example 1:
[image: Screenshot 2023-04-03 at 7.39.18 AM.png]

[image: Screenshot 2023-04-03 at 8.41.00 PM.png]

example2:
[image: Screenshot 2023-04-03 at 8.46.06 PM.png]

[image: Screenshot 2023-04-03 at 8.46.51 PM.png]

example3:
[image: Screenshot 2023-04-03 at 8.48.36 PM.png]

[image: Screenshot 2023-04-03 at 8.49.10 PM.png]

In the example 1, the shape seems to be distorted, but it is still
consistent with the 3 vectors. the twist in this example can be corrected,
by aligning the section and the slices can be removed

[image: Screenshot 2023-04-03 at 8.54.47 PM.png]

On Mon, 3 Apr 2023 at 13:22, Sanjeev Prabhakar sprabhakar2006@gmail.com
wrote:

I am not sure that it makes sense to say you can "separate the orientation
problem from the path extrusion problem".    When I wrote the code in BOSL2
I followed the approach from list-comprehension-demos, a library that was
written in 2015 when list comprehensions were first introduced.  The core
operation there, which they (and BOSL2) call "sweep" is to take a 2d
profile and a list of transformation matrices.  Then you write
sweep(profile, transform_list) and an object is created by linking together
successive transformed copies of the profile.

This is a very powerful framework.  The transformations can potentially do
anything (that gives you a valid polyhedron at the end) so you can change
the scale, you can add twist if you want, or really rotate any direction,
and you can translate around.  This sweep() function is also pretty easy
to write.  It seems like if OpenSCAD were to add a native function in this
general problem area, this is the one that should be added.  If we had
this form of generalized sweep but it could handle self-intersections that
would be great.

The path extrusion problem is a special case of this more general problem,
but what distinguishes the path extrusion problem is that you need to
construct the set of tangents and normals to the path.  This is the
orientation problem, and I think that the path_extrusion problem IS the
orientation problem, so I disagree that you can separate path extrusion
from orientation.  That a path in OpenSCAD is discrete really doesn't have
a huge consequence here.  You can estimate the tangent and normal from a
discrete path.  The huge complication is that the problem is ill defined,
so you may not know what the normal should be.  If you don't know what the
normal should be...you're going to have a hard time calculating it.

So yes, you can toss that ill-definition off to the user and force the user
to supply the normals.  That makes a lot of work for the user.  How is the
user to manage this?  And what is gained by separating it as an extra step
for the user?  It seems to me like it's a complication, and deterrent to
using the code.  I submit that the user usually doesn't want to bother with
having to figure out how to construct the orientation and would rather have
the path_sweep module do it.  In BOSL2 I do give you the option of
actually specifying the normals yourself.  There is an example of using
this feature to extrude a shape on the surface of a knot.  But in truth,
the user wants help making these normals, and that is the job of the
path_sweep() operation in BOSL2.  Basically what path_sweep does in BOSL2
is it computes the tangents, somehow constructs normals (depending on the
selected method), applies user specified twist, and if necessary, adjusts
twist so that closed paths meet up properly.  Then it passes that info to
sweep().

On Mon, Apr 3, 2023 at 9:31 AM Rogier Wolff R.E.Wolff@bitwizard.nl wrote:

If I want to create an stl file, say, of a nut and bolt, then I use
NopHead's library. But generally i use purchased hardware, and only need
to create the parts to fit them to, which is much simpler than fiddling
with thread profiles in 3d printing.

Hexagon heads /sockets created by a cylinder with $fn=6. Plenty of
charts available for the standard sizes, just allow something for
fitting tolerances, if inserting nuts. But, use insets, or self tappers.
The standard machine screw threads are not designed for soft materials.
(assuming the end product is 3d printed, not machined from solid)

On 04/04/2023 10:32, Karl Exler wrote:

Why do libraries make you really nervous?  BOSL2 can create nuts and bolts
with a single library.  If you want to use list-comprehension-demos you
need a version that's been patched to work with recent copies of OpenSCAD.
Your error message above suggests that your OpenSCAD might be too old for
the library you're trying to use.  (The line 507 looks OK, but has a
function literal, which I think requires 2021.01.

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

Another solution is threadlib:  https://github.com/adrianschlatter/threadlib,
which also depends on several other things, but not agent-scad

On Tue, Apr 4, 2023 at 5:32 AM Karl Exler karl.exler@meinklang.cc wrote:

The line you showed has a function literal in it, so you need at least the
latest release of OpenSCAD because it is a fairly recent addition.

On Tue, 4 Apr 2023 at 11:17, Adrian Mariano avm4@cornell.edu wrote:

Dear Gentlemen
Due to your help I made big advances yesterday and I produced my first
set of screw and nut today: (jupidoo)

threaded_rod(d = 13.65, l = 12.45, pitch = 7.82/4, anchor=BOT, bevel =
true, $fn = 64);
threaded_nut(nutwidth=17, anchor=BOT, id=13.65, h=5, pitch=7.82/4,
$slop=0.05, $fa=1, $fs=1);

And -as expected- it doesn work at the first time. It is really perfect,
but too narrow.

Is it d and id?? I took the same values, but I guess, that was not so
good. Is there a factor with whoch I have to make the "id" smaller than
"d" ?
Many thanks
Karl

Am 04.04.23 um 13:12 schrieb nop head:

I often include a variable called "slop" which I set to 0.1 or 0.2 in
order to get printed parts to fit correctly.  I subtract it from ODs and
add it to IDs.

Jon

On 4/5/2023 3:12 AM, Karl Exler wrote:

You have not given any allowance of space between the parts.  Real screw
standards will specify that the rod is undersized compared to nominal
diameter to allow for clearance so that the screw can actually fit
together.  If your printer was VERY accurate you might get away with the
.05 allowance you're introducing using the $slop parameter, maybe.  But you
might try printing test pieces that are simple cylinders and holