donmilne

I've been printing with relative success for the last week or two, but last night I seem to have had some kind of disaster. I was giving KISSlicer a first test print, printing a kind of cylinder (a motor mount / noseplate). I went off to get something to eat, when I came back the PLA print had detached from the plate and stuck to the nozzle, which was waving it around. It was about 1mm thick by that point, but a lot was caked around the nozzle. I pulled all that off the nozzle while still hot and then aborted the print. (The KISSlicer UM1 profile I tried didn't seem to heat the bed, hence the print came loose). I was going to upload pictures, but if there's any way for mere mortals to upload pictures on this site then the method escapes me. I can view my own (empty) gallery, but damned if I can see any way to add to it. After the KISSlicer failure I decided to revert to Cura. It started off ok, and again I went off to do something else. When I came back about an hour later, the UM2 was "air printing". Head moving around, bed had decended about 20mm, no filament coming out. The feeder was a bit dusty but not excessively so. It had printed the first 5mm of the fairly dense base, so it filament flow hadn't stopped immediately. Quite strange. I aborted the print of course. Next day (today) I tried to change the filament. The head heated up, the feeder motor went into reverse, but the filament didn't budge, not even with my assistance. I then tried following the troubleshooting advice in the manual, i.e. printing at 260C for 10 minutes. Nothing came out during the print, and I still couldn't change the material after. Does anyone recognize this failure, and tell me what the cure is? Telling me how to upload pics would be useful too.

I downloaded FreeCAD and attempted to open an STL. It locked up. I then tried it on a nice simple DXF that my own software has no problem with. FreeCAD locked up. I thought about creating new models from scratch but found the user interface completely opaque. Very bad first impression, I'll probably never look at it again. Definitely not ready for primetime. So far, I love OpenSCAD, and especially I love the fact that I can tweak the dimension on anything in seconds. It's a pity that "Thingiverse" objects are mostly in "compiled" STL rather than in a tweakable source form a la OpenSCAD. Starting with zero previous 3D user experience I unboxed a 3D printer and created my own mid-complexity 3D object and printed it - all in the same day. I'd say that speaks for itself - though I admit I am a software engineer and hence am not intimidated by programming languages (unlike 3D graphical interfaces!). If only OpenSCAD didn't grind to a halt as the model starts to get even slightly complex, I'd be a very happy bunny. I'm seriously tempted to write my own near-clone from scratch.

I appreciate that this is not the place to expect support for Simplify3D, I mentioned it only for information. My question really was whether this outcome was normal for the UM2 printing something as thin as the top of a 2mm curved wall. My expectation was that it might be, because I assumed that the extrusion width was fixed, and the placement precision finite, so maybe the nozzle just couldn't get in there - and you're telling me this supposition was wrong. I was thinking somebody would tell me "yes, that's because the bed is too close to the nozzle, or not close enough - and here's how to fix it". Something like that. Simplify3D does have an animated gcode preview feature, if that's what you mean, but it would take a long time to get to the top of the print... and I'm not sure how accurately it would simulate the UM2 anyway.

Can the UM2 vary the extrusion width? I had assumed that was fixed by the nozzle - which I believe is 0.4mm though I have no practical way to measure it.

Quite so. Ah, if only I didn't live in the north of Scotland, and it wasn't the middle of winter. Are you having a really sunny winter there in Sweden?

Let's call it a cylinder - a tube. It has a inside, outside, top and bottom. I expect this to printed as a solid shell x mm thick with the core filled with whatever is configured. Instead, I find that with the tube set to 2mm wall thickness at the top of the tube I'm seeing an inner and outer wall and a narrow gap in between. I'm not sure that it has an explicit top surface at all. I assume it's a problem of the narrowness of the gap once you subtract twice the shell thickness - it looks like it could be close to 0.4mm.

You might also want to look at OpenSCAD. If you're the right brain type who prefers words to pictures then you'll get on very well with OpenSCAD's scripted approach.

Thanks for telling me how great cell phones are these days, I'd never have known. Unfortunately I don't own the generic mobile phone market, I own a specific Nokia phone, and the camera on my phone is not very good in poor light. Anti-shake doesn't work very well on poor contrast images either. Perhaps after I get myself a nice lighting rig I'll start posting photos around here.

Thanks for the tip - I'll give that a try.

The only camera I have handy is the one in my phone, and I'm not sure a fuzzy picture in poor lighting would really be all that useful to see the fine detail I'm talking about. I'm happy to share the model, but it's really nothing special: a simulation of a piece of 40mm diam 5mm wall thickness tubing that's been turned down to smaller diameters here and there, including at both ends where it's "turned" down to 2mm. The original the model is copied from really was created on a lathe.

