geert_2

The PET I have does similar things: - While printing, it accumulates on the nozzle. This goo then gets brown, and sags onto the model, and is deposited as big brown blobs. The brown color is clearly visible on white, but it may not be visible on black. These blobs also tend to cause fine hairs in the print, both when the blob sags from the nozzle, and when the nozzle passes through previously deposited blobs. - When bridging, it does not pull a nice bridge. But the strand rather tends to snap like chewing gum and fold back onto itself, causing a blob under the nozzle. This is then deposited onto the next wall upon arriving there. - When traveling through air, the nozzle leaks a little bit (overpressure in the nozzle that is releasing), also causing a little blob or "insect antenna" onto the next wall. It looks quite similar to what you have. Watch carefully while printing, and you can see it happen. Printing slow and cool reduces these effects in my models, but I can't totally eliminate them.

Yes indeed. My models typically take 2...3 hours to print; I rarely have longer printing times. So I don't know how it would hold up for long prints that take a day.

Yes indeed, "insect antennas". I see this on PET too. Printing slower and cooler helps a bit for me, since there is less pressure build-up in the nozzle, and it leaks less. But I don't know a way to eliminate it.

I print PET on clean bare glass. Sometimes I use the "salt method", however this does not increase bonding for PET (contrary to for PLA), but rather reduces it slightly, but it makes it a lot easier to remove parts. To prevent warping, I use no cooling fans, or the absolute minimum. But no fans makes it more difficult to bridge gaps, so this is not suitable for every design. Most of my designs are long flat models without bridges. I tried dilluted white wood glue too a couple of times (ca. 10% glue in water), but this bonded way too strong: it tended to chip the glass while cooling, even before I started pulling the model off. So if you need a strong bond, this should work. :-) But the bed has to be hot enough: 60 or 70°C won't do for me. I need to set it to ca. 80...85°C. The photo below is what the bottom of my PET parts looks like: no corners lifting, and no problems making tiny holes. These are small models (see ruler behing in mm and cm), but it looks the same for my bigger models. Printed with the salt method to facilitate removal: you can see the tiny pits caused by the salt (looks like corrosion pits).

What about changing your bonding method, so your prints can be removed easily? For PLA, I use the "salt method": wiping the glass with a tissue moistened with salt water, prior to printing. No glue. This gives good bonding while hot, but absolutely no bonding when the glass is cold. So, models can be taken off very easily. But this is only for PLA, and for low flat models (not for "lantern poles"). For my old manual, see here: https://www.uantwerpen.be/nl/personeel/geert-keteleer/manuals/ User neotko uses hairspray for bonding, and he removes his models by adding a few drops of window cleaner (if I remember well) when ready. This creeps under the models and dislodges them easily. This seems to work very well for his big models too. For wood glue, some people put the model and glass bed in the fridge, and then it pops off. Probably there will be other methods that are worth trying? With the salt method, I can easily print these models at 100% infill, without edges lifting, and without damaging them when taking off after completion.

Another trick is to model in your CAD design a few dots outside of your model, sitting 1mm lower than the real model. So Cura will drop the dots onto the bed, and then the real model will float 1mm above the glass. This of course will require you to use supports for the floating part, otherwise it will print spaghetti.

I guess the snap-areas are the two "keyhole" openings? First I would make the splits a bit wider, so the clamps would deform less while snapping over the rails. That should still give enough retention. Depending on the load it has to see, I would also use a higher infill, maybe 50...70% or so? And the layer-lines should definitely go in such a way that they cause no weaknesses. Just like when you cut wood along the grain. You also need to print hot enough, and slow, to get a good layer-bonding and no underextrusion. Personally, I have no problems with PET snap-fits, like these chains. In PLA they crack after some time, due to the PLA being too brittle and stiff. But PET is flexible enough. The cream ones are PLA, and they begin to crack and deform after repeated use. The green one is PET: no cracks, no deformation. For reference: text caps-height is 3.5mm (=hollow watermark in the model). I printed these solid 100% infill, at 215°C, and slow at 25mm/s, 0.1mm layers.

