scottmg

Yes please! Is there any way at all to see what the fan speed will be exactly, other than looking at the gcode?

I have a theory that thinner layers and thinner extrusion leads to more warping. The basic thought I have in my head is that if you're printing thin layers, near or below the nozzle width, then you're smearing and stretching the filament, leading to higher stress and potential warping. It's easy enough to find a print that's prone to warping, print with different settings and then eyeball the result. But I'm wondering if anyone has taken a truly scientific approach with this and has good numbers? Maybe even measured stress with some fancy hardware? What do you think? Agree or disagree? Did I miss something obvious or something counter-intuitive? Know of any existing hard data?

Here's a part I threw in as a test. The test.failed.3mf fails every time I try. I turn on adaptive layers and it slices in a second. Turn adaptive layers back off, and it never completes. No CPU activity, just sits there (see image below). Adaptive layers isn't the key, lots of settings will break it, lots fix it; I don't see the pattern yet. I cleared out the log, tried the one that fails, attached the project and log. Then I closed Cura, cleared the log, reloaded, tried the one that succeeds, and attached that project and log. I still have Cura 4.5 installed alongside 4.8. I see this all the time with lots of parts in 4.8, I have never seen it in 4.5. My machine is Windows 10 Pro. Core i7-6700K 4GHz, 16 GB RAM, about 9.2GB free. I have about 249 GB of free disk space one a single SSD. If you need to know more, let me know. Thanks Scott test.succeeded.3mf cura.succeeded.log cura.failed.log test.failed.3mf

I have Cura 4.5 and 4.8 both installed. After switching to 4.8, this happens to me all the time - there is definitely something wrong with 4.8. It'll be slicing just fine, then I'll change a setting, and then the slicing never completes. I change the setting back, and it slices again. It seems many settings can trigger the break - nozzle width, compensate wall overlaps, adaptive layers, etc. I haven't come up with a common theme yet. When this happens I'll either change settings until it works again, or just go back to 4.5. This is so frequent for me I have a hard time believing Ultimaker doesn't know about it. But if anyone needs a profile and STL to repro the issue, I can get those things.

Yes, you got it. As I mentioned I believe linear advance in Marlin is intended to help, but I've never seen anything that leads me to believe the Ultimaker supports those codes. I've tried turning up the acceleration and jerk settings so that it flies faster through turns, but the effect seems pretty marginal. It might make for larger files, but this seems like it could be something fixed in Cura as well. I guess maybe it's time to look into creating a plugin myself. Thanks for the info. Scott

Imagine printing a 'T' shape. The skin for the current layer is printed along the middle of the T (|||||). So as it's printing there's a short up and down path at the top of the T, and a long up and down path through the main part. When printing the short path, constant back and forth, it's too thick. When printing the long path, it's too thin. I assume this is what linear advance/flow compensation is for. But I've experienced this to some extent with just about every printer and material I've owned. The printers that support linear advance seem to have minor improvements, but going really slow is the only thing that seems to work all the time. So a couple of questions: 1) Does the Ultimaker 3 support any kind of linear advance/flow compensation? 2) Does Cura have any settings to fix this? This seems like an imminently fixable problem. Increase the flow for a long, fast straight away, slow the flow in high acceleration/high jerk areas. I took pictures, but they're terrible and dark. If my description is unclear, I can do another print specifically to demonstrate the problem. Thanks for any input, Scott

I experimented with lots of materials. A good chunk was Onyx (beautiful stuff, not as stiff as I'd like) printed on another printer. The tension arm and base were 3dxtech carbon fiber polycarbonate. For the gear Polymaker CoPA (Nylon 6 + Nylon 6,6). That's a story in an of itself. Making a proper involute gear is a pain, https://geargenerator.com/ was very helpful. I tried, repeatedly, to get a good one out of acetal. I'm pretty sure that stuff can warp the fabric of spacetime. I even have an old Makerbot converted with a heated chamber that will go up to about 75c, and still just couldn't get the results I wanted. Maybe some day I'll figure it out. The Nylon is pretty slick, added in some PTFE SuperLube and it seems to be doing well. I only have 10 hours or so of prints on the current setup, but I don't see any signs of wear on either gear yet.

Here's a test print with the custom feeder that finished a bit ago. The photo isn't great, the Carbon Fiber PETG (3dxtech) is doing funny things with the light. But I'm very pleased, it looks great. Even the threads inside look good. Maybe a bit over-extruded, but that's easily fixed.

https://www.thingiverse.com/thing:4235488 The most expensive parts are the BondTech gears ($27 USD), bearings ($12), and push-to-fit connectors ($16 for a pack of 10). I listed my source for the hardware, bulk (packs of 25-100 pieces) stainless hardware, and that's pretty expensive if you go that route. If you can go to a local hardware store, you could probably finish it up for around $10, so you're looking at $65 USD. So unless you're a hobbyist, the reality is the bondtech ones are probably better for a little bit more money. But if anyone is still interested, its out there. Feel free to make suggestions. This was built based on parts I had on hand, moreso than ideal parts or cheap parts. There is definitely much room for improvement. I wasn't even thinking about showing anyone else until I ran across this thread, and made the original post kind of a whim. The main things I like: 1) The oversized tension arm is way easier to push than the little sliders on the side of the original (and the bondtech looks to be similar) 2) You don't have to update firmware. I'm pretty sure I saw that in the bondtech installation instructions, which is what initially put me off of it and starting me thinking about building my own. 3) So far most profiles have been good. It was difficult to match the feedrate exactly of the original (bondtech gears are a different diameter than the original knurled drive, I had to calculate a new gear ratio, and the ideal was between 39 and 40 teeth, went with 39). Maybe some over extrusion on some, but not much. 4) It was a fun project!

I'm sure I'm not the first to do this, but I designed and created my own feeder for the UM3 that's a bolt-on replacement for the existing feeder. So far it's worked great. It uses Bondtech gears (the whole purpose was so I could print abrasive filament without worry). If anyone is interested, I can clean a few things up and post .stl files and the source designs (Sketchup 8 (not exactly modern, but free)). You'll have to buy some of your own hardware (gears, a few bolts, a couple bearings, push-to-connect fitting). And obviously it would be an unsupported, DIY, at-your-own-risk kind of thing. The nice thing is, if you don't like it, just bolt the old feeder back on. I suspect most with a UM3 will want a professional, supported solution, but figured I'd throw this out there.