Jump to content
Ultimaker Community of 3D Printing Experts

Uglier overhangs with finer print resolution?


Recommended Posts

Posted · Uglier overhangs with finer print resolution?

Blam 0.1 layer resolution on the overhang test, almost perfect except for a little bent corner on the last layer; I watched the whole thing print. Apparently the UM2 is full of magic, I'm going to bed!

d2MdPRZ.jpg

dibyoFW.jpg

Still don't know if this is going to work on the gear though.

 

  • Link to post
    Share on other sites
    Posted · Uglier overhangs with finer print resolution?

    A few things to note...

    The effect happens at corners, not overhang edges. The effect moves slowly from the corner to the whole outer edge but the problem is a corner issue. So at the tip of the gear teeth and the 2 corners of Aaron's overhang tests.

    That the inner border "repairs" the problem is fascinating - I hadn't noticed that - I guess maybe because I never went down to 5mm/sec? Don't you print inner shell before outer shell? Is it the next layer up inner border repairing the blob?

    The effect is cumulative as Aaron and I both mention so it's not surprising that it looks good for a while and then you start to see blobs. The effect is visible earlier but not until you get blobs does it really stand out.

    It looks like there is not enough material on the previous layer to grab onto the new material needed to form the overhang,

    Yes! One might think the effect would be smaller with smaller layers because the overlap is lessened (the layer above touches more of the layer below). And it is. Except the problem is cumulative and with thinner layers you have more layers for the problem to build up. So thinner layers can be better or worse... depending... on I'm not sure what.

    Great Stuff Aaron!

    POSSIBLE CURA SOLUTIONS

    1) move away. Inner overhang corners are fine. So if the nozzle could leave the corner moving away it might help. But you would need to do a retraction to avoid a string. Seems crazy. I don't think this would be good but it's a possible idea.

    2) Print inner shell last. Maybe if the inner shell is printed second it will keep the corners from lifting. Unfortunately this might have to be done at 5mm/sec on the corner.

    3) Non printing push down corner repair. After the outer shell is printed for a layer, the head could retract filament, and then revisit every corner printed on that layer and remelt it by hovering over each corner for some amount of time (.5 seconds?) or re-tracing the inner shell at 5mm/sec within 1mm of each corner?

    4) Do option 2 but only slow to 5mm on every N layers. Maybe every 3rd layer is enough to keep the problem from getting serious.

    One could test option 2 by setting the jerk speed to 3.5mm/sec which would mean the head would slow to 5mm at a 90 degree corner and then also set the acceleration to 37.5mm^3/sec (wow that's low!) it will still be under 10mm/sec 1mm away.

    But again - the old cura, cura 13.04 let you choose the print order so you could do outer shell before inner shell:

    http://software.ultimaker.com/old/

    Maybe that alone would help. I'm pretty sure this "inner shell" pass that Aaron shows in the video is on the layer above - doing it on the same layer might help even more.

     

  • Link to post
    Share on other sites
    Posted · Uglier overhangs with finer print resolution?

    The effect happens at corners, not overhang edges. The effect moves slowly from the corner to the whole outer edge but the problem is a corner issue. So at the tip of the gear teeth and the 2 corners of Aaron's overhang tests.

    Yeah you are definitely correct, when I started this thread I hadn't spent as much time watching the output as I should have. The edges look fantastic, this machine is quite something.

    That the inner border "repairs" the problem is fascinating - I hadn't noticed that - I guess maybe because I never went down to 5mm/sec? Don't you print inner shell before outer shell? Is it the next layer up inner border repairing the blob?

    Yeah, the inner shell prints before the outer shell does on that same layer, and on the next layer up the inner border is repairing the blob. I think because the nozzle is spending more time at the corners during the inner boarder move, heating up the blob created on the outer boarder move, then it sticks to the nozzle and gets dragged back into the print.

    The effect is cumulative as Aaron and I both mention so it's not surprising that it looks good for a while and then you start to see blobs. The effect is visible earlier but not until you get blobs does it really stand out.

    Yeah, you can start to see very tiny lips the curl upwards form at the beginning of the print, for a couple of layers it looks like the layers are going to be smoothed out again, but the curling gets compounded until the curl disrupts the outer border.

    Yes! One might think the effect would be smaller with smaller layers because the overlap is lessened (the layer above touches more of the layer below). And it is. Except the problem is cumulative and with thinner layers you have more layers for the problem to build up. So thinner layers can be better or worse... depending... on I'm not sure what.

