I assume it's slow?

First, you need to translate the differenced cylinders so that they start below the cylinder you're punching holes in, and make them longer so that they poke out part the top.
That should make it render faster

With a FOR loop you only have one set of commands, repated as many times you want around the rim.
(Sorry, don't have time to check the correect syntax right now. I have a CAKE that needs to be eaten... )

And finally, to clean it up a bit, use indentation.

ROTATE ()
TRANSLATE ()
OBJECT();

Notice that there's no need to use the curly brackets because both ROTATE and TRANSLATE expect an object of some sort to follow.

Trygve

Den 28. august 2020 kl. 14.23.47 +02.00 skrev AndyY via Discuss discuss@lists.openscad.org:

It is faster to do in 2D and then extrude to 3D

diameter = 34.0;
linear_extrude(10)
difference() {
circle(d = diameter + 10);

    for(i = [-1 : 1])
       rotate(i * 120)
            translate([0, diameter / 2])
                circle(d = 6);
}

On Fri, 28 Aug 2020 at 14:02, trygve@totallytrygve.com wrote:

Very slick.

On 2020-08-28 9:17 a.m., nop head wrote:

--
Ron Wheeler
Artifact Software
438-345-3369
rwheeler@artifact-software.com

AndyY:

Nophead and Trygve gave some good hints for improvement, but your
original really wasn't all that bad.

What was it that was concerning you about your result?

The only thing I see to complain about in the render that you got is
that the punched-out circles are decagons, and the big circle is still
pretty obviously a polygon.  (It's a 30-gon.)

$fa, $fs, and $fn control the number of sides used for a circle.
https://en.wikibooks.org/wiki/OpenSCAD_User_Manual/Other_Language_Features#$fa,$fs_and$fn

I'd try

$fs = 1;
$fa = 8;

to get a better approximation without going wild.  Mostly, $fs will
control the small circles and $fa will control the large one.

The render tends to make circles look more polygonal than they really
end up.  Let's do some math.

a_based=360/$fa;
s_based=r*2*PI/$fs;
n=($fn>0?($fn>=3?$fn:3):ceil(max(min(a_based,s_based),5)));

OpenSCAD "circle" radii are measured from the center to the vertices. 
The distance from the center to the middle of an edge is r*cos(360/n/2).

For the default values ($fs=2, $fa=12):

Sides
Vertices

Diameter
Edges
Diameter
Error
Error
Percent
Outer
30
34
33.81
-0.19
-0.5%
Inner
10
6
5.71
-0.29
-4.9%

With $fs=1 and $fa=8:

Sides
Vertices

Diameter
Edges
Diameter
Error
Error
Percent
Outer
45
34
33.92
-0.08
-0.2%
Inner
19
6
5.92
-0.08
-1.4%

If you want to reproduce those numbers:

$fs = 1;
$fa = 8;
function v2e(r, n) = r * cos(360/n/2);
for (d=[34,6]) {
    r = d/2;
    a_based = 360/$fa;
    s_based = r*2*PI/$fs;
    n = ($fn>0?($fn>=3?$fn:3):ceil(max(min(a_based,s_based),5)));
    edges = v2e(r, n);
    error = edges-r;
    percent = error / r;
    echo(n=n, vertices=r*2, edges=edges*2, error=error*2, percent=percent*100);
}

Note that the *2 in the final echo is because the calculations are in
terms of radius and you specified diameter.

Hi,

Referring to Andy'Y's original solution, I thought I'd have a look at
it. Bear in mind that I only discovered Openscad a month or so ago, I
found the sort of backwards progression a bit of a nuisance. From here,
I found AngelCad, and thought I'd attempt the 'flange' using that
scripting language, which for me, I can probably still read and
understand in a month or two's time.

double diameter = 34;

solid@ disc()
{
   return cylinder(10,(diameter+10)/2);
}

solid@ hole()
{
  return  cylinder(10,3);
}

solid@ threeholes()
{
   solid@ holer = translate(diameter/2,0,0)*hole();
   return  holer  + rotate_z(120)*holer + rotate_z(-120)*holer;
}

solid@ all()
{
   return disc()-hole()-threeholes();
}

void main()
{
   shape@ obj = all();
   obj.write_xcsg(GetInputFullPath(),secant_tolerance:-.01);
}

To me, that is quite a logical approach - make the disc, make a hole,
make the three holes positioned where you want, then drill 'em in the
disc. A straight forward sequence, that would be as you machined the
solid. By adjusting the secant tolerance, it regulates the number of
steps in forming a circle, for example.

Where's the naughty step?

The code I posted was via gmail and I simply set the font to Fixed Width to
preserve the formatting.

On Fri, 28 Aug 2020 at 20:12, AndyY via Discuss discuss@lists.openscad.org
wrote: