Jump to content

lars86

Member
  • Posts

    351
  • Joined

  • Last visited

  • Days Won

    2

Posts posted by lars86

  1. Okay, thanks Owen.

    I haven't played with Netfabb much at all yet, because I haven't had the time to figure it out yet. I like that it is more customizable in print settings than Cura, but it is not very intuitive, and the lack of tool tips make it a pain to get going quickly.

    Do any of our Ultimaker gurus here have some print profile settings they could start me out with?

    It seems to be overextruding for me as well. Very visible on the first couple layers of perimiter passes, where you can observe a bunch of overlap.

     

  2. Having rapid movements doesn't reduce stringing much and there are so many other things to control - it's easier to just speed up and slow down everything. If you print from Cura you can control 3 or 4 separate speeds I believe (walls only, fill only, etc.).

    You also can control "flow" live as you print. This is handy e.g. if you want the bottom layer perfect (no gaps, no overlap).

     

    I notice a big difference in stringing in slow vs faster rapids. It makes perfect conceptual sense too: the faster you pull on that dwindling filament strand, the more likely it is to shear away. If you pull it slowly, you will get plastic deformationas you stretch it out.

     

  3. Marlin tries to move X,Y,E (and occasionally Z) together. the gcode sets the desired speed in X,Y,E and marlin splits the move into 3 sections: accel, constant speed, decel. Often it never gets to the desired speed so there is just 2 sections: accel, decel. Marlin looks at things like jerk and max acceleration and picks the limiting acceleration (usually XY, but possibly E) and limits acceleration on *all* axes so that as you might think of it, it travels in a *straight* line from one point in 4 dimensional space to the next point. If it didn't make it a straight line, then you would get overextrusion in one section and underextrusion in another section of a given segement.

    So I don't think the problem is with the software.

    However, there is a lag or delay in when you tell the extruder to move and when the filament comes out or stops coming out. The extruder controls the pressure in the head almost more than the actual amount of extrusion.

    Maybe someday Marlin will model this but it's complicated as it depends on the type of filament, pressure, hole size, etc.

     

    I am definitely noticing that disconnect between extruder drive movement and physical extrusion. It's definitely making me want to suspend the extruder drive above the machine, cut the bowden tube length in half, and have a much more direct, lower friction, filament path.

     

  4. I think the over-extrusion is related to excess pressure and ooze, rather than acceleration. The move planner does keep the x,y,x and e axes in step as acceleration occurs, so I'm pretty sure it's not related to that.

    I'm struggling to understand how stringing can be occurring on approach. It doesn't pre-emptively start extruding and even if it did, there's nothing for the plastic to grab on to, so worst case you're likely to end up with a blobby mess at the point where the head reaches the printed object, as any extrusion gets wiped off onto the object.

    There is an optional feature called 'Advance' I think which does do some sort of 'intelligent' extrusion management - but it should be off by default. You might want to check the config files of your custom build. Maybe that got turned on?

    Can you perhaps shoot some video showing the problem, so we can better understand the effect?

     

    If it was just an issue of ooze, I don't understand why it could be cutting back and forth, laying down infill in a 20mm wide area perfectly. Then when that width shrinks to say 5mm, massive overextrusion.

    On the next print that I notice the weird stringing issue, I'll record it. It's not a huge deal, but since those rapids occur mainly in the same XY position, I'll get a series of warts on that wall.

     

  5. A couple things that I've noticed so far in printing:

    • [*]In areas with very quick reversals (<5mm or so) the machine overextrudes. This is made even worse with lower acceleration values. I could be wrong, but it seems like it doesn't take XY acceleration into account when deciding extrusion speed, so for fast XY speeds wiht low XY acceleration, it's pretty bad.

    • [*]The feed rate override on the main display modifies both extrusion feed rate, as well as rapid movements. I can't imagine why you would want these two tied together. Having the main override control only extrusion moves would be much more useful. This would let you tune printing speeds for quality, while maintaining fast rapids for shorter print time and less stringing.