    Great point, yeah it's hard to see the effect happening in real time with thinner layers; but it's definitely a cumulative effect and not really apparent right away.

    POSSIBLE CURA SOLUTIONS

    1) move away. Inner overhang corners are fine. So if the nozzle could leave the corner moving away it might help. But you would need to do a retraction to avoid a string. Seems crazy. I don't think this would be good but it's a possible idea.

    2) Print inner shell last. Maybe if the inner shell is printed second it will keep the corners from lifting. Unfortunately this might have to be done at 5mm/sec on the corner.

    3) Non printing push down corner repair. After the outer shell is printed for a layer, the head could retract filament, and then revisit every corner printed on that layer and remelt it by hovering over each corner for some amount of time (.5 seconds?) or re-tracing the inner shell at 5mm/sec within 1mm of each corner?

    4) Do option 2 but only slow to 5mm on every N layers. Maybe every 3rd layer is enough to keep the problem from getting serious.

    Very interesting suggestions:

    1. Do you mean, lift the head at the tip of a corner?

    2. I'll take a look at Cura 13.04 and see if reversing the shell print makes a difference.

    3. Yeah, the head would have to remelt and lift then you would have to press down rather quickly, I've tried to mend the corners right after the head moves over them with a metal rod but the plastic solidifies too fast and perhaps the metal rod wicks away heat too quickly. Maybe a warm metal rod?

    4. I see what you mean here, every nth layer gets repaired so the cumulative effects don't build up.

    I like your suggestions for the jerk and acceleration, I'll give them a try as well.

    I'll report back my findings; what are your opinions on doing these tests on the overhang piece? I've modeled it so I don't have to wait for the entire bottom gear piece to print out, I was trying to find a quicker way to test different ideas out.

     

  • Link to post
    Share on other sites
    Posted · Uglier overhangs with finer print resolution?
    what are your opinions on doing these tests on the overhang piece?

    The faster you can learn results, the more tests you are likely to do. So it is very important to have something that gives you results as soon as possible. I would like to see a trapezoidal piece where two consecutive corners are on walls that are vertical and overhang, with a 3rd corner that is doubly overhung like your gear. I'll mock something up right now and repost in a few minutes...

    1. Do you mean, lift the head at the tip of a corner?

    No - I mean keep moving in X/Y plane but stop extruding. This is impractical as you will get oozing/stringing or something bad. I thought maybe it could go beyond the corner and do a 270 degree marching-band-turn and come in at another angle. I don't want the head pulling plastic inward - better to pull outward than inward. But I think this would be worse and introduce other issues. Still - what you haven't seen yet - is that sometimes this lip gets worse and worse until it is sticking up 2mm and the head hits the piece and it gets knocked off the bed and the entire print fails! I guess if we could make travel moves approach the part at a point that didn't have an overhang that would help.

    Regarding pressing down with metal - I had success with that when the heated bed was at 70C and I left a putty knife on the bed when not in use and I don't think it was so much the heat of the knife but I bent the entire piece down enough to keep it from getting damaged when the head came back.

     

  • Link to post
    Share on other sites
    Posted · Uglier overhangs with finer print resolution?

    Nice! I like the overhang piece you generated. I'll test it out later today. I'm about run some errands, but I have a couple of photos of the results from your suggestions above and also of more tests I've been running on my overhang piece. I wanted to comment on your clarification of suggestion 1. I think this is a really good idea. I have been messing around with the lift cool head settings and watching my nozzle while it sits and waits for the part to cool and have found a fairly wide temperature band where there is no oozing of filament from the nozzle while it waits a for several seconds, this gave me better surface detail and no blobs upon reentry of the nozzle to the part. How would one go about trying to implement something like this? Are requests of this nature sent to Daid, he seems pretty busy with his RCs I think he's on number 6 right now, and everyday there seem to be more bugs found.

    Yeah, I can imagine the lip getting fairly high fairly fast, and when it cools it will definitely hit the head; in the video I posted of my overhang test the nozzle is buried into the part by 1+mm. When I was printing out my overhang test in post #26 I was able to keep the curling to a constant minimum, and the only reason there was any real curling towards the top was because I printed it with zero infill and forgot to uncheck top fill in expert settings. I think if you alter the code for delicate parts that tend to curl and apply those settings to only those layers you can substantially reduce the curling building up over time.

    I'll post some better prints than the one in #26 and the settings I used to make them, when I get back home later today. The picture in post #26 came out as flat as it looks in the picture, I didn't do any post processing to the piece after I pulled it from the bed.