Ok, second print completed. This time I waited until the printer told me it had cooled down. I'm seeing some problems with the print quality however. It's basically a cylinder that's various diameters and wall thicknesses along its length. At the end of the tube it's down to a wall thickness of 2mm. This prints as two walls with a narrow gap in between. Is that normal? Also, there's a kind of shelf halfway down the cylinder. Instead of being solid I see the fill pattern with hairline cracks between each run. Can this be fixed? I should mention that I used Simplify3D to slice this model because I was concerned by reports of poor support structures with Cura (that internal shelf again).

Possibly I jumped the gun on the print cooling. Once it got down to around 32C (I can't now remember which temp that was) I reckoned that I wasn't going to burn myself on that, so I took the glass plate off. Hey, I was keen to get a close look at my first print. When running the robot I'm not sure what temp it used - I'm still learning what the display is telling me. For my own model (same UM PLA) I chose 230C. That being around the middle of the recommended range on the PLA packaging. What are the downsides of being at the higher end of the temp range? Object not cooling quickly enough?

I started doing something similar, though using $fn only. $fn~=40 gives me something that resembles the default. $fn=100 gives nice results for what I've done so far, however it does have to be tuned for the object diameter. Even so, even at teh lowest quality OpenSCAD is pig slow once you have several diameters and holes in play.

It seems to have snapped quite cleanly along a single layer, where the robots feet (pedestal?) meets the legs. There's no obvious crumbliness above or below it. I think I just pulled too hard and it snapped along a weak plane. Is there a trick to separating from a glue surface without using too much force? Run it under warm water perhaps? Before I even got my printer I'd heard that PLA was quite hard - and therefore brittle. As opposed to ABS which is a bit softer and a lot tougher. Is this not fair comment? The flexibility to use ABS was one reason why I went for Ultimaker.

After a couple of false starts with my Ultimaker 2, I finally seem to be making some progress. I started unboxing around midday, it's now 18:40pm. First setback: I followed the first start wizard up to the point at which the feeder was supposed to pull in filament - but it just got stuck and the knurled wheel ground out a divot from the PLA. Strange thing was that this seemed to be same blue PLA filament that the test print was done in. I could see no way to advance or remove the filament, it seemed to be stuck tight. Ok, there was four prominent screws on the feeder assembly, so I took it apart to undo the jam manually... and several internal components plonked onto my carpet (and another component on the inside, but I didn't know that yet). Ok, with reference to photographs on this site I worked out how to put the feeder back together... then when I tried to put the screws back in they just spun in the holes... of course I had no idea that the same screws hold the motor on in the inside. So, remove the white cover. Now I need one hand to hold the motor, one hand to hold the feeder halves together and in place, and a third hand to manipulate and screw in the screws. Oops! That's more hands than I have. I solved the problem by taping the feeder halves together and somehow made two hands work for the rest. Finally! The feeder mechanism now seemed to be working - the filament shot along the tube... and five minutes later nothing seemed to be happening. Based on a message here about the filament sometimes getting caught up on it's way to the nozzle, I withdrew the filament again (after all that), trimmed a pointier end, and tried again, holding my breath. It worked, yay! 40 minutes later I had printed out the robot example, which unfortunately snapped as I was trying to get it off the plate. I've heard that PLA is brittle. Would ABS be less likely to snap? Right now the UM2 is printing a model I made from scratch using OpenSCAD. It seems to be moving along nicely. I think that feeder mechanism needs a redesign. IMHO the motor should have it's own mounting on the inside, allowing the feeder to be removed for maintenance. Also the two halves of the feeder should screw or click together somehow so that it holds itself together while I mount it in place. Also I think the first start wizard should be divided into two seperate steps: levelling of the base, and first time material loading. As it is at present, when one of these steps went wrong I saw no way to backtrack, the only option I saw was to cut the power and start again from the beginning. Don't get me wrong, I'm still enthusiastic, but I'm hoping you guys are open to fixing these teething problems in a new product.

I'd say it's still worth your while looking at it, because it's such an effective way to create 3D engineering models. With no prior experience I was able to start creating a complex model within minutes, and without being hamstrung by a poorly designed GUI. If you don't need lots of nice curved surfaces and hole then perhaps you won't need to worry about the speed constaints. Speaking of which, I assume it's a combinatorial problem (searching for possible intersections?), and I think they just chose a particularly slow computational graphics library, CGAL. Reading up about it, the library seems to be using variable precision data types for coordinates. What's more they claim that they decide after each calculation how much precision is needed to store a result. Sounds clever - and slow. In most calculations I don't need massive precision, I just want a quick rendering of the model so far. The heavy duty stuff could wait until a "publish" step or similar.

