Delamination during print

Discussion in 'Calibration, Help, and Troubleshooting' started by Spoon Unit, Dec 13, 2016.

  1. Spoon Unit

    Spoon Unit Well-Known Member

    Joined:
    Sep 6, 2015
    Messages:
    1,095
    Likes Received:
    322
    I'm looking for some advice on what to check in regard to a print which is delaminating and falling to pieces mid-print.

    First clue
    • I can print one copy of the item perfectly
    • If I print 4 copies, delamination occurs (tested twice)
    Here's an image of a completed single print.

    vlc_2016-12-13_09-56-15.png

    And here's an image of the 4-copy print just after failure:

    vlc_2016-12-13_09-57-35.png

    So, first question. What does a single copy success and multi-copy failure hint at for a problem?

    I've printed other items in large numbers of copies with no issue, so I'm wondering if this could be model-specific, even though I can't imagine why?

    Clearly a longer running print runs the motors for longer, so could this be an indication of motor failure, or overheating? It's almost as if Z movements are overshooting at layer where we see the beginning of delamination (layer 10).

    The initial red images above are with a layer height of 0.29999 just incase S3D was slicing wierdly. But here's a previous print of just two copies which went wrong on a lower layer:

    vlc_2016-12-13_10-05-54.png

    And here's the point when things first went wrong where, after printing one item successfully I tried to print a whole bed:

    vlc_2016-12-13_10-07-33.png

    The successful print at the top, by the way, was printed after the failures at the bottom.

    So I'm scratching my head wondering what's happened, or what's failed, what I can do to fix it, and what I can learn from this. Ideas anyone?
     
  2. mike01hu

    mike01hu Well-Known Member

    Joined:
    Apr 22, 2014
    Messages:
    957
    Likes Received:
    166
    I am wondering if it's nozzle temperature and extrusion rate as your extrusion looks pretty round with no deformation to indicate that interlayer adhesion is taking place. Have a look at your temperature graph to see how the nozzle is performing and consider upping the flow rate a bit . . . or are you overcooling the work-piece? I think a 0.3mm layer height is a bit high for a 0.4mm nozzle, if that's what you're using, but that's just me as I like to keep well under the suggested 75-80% nozzle to layer height ratio. :rolleyes:
     
    Spoon Unit likes this.
  3. Ephemeris

    Ephemeris Well-Known Member

    Joined:
    Nov 25, 2015
    Messages:
    296
    Likes Received:
    182
    I agree with Mike. I think the temperatures are marginal and the multiple prints have more time to cool and that pushes you into fail territory. By the way is that ABS ?
     
    Spoon Unit likes this.
  4. Spoon Unit

    Spoon Unit Well-Known Member

    Joined:
    Sep 6, 2015
    Messages:
    1,095
    Likes Received:
    322
    PLA. Incidentally I've printed full plates before with 0.3 and 100% cooling from layer 3. Thinking about it, there are no overhangs or scary geometry here by design, so, I may try another 2-print with minimal or zero cooling. I could switch layer to 0.25 also as that won't add overly and, as you say, even though 0.3 is my go to, perhaps I'm revealing an issue with this particular model in some way.

    Print temp is: layer 1, 210, layer 3 onward 200.

    Thanks for your thoughts so far!
     
    mike01hu likes this.
  5. A Dragon In A Pie Costume

    A Dragon In A Pie Costume Well-Known Member

    Joined:
    Jul 6, 2016
    Messages:
    202
    Likes Received:
    15
    maybe try a different slicer, some handle multiple objects differently
     
  6. Spoon Unit

    Spoon Unit Well-Known Member

    Joined:
    Sep 6, 2015
    Messages:
    1,095
    Likes Received:
    322
    Thanks for the ideas. Switched layer height to 0.25, kept temp up at 210, fan down to 15% and have 4 prints exactly as I would want.

    chrome_2016-12-13_22-00-58.jpg
     
    mike01hu likes this.
  7. mike01hu

    mike01hu Well-Known Member

    Joined:
    Apr 22, 2014
    Messages:
    957
    Likes Received:
    166
    It never ceases to amaze me how inconsistent 3D printing can be with only a small change in a parameter or two. Was this a change of colour too, as I find quite a variance with PLA of different colours, white in particular, that I suspect is due to colour fillers? It is why I save the factory files for each job and comment on filament colour and supplier, but I don't always win!
     
  8. Spoon Unit

    Spoon Unit Well-Known Member

    Joined:
    Sep 6, 2015
    Messages:
    1,095
    Likes Received:
    322
    The thing is, I've been printing since April with 0.3 layer with no issue at all. Full bed prints; no problems. So this is really a completely new problem. Thinking about it, I've never printed a full bed with either of these colours, so that could be a possible source of problem, but it's still weird. Still it's something to hold on to; perhaps 0.3 will still be OK with the materials I've printed successfully at that height.