     

  • Link to post
    Share on other sites
    Posted · Uglier overhangs with finer print resolution?

    Never mind what I said about your suggestion number 1. I see what you mean, if without a retraction the filament would still have a tendency to come out and with a retraction the filament already deposited could get sucked back into the nozzle.

    Edit: I don't know still, I think it would be at the very least an interesting idea to try and implement.

     

  • Link to post
    Share on other sites
    Posted · Uglier overhangs with finer print resolution?

    Alright so I just got out of the shower, and am really going to run those errands; I have almost no food in the house :shock:. Can someone provide me an .stl file for a sandwich :lol:.

    But I was thinking what if there was a setting to do something like cool head lift does and then the head comes back inside the corner and sweeps it to the outside like you suggested, but while still not extruding any material, then it comes back in onto the inner shell, after the sweep, to continue it's business while avoiding the outer shell edges. Sort of like the lift cool head actions again, but this time when the nozzle reenters the piece it doesn't come in from the edges and instead drops down onto the inner shell layer.

     

  • Link to post
    Share on other sites
    Posted · Uglier overhangs with finer print resolution?

    By the way, in my above print, it makes a big difference if you print clockwise or counter clockwise. In either case the corner you hit first I suspect comes out better than the second corner. Or at least it looks different.

    So having Cura alternate clockwise then counter-clockwise on each layer would help if the overhang point isn't symmetric. For teeth of Aaron's gears it makes no difference.

     

  • Link to post
    Share on other sites
    Posted · Uglier overhangs with finer print resolution?

    Hey gr5, I tried the older version of Cura, boy has that program matured! Printing the outside layer first didn't do anything to help the print. I even printed at the settings I posted at the bottom of this post but my overhang test piece would curl up like I was printing with normal settings. I still have yet to try your acceleration reduction, although I'm not sure how well it will work with such a cold nozzle; I think I'm stuck printing out the overhang layers really slowly. Will report my findings.

    Alright so here is how I figured out how to minimize curling around the corners of my prints. I hope this can help others out there and I'm very interested to hear your feedback/success/difficulties with the technique.

    So to start out with here are some pictures comparing the results I was getting, these are only a fraction of the little overhang pieces I printed out.

    Before I knew what I was doing, we have 0.2 mm layer thickness on the very left and the remainder are 0.1 mm layer thickness, all with a bunch of different print settings but the same g-code:

    8slPjNH.jpg

    yJPD2nC.jpg

    And after I watched enough of these stupid things print out, same as before with a 0.2 mm layer thickness on the left and the remainder being 0.1 mm layer thickness all with varying printer settings centered about specific thresholds, and all with the same g-code:

    ao4YNft.jpg

    rj4DzdS.jpg

    That little lip doesn't get any bigger than what you see in the image above. Also, if I were to abort the print while it was printing the inner of two shell layers, the lip would probably be even smaller.

    Alright, now for gr5's test print (0.1 mm thickness); the distorted ring near the base is because I put a folded up piece of stock card too close to the print which resulted in cooling the head too much. I was trying to use the card to focus the a video I was recording, and noticed I was getting a tiny bit of under extrusion almost instantly when I moved the card too close; yeah that's how cold the nozzle was, I call the settings my Ultra-Low settings:

    gONQGiy.jpg

    vERU4GL.jpg

    Here are a couple of videos of gr5's test piece printing out, notice the same blob grabbing phenomena as in my previous videos. I accidentally had three shell thicknesses from a previous test in Cura so the first inner shell print does nothing. To recap from my previous video, I noticed that if I printed really slowly the nozzle grabbed the blob that got generated on the outer shell move, and dragged it into the inner shell where it couldn't accumulate. In my previous attempts to control this action the piece still wasn't turning out as well as I wanted with the corners melting too much sometimes.

    So I decided to take things as far as I could, take a look at the videos, the piece is printed with a 0.1 mm layer thickness:

    This video shows the amount of curling in the middle of the print, and the next two videos show how the curling stayed constant and didn't build up over time.

    This video shows the blobs being re-absorbed into the print, like my previous video does, but the results are much cleaner.

    This is a video of following the nozzle as it prints the shells of the test piece, remember I accidentally set Cura to do three shell passes, so the first pass does nothing.

    So from the video you can already tell what one of the settings is, it's to print slow; Ultra-Slow. I have tested the speed and get good success from 2-4 mm/s any faster than that and the prints start to curl too much.

    The other setting if you haven't already guessed is to print Ultra-Cold, I had my nozzle zeroed in at 175C. I could go to 170 but the nozzle was so close to under-extrusion that if it got just a couple of degrees C lower, I would get under-extrusion. I tested all the way up to 180 and saw good results; around 190 was too warm and would start to melt too many layers at time.

    The fans were turned on to 100% after the first layer, and the heated bed started at 60C and was brought down to 40C after the first layer was applied (a tip from Yellow-Shark, they suggested in this thread).

    To figure out how cold I could print with my red PLA, I printed my overhang test, starting at 210 and went down all the way to 175 in 5 degree increments waiting to see under-extrusion.

    Anyway, I'm sure there is more than one way to skin a cat and would like to know how other people wrestle with gr5's and my overhang test pieces. Also, if you change specific layers in your code to reflect the settings posted, I would really be interested in the results.

     

  • Link to post
    Share on other sites
    Posted · Uglier overhangs with finer print resolution?

    Brillant videos Aaron!

    Your experience is similar to mine, like I expressed in the thread gr5 referred to (http://umforum.ultimaker.com/index.php?/topic/4094-raised-edges/?p=33922).

    If the part is small, reducing speed (and lowering temperature) is the best way to avoid the issue. But if the part is big and includes small islands (e.g. Meshmixer support structures), you will face the same problem but you don't want to reduce overall print speed.

     

  • Link to post
    Share on other sites
    Posted · Uglier overhangs with finer print resolution?

    Hey gr5!!

    Alright so I tried the acceleration and jerk settings, I brought them down as close to your recommended settings as I could, then brought them down to the lowest settings that were possible, while printing at 20 mm/s and there was no improvement on the curling.

    BUT...

    I tried out your suggestion number 4 with some interesting results.

    POSSIBLE CURA SOLUTIONS

    4) Do option 2 but only slow to 5mm on every N layers. Maybe every 3rd layer is enough to keep the problem from getting serious.

     

    The one on the left is my manual attempt and the one on the right is with the use of tonycstech's Modify g-code program found here: http://umforum.ultimaker.com/index.php?/topic/4185-tweak-at-layer-number-windows-application/, I had a print speed of 20 mm/s and modified the g-code to print out at 100% every odd layer and 20% every even layer starting at the 6th layer.

     

    ekv2HJg.jpg

     

    EJgUd6w.jpg

     

    Here are a couple of videos, the first of me trying the every other layer thickness manually and the second using tonycstech's awesome program.

     

    Manual:

     

    tonycstench's g-code modifier program:

    But if the part is big and includes small islands (e.g. Meshmixer support structures), you will face the same problem but you don't want to reduce overall print speed.

    I'm not sure what you mean about the first part of your sentence, the curling is maintained and does not accumulate with the settings I posted earlier. You are right about the print speed though, it would be silly to print out a large piece at such a low speed just to take care of some curling, but as a compromise between stopping curling and maintaining a reasonable print speed, I think doing the slow passes every other or maybe every 2nd layer would be worth it. Especially if you don't want to babysit your print to prevent the curls from hitting the printing head.

     

    I've sent a request to tonycstech in his thread, and hope it's not to difficult to make the appropriate addons. Here is a link to what I posted in his thread: http://umforum.ultimaker.com/index.php?/topic/4185-tweak-at-layer-number-windows-application/?p=41931

     

     

     

  • Link to post
    Share on other sites
    Posted · Uglier overhangs with finer print resolution?

    I also tested the settings posted above on a 4 mm diameter 0.1 layer thickness cylinder protruding from the build plate at 60 degrees. The lip did not accumulate in this print either; the print was designed to simulate what a support structure would look like at this angle. The lip did not accumulate throughout the print, just like the other overhang tests.

    1EFsvam.jpg

    ev0DGMc.jpg

     

  • Link to post
    Share on other sites
    Posted · Uglier overhangs with finer print resolution?

    When your printer diameter is 0.4 and layer thickness is 0.3 and print angle is over 30 degrees, how do you expect the next layer not go over the edge (well almost over the edge) and either fall over or curl up ?

    Print that layer at super slow speed or make layer height 0.1-0.2

    Cooling will not help much because blower will move away before it manages to cool anything.

    Speed will help far better. Slow down to 20% or even 10% and observe.

     

  • Link to post
    Share on other sites
    Posted · Uglier overhangs with finer print resolution?

    cylinder protruding from the build plate at 60 degrees.

     

    What speed did you print at?

    Keep in mind you did 60 from horizontal but you can get almost horizontal. You can print about 10 degrees from horizontal. The surface will be very rough but you can do it.

