I personally would do it using the BOSL2 library, which provides chamfered
cylinders.  Or you could write your own chamfered cylinder if you want to
work more.

https://github.com/revarbat/BOSL2/wiki

include<BOSL2/std.scad>
$fn=64;

ChipBase=44;
x=3 ;
y=3;

module Chip() {
difference() {
cube([ChipBase,ChipBase,5]);
//translate ([ChipBase/2,ChipBase/2,1.7])

translate([ChipBase/2,ChipBase/2,1.7])cyl(r=20.30,h=4,chamfer2=-2,anchor=BOTTOM);
translate([ChipBase/2, ChipBase/2, -0.5]) cylinder (r=10, h=3);
}
}

for(i = [0:ChipBase:(xChipBase)-1]){
for(j = [0:ChipBase:(yChipBase)-1]){
translate([i,j,0]) Chip();
}
}

SUggy62 wrote

--
Sent from: http://forum.openscad.org/

On 2/7/2020 8:50 PM, lar3ry wrote:

Yep, though I thought in terms of a cone rather than a rotate_extrude of
a triangle.

Also I couldn't resist factoring the translate-to-center out.  Nor could
I resist extracting all of the variables out.  Note that we don't need
separate variables for the thickness of the negative cylinders, because
they only have to reach above the base; the thickness of the base (+2
for the through hole) is plenty.

$fn=64;

ChipBase=44;
x=3;
y=3;
thickness = 5;      // Base thickness
thickness2 = 1.7;  // thickness inside chip hole
chip_r = 20.3;      // Outer hole
hole_r = 10;        // Inner hole
cham_h = 1;        // Height of the chamfer
cham_r = 1;        // Radius (in addition to the hole) of the chamfer

module Chip() {
difference() {
cube([ChipBase, ChipBase, thickness]);
translate([ChipBase/2, ChipBase/2, 0]) {
translate ([0, 0, thickness2])
cylinder (r=chip_r,h=thickness);
// -1 and +2 are to avoid Z-fighting.
translate([0, 0, -1])
cylinder (r=hole_r, h=thickness+2);
translate ([0, 0, thickness])
chamfer();
}
}
}

// z=0 is at the top of the hole, so we generate a chamfer down from that.
// Note that this module controls all of the geometry of the chamfer.
module chamfer() {
translate([0,0,-cham_h]) {
// cham_h2 gets us out of Z-fighting.  cham_r2 keeps the
// angles the same.
cylinder(h=cham_h2, r1=chip_r, r2=chip_r + cham_r2);
}
}

for(i = [0:ChipBase:(xChipBase)-1]){
for(j = [0:ChipBase:(yChipBase)-1]){
translate([i,j,0]) Chip();
}
}

On 08.02.20 06:20, Jordan Brown wrote:

I usually don't bother with the adjustments by making the
cylinder big enough:

cylinder(r = hole_r, h = thickness * 3, center = true);

ciao,
Torsten.

Chamfers in BOSL2 are actually fairly convenient in this case.  A cyl() with a negative chamfer2= value makes an external top chamfer mask.

include <BOSL2/std.scad>

$fn=64;
x=3;
y=3;
ChipBase=44;
thickness = 5;      // Base thickness
thickness2 = 1.7;  // thickness inside chip hole
chip_r = 20.3;      // Outer hole
hole_r = 10;        // Inner hole
cham_h = 1;        // Height of the chamfer

grid2d(spacing=ChipBase, cols=x, rows=y) {
difference() {
cube([ChipBase+0.01,ChipBase+0.01,thickness], anchor=BOT);
up(thickness2) cyl(r=chip_r, h=thickness-thickness2+0.01, chamfer2=-cham_h, anchor=BOT);
cyl(r=hole_r, h=thickness*3);
}
}

RevarBat wrote

Tried that, and got:

ERROR: Assertion 'false' failed: "Bad arguments." in file
../Users/Admin/Documents/OpenSCAD/libraries/BOSL2/vectors.scad, line 120
TRACE: called by 'vector_angle', in file
../Users/Admin/Documents/OpenSCAD/libraries/BOSL2/coords.scad, line 151.
TRACE: called by 'rotate_points3d', in file
../Users/Admin/Documents/OpenSCAD/libraries/BOSL2/attachments.scad, line
156.
TRACE: called by 'find_anchor', in file
../Users/Admin/Documents/OpenSCAD/libraries/BOSL2/attachments.scad, line
253.
TRACE: called by 'orient_and_anchor', in file
../Users/Admin/Documents/OpenSCAD/libraries/BOSL2/transforms.scad, line
1053.
TRACE: called by 'if', in file
../Users/Admin/Documents/OpenSCAD/libraries/BOSL2/transforms.scad, line
1052.
TRACE: called by 'if', in file
../Users/Admin/Documents/OpenSCAD/libraries/BOSL2/transforms.scad, line
1033.
TRACE: called by 'grid2d', in file OpenSCAD, line 13.

--
Sent from: http://forum.openscad.org/

It's possible that you have an old version of BOSL2.  It's still alpha code, and I've been actively reworking it for months now. I suppose I shouldn't really be referencing it yet, but it's already so much nicer than BOSL1 that I just don't use BOSL1 anymore.

• Revar