    • [*]Most of my stringing happens on the appproach, rather than the departure. I am not using the "extra length on start" feature in Cura. Is there some sort of lookahead that preemtively starts extrusion?

    Thanks, I'm having a blast with this thing so far!

    Lars

     

  6. Markus, this is really cool!

    I just got my UM a few days ago, and the pulley runout has been bothering the hell out of me!

    Please keep us updated, and I would definitely be interested in purchasing a set.

    Thanks,

    Lars

     

    Finally they arrived.

    A nearby CNC turning shop produces some thousand gas burner jets for my wok burners each year. So I asked them, if they see a way to produce tailored pulleys for a reasonable price. For this reason it had to be a one-part-pulley instead of the three-part-pulleys, which are commonly available - and fall into pieces when drilled to 8mm, as I had to learn...

    Production samples arrived today.

    From left to right: Ulimaker pulley, standard pulley with pilot borehole, "new" pulley

    ulitmaker-pulley-vs-standard-pulley-vs-lathe-pulley.jpg

    Ultimaker pulley, new pulley

    ulitmaker-pulley-vs-lathe-pulley.jpg

    We tested the fitting by hand with an Ultimaker short belt, which I don't need anymore (thanks Calum). You can feel no play, whereas the original pulleys slightly move with a belt wound tightly around them.

    The pulleys feature a 8h7 center bore and two M3 threads. The small gap between the cogging and the flange is a trade-off of the one-pice design. I'm not sure whether this will increase belt wearing (if so, it should be possible to design and mount some split spacers to fill the gap). But compared to the wobbling original pulleys (hey, really ugly in closeup), which even seem to feature a slightly contorted cogging, it is definitely worth a try.

    I will be mounting the new pulleys the next days and keep you informed.

    Markus

     

  7. Sanjay:

    Thanks for the feedback!

    Yeah, the threaded size has competing properties. Larger threads have more surface area, which enhances conduction from the tip/heater block, but that heat must now travel a longer distance through the insulative SS to reach the up tube. I'll play with both configs and see which wins.

    I also tried bumping the 1mm section length to 3mm and did not see an appreciable difference.

    I did a few hand calcs to compare the brass vs aluminum nozzle. Aluminum actually has a much better heat capacity (mass based) than brass. If you pretend to be volumetrically limited (which we really aren't) and factor in density, I find that brass has ~26% higher volume based heat capacity. But aluminum is at least 63% more thermally conductive. I think that aluminum wins in this regard, especially considering that you can bump the aluminum nozzle size up 26% to achieve the same energy density, along with much better conductivity.

    I am playing with a design that has a 1" diameter nozzle with the heater built in. Even with a simulated layer of insulation on all exposed nozzle surfaces, I am seeing temps ~5*C higher on one side of the melt zone than the other in transient simulations. The single heater cartridge is inherantly prone to this. My original idea was to have a flexible heater element that would do at least one wrap around the perimeter for even heating. I really think that even twin smallers heaters, 180* opposed would be enough.

    Sourcing electrical components that can be driven with the existing power supply circuit, and will provide similar heat output, is out of my realm of expertise. I'm a mechanical design guy! Can someone recommend a twin heating element solution that would be a plug and play replacement for the Ultimaker cartridge? I guess roughly half the wattage and twice the resistence , wired in parallel?

    Lars

     

  8. Sanjay:

    Good to know that 0.4mm is plenty strong... obviously, the thinner, the better for this piece!

    I'm glad my simulated conductive transfer is in the ballpark too. The stock UM hot end uses a brass up tube which my simulations show to conduct a vast amount more heat. I put a few images up in a gallery from my tests. The first compares a brass nozzle and uptube, to a stainless steel up tube, and finally stainless tube with heat sink. The next compares the original 3 fin heat sink I modelled to the heat sink I made yesterday.

    All tests were run with 50 W/m2K convection on exterior surfaces, 40W heater power added to the cartridge bore, and 4W lost to the filament in the melt zone.