Having played with OpenSCAD (generally just perfect for me) and seen how it grinds to a halt when the number of facets is still just a few thousand, I take it back: there's a clear argument for expressing curves directly rather than as a zillion facets!

Yes, I already acknowledged Daid's mention, and I do know who he is. I specifically thanked you for your description of OpenSCAD, as that's what raised it above the level of "yet another app to look at". It sounds ideal for me, though the docs seem a bit lacking so far. My source of Sketchup info is the Wikipedia article, which says that Google sold it to Trimble in 2012. Obviously this puts a big question mark beside it from the pov of someone selecting an app to invest learning time in. [Edit] I found the OpenSCAD online manual linked from the Wikipedia article.

@gr5: Oooh... just noticed your description of OpenSCAD. That sounds very appealing, I'll definitely look at that.

Yes, I realize that smaller facets is how I would get better curves. In fact there's an argument for saying that there's no point in going for anything more complex because the motors have finite precision anyway. Do any other standards handle variations in density inside the model? And does Cura support these? Thanks for the OpenSCAD and DesignSpark_Mech tips. As it happens I'm already signed up with DesignSpark to use their PCB package, so hopefully I can access that package without going through the hasse I had with the latter! :-)

To give you a background - I'm an engineer, not an artistic type, I plan to make only practical objects in my 3D printer. I'm used to making things for myself: I have a hobby workshop of my own, and I own a lathe and a milling machine. I expected to be able to do anything in a 3D model that I could on either of those two machines: for example start with a cylinder, drill a hole and thread it. I've seen mention of Sketchup in my searches. I believe Google has ditched it, and from what you describe it sounds quite limited anyway?

I'm a total noob to 3D printing and I must admit that I had no idea that Cura is only used for printing 3D models, not for creating them. So, having learned my mistake I've been having a look at what 3D model creation software is available for a reasonable price - ideally try for free - that isn't too complicated. (Recommendations welcome). Frankly I'm so far finding that software seems to fall into two categories: either bug ridden garbage, or very flashy but intimidatingly complex. Interestingly, I had a look at the STL format itself and discovered that it's very easy to understand. I'm a C programmer myself and it so happens I did a lot of 3D graphics calculation work back in the 90's, so the purpose of everything in the STL file is very clear to me - I reckon I could "roll my own" STLs far more easily than learning how to get a complex 3D CAD package to do it for me. However, the implications of such a simple 3D format (that Cura seems to use by default) then occurred to me. First, everything is made out of facets - no true curves. Doesn't this make a nonsense of high printer resolutions? Also, the facets will let you distinguish between "inside" and "outside" of an object, but not vary the density (fill factor?) of parts of the inside. This scuppers a plan I had in mind for my printer, which was to make custom enclosures for certain electronics projects, with strengthened lugs where (say) the enclosure is to be bolted down. Of course I can still make the whole thing at the higher density, I guess that's the cost of simplicity. Maybe the tweakatz plugin would be useful for this in some projects too. Anyway, I'm curious: is this simplicity just because STL is old and predates 3D printing, or are other 3D object description languages equally limited?

Good call. I found updated drivers for my chipset on the Intel graphics site, installed them, and everything is fine now. The robot is in yellow, and the "Ultimaker 2" logo is visible in the background. I was a bit wary about replacing OEM drivers with generic Intel ones, but it worked out well (I had an Acronis backup to revert to if it failed). Thanks for your help.

Thanks for the reply. Hmm. I don't know how well tested the OpenGL wrappers are on this machine. I've never noticed a problem before... mind you I haven't used many 3D apps. I'll check if updated drivers are available.

I've just installed Cura 14.01 on an XP SP3 host with a view to learning it before my recently ordered "Ultimaker 2" arrives. When I load the example robot model the background platform is rendered ok (except no "Ultimaker" logo in background), but the robot itself is completely black - all I see is a sillhouette. Surely that's not the way it's meant to be? The manual shows the robot in yellow and has visible detail too (I assume this is the color set in model preferences). I was going to upload a picture, but I don't see a direct picture attachment button here (I don't use a data service so I can't use the URL option). I'm guessing the software is making unknown assumptions about DirectX version or shading support. When I run the dxdiag Direct3D test I pass the test for DirectX 7,8 and 9. DirectX10+ is Vista and later only, I assume that's not required since XP support is claimed. I see nothing in the Cura manual, system requirements section, about what kind of graphics support is needed. I see no troubleshooting section. I don't see anything appropriate in the Cura settings either. Suggestions anyone?