    In any case, 0.3 is close to the rule of thumb max at 0.32, so I'll move to a new go-to layer height of 0.25 or perhaps i'll fit a 0.6 and move to 0.4. I just couldn't think why I'd suddenly have problems doing what I'd been doing. Considering the material is a good new thought and feels more illuminating than just switching to a more crushed layer height, thought I can see the thinking there too. The 4 prints in the final picture all work really well. Incidentally, it's this model:

    http://www.thingiverse.com/thing:1959745
     
    mike01hu likes this.
  9. mike01hu

    mike01hu Well-Known Member

    Joined:
    Apr 22, 2014
    Messages:
    957
    Likes Received:
    166
    Yes, I remember one of your other full bed prints and was impressed, so I was surprised by your current problem, which made me think about my issues that related to colour and respond to you; white PLA from FilaPrint is my curse filament. I like your bag clips by the way!
     
    Spoon Unit likes this.
  10. Henry feldman

    Henry feldman Well-Known Member

    Joined:
    Sep 5, 2015
    Messages:
    969
    Likes Received:
    389
    Almost all white PLA filaments suck. Not sure what the white pigment is (@Sanjay can we get a chemist's opinion here) while silver seems to adhere the best. When I absolutely need a print to work, it's silver. I found the matte-fiber HTPLA in white seems to work fine. Some of the white PLA/PHA mixes are better than straight PLA, but never as good as silver. Other colors vary.
     
    mike01hu likes this.
  11. Spoon Unit

    Spoon Unit Well-Known Member

    Joined:
    Sep 6, 2015
    Messages:
    1,095
    Likes Received:
    322
    Silver, at least if it sparkles a little, has some kind of particulate added. The E3D Everyday White seems fine though. Actually I started a print last night with 0.25 layers and again began to see delamination. Aghast, I looked down to see my 100% fan and, given the geometry I was printing, killed the fan to 15%. The remainder printed perfectly and I can clearly see how the layers change consistency and I can feel (and hear) the difference in strength. I'm beginning to wonder if my part cooling fan has sort of loosened up, or attained it's true maximum speed and is now kicking a little too much ass. I'll certainly be altering my default cooling options down somewhat.
     
  12. Spoon Unit

    Spoon Unit Well-Known Member

    Joined:
    Sep 6, 2015
    Messages:
    1,095
    Likes Received:
    322
    OK. This is miserable now. Back looking for more ideas. I really think there's a major problem here now.

    This is the current state of the LCD:

    2016-12-17 17.02.47.jpg

    And here's the mess above:

    2016-12-17 17.02.59.jpg

    As you can probably make out, things are not quite right here. Not right at all. Print settings are 0.25mm layers, nozzle temp 200. First three layers seemed to print just fine. Switched ink and left the printer at it. When I came back, that's what I saw.

    When this first began, my first guess was at a problem with Z movement. In the images above, the LCD is showing a Z of 2, but as you can see, the nozzle must be 4mm+ off the bed. My next step will be to do some basic movements with the LCD in see whether it's moving to the right position. However, I'm going to remain flumoxed that I just printed last night a pretty large item at 0.25 layers, and then today I get this.

    Intermittent issues are the worst to solve. Would love some extra eyes and brains on this. Ideas?
     
  13. mike01hu

    mike01hu Well-Known Member

    Joined:
    Apr 22, 2014
    Messages:
    957
    Likes Received:
    166
    It's almost as if the slicer is changing a parameter after your filament change. Are you running this as two processes, as this may be where the hangup occurs? This is where a video might help but with black filament it makes it tricky.
     
    Spoon Unit likes this.
  14. Spoon Unit

    Spoon Unit Well-Known Member

    Joined:
    Sep 6, 2015
    Messages:
    1,095
    Likes Received:
    322
    Thanks for joining in again Mike. I can't imagine what parameter that might be though. When I watched the nozzle immediately following ink change it laid the first layer down perfectly on the left. But it doesn't seem to have done that on the right. There are two processes, as you guessed. Both are identical apart from start and end heights.
     
  15. mike01hu

    mike01hu Well-Known Member

    Joined:
    Apr 22, 2014
    Messages:
    957
    Likes Received:
    166
    Clutching at straws here, and I know you will have done this, but I assume you have attempted to measure the overall height of the area that survived to see if there is any height variation from the indicated value. The second straw that I would pluck is the G-code at layer change to see if the Z data is actually what you expect i.e. 0.25 mm more than the last one. I think @Francesco has been having intermittent height issues too, so there may be an underlying firmware issue but he has not resolved his issue yet, so for the third straw, is the delamination error consistent in that direction i.e. all the right-hand elements at fault? If so, then the levelling algorithm may be fooled and not making the adjustments consistently, or that Z motor may not be responding correctly.
     