    For the time being, most parts will be made on the manual lathe. I made a custom 1mm slotting tool yesterday and created a heat sink with it. The up tube will start as a stainless steel 8x1.25 bolt, and I'll machine it to my design shape.

    Composite.Stock.SSResponseGraph.Stock.SS.LarsIMG 20130410 175437Response Graph 3.vs.6.finIMAG0043Capture

     

    Hi all again,

    We have ordered some really nice looking POM (Delrin) push fit pneumatic fittings. These definitely have a permissive OD for the filament, and POM is a very low friction material so these should work nicely, just need to get a hold of them for testing now.

    Calinb:

    From the point of view of making your own on a one-off basis this is all well and good, but from a marketable retrofit version you have to consider that our heatsink is designed as an efficient heatsink, and that plate is designed as.. a plate. That said, we have had really good results with a plate as a combined heatsink and X-Carriage, with a 3mm SS heatbreak poking right out of the bottom of the plate, leading directly to the heater/nozzle. This gives really low Z profile which is nice. We did need to add a PC heatsink/fan combo to the plate, just like am001 has in his photos.

    Lars:

    0.4 is indeed quite thin, but no challenge for most modern production companies - it is also perfectly strong enough, I cannot break the SS part by hand - I have to use a spanner and the chuck of a lathe to grip it firmly enough so I am not worried about mechanical strength at all.

    7W seems about right for you CSA and length with stainless - this is quite a lot considering that your average hotend needs about 10-15W at idle to maintain temperature. Pushing more power into it to compensate shouldn't really be a problem, but remember that heat you then have to get rid of (quite rapidly and sharply) at the heatsink. My advice would be to consider going longer or thinner. How are you planning to manufacture?

     

  9. Righty!

    Thanks everyone for the informative replies and suggestions:

    Lars:

    We too are modelling/simulating in SolidWorks. Very nice.

    Our wall thickness is 0.4mm. You have to consider both the wall thickness and it's length obviously. Fouriers thermal transfer equation is most useful in this regard. If your break is short (Like ours) then you also must consider convection and radiation. This is more of a black art and really needs to be done empirically rather than calculated.

    What length of 1mm wall section do you have? How much wattage is conducting up that break?

    Brass for the nozzle is for a couple of small factors - It machines much more easily for tiny holes, and the finish is less "burry" and more smooth, which makes a real difference when you want a quality exit orifice. The other is that aluminium nozzles have a very low specific heat capacity - you actually want to store some heat, for sudden extrusion bursts where melting lots of plastic sucks out heat and for keeping a smooth temperature when fans and airflow are involved. When you have a fast moving head like in an UM then you are always experiencing significant airflow.

    We did experiment quite a bit, and look at the results others are getting with various nozzle orifice geometries. Our lead in angle is largely dictated by the angle of the drill bit tip - there is some room for adjustment here, but not significantly so. The main thing to get right is the ratio of diameter to length. You need to induce laminar flow in the fluid plastic for nice results, which requires length to be at least 2x orifice diameter. The other consideration here is ease of flow/backpressure versus oozing/dripping. We went for 2.4x diameter for a length of 1mm which gives a nice balance between all these factors.

     

    Great info Sanjay!

    Wow, 0.4mm is THIN! My 1mm thick section is 2mm long. In that zone I plan on using a stainless steel washer, backed on both sides by isulative ceramic tape. The tape should keep it relatively isolated from conductive and convective heating and the SS washer should act to reflect some radiation too.

    According to SW, the power conducted up my stainless tube is ~7W, I'm not sure if that is correct, or if I need to make some reference geometry to get a better number. What are you seeing conducted up your mini,mini tube?

    Interesting point on the heat capacity. Have you been able to demonstrate that issue with an aluminum nozzle? I ask because I did some hand calcs for how much power is consumed to melt a mass/volumetric flow of filament and found a very low number. I used a feed rate of 2.5mm/s of 3mm ABS (which I have since been told is about double a normal extrusion rate), and came up with 5.3W consumed to bring it from 25*C to 220*C. That seems a modest amount of power, so I'd expect the fluctuations to be even smaller. Having access to very fast thermal conduction from aluminum and a powerful heater would seem to negate any issue with nozzle temps dipping. Thoughts?

