stevegt

What Jean-Baptiste is illustrating here is what I have in mind -- adding a second heat sink and a second fan to the front side of the existing head design. It means having to replace all of the existing aluminum parts, but should solve both the dual-extruder overheating as well as the single-extruder teflon cooling issues Kris discusses.

Hi nallath, can you clarify the license on those files, or find someone who can? The hardware files are released for "non-commercial" use only, using this license (per the LICENSE file on github): http://creativecommons.org/licenses/by-nc/3.0/ This license seems to mean that: Parts for the UM2 will never be able to be legally sold by any third party. If I pay a third-party machine shop to make a part, the machine shop is in violation of the license, right? The sale of derivations, including improvements based on existing parts, also seems to be prohibited by that license. If someone wants to sell an improved part for the UM2, they can't legally do that, right? Am I reading this wrong? Steve

Hi Librarium, What build date and firmware version is your printer running? (Maintenance|Advanced|Version) I ask because we've recently made some bug fixes that could help with this. Is the tiny blob coming out while the nozzle is against the build plate, or is it not priming correctly while the nozzle is still far away? Steve

Daid, Sander, do you think either of you could get a github repository set up for the dual extruder design work, or is that something the community should initiate instead? I know this raises some difficult issues. I think "not yet" is a valid answer, but I also think it's worth talking about the big, smelly elephant stomping around in the room. I think it's not correct to beat ourselves up over past decisions; we all do the best we can with what we know at the time, and try to learn from that experience. For the current situation, a post-event analysis could go something like this: If the UM2 design had been done as an open project, then we would have seen more people working on dual extruder designs from the beginning, that thermal error would have been caught early, and Ultimaker would probably be shipping dual extruders today. Paradoxically, the post-production STEP dump enables verbatim copycats, while the file format and timing both prevent the sort of collaboration which would make the UM2 a better product overall (what does a pull request on a STEP file mean?) This combination gives users and the company the worst attributes of both the open and closed source development processes. A more transparent design process would address these issues, increase confidence, lead to more sales and lower support costs, etc. As a business owner myself, I know IP-related decisions are scary, fraught with irreversible pitfalls, have implications throughout the business, and are impossible to make in ways that satisfy everyone. I also know that some customers will just never be happy, no matter what you do. :wink: The question is, should we try a more open design process now? Designing a subassembly like the dual extruder head would be a good test of the process, with limited risk. The other obvious candidate would be a new feeder mechanism, which is already in play and needs to be touched for the second extruder anyway. There are still problems which would need to be overcome -- for instance, file format would need to be something more upstream than STEP, and would probably collapse to openscad or freecad as a lowest common denominator, which I'm sure will make anyone scream who already paid thousands for something more capable. And so on. Steve

Whups, it was late. Reverted back to G92, now triggering end Z move on M25, pushed to github.

Heh. I just saw that Jemma updated her profile -- she's got a UM1, and I have no idea how compatible it will be with my script -- I have a UM2 and the script assumes UltiGcode. It should be a start, at least. I wanted to tackle this anyway, not for well plates, but for similar situations I've found myself thinking about -- at least I have a good feeling now that it's doable. If I ever have to touch Cura for anything else, though, I'm probably going to add some better hints in the gcode comments to make life easier for plugin authors. Steve

Okay, now delimiting on G10, and added the Z move at end. (Helped to delimit on G10 to get the latter anyway.) Pushed to github, but completely untested...

What about delimiting on the G10 then? EDIT: Whups, and we also need to add a high Z move at the end, or else we do a perfect print of everything and then knock the well plate silly when we go home. (I think everyone who owns a CNC machine has made that mistake at least once. It's a great way to break a bit while destroying a perfect, finished part...)

Okay, this python script might work -- not tested much, so be careful until you know it works: https://github.com/stevegt/UltimakerUtils/blob/master/well-plate-z.py ...someone might use that as the basis for a plugin to do the same thing.

...looks like maybe we can use either the "G92 E0" or "G10" lines as a delimiter in the meantime... Fragile.

George, the head moves to the X/Y position of the next part before the LAYER:0 line. This means the regex fu that would be needed in a plugin is nontrivial, because it has to find that LAYER:0 and then backtrack to find the first G0 relating to the next part. My first attempt, trying to match all of the "G0 ... LAYER:0 ...G0' part delimiters while extracting the relevant X/Y/Z numbers, never was able to complete. I finally killed it after several minutes: re.finditer('(.+?)\n(G0.+?)\n;LAYER:0\n(G0.+?)Z(\S+)\n(.+?)', gcode, re.DOTALL) Note to self and lurkers: We need a well-defined PART:n line emitted by fffProcessor, to cleanly delimit between the gcode of one part and the next. This hinting and others like it would enable easier plugin authoring.

Okay, here's a Quick Hack: Print a tall tower, outside of the well plate, that is higher than the well plate. Make sure to arrange things so that the tower prints first. Here's why: Cura always writes a Z move at the end of each part that is as high as the highest part. There's still no high Z move before the first part, which is why that tower needs to be the first part printed; that will keep you from crashing into the well plate when approaching the first well. Related code is at https://github.com/Ultimaker/CuraEngine/blob/b27a51e031c9dce3055f8931403d544b892b93df/src/fffProcessor.h#L509 (Note to self and lurkers: maxObjectHeight really should be settable in the UI and passed through from Cura/util/sliceEngine.py to fffProcessor, to allow users to solve this sort of problem without printing a tower. Right now it's hardcoded to 0 in fffProcessor, then increased via max() as each part is generated.) Steve

Ohhhh.... Placing an existing well plate on the glass, and then printing objects down in the little wells? Wow, I'm surprised that works at all... :wink: Hm. You could do it, but unless someone who knows Cura's internals better jumps in here with a magic setting, I think it would need either post-processing on the G-code (probably with a python script) or a plugin. A python script to do this looks easy -- look for each ";LAYER:0", then find the next "G0 ...", and insert something like a "G0 Z100" before it, then insert a copy of the "G0 ..." without the "Z..." parameter, then print the original "G0 ...". EDIT: Whups, hang on, it's not so easy -- there's an X/Y move before that LAYER line that sets you up for the next object. Looking, hang on...

Hi Jemma, If this is for well plates or other labware, you might try T-Glase ("tee-glass"). It's a PETT material that is FDA approved for food contact; details at http://taulman3d.com/t-glase-features.html. I'm assuming you're either discarding them after use (in which case porosity doesn't matter as much), or autoclaving them? If you're in the States, one place that stocks T-Glase is plenty of conversation around here about it. I'm assuming you know about the NIH 3d Print Exchange? Good luck, and let us know how it goes, Steve

Ah. I get it. After seeing her other posts talking about cell growth, I think Jemma's talking about well plates, the plastic trays with arrays of wells in them, used in cell cultures, DNA sequencing, etc. http://goo.gl/NjBQUL Jemma, printing a well plate should be straightforward. Printing more than one well plate at a time should also work; you probably want to go into Cura, and select "Tools|Print all at once", like gr5's saying. That way it will do all of the objects' bottom layers first before moving up to the next layer, etc. to avoid collisions. I'll post an idea about PLA alternatives over in your MSDS topic. (Short version -- try t-glase). Steve

I use the Makefile build. http://goo.gl/xSQnbc is my latest; it works with arduino 1.0.1 and firmware version 14.06.1 and 14.06.2. (Just noticed I need to update the arduino version comment at the top -- that'll be in my next pull request to Daid.) I'm using Linux, but don't know of any reason why it shouldn't work on other platforms. Since Daid doesn't use the Makefile, I intend to keep this updated in https://github.com/stevegt/Ultimaker2Marlin, and expect to continue forwarding pull requests upstream to Daid.

Funny -- while thinking up what spare parts I want to stock, printing that housing was near the top of my list. ;-)

I think both Kris and 3Poro have done tremendous work toward understanding UM2 hotend heatflow, in particular in regards to teflon temperatures. This work provides a pretty good foundation for improved hotends, whether single or dual, factory or aftermarket. Kris' thermal simulations are a must-read for anyone thinking of improved teflon cooling.

Ah Hah! Sander, this is the third printer I've heard of (including mine) which was packed with the glass on the left side and axis rod cardboard under the bed. It sounds like this is standard packing now. It didn't cause any problems with mine, and it sounds like jbeals' is okay, but ZReed's printer sounds like it may have a broken y limit switch, possibly from the bubblewrap snagging on the switch lever. I don't remember seeing anything in the packing method that would prevent this. It might be worth looking into. I'm trying to remember -- is the glass bouncing around on the bed temperature cabling when packed this way, or is there a horizontal piece of cardboard across the top of the bed that would prevent that? Steve

Ok, at 14.06.2 you have the z homing fix, which is good, because with a falling bed you'll have needed it. You have the new priming behavior, where it deposits a blob on the front left as an anchor before print start. Whether you have Simon's later priming and calibration height fixes would depend on the build date; I don't know whether Daid built the official 14.06.2 before or after pull #35. If you have a build date earlier than Jun 23, then you don't have the #35 fixes yet. If you don't have those fixes, then: You'll hear about three skips of the feeder stepper motor during priming, you'll want to make sure your feeder is adjusted tight enough to cause those skips, not grinding, and you'll notice other odd priming behavior at times. You'll also want to use the thinnest possible shim for bed height calibration, and then always cycle the power after calibration, before starting a print. Details on both of these things are in the more recent posts of that topic I referenced. Steve

Hi jbeale, and welcome! Since you just got the machine, you'll probably want to also check your firmware version -- there are a couple of bugs that we've recently fixed, one of which seems to affect the newer machines more. I just got my UM2 in May, and both of these fixes greatly improved the consistency of prints for me, particularly the first layers. Go into Maintenance|Advanced|Version, and make sure you've got at least 14.06.1 installed. That will take care of a z homing bug that was causing a lot of folks (including me) all sorts of confusion. If you want improved priming (so you don't have to grab the thread at print start), and another z-related bug fix, go get the pre-release firmware Simon/illuminarti posted later in that same topic. If you'd rather wait for an official release, keep an eye on http://goo.gl/s6OuYX -- specifically, you want something after pull request #35. I don't know if Daid's going to keep the 14.06.02 version number, or increment it further; either way the official release build date should show Jun 23 or later for these fixes. I was able to get the filament guide to click into place on mine, but yes, it did take a lot of force. The consensus around here seems to be that it's not needed in the first place. I've not seen a difference between printing with and without looping the filament around it. I like the idea of putting the spool on a lazy susan on the floor, but haven't tried that yet. I've swapped back and forth between PLA and ABS several times so far -- never any issues. I think Sander was just trying to scare you into making sure you really do purge all the ABS before you turn the head temp down to PLA temp. :-P I'd be curious to find out what version of firmware your printer shipped with, and if your z axis shows the same low friction as mine -- does yours also fall to the floor when power's off, or am I one of the few lucky ones? :wink: Edit: Oh yeah, and where was your glass packed? Was it under the bed, or was it over on the left side, with the second piece of rod cardboard stuffed under the bed instead? Thanks, Steve

Hi Daid. If this is still a problem, and in case nobody's noticed this solution yet, you might pass this idea along... While staring at my printer just now, I realized there's all kinds of cubic real estate available on the front of the printhead. A new set of aluminum blocks, with a second radiator and fan on the front, with both front and back fans blowing *out*, not in, will not only provide enough cooling but also should help suck some cool air in to the teflons as well, rather than cooling them with preheated air the way we're doing right now. Just a thought. If the problem's already been resolved, then disregard. Steve

Simon, you might recall me mentioning this to you and Sander at Bay Area Maker Faire -- recent UM2s, including mine, are shipping with the glass bubble-wrapped and stuffed vertically in the left side of the printer, nestled against the left inside wall, rather than under the bed. There's only cardboard around one of the axis rods -- I think it was the right-hand y axis rod. The other piece of cardboard is for some reason folded flat and mashed under the bed, where the glass is supposed to go, rather than being wrapped around an axis rod. All of my recollections match what ZReed's saying. I'm not sure if the bubble wrap around the glass could have snagged ZReed's y-axis switch lever, but it seems possible, particularly if the box gets inverted or hits some serious bumps in shipment. Steve

Water cooling!!! :cool: I might actually be serious.

Okay, I decided to put my money where my mouth is, and have ordered a first-draft nozzle blank from shapeways; $30 including shipping. I think I have the dimensions right for drilling and tapping, but won't know until I try it. All I've done so far is increase the nozzle OD where the external threads go -- it looks like the original STEP had the thread minor diameter instead of major, although I might be completely confused. Files at https://github.com/stevegt/Ultimaker2/tree/master/1301_Integrated_nozzle_heater_block_3mm_filament_(x1).