kfsone

I've just gotten done recovering with leveling issues on my UM3, my core findings being this: - Glue, even fingerprints on the bed, can affect the leveling results, - There are little soft-material covers by the print head. They're not solid, and material can get pushed up behind them. Take your printcores out and give them a thorough inspection, as this can throw leveling, - When I added a buildtak flex bed, I didn't realize just how much it would throw leveling, or that it would render auto-leveling useless, - Manual level OR auto level, don't expect both to work The auto leveling seems to have certain limitations that go out of whack the instant you touch any of the screws. There also seem to be several bugs in the manual leveling process. In order to successfully level my printer, I totally ignored the display and did my own thing, because so many of the prompts were unhelpful (and the lack of a "go back" or "cancel" option is totally frigging annoying with the U3s oops-I-dialed-when-you-clicked dial). I would recommend you do a fresh manual level: Start by turning all of the screws to make sure they don't feel super loose - don't over-tighten them either. I changed mine from being so lose I could brush one finger tip past and turn them to where I don't need to grip them but I need a fingertip on both sides for them to turn. Next, run the manual level and for the "1mm away" step, use something like a credit card. I used an old one and made sure that the raised print fit under the nozzle. No accuracy is really required for this step but I found it helped if all 3 checks were about the same height. For the calibration card check: my leveling had resulted in the nozzle being so low that nothing came out. So, I replaced the calibration card with a thin sheet of card, about double+ the thickness of the calibration card. This had several advantages, not least being that its flat without requiring any pressure which might invalidate the calibration. When you do this, you're not looking to feel like the card is being caught on the nozzle. You should be sweeping the card/sheet back and forth and then adjusting until you can *just* barely tell that there might be contact with the card. I found this so hard to judge that what I actually did was reach the original point I considered "a little resistance" and then give the screw a 1/16th turn in the opposite direction, which put me at a place that ever-so barely felt a little different than moving the card away from the nozzle. Then I did a fairly simple test print of a single-layer circular face with a rectangular cut-out in the middle (15mm diameter, 8x5 rectangle). I still ultimately wound up making a couple tiny screw adjustments, but I also discovered that running an autolevel after this made things worse again, so I repeated the manual level and I've been getting great prints again. I'm about to apply GR5's tip on setting "Combing Mode" to "not in skin" in the hopes I stop getting horrible, messed up first/top layers.

Bump with thanks to @gr5 -- great info. Thinking about how my printer routinely drags the prime blob into the print (it looks like it tries to wipe the nozzle but the nozzle is still too hot) and wanted a repl to experiment with - this seems like it might be close enough. You mentioned a web service, and I guess it'd be easy enough to flask something, but is there any real interest in poking at this kind of level by other people? If so, maybe I'll look at making time to write a Jupyter kernel (https://github.com/jupyter/jupyter/wiki/Jupyter-kernels)

Unleveled bed leads to two extremes: nothing leaves the nozzle, or nothing sticks to the plate. After levelling, how do you know which way you've gone? Well, you print something. So give about 1 minute for the leveling process, the time an autolevel takes plus the time it takes to print something simple, and you end up wasting a lot of time. So cut it short: Incorporate a simple option to print a test square with diagonals from the tail end of manual levelling. Thus flow becomes: Manual Levelling > OK > Back position > Front Right > Front Left > Z-Offset > Auto-Level? > Test Print? > Repeat? It'd be really helpful to have the print try and print a couple lines, at least, after levelling. I'd personally be inclined to ask the user questions that would help me adjust the offsets.

Bought an UM3 couple months ago, and the transition from higher-end hobby printers to a printer I trust ... just amazing. Earlier this week a print with colofabb nGen went wrong long enough to goop up both print heads with nGen and require me to remove them and clean them off. When I put them back, the bed was out of level and autolevel wasn't helping, so I did the manual level and tried to do the XY calibration but either the print head would be pressed so tightly against the bed that nothing came out or it just wouldn't stick. Using the original UM green, did a cura connect reset, wiped my cura profiles, no help. So I turned all of the levelling knobs a 1/4-1/8th counter-clockwise and now I get prints but the first layer is bumpy and I get elephant's foot and curling, but mostly, part way thru the first player, the print head crosses and catches on something and the print starts to come undone (yes, I have combing and avoidance enabled, I reset to defaults) I'm wondering if the honeymoon period is over now, and if from here on it will only ever print as well as lower-end/hobby printers I'd had before, or if there is a guide to restoring that factory fresh reliability?

I just threw all my money at the screen, did any of it arrive? 🙂

I've never found any way to get gradual support to build viable support, it never graduates between the original "lazy" support and the next layer up: Might produce something like this: I expected that the individual lines of support would be further apart, but instead it just seemed to produce half the support, leaving a whole half of each subsequent tier floating in thin air: (Yes, I've tried different patterns, concentric is where it is easiest to see when you are controlling the camera view) Anyway, this suggests to me I'm doing it wrong?

UM3 bed got out of level and I wound up making settings tweaks that didn't help, and now I'm trying to back them out, but as I scrutinize how the build gets printed, some of it is counter to my *novice* intuition. If you watch the first layer, it starts laying down the wall around the "window" panels and the speeds are kinda weird: then it makes an odd choice about where to start the next panel, resulting in a bunch of extra travel that could have been eliminated if it started at a different point: the order in which it does these next pieces is a bit puzzling too - in that it works left, bottom middle, right, center, top right, left, for all of which it only does half the lines. Then it does the outer perimeter in full before going back and finishing the window-walls; then it fills the rest of the face: and even then, it goes back around and puts more lines directly around each wall. The other issue I'm having is that there are weird patterns on the bottom surface because of the way it tries to fill the base resulting in a surface that (in a kind of cool way) has a panelized look, like an 80s starship surface heh. And whether I use lines or zigzags, the way in which seams form between the contiguously printed panels and resulting travel significantly affects the smoothness and rigidness of the bottom surface: the marked lines show up as clear seams/distinctive "panels" on the bottom surface: Is this where perhaps I should be using concentric? I gave that a shot and it had the same outlining wastage and then it did this: It seems a lot like Cura could have made things easier on itself if it had realized that it's dealing with a round/symmetric design? The resulting seams are actually kinda neat: But they still somewhat marr the finished surface which - went it adheres properly - is a beautifully flat, smooth surface. I've attached the print and the settings I'm using, I'd love any input on what I'm doing wrong and how I can produce a better print? -Oliver Insert image from URL spool.3mf spool.curaprofile

Printing any kind of accurate detail depends so desperately on line-width and layer-height that most 3d modelling is doomed to failure. A 3d friction join will test print beautifully, but then when you print the full model, the raft shifts the positions of the joints so they're not multiples of 0.35 and none of the friction joins work. So surely there has to be some software out there that models in ratios of lines and layers?

When Cura prints a prime blob or performs a nozzle wipe, it always seems to do it *immediately* before moving the print head towards the model. Wouldn't it make more sense to move it away? I've found that things that need the prime tend not to adhere so well, and the printhead routinely drags the prime blob if not into the part, somewhere that it's going to wind up interfering with the part. The same thing with skirt: I've watched it lay down a beautiful line of skirt that adheres perfectly, but then it travels straight to or across the part footprint with a tiny little dangle of ooze that falls off in the least convenient part of the adhesion layer, or it turns out to be the one place in the print that it's going to string. Lastly: is it possible to move the print head outside the bed? I just wonder how many prints would go better if the the head couldn't traverse one of the corners of the bed, with and then lower to { first-layer z - first-layer-width/3 } at the very edge of the bed moving - slowly - to first-layer z as it crosses up to the bed... If anything is hanging off the head, this should knock it off, and the trip along the side and round the corner should help to drag any ooze or string off the plate? (Yeah, it sounds too easy, so I suspect it's not or you'd have done it 🙂

