Identical feature printing diffidently on each end

Discussion in 'Materials' started by tebee, Feb 23, 2011.

  1. tebee
    tebee Well-Known Member
    I've just got back a model of a locomotive I had printed, and while I'm generally pleased with it the radiator details nave printed out incorrectly.

    The strange thing is it has an identical radiator at each end but both have printed differently. It should be slightly recessed.

    On end 1 it's slightly raised , though there is a recessed line marking the edge [​IMG]

    On end 2 it far too deeply recessed
    [​IMG]

    original STL file attached
     

    Attached Files:

  2. GWMT
    GWMT Active Member
    Very nice model tebee! Is that in HOn30? What is the diameter and height of the rivets on the cab?

    I wonder if the problem occured because the radiator faces didn't line up exactly with a material layer for printing so the machine 'rounded' the radiator surface up or down to the nearest matching material layer (assuming the shell was printed from left to right as shown in your pictures).

    I 'm new to Shapeways and have only received one print so far; that print had a tiny error on one face in the X-Y plane. Nothing printed on that face where it intersected a perpendicular face, leaving a shallow recess the width of the perpendicular face one print layer deep (0.2mm deep by 0.7mm across).

    A question for Shapeways staff: what is the exact thickness (no rounding off) of the print layers for White Stong & Flexible material? If we design our models in multiples of that thickness everything should print exactly as it was drawn.

     
  3. Youknowwho4eva
    Youknowwho4eva Well-Known Member
    Interesting, do you have the model public? If not, do you have a model we can look at? Have you viewed your upload in 3D to see if perhaps it was a conversion or upload that caused a change? My hypothesis is, if you subtracted the recessed area, an invisible inverted shell could have remained.

    As far as GWMT's question about designing to the print layers, that's an idea. The problem is 1) shapeways doesn't currently let you choose the print orientation, and 2) unless the model is lined up perfectly, all the work of building it per build layer would be for nothing. As far as the thickness being rounded, it's impossible to give an exact thickness. For wsf, each layer is spread on, and fused. The layers consist of granules of material that won't ever lay exactly the same every time. It would be like trying to make a perfectly level bed of sand (larger granules).
     
  4. tebee
    tebee Well-Known Member
    The model is attached to the first post - viewing it in Netfabb it looks right - the radiators are recessed the same both ends and there is a simulated mesh.
    I was quite prepared for the mesh not to appear here as it's on Shapeways minimum detail level, but I'm trying the same model somewhere else were it should print.

    Yes it's HOn30.

    The rivet are 0.25mm hemispheres, but the ones that where the same size on the side of the cab did not print
     
  5. GWMT
    GWMT Active Member
    Here's a screen shot of the STL tebee posted showing the rad grill:
    tebee Loco Face.jpg

    I've been designing models using a 0.007" x 0.007" x 0.007" grid (0.2mm x 0.2mm x 0.2mm) based on info from Dimension ( http://www.dimensionprinting.com/3d-printers/printing-produc tspecs-elite.aspx ). At its finest setting Sandstone (Z Corp) prints on a 0.0035" x 0.0035" x 0.0035" grid ( http://www.zcorp.com/en/Products/3D-Printers/ZPrinter-650/sp age.aspx ).

    I try to fix the model in relation to the Origin Point (in Solidworks) so all surfaces are some whole multiple of 0.007" away from the Origin Point. I assume the 3D printer software requires the operator to place the origin point of your model within the build volume at a grid point based on the minimum layer thickness of the material used - is this true?
     
  6. tebee
    tebee Well-Known Member
    The front of the grill should be recessed 0.48mm and the grill is 0.22mm thick. The grill consists of 0.3mm square holes surrounded by 0.2 mm wide bars if you see what I mean.

    I've just dug my vernier out and on end 2 above it's recessed 0.67 mm so it's conceivable the grill just hasn't printed at that end but that does not explain the protrusion on the other.

    I've been checking through and I've got a couple of other models in the same batch were things don't seem to have printed the right depth/thichness , though they are not as noticeable as this one.
     
  7. stop4stuff
    stop4stuff Well-Known Member
    There's a lot of information about what's possible, what's not and deviations on the Design rules & detail resolution for SLS 3D printing page.

    I did take a look at your model there tebee, in AccuTrans there were inverted normals and in NetFabb there were numerous holes and a overall negative volume too. Perhaps in fixing the model, Shapeways' MeshMedic created more issues?

     
  8. GWMT
    GWMT Active Member
    Thanks, shop4stuff! I haven't looked at that page in many months :blush: - it's bookmarked now.

    From the page https://www.shapeways.com/tutorials/design_rules_for_3d_print ing the layer thickness for White, Strong & Flexible is 0.15mm (0.0059").

    I'm going to design my next model on a 0.15mm x 0.15mm x 0.15mm grid and see what that does.

     
  9. tebee
    tebee Well-Known Member
    OK if it's on a 0.15 mm grid that could sort of explain the one that's too deep , but the one then is slightly extended instead of being recessed is more than .48mm out so it would have to have jumped 3 grids.

    So that still leaves me with the question - what went wrong?
     
  10. GWMT
    GWMT Active Member
    stop4stuff said he opened your STL in AccuTrans and saw some inverted normals in the model. If they were in the grill area could Shapeway's MeshMedic have gotten confused and extruded the grill recess and mesh outwards rather than inwards at one end and cut the grill mesh deeper into the rad recess instead of extruding it outward on the other?

    Maybe correcting all the normals so they point in the same direction will fix the print problem.
     
  11. stop4stuff
    stop4stuff Well-Known Member
    It wasn't so much the inverted normals, but more on the normals along with the holes in netfabb side of things.

    Having said that, Tebee, can you identify the print orientation from your model?
    I ask because it could be that the area and deviation of depth in the radiator regions were printed horizontally (as the printer lays and sinters the material) there's a bit of sag going on. Does the roof look like it has layers from side to side or lengthways?
     
  12. tebee
    tebee Well-Known Member
    It's layered across, from side to side - so sag sounds a reasonable explanation.Next question is that my fault or Shapways , what is the solution?
     
  13. stop4stuff
    stop4stuff Well-Known Member
    TBH, I couldn't say who the fault lays with. Did you ask the Shapeways people about your model yet? (service@shapeways.com)