For my (older) UM2 and PLA and PET materials, 25...30mm/s is slow enough to get good quality. At 50mm/s (default for PLA) quality is okay for most models, but not optimal for high details. Printing at 10mm/s is too slow: then the material is sitting too long in the nozzle and it starts to discolor and decompose (gets brown). It also depends on the temperature. Lower speed gives less ringing, cleaner corners, and better layer bonding. I would say, make a small test model with your typical features, and print that at various speeds and temperatures, and closely watch what happens while printing. This may cost a day of testing, but you will soon win that time back.

Some materials do not print well over gaps: instead of pulling a nice string, they snap and fold back into a blob on the nozzle, thus effectively printing nothing, and not closing the gap. That blob is then deposited on the next wall the printer encounters, causing a weird sort of "insect antennas". Could it be that you have this effect too? In addition to the fact that you are not printing diagonal lines over the gaps, but concentric lines, which have nothing to hold onto.

Hello, During the years the site has been getting better and better, so this is a good evolution. But I think it could still contain a bit more diagnostics manuals and tips and tricks that come back very often in the forum. For example: - A detailed list of all existing and well working bonding methods to the glass bed: the "official" glue stick, the official glue stick + wiping it with a wet tissue to egalise it, 10% wood glue in water, hairspray, my salt method (PLA only), 3DLAC, sheets, etc... With all their advantages and disadvantages. So that people can try them, and choose the one that works best. - A detailed list of possible causes of underextrusion, like the one gr5 has (ask him if you can reuse it). - Etc..., just all the things that keep coming back on the forum.

Yes, obviously an atomic pull by rotating and wiggling is not going to work on split channel nozzles. :-) At the time of writing that manual, I didn't even know such nozzles did exist. For temperatures, I think the best is to try lowering and increasing temp in steps of 5 or 10°C on the fly, while doing a testprint. And then see when it starts to underextrude, or get too liquid. And then find something inbetween that works best for your typical models, speeds, materials, nozzle, and printer model. All these have an influence. Start from the default values for your situation. Thermal paste is thermally conductive paste, to get a better thermal contact between components, to heat up or cool down things more efficiently. Often people use thermally conductive but electrically insulating paste between microprocessor chips and cooling plates in computers. This is usually white or grey paste. Sometimes we use paste that is both thermally and electrically conductive. This is often copper paste, which is also used to prevent bolts and nuts to get stuck and corrode. Copper paste lubricates a bit, and makes movement easier, *but it can cause short circuits!* So it depends where you want to use it. Copper paste is more often used on screws in bolting together heavy equipment, like gear boxes, engines, pumps, piping in chemical plants, etc... I have copper paste between the nozzle and cooling plates on my UM2. It is the dark color shade (barely visible) at the bottom of the inox ring with round holes, just above the lowest thick aluminum plate, and just below the tiny red marker line.

Maybe you could circumvent this by adding four little dots of 1 layer thick in the corners, outside of your real model? So they define the borders? And always keep these dots in the design?

@silver-girl: are you sure you have a standard nozzle? I think there are a few printers with split internal channels in the nozzle, like a snake-tongue. Maybe you have such a one? Standard atomic pulls should look somewhat like these below. Here the big blob in the white indicates that the teflon coupler (UM2-printer) is worn out (=internally deformed) and needs replacement. The orange at the bottom is what an optimal result should look like. Orange/white colors shades are due to doing the atomic pulls as part of a filament color change. @PyramidHead76: yes, underextrusion can be the result of the sum of lots of little things. Added up resistance in the unwinding of the spool, spool friction, bowden-tube friction, nozzle friction,... What I usually do is unwind a bit, straighten it by rolling it up in inverse direction on a skater wheel (7cm diameter), immediately release again, and wind it loosely up again on the original spool. So the tension is off. Indeed, often I see small cracks due to the straigthening. But I do *not* let it sit under straightening stress for longer periods of time, because that causes these cracks to grow and break the filament. So after straightening, you should immediately release the stress. Especially for hard and brittle PLA filaments. Microcracks in straightened colorFabb PLA/PHA (color: natural) filament: Straightening filament with a skater wheel, then releasing it. So it sits very loose around the spool, and is prevented from falling off by the yellow slider on the edge of the spool. The bending radius is now almost identical to that of the bowden tube, causing very little friction in the tube and nozzle. This especially helps for the old UM2 (non-plus). My prints normally take 2...3 hours, and consume 2...3 meters, so I can easily do this (might be inconvenient for day-long prints). Try manually moving both unstraightened and straightened filament through bowden tube and nozzle to feel the difference in resistance.

Usually I do the cleaning with desinfection alcohol (I think that is 70% isopropylalcohol + 30% water?), then clean again with pure luke-warm tap water only. Then prior to printing, I wipe the glass with a tissue moistened with salt water. This leaves a thin, almost invisible mist of salt stuck to the glass. This greatly improves bonding for PLA, compared to printing on bare glass, as long as the glass is warm (60°C). After cooling down, models pop-off by themself, and there is zero bonding. So this method does not damage delicate models. After that, I just re-wipe the glass with the tissue, without taking it out, and without cleaning. I just clean once every 6 months. The rest is wiping with the tissue moistened with salt water only. This salt method only works for PLA. For PET, it does not improve bonding, but does reduce the risk of chipping the glass. For ABS it does not work; and I don't know about other materials. Also, this method works really well for my long and low models. But it is not optimal for thin high models like lantern poles: they tend to get knocked over. Maybe because the salt can not absorb shocks, contrary to a more flexible glue layer? For me, the "salt method" works better than the gluestick which I tried in the very beginning. But maybe I applied too much glue, and I did not wipe it smooth with water afterwards, which seems to give much better results for others. For an old manual (and too long, I know), see: https://www.uantwerpen.be/nl/personeel/geert-keteleer/manuals/ Other bonding methods: - 10% white wood glue dissolved in water (user gr5), - hairspray (user neotko), - glue stick, - 3DLAC spray (and other similar products), - very clean bare glass (a bit unpredictable), - dedicated adhesive foils (for PP, PE,...), - painter's tape (for non-heated beds), - others I don't know? I would suggest you try different methods, and variations within methods, until you find one that works well for your environment, your filament, and your models. As far as I know, there is no perfect method that always works. But there are several methods that are very good for most purposes. Stay with the printer and watch closely when doing these tests, so you can exactly see what happens, and you can abort when necessary. Also, do the tests with a very hard test model like this below, with a very small contact-area and huge overhangs which tend to curl up and try to warp the model. If the test works for this model, then you have some safety margin for more moderate models. 1. The central round area is to reduce the risk of the model coming off totally. 2. If this model would come off, the area around it prevents it from sliding around. 3. This is the usual effect if bonding is not perfect: edges warp. Notice how much corners of overhangs tend to curl up. 4. Similar model after printing: the edges did warp and lift, but the test could still be completed. In poorer bondings, the model would have fallen off. In better bondings, the ends would not have lifted. 5. Spaghetti after this model got knocked over, due to the head banging brutally into the curled-up overhangs. 6. Absolutely no warping/lifting for normal models.

If you have an oven, for example to dry filament, you could also pre-heat the cleaned glass in there. This should reduce the warming-up time a bit. This is just a theoretical idea :-) I never tried it myself, since I never take the glass out of my printer: I let it cool in the printer until the models pop off, then wipe the glass with a tissue moistened with salt water, and start the next print.

Have you checked these? - Check if the little nozzle-cooling fan at the back of the head still works? If not, or if too slow, this would cause the heat to travel up into the filament, soften it, and make it hard to get through the teflon coupler. Sometimes filament strings and hairs get sucked into it, slowing it down. - Does the feeder wheel not slip on the drive axis? Write a colored mark on both, and see if they stay aligned? I have read that this occasionally happens. - With a fine needle (with rounded edges so you don't damage the nozzle) gently and carefully poke through the nozzle? - Bad filament that is too hard to unroll near the end of the spool? This could act as a very strong spring, trying to pull the rolled-off filament back onto the spool. Also, filament with a too tight bending radius causes very high friction in the bowden tube and in the nozzle. This is why near the end of a PLA spool, I manually unroll a few meters of filament, straighten it, and roll it back onto the spool very loosely. - Bad filament with incorrect (too thick) diameter? - When manually heating the nozzle, I guess the temperature readings are okay and stable? But if you can easily push filemant through manually, it should be okay. PS: if you would like a more gently method of atomic pulls, without brutal pulling and risk of displacing nozzle-components or bending rods, you might try my method: disconnect bowden tube at front, manually heat the nozzle, manually insert and extrude some material (preferably PLA), let cool down very well until at room temp (blow compressed air to speed this up, if available), then gently wiggle and rotate the filament to dislodge dirt, heat up again to 70°C, and gently rotate while gently pulling the filament out. No brute force. For me this works equally well as the traditional atomic pulls, but it is easier on the machine. For the full manual and photos, see (and then scroll down a bit): https://www.uantwerpen.be/nl/personeel/geert-keteleer/manuals/

From the gcode, can you write text to the LCD-display? If yes, you could insert status messages every few lines in your custom code, and then let it wait for a second (so you have time to read), before doing the next command? Might make debugging easier.

Ask Simone Gierz, she can definitely make one for you. :-) See her TED-talk: https://www.youtube.com/watch?v=c0bsKc4tiuY

Dental composites are extremely hard; these are the white fillings. They are particle filled (=sort of sand), methyl metacrylate based epoxies. But they are also extremely expensive, and light-cured, so you need special equipment (blue power-LED) and thin layers to cure it. And you need water-cooled diamond disks to cut and shape it after curing, you can't get through with a knife or steel bit. So these materials are not practical for other uses. But maybe you could simulate this concept? Print a mould, coat it with release spray, and then fill it with a mix of fine sand and epoxy? More sand gives a stiffer model. For a good bonding of the particles, the sand needs to be chemically treated or washed in a certain way to activate it, but I don't know the details. But if strenght is not critical (e.g. no pulling forces) and only the hardness is of importance, you could leave this treatment out.

I don't know much about gcode, and even less about firmware, so this is just guessing. Could the problem be that "padding thing", that you see earlier in the gcode, at the end of the normal file? Try copying and inserting that padding at the end of your push-off sequences too? Apart from that, I think any resetting or homing is done at the start of a new print job, just like any normal print. In normal use an UM2 can start from whatever situation it was left in (warm/cold nozzle or bed; head or build-plate sitting anywhere), so that shouldn't cause problems I think?

Maybe just remove the filament, so it "prints" empty?

The brown spots on your print - if they got there while printing (not afterwards due to post-processing) - are likely material that got accumulated on the outside of the nozzle, then partially burned or decomposed, sagged and got deposited on the print. Or material that got partially burned inside the nozzle and then extruded. If your purpose is to make a model as a base to create silicone moulds later on, I would also recommend trying PLA. I have good experiences with Ultimaker Pearl PLA, and with colorFabb red and orange PLA/PHA. And smooth this later on, so the mould is smoother and can be removed easier. ColorFabb's Traffic Red can be smoothed very well with acetone, if desired (see the dedicated thread on acetone smoothing PLA from user cloakfiend, with lots of good photos). I don't know how Ultimaker PLA would smooth, as I never tried that. With PLA you are far less likely to run into problems like delamination, warping and clogging. I never had delamination and never had real clogs, so I can't say much about the gap in the last photo. Except this: keep watching closely while printing, and use magnifying glasses. Then you can see most of the causes of the problems as they are happening. This takes some time in the beginning, but gives a lot of extra understanding. And try lots of small testpieces, each with your typical problem aspects (depending on your designs and complexity), before doing big prints. Consider this part of the learning curve and cost. I still do this today, even after thousands of prints: if I don't really know how a new design feature is going to come out, I make a small test piece with only that feature, and with small variations on that feature.

If you are not sure it will print well, maybe in CAD you could cut out a small area of the top-side (=with the red faces). Just a tiny cube that has all the problem-aspects? And print only that, and refine settings if required? So you don't waste too much material.

I don't know this material, but if I would see this effect in PLA on my UM2 printers, this would be overextrusion. Is there a specific reason why you prefer ABS? Because it is known to be more difficult to print than materials like PLA, tough PLA and PET. (Although PLA can't handle warmer temperatures, and PET is more difficult to glue and paint than ABS.)

In addition to the load of the supply: If the heated bed takes a very long time to heat up, for example for materials that need a higher bed temperature like ABS, a side effect of heating both nozzle and bed together might be that the molten filament in the nozzle gets burnt and clogs it up by the time the bed is ready. So, first bring the bed up to a stable temp, and only then start heating the nozzle to prevent product degradation and clogs. Especially with products where the melting temp and burning- or decomposition temp are very close together.