     

  • Link to post
    Share on other sites
    Posted · Uglier overhangs with finer print resolution?

    Hey tonycstech, I do expect the edge to curl up; but I've managed to control the curling from accumulating as the print progresses with great success. I've tried speeding up the print with the lower nozzle temperatures (175-180C) and I still get the curling accumulation throughout the print. Only when I slow down the print to a snails pace of about 4 mm/s can I consistently manage the curling of the edges.

    For example I tried to print the bearing piece attached to the string in the image below with many different settings, cold nozzle fast and sow print speeds, and hot nozzle fast and slow print speeds; and the only way to prevent curling was with cold nozzle slow print speeds. The slow print speed seems to give the material enough time to evenly heat and then consequently evenly cool down, preventing the curling from happening. Without these settings I would start to get a bowl shape after a couple of layers just like in foehnsturm's print video: http://umforum.ultimaker.com/index.php?/topic/4094-raised-edges/?p=33464

     

    merg6ot.jpg

     

     

     

    @gr5 I printed the cylinder at the same settings I posted above, 4 mm/s. Without going this slowly I would get the same type of lip accumulation foehnsturm would see in his prints. I picked the arbitrary angle as a function of foehnsturm's comment. I thought the portion of the sentence I did not understand in his comment was in reference to my settings not working on round support structures that have an oblong cross section. I just wanted to print out a simulated support structure to show that the lip curling effect does not accumulate with the settings posted earlier.

     

  • Link to post
    Share on other sites
    Posted · Uglier overhangs with finer print resolution?

    Great research.

    Just to clarify my comment. I wanted to say that printing slow (more exactly: as slow as perfect overhangs require dependent on the available cooling right at the moment of extrusion) seems to be the best (and so far only?) cure. But if you have a large part with say 95% uncomplicated geometry and only 5% of those critical overhangs printing with snails pace is a huge price to pay.

    Printing slowly only every second or third layer could be a solution. The other idea would be to identify the critical path segments within a single layer and change speed, acceleration, jerk or whatever just for those segments. Cura knows about overhangs, so that should be possible - in theory.

     

  • Link to post
    Share on other sites
    Posted · Uglier overhangs with finer print resolution?

    Hi Aaron

    1. Do you use glue to make parts stick to surface?

    2. These settings you are talking about, you determine them when producing the geode or afterwards at the machine when before print starts (or even after) ?

    Celso

     

  • Link to post
    Share on other sites
    Posted · Uglier overhangs with finer print resolution?

    I think Aaron has currently the glue stick as an adhesive. For PLA, there are already a lot of possibilities to improve the adhesive properties. This is especially important for the very first layers of prints, so that initial separations and distortions can be avoided. Also, the only way to print at very low bed temperatures, for lowest distortion of an object. There are also adhesive left almost no residue, thus obtained the best possible surface quality.

    Search in the forum or elsewhere by: "direct printing on glass"

    So You'd very quickly find the best adhesive, and can find more information on this topic.

    Markus

  • Link to post
    Share on other sites
    Posted · Uglier overhangs with finer print resolution?

    Hi Aaron

    1. Do you use glue to make parts stick to surface?

    2. These settings you are talking about, you determine them when producing the gcode or afterwards at the machine when before print starts (or even after) ?

    Celso

     

    Can you update your profile with the printer you have? My recommendations are for the UM2, since that's what I have.

    1. The easiest method I have found to have my parts stick every time is using some extra strength hairspray. NOTE: DO NOT SPRAY DIRECTLY INTO THE UM, you will wind up getting sticky hairspray into the gears and pulleys and belts (oh my). What I do is spray a good amount onto a paper towel and then wipe it onto the buildplate. It dries instantly leaving a thin sticky layer that works great. You can also pull the glass plate out and directly spray it then, but that's just more of a hassle.

    You can also use the gluestick, most people seem to recommend spreading some on the glass then using a slightly damp paper towel to spread it out into a thin layer. Other methods include the ABS slurry/glue, blue tape, and kapton.

    2. The print speed settings are set in Cura, the temperature settings (when using UltiGCode) are set on the UM2 under the "Material" section. You can also manually adjust the temperature and speed when printing using the "Tune" menu.

     

  • Link to post
    Share on other sites

    Create an account or sign in to comment

    You need to be a member in order to leave a comment

    Create an account

    Sign up for a new account in our community. It's easy!

    Register a new account

    Sign in

    Already have an account? Sign in here.

    Sign In Now
    ×
    ×
    • Create New...