    I'm with you on the nozzle geometry too. It would seem like making a customized tool that cuts a transition profile from 3.2mm to 0.4 could be a big help. Having the 0.4mm entry comprise a steep angle (118*), with a sharp edge is certainly not helping produce laminar flow. If you could move from a very shallow angle, through a radius, and in... that would be awesome. Tiny fuggin tool though! I just got a couple 0.4mm bits in the mail and couldn't tell two were stuck together at first! hahaha.

    Lars

     

  10. Your hotend already has a heatsink but my Ultimaker already has the plate! ;) The plate may not become my entire heatsink, but the surface area of the plate could be increased somewhat with some milling--perhaps even surpassing the surface area of your hotend heatsink. I don't know yet.

     

    Having the print head plate serve as your "heat sink", is a bad idea in my eyes. This means that any heat that you intend for it to dissapate, has to be conducted up the entire tube, which leads to a very large zone which is well above the glass transition temp of the filament.

    I think the idea of minimizing this zone, in the pursuit of having cool, rigid filament right up to the melt zone, has a lot of merit.

     

  11. Just getting started in Cura, and I'm curious what the significance of the per layer colors are. When it shows a solid color (no beads) does that just indicate solid fill? Why different solid color triangles within a given layer?

    Thanks!

  12. Hey Sanjay, that's a beautiful part you guys have there!

    I've been spending some time fiddling in Solidworks, designing my own hot end and have ended up somewhere similar to you guys. My stainless thermal break necks down to a 1mm wall thickness right now, and the loading sim tells me it will be strong enough. I'm curious how thin you have gone.

    My thermal simulations are showing a very sharp transition zone in mine too, even without active cooling (certainly more so with it).

    I'm also curious why you have chosen brass over aluminum for the nozzle.

    I haven't even gotten my Ultimaker kit yet, so you'll have to get mounting insight from the experienced guys!

    Cheers,

    Lars

     

  13. Bump:

    Hey guys,

    I'm new to the 3d printing scene, and am hoping my Ultimaker kit ships in about a week, so I can get started!

    I'm a mechanical engineer and avid designer, so not having the printer hasn't held me back much from thinking about ideas for upgrades (for better or worse, hahaha)

    Today, I spent some time looking at hot-end designs and thinking about possible improvements. Some of which may be off-base, since I don't have anything in my hands yet. It seems that key points would be to have a very well controlled and uniform temperature hot zone, which transitions as quickly as possible to the cold zone. The current hot end with a single power resistor, thermistor close by and no insulation, seems like it would lead to inconsistent filament heating when fan speed / ducting, print head speed, extrusion speed, or ambient air temp changes.

    What if the aluminum heater block was cylindrical instead, with a u-groove cut in a helix around the outer surface. Then a flexible heating cord like this: http://www.omega.com/ppt/pptsc.asp?ref=HTC&ttID=HTC&Nav= was wrapped in the groove. You could then wrap an insulative sleeve on top of that, and have a very evenly heated hot-end with less heat loss (radiant and forced convection from the cooling fan).

    For the transition to the cold zone: The heater block threads onto the nozzle (like normal), moving upwards you have a short thin walled section (of nozzle) to reduce conductive transfer, then back to OD threads to which you would attach a small heat sink similar to this: http://media.digikey.com/Photos/Tyco%20Photos/4-1542005-0.jpg . You could then redesign the fan duct to create a small airstream aimed at this cold zone heat sink. It makes sense to also use an insulative washer between the heater block and heat sink.

    I'm just throwing stuff out there... hopefully it turns into a great discussion with contributions from the very experienced on here, and eventually some upgraded parts!

    -Lars

     

×
×
  • Create New...