Had my U3 for about 3 weeks now. PVA had been working fantastic, then I spent a few days tinkering with ABS before coming back to PLA+PVA. Now I just can't get PVA to print: - During priming, the PVA pops/snaps like bubbles bursting, - When it does get adhesion, it's stringy, but not the gossamer-stringy I'd expect from excessive temp, and it happens at lower PVA temps, the string gets in the way, and the print is ruined, - Prime blob is huge, gets burned at the top, and doesn't get any adhesion, so the print head drags it off into the model, smears it around a bit; when it's finished the layer it retracts and travels smoothly and cleanly to the prime tower, does a prime, and drags a string back to the model, which invariably catches on printed part, - Without raft and a prime tower, I generally get no adhesion, - If I print PLA raft, the PVA won't adhere to it at any temp, - If I print PVA raft, the PLA sticks to it, put then any PVA printed above the PLA is really inconsistent - bits of it are lumpy and brittle like the head was way too cold, while other bits are silk-stringy and almost magnetically attracted to the nearest PLA wall where all the PVA that can get to forms a nice little blob, This beastie is a prime tower from a print today (the model didn't fare nearly so well): - Yesterday the PVA got kinda ground up in the tube, and was making a nasty, loud, cracking sound like someone snapping the a line of filament, and sure enough it stopped and looked like it had been chewed up in places, Is this normal? I have a shield around the printer to try and prevent temp fluctuations, but maybe I need to put a proper enclosure around it? Is it moisture in the PVA? Is it the print head? Have I lost bed calibration 0 I've seen similar wiry-string when I had a model that had a part that was 1 layer off the bed and I wasn't printing support: PLA seems to print/adhere fine, it just seems to be my PVA that's having this problem now tho.

The attached model (it's actually a sort of lazy-susan for jewelry boxes) has been throwing me a lot of interesting curve-balls, it was mean't to be difficult 🙂 Maybe not quite this difficult... The one I can't seem to fix: unsupported supports that invariably droop and take down parts of the model. I can fix any given specific case by switching support pattern, and while most of them are between layers 195-198, no one pattern fixes them all. What's worse is that one that looks like a dangling end? That's actually the *start* of a line. Concentric produces this totally unconnected/unsupported circle; I hoped it would just fall in, but both times I tried printing it drooped into the next support in while still hot enough to buckle enough to ruin the top of the print. Layers 196 and 197 in Grid: (emphasis on the disjoint ones) "Cross" at 40% almost pulls it off, until this happens when the sheath-column tapers out: Those inner "wings" are only a couple layers from being supported, but it's just enough that they say and buckle the top 🙂 ZigZag gets really, really, really close on some percentages, but there's always something, and sometimes its quite drastic: Given most of these happen at the 196-197 transition, I get that it's most likely a problem with the model - but what? 🙂 Cura: 4.0.0-BETA Ultimaker 3, PLA (Green), PVA (Natural), vanilla "Normal" config with support, w/ and w/out adaptive layers, prime tower, raft, no support interface, various support densities: support on extruder 2 (pva) and raft on either extruder. Mount.3mf

That was what I suspected, which means you also need to design the part accordingly. That seems like something that might be worth focusing on once I get to plugin development 🙂

Does Cura make any effort to organize sections of infill? It often seems to make bad "choices" about the alignment of the infill to the model itself, as in this example: If the infill was one line to the right, it would be providing balanced support. Instead, this pattern is going to provide uneven support.

Spent an hour trying to figure out what one of the UI buttons did, but nobody else seemed to have it, before it dawned on me it was probably an extension. Would be helpful if: - There were tooltips over the UI buttons in the left-toolbar, - Extension UI components were visual distinguishable from built-in - plus, if it's possible for an extension to imitate a built-in, it will eventually be abused

Just got a brand new UM3 from MatterHackers, and the camera wasn't accessible, just shows up as a broken link. If I copy the "image address" from the broken-link icon to the address bar, it's 127.0.0.1... Change the IP to the printer's IP, and voila: camera. Have you considered just running haproxy on the device so you can serve everything on the same port under different folders?

Just got a new UM3 from Matter Hackers, had the same issue, but the problem is that you are linking to the stream at 127. 0.0. 1 instead of the UM's ip In the attached images, I right click on the broken-image icon and select "Copy image address", then paste it into the address bar, and it's *127.0.0.1:8080". If I replace the localhost ip with the 192 address, l can see the camera stream...