perimeter separating

I have been experimenting with cookie cutters, trying to find the ideal balance between wall thickness and strength.
So I made a test object (in Rhino) consisting of 5 intersecting thin-walled objects (very much like the .5mm thin wall calibration object). These objects vary in wall thickness from .4mm (my nozzle size) to 1.2 mm. I set slicer to 100% infill, layer height .25mm. For good measure I also set solid infil every layer. I'm printing PLA at 205C with a E3D hot-end, when I go lower the layers don't stick together well enough.
The .4mm thick wall measures .5mm after print and is build from one perimeter. the .6 is .8, also one perimeter. The .8mm is 1 mm, still one perimeter.
The problem begins with the 1 mm thick wall. This is build from two perimeters, but these do not seem to be connected. When I check my gcode at gcode.ws it shows infill between the two walls, and while printing the head goes over the area three times, but there is a gap between the two. I can easily put my fingernail inbetween the two perimeters and when I try to bend the object the two wall parts come apart.
I figured maybe my nozzle is less than .4 mm causing the perimeters to turn out narrower than they should be, so I created new gcode with the nozzle width set to .35. Now the .8mm wall is build from two perimeters, still not connected...
My extruder is calibrated, extruding 50 mm. of filament results in 50mm plus or minus .1 mm. so I don't think that can cause this problem.

Any ideas?

Interesting observation, if I print a solo 1mm. thick wall square the perimeters are fused together nicely, no matter what settings I use in slic3r. The wall is 1.5 mm. thick though where it should be only 1mm....

Adjust your slicer settings so that you get 0.7mm wall thickness for 0.4mm layer height. It could be that you have too high a value in the filament thickness setting. Not familiar with Slic3r enough to tell you where that is.

what you are seeing is backlash. because those walls are approached from a different direction, the lash in your system leaves a gap between them. I suggest you attempt to adjust your mechanics (tighten belts etc). you are using machined pulleys? if not, do. and if all else fails compensate for lash in your slicing.

Been away for a while, sorry for the late response. I adjusted Slic3r settings and my prints are much better now.
Thanks again,

Arjen

Replace `2d_shape` with your cookie cutter shape. I could be set with `import`, but afaik it just works on .dxf(well for 2d) Dont know how to convert .png etc to .dxf decently, inkscape? (no SAAS allowed for me, and i kindah just want it via commandline)

h=10;
w=40; h=40;
t = 1;
bt = 4;
bh=2;

module 2d_shape()
{   square([w,h],center=true);
}

module shape_to_wall(t=t,h=h)
{

    linear_extrude(height=h) difference()
    {   minkowski(){ 2d_shape(); circle(t); }
        2d_shape();
    }
}

module shape_cookie_cutter(t=t,bt=bt)
{   union()
    {   shape_to_wall(t,h);
        shape_to_wall(bt,bh);
    }
}

shape_cookie_cutter();

Would kindah be handy to be able to pass a module as an argument here maybe i should try use ImplicitCAD, but as haskell this time. (Openscad making its own language might well have been a bad idea. Well, it being c-like and not java helps with popularity, and i dont know what options they really had.)

What ended up being the solution to your issue? I'm currently facing the same issue where my perimeters aren't properly fusing together and the parts are printing out ever so slightly larger than they should be when measured with a micrometer. I ran out of ideas and google led me to this thread.

Lowering the various extrusion widths (In particular `--solid-infill-extrusion-width`, `--first-layer-extrusion-width`) can help. Basically it makes more thinner lines, so it might more quickly put a bit of solid infill in between.

> When I check my gcode at gcode.ws it shows infill between the two walls, and while printing the head goes over the area three times, but there is a gap between the two.

This seems a bit strange to me though, you'd say with backlash you would see evidence of some print moves between the layers? It could be a plain move(non-extruding move) gcode, in particular if `--avoid-crossing-perimeters` was enabled.(In which case it is more likely to follow the profile)

(attachment is in case people dont feel like using openscad just to see what i posted)

Poked around with some of the suggestions in here last night and I think Greg may be at least partially right about backlash. Going to chase down any slop in my mechanics before I mess around with any more slicing or firmware settings.

Thanks