    Spoon Unit likes this.
  16. Spoon Unit

    Spoon Unit Well-Known Member

    Joined:
    Sep 6, 2015
    Messages:
    1,095
    Likes Received:
    322
    I've also been following @Francesco's issues. He posted in the chat box that restarting his machine and loading failsafes had some success. I'm not sure if that success will continue or if it's temporary. Meanwhile I decided to go for a full reset myself. Stop machine, remove bed, clean bed, replace bed, start machine, apply new sticker for Z stop (I'm still on the original glass), level bed, add-UHU and hairspray (I know right .. but it really seems to be good - possibly oversticky). With that all done, I ran a test single-layer print to home in the Z-Bed and then a print running. First up - 6 tiny clips - all perfect. Then a pair of eyebrows (don't ask) fine, and finally a small gun in two parts. Right up to the gun, it was all great. However during the gun print (which is tiny) I swear that layer 10 or 12 was higher than it should be and that continued for 5 layers or so, than it seemed to be fine again. Because the model is so small, the print completed and I could observe it return to normal operation on higher levels. That got me thinking toward a z-motion issue, but I've checked this out and it's completely smooth. I also ran a couple of tests this morning after the reset to check the LCD-driven Z moves to 30, 40, and 100 all moved the right amount and that was just fine.

    I didn't check the heights of that previous model. It was so far from correct I just ditched it in disgust.

    The layers in that failed gcode look like this:

    M117 Layer 1, Z = 0.25
    M117 Layer 2, Z = 0.5
    M117 Layer 3, Z = 0.75
    M117 Layer 4, Z = 1
    M600
    M117 Layer 5, Z = 1.25
    M117 Layer 6, Z = 1.5
    M117 Layer 7, Z = 1.75
    M117 Layer 8, Z = 2

    The gcode checks out too. Retractions etc all work with the layer indicated. So it's not doing anything weird. I'm not using S3Ds layer change code for the M600; that's inserted manually after the GCODE was produced.

    I only recently check the stepper motor votage, but I will have to do that again to be sure it's where it should be.

    The other thing about a Z motor issue is that it should, sooner or later, create a very unlevel bed. It's highly unlikely that both stepper motors for Z are failing in exactly the same way at the same time. So perhaps the focus moves to the Rumba board which is. I've had so little issue with the machine in general but this is now just irritating. Replacing the Rumba is one option, or possible moving to an alternate.
     
  17. mike01hu

    mike01hu Well-Known Member

    Joined:
    Apr 22, 2014
    Messages:
    957
    Likes Received:
    166
    It's frustrating, yet most of us have no issues at all. I will be investigating 32-bit systems in the future so that there is no issue with processor overload but I am not pushing the envelope for fast times and extra features. I am still on Alex's final Marlin release before he deserted us :( for Duet and have not plucked up the courage to shift up to RC8 whilst things are OK for what I do.
     
    Spoon Unit likes this.
  18. Henry feldman

    Henry feldman Well-Known Member

    Joined:
    Sep 5, 2015
    Messages:
    969
    Likes Received:
    389
    I assume you've checked that bed-level compensation is actually happening. I got something like this, and noticed that while there was a perfectly valid mesh loaded after a bed level, the z-motors were not moving as the head moved around to compensate for the mesh. Never figured it out, but it's happened 3 times, and so far the only way to make it work, is to reload firmware. Ugh.
     
    Spoon Unit likes this.
  19. Greg Holloway

    Greg Holloway Administrator
    Staff Member

    Joined:
    Sep 4, 2015
    Messages:
    565
    Likes Received:
    397
    If you have z-hop enabled try turning that off.

    Also check that Z moves as you expect it to. Move the bed down 100mm then back up 100mm and make sure it returns to the same place, try doing it a few times too.
     
  20. Spoon Unit

    Spoon Unit Well-Known Member

    Joined:
    Sep 6, 2015
    Messages:
    1,095
    Likes Received:
    322
    The old adage of "If it ain't broke, don't fix it" is not an old adage for nothing. That said, since tohara release the firmware updater I've been on RC7 + my own variable overrides and this has got me to a point of feeling very comfortable that I can try something new, yet get back to the same thing that was working with no hassle. If you haven't tried that tool, it's really worth it as it free you from being on a fixed codeline that's not being updated and give you the comfort to change variables as required. Obviously, that still requires that you get it lined up first and the variable values that work for you will probably be different to others. However, if you know you can instantly get back to that previous release from Alex, it's really just a question of whether you want to spend the time on this over other activities.

    I haven't, partly because that particular problem hasn't been a problem for me since I moved off RC6. I had that some problem and it was incredibly frustrating. At this point, I can happily turn the machine off, come back to it, turn on, and know that the mesh is there and it's ready to go - current issues aside. Still, it's a good point and worth a double check.

    No idea what that is. Is that just the z movement you can setup after a retraction? I do use that, and there's no way a PLA print would work without it. However, I do recall someone spotting that this was required in fact for an Edge print to get rid of stringiness.

    Yep, did that one. Not 5 times in a row though for example. That's worth doing to check repeatability though. Will be done.
     

Share This Page