened cold, it could leak. But I am not familiar with an UMO, so it is best if you google for a manual how to disassemble the UMO-nozzle.

I like the conceptual idea of "drying on the fly" very much, but I have the same doubts as Ishy. Normally people recommend to dry plastic for at least *several hours*, preferably in the presence of desiccant (to remove the dislodged moisture). My printer usually consumes about 1m of filament per hour. So a heater would have to be 2 to 3 meters long to meet that spec, and it would need to have escape-routes for the moisture. So I am not saying this can't work, I hope it does, but I do have doubts. Also, this seems to be only for 1.75mm. Along the same line of thinking: in the chemical industry they have heater systems for heating thin pipes and sensor lines with fluids, so they do not freeze in winter. These sort of "socks" do encapsulate the pipes. Maybe you could have a look into such systems, and try to adapt one? Maybe then you could make a longer one? And at places cut holes to allow moist air out and fresh air in? Not sure if that would work either, but it could be an interesting experiment?

In the past I have seen people doing this by just watching the printer, and when it reaches the desired layer, manually pause it, and manually change filament. Then continue. This may not be the most convenient method, but if it works to get the job done... :-)

Does this also happen with a slightly larger block with *plain* walls (=without any text or indents)? Just a clean cube? In my UM2 printers, when printing small objects with very sudden changes in layer area, that also shows up as "horizontal lines" in the print, due to the differences in layer cooling time. See the images below. For me, printing slower and cooler helps, or for small objects, printing a dummy "cooling tower" next to the real model. But this will only work for you if this is the cause, of course. I could imagine that an irregular feed also would cause this problem: changes in flow rate, filament diameter, kinks in filament,... Or irregularities in nozzle temperature, or air flow around the printer. Or vibrations in the X, Y and/or Z-axis? Sift for the laboratory sink: Tiny part (see the irregularities in the side-wall in front); the ruler is in mm/cm:

I thought all countries world-wide had switched to the metric system? The UK has switched gradually, with the latest units changed around the year 2000, I think? Today in the UK the old imperial units are only allowed for "drinking and driving", as one Englishman said on another forum. (=for road signs and beer bottle labels). But correct me if this info is incorrect. :-) The USA has officially adopted the metric system in 1875 (yes, 18..., not 19...), although most politicians, government officials, and press people are not aware of this. So they keep converting these metric units back to imperail units for "convenience", although it is very inconvenient for them, and it slows down conversion in everyday life. This causes big headaches for US-technicians, engineers, industry, and all other internationally oriented businesses. I thought the handfull remaining countries in Africa and Asia, previous English colonies, had also switched recently, except maybe Burma/Myanmar? Although in daily life people tend to keep using their old habits of course, just like we still use horse-power for cars instead of kilowatt. I guess because the number in horse-power is 1.3x higher than in kilowatt, so it sounds more impressive. :-)

Did you print this with an enclosed front? If not, I could imagine that too much cold ambient airflow would cause this? Or maybe too much cooling fan? I have no experience with this material, but generally high-temp materials need an enclosed front and little or no cooling fan for good layer-adhesion. Otherwise the new layer won't melt into the previous layer.

Originally I also used a piece of plain paper, but now I just do it visually: I print a border of 15cm x 15cm, one layer thick, with a skirt of 10 lines, and then adjust on the fly while it is printing this skirt. This works well for my UM2; I don't know if this is recommended for later printers like UM3 or 5.

Yes, that is how I understood it. :-) To be totally sure I just tried it on my old Cura 14.09, and there the function is named "Split object into parts". It works, and you can indeed delete each of the parts separately afterwards. So, you could save the remaining as gcode, and print them one by one in this way. However, it does have side-effects: if there are hollows in an object, like my watermark text, these will also be extracted as separate objects... Lol. Which should have been this:

This is a good idea. Even though I printed so many "dummy cooling towers" next to my models, this has never occured to me. I am going to borrow this idea. :-)

I think in older Cura versions it was possible to "split object in parts" or something similar? This would split a model into all its parts, and these would all be distributed on the build plate separately. Then you could delete the unwanted parts one by one, and only keep the desired ones. You would have to redo this whole sequence of splitting and deleting for every part, so it is a bit cumbersome, but it could be a temporary workaround to get a job done. I don't know if this is still possible in the newest versions of Cura (I haven't tried)?

I guess that your model has not enough cooling. Print this slow, as cool as possible, and print a dummy model next to the real model. Try a dummy cube, or an inverse of the model. So that the hot nozzle has to spend some time away from the real model, and the real model gets time to cool. But as said, it is a guess. See this conceptual example: Edit: it looks like this effect with and without dummy cooling tower: