geert_2
-
Posts
2,084 -
Joined
-
Last visited
-
Days Won
34
Content Type
Forums
Events
3D Prints
Posts posted by geert_2
-
-
A couple of years ago this has been discussed here, with good explanations and photos. If I remember well, the most common was that the connections at the back of the bed broke, and how you could repair them, but I am not sure. Maybe you can find these posts again?
-
I do understand this particular situation, but generally confirmation-dialogs are there for a very good reason. On my UM2 I have occasionally pushed the "abort" button by accident when scrolling through the menus, while fine-tuning the printing. That scroll/push wheel occasionally skips to the next option while pressing the button, if you don't hold it perfectly still. I am glad that confirmation-dialog was there...
Idem for deleting files in Windows. Potentially destructive actions should require confirmation.
That is why in UI-design the rule is that the confirmation-dialog should repeat the action-word. Thus not a simple "Yes / No" choice, because people want to start an action, that is why they pressed the button after all. So they tend to say "yes" to the action: "do it". Only, it could be the wrong action if they pressed the wrong button.
A better wording is a choice between "Abort / Cancel", or "Delete / Cancel", or "Format / Cancel", or "Left / Right", or whatever appropriate term for the action, where the action-word itself is repeated. This stops their automatisms, and they have to think: "Abort? Huh, what the... No, no, I did not want that..."
-
I don't know your printer, but to see if there are no *mechanical* blockings, try switching it off, and move all things by hand. They all should move smoothly, without requiring brute force, and they should have no hard stuck points. In this way, you can often feel if something is wrong: broken bearings, incorrect axis-alignments, gears or belts blocked by dirt, too dry rods and linear bearings, too much play, etc... Things should have a precise fit with no play, but still move quite freely and smoothly, similar to cameras and microscopes, or like the mechanics in car engines. It are precision instruments. In this way you can separate mechanical from electrical/electronical issues, hopefully.
-
Or design your custom supports in CAD, as part of the design, integrated into it. Then switch off supports in the slicer.
Standard supports are good for standard situations. But special situations might require special supports. You can make custom gaps between supports and model, you can include holes and other features to insert pliers or hooks to make removal easier, etc...
A few examples:
Custom orange and pink supports, with custom brims, that can easily be grabbed by pliers and wiggled out. These are tiny items, too small to get in there with a knife.
Free hanging supports, easy to remove, and they don't damage the bottom. These are a bit wider than the part they support, so I can easily push them down with a tiny screwdriver:
Idem, front view. Width of the gap is ca. 5mm, ribs on the support are 0.5mm, inverted stairs at bottom of the support are 1mm:
Try your concept on a small test model, before doing a big model.
-
I think that bridge came out quite well too. A bit of cement-color paint in the grooves, and a bit of stone-like paint overall, and it will look great in the landscape. I have seen far worse bridges in HO-landscapes. :-)
-
You might get more satisfactory print results if you specifically design or redesign it for 3D-printing. I gave the idea a quick and dirty try. These wings are way more stable and should print better, with or without brim. And the eye for the wire is less likely to break off. Also, it has less steep overhangs that could cause trouble. This is a very simple model made by adding and subtracting basic shapes, and then rounding and chamfering a few edges. Created in DesignSpark Mechanical. ... Okay, it looks more like a bomb than a rocket now, but hey. 🙂
Here the STL file, and the DesignSpark Mechanical native RSDOC-file (zipped). Feel free to modify and play around with it. Print as slow, cool and in thin layers as possible.
-
1
-
-
Thanks for that info. Yes, that all makes good sense...
Is that titanium plate smooth and high gloss, or rather sand-blasted? Could you print PLA or some materials without any bonding at all, like on glass?
As for a 2mm version: maybe you could glue a couple of tiny 1mm plates on the titanium plate under the clips, so that you do not have to rework the clips, but you can still use a lightweight 2mm plate (if that would be stable and stiff enough)? In this way you would still have the option to use the original glass.
-
Maybe you could switch off prime towers in the slicer, and make your own in CAD, as part of your design? Then you can make them exactly as you want?
Similar to the way I make my "dummy cooling towers" in CAD, to move the nozzle away from small objects, so they get more cooling time.
The green cube here is a dummy cooling tower, otherwise the top area of the yellow part would not print well, it would deform because the heat can not escape. These models are quite small: for reference, text caps height is only 3.5mm, and text legs are 0.5mm wide. The dummy cooling tower has a custom brim, also designed in CAD, and a hollow area at the bottom where no cooling is needed.
-
For long-term stability, I prefer to design the mesh into the model in CAD. Future slicer versions, or different slicer brands, might treat infill patterns differently. Or they might give the infill a lesser density than the walls. If the mesh is designed in CAD as part of the model, it is more stable.
Usually I make a solid block as the base first, and then create one hole into that block. And then I create the mesh pattern by multiplying that hole horizontally and vertically as many times as I need it, using a function "create pattern" or similar. See the sift below, for use in the sink in my lab. Or you can create the mesh separately in this way, and store it as a separate file. And then later "union" it with other designs. Similar like you would store your logos and watermarks as separate files, and union them into your designs later on.
Chose width of the ribs so that they are printed well. Not too thin. I often make ribs and legs of text 0.5mm for a 0.4mm nozzle. Because in corners the smooth bendings in CAD are replaced by straight lines in STL. If rib-widths in a curve would be exactly 0.4mm, then after conversion to STL, it would variate between maybe 0.39mm and 0.41mm due to the straight line segments. The lines of 0.39mm might be unprintable, and be left out. Or they might require tricks to print well.
Photo: sift for lab sink.
-
Out of curiosity: is there a specific reason why you have chosen titanium? Instead of ceramics, steel, aluminum, filled composites, or other materials? And what did it cost, as titanium is not the cheapest?
-
I usually set top- and bottom thickness equal to wall thickness. For a 0.4mm nozzle, wall thickness often is 0.8mm or 1.2mm. So maybe try that first, and adjust if required? I have no idea what value it uses when you fill-in nothing.
-
On a standard UM2, the first layer is printed at 20mm/s, while the standard speed for the rest is 50mm/s. Speed gradually increases (at least it seems so) from this 20mm/s to 50mm/s during the first couple of layers. But at 20mm/s the material is sitting 2.5x longer in the nozzle than at 50mm/s, so it is much more liquid. And the pressure on the feeder wheel is less, making that feeder slip less onto the filament. The square pattern on the feeder wheel translates into a *diamond* pattern in the filament due to this "partial slipping". The square holes are being stretched out, which gives a lower flow than set if there would be no partial slipping. So, at lower printing speeds = more time to melt = less pressure = less slipping = more material fed = higher flowrate = better filled layer. That, combined with a nozzle that is often set a little bit too close to the bed (for a better adhesion and smoother bottom), gives more dense first layers. And less dense filled upper layers.
If the first layer is overextruded, due to this lower speed, or being too close to the bed, it contains too much material. That overfilled-effect shows in the next layers, they get a bit overfilled too.
These things could cause the first layers to be different from the body of the model. At least as far as I have seen.
I don't know how all this translates to pellet printers, which I guess work with a sort of conical worm screw? But I have no idea how to calibrate such a thing. I think you should first get the *higher* layers good, thus the body of the model, until these higher layers get a smooth and closed surface, without underextrusion. And then go to adjust the settings for the first couple of layers. That is how I would approach it, probably.
Anyway, consider all this as "educated guessing", at best, not as truths. It's hard to give good advice on something you don't know... :-)
-
Printing multiple models improves cooling, but increases the risk of the curled-up parts being hit hard when the nozzles comes back from the other object. It depends on the geometry of the models.
This testmodel was supposed to become a bridge, or table. The inverted triangles are only for support, and need to be removed after printing.
It got knocked over due to the edges curling up quite a lot, about 1.5mm, and the nozzle hitting really hard:
Close-up:
Now with redesigned supports, and a bigger footplate.
Inverted stairs gave less curling-up than inverted triangles, in my tests. This was just a test model to try-out the concept of free hanging supports, instead of traditional supports going all the way down to the bottom.
-
These are in transparent PET (which becomes translucent after printing, due to the entrapped air in-between the sausages, breaking the light). I believe transparent PLA should even work better than PET as it flows better, but I don't have that... If you need a higher temp-resistance than PLA, then PET or CPE would be best, I think.
Translucent materials like colorFabb "natural" PLA/PHA also work with bigger watermarks, but not with such fine text as this. For fine details, you need a really transparent material, and you need to print it slow, cool and in thin layers for maximum effect. Make small test models to find the best settings for your model.
-
Such steep overhangs curling up is normal. In my experience, printing cool and in thick layers helps a little bit, but does not eliminate it. More than half of the extruded sausage is printed in the air and has no support to stick to. So it sags when extruded, and then it curls up when cooling...
-
I am just guessing here, so...
The wavy aspect looks like the previous lines shining through. Thus: smooth previous lines will show through less than lines with big gaps.
Have you checked if layer-height, speed, temp and cooling fans are the same everywhere, on all layers? Often they differ for the first layer(s). If they are all the same, I would guess that the bed is too close. At least if it would be a standard 3D-printer; hard to say about a custom one. I don't know if effects of pellets not molten enough might play?
Long ago we had a plastic company in our neighbourhood: they extruded pellets into lollypop straws. Each time they started up the machine, after warming up, they had to extrude and waste some material, before the flow was steady enough for production. I could imagine that you have similar startup effects at the beginning of each print: different temperature and viscosity of the melt? If you can, try manually purging some material, and then immediately start the print? It could help diagnose things?
-
Just a question: do the flow channels have a special irregular shape? Or are it just straight channels? If straight channels, maybe another option could be to print them in a very strong material like nylon, cast the gel around it, and then pull-out the nylon strings? But this obviously won't work with irregular channel shapes...
The problem with dissolving PVA could be that you need a good flow in the channels, to wash away the dissolved material, and to apply fresh water. I don't know if such a flow is possible, it may depend on the dimensions?
-
This is the maximum I ever got. From my tests: print as slow as possible, as cool as possible, and in as thin layers as possible. The best here is 0.06mm layers, 10mm/s speed, and around 200...210°C. This is transparent PET (brand no longer available). Top row 50mm/s, bottom row 10mm/s; layer-height from left to right (mm): 0.4; 0.3; 0.2; 0.1; 0.06. Blocks are (in mm): 20 x 10 x 10, with my hollow watermark sitting halfway.
Do similar small tests to dial-in optimal settings for your system and material.
Sanding, polishing, or chemically smoothing with dichloromethane may be required to reduce layer lines. Otherwise they act as separate little lenses, destroying the clarity.
Another option might be to make a mould, polish that, and cast some clear epoxy in it. This will give far better results. Or print an "original" in non-transparent material, polish that very well, make a silicone mould around it, and use that to cast the epoxy. If you need the light shining through, like in car rear lights, 3D-printing might be good enough. If you need a clear lens and a beam of light, like in car headlights, you will need to cast.
The testsamples:
The best: 0.06mm, 10mm/s, 200...210°C:
Left one is the best as printed, right one is after some sanding and polishing, so the outer layer-lines got removed. Inner artifacts caused by entrapped air are obviously not removed. Text is 3.5mm high, text-legs are 0.5mm.
One part is as printed, the other chemically smoothed with dichloromethane. This is PET too:
-
Hoi Sander,
I would be interested, but I have a few questions:
- Can this PETG be smoothed with dichloromethane too? How well does that work?
- How transparent can you get parts, with which optimised settings?
- How is UV-resistance, when sitting outside in the sunlight?
- Over time, does printing with this PETG accumulate dirt in the printer-nozzle? The brand of PET that I use now, does leave a sort of glossy varnish coating in the nozzle, which is hard to remove. Contrary to PLA that rather leaves powder-like ashes that easily come off.
- When parts fracture at overload, how does that go? Following layer lines, or going diagonally through all lines; and with or without warning (whitening, microcracks, permanent deformation)? The brand I have now, fractures without any warning, no whitening, no permanent deformation prior to the failure.
Could you post a few photos of these aspects? See these pics below for comparison.
Transparency tests with another brand of PET: top row printed at 50mm/s, bottom row at 10mm/s; layer thickness from left to right (mm): 0.4; 0.3; 0.2; 0.1; 0.06. At 10mm/s and thin layers, browning begins to occur, due to sitting long in the nozzle, even at reduced temperature. These blocks are 20mm x 10mm x 10mm, with my hollow watermark sitting halfway.
Other PET-brand: snaps diagonally through layers at overload. See the radial pattern. No warning, no whitening, no deformation prior to failure.
Other PET-brand: one item smoothed with dichloromethane, the other is as-printed.
-
Do *not* use SketchUp: this might seem to work well in the editor, but it will be an endless nightmare, as soon as it comes to printing, because it always creates defective models. Big no-no.
I use DesignSpark Mechanical: this is freeware, only requiring registration. Good for geometric shapes like machine parts. It is easy to learn, and there are lots of good tutorial videos on Youtube. Not suitable for organic shapes.
Other people prefer other programs.
Search on Youtube for 3D CAD software tutorials, and try a program of which the workflow appeals to you. Expect some learning curve: 3D-software is complex. But once you can create your own things, it is very rewarding.
-
Indeed, PLA changes crystal structure into more crystaline as it ages, becoming more brittle; this is reversible. Also, it degenerates due to water-absorption, this is not reversible.
I also found that if a string of filament is stretched for a longer time, like in the bowden tube, micro-cracks tend to form. This also makes it brittle. Probably this is what you have?
Also, some colors and brands are worse than others.
So, I remove the roll after printing, and store it in a plastic box with bags of desiccant.
These bags can be re-used by heating them. They are sold in car-accessory shops, used for drying cars and campers in winter. The blue dot turns pink when wet.
Micro-cracks in filament:
Idem, seen through a microscope:
-
Yes, you can glue PET parts with dichloromethane, on the condition that they are touching each other. Dichloromethane dissolves the outer layers and makes them bond. But since it is a very thin liquid, it has no "filler" properties, contrary to epoxy glues. So the parts have to mate almost perfectly. I mate them first, and then let dichloromethane drip and seep inbetween both surfaces, using a brush. Bonding strength is comparable to contact glues: not very strong, by far not like composites, but in most cases good enough. Smoothing/bonding also helps to seal tiny holes and makes it water-tight.
Concerning licenses: in Belgium it can be bought without special license, contrary to chloroforme. At least for companies and universities, I am not sure about personal buying. Ask a chemical company.
Use it outside, or in very well ventilated rooms with fume extraction (I have a fume extraction cabinet in my lab). It is not healthy. And wear eye-protection: it whitens the skin just like acetone, so we don't want it in our eyes.
This part is smoothed and bonded with dichloromethane. You see how both colors blend around the bonding surface. This is PLA, but for PET it works the same, I just don't have photos of it.
-
1
-
-
Mine are both: Sandisk, SDHC, 4GB, speed class 4. Any speed-class should do, as they are all faster than the printer anyway.
With most of my gcode-files being 2 to 5MB, it could contain about 1000 files.
-
If I had to make such a thing, I would print a mould in PLA, or in a dedicated mould-making filament. And then cast a hard and pressure-resistant material in it, some sort of sand- or metal-filled epoxy or so? Be sure to use huge amounts of release-spray, or it will glue like hell to the mould.
Or 3D-print an original model in PLA, make a silicone mould around it, and then a stiffer harness. And cast the epoxy in the silicone mould. Saturate the mould with silicon oil prior to casting, to extend its life.
In this way, you can make the mould from multiple pieces, glued together. But the cast is a one-piece, without seams and glue. So you have the good things of both 3D-printing (self-made custom models) and casting (strong, single piece, no weak points due to cavities, layer-lines, seams, glue,...).
Of course, all traditional mould-making tips and tricks apply: no undercuts (unless you cast very flexible rubbers), select "natural" seam lines in the model, make sure it can be filled, predict the flow of the casting material, no dead ends where the material can not get into, provide pouring and venting holes, alignment-keys, clamping features to keep both mould-halves together, provide indents where you can insert a compressed-air pistol and screw-drivers to pry apart the moulds after casting, etc...
A silicone mould can handle heat up to 200°C usually. A PLA mould not... So use low-exotherm, slow curing epoxies in that case.
Best UPS for UM3 ext.
in UltiMaker 3D printers
Posted
It is 20 years ago since I last selected an UPS. A few things I vaguely remember:
- Find the continuous power it has to be able to deliver. See the name-plate on your machine. Multiply that value by 2x, to account for variations, little add-ons like a LED spot, and system-degradation.
- Find the peak-current it has to deliver. Some devices may draw a fairly low continuous average power, but with occasional high peaks. Multiply that peak by 2x, just to be safe.
- Chose the time the UPS has to hold up these values.
- Look up the dimensions and the prices of UPSses. And then fall on your back from astonishment. And reduce your up-time to more realistic and affordable expectations.
Most power-outages do not last longer than a few seconds to minutes. If it lasts much longer, then a failed print probably isn't going to be the worst side-effect you are going to have. So I would go for maybe 20-30 minutes up-time. After a couple of years the batteries will degrade, and you will still have 10-15 minutes.
If you expect frequent power losses for a much longer time, like in California whenever there are high winds, storms, thunders, draughts, freezes, wildfires, no-winds (=no wind power), etc..., then you'd better get a diesel generator too to power your house. The UPS then is to get over the first minutes so the print doesn't abort, and then the diesel takes over. Use non-bio diesel or fuel, because the bio-fuels degrade fast due to algae and bacteria growth, and they lubricate less and are chemically more agressive, shortening the engine-life.
An UPS - even a small one - is also good in case of unstable power, or overvoltages, as it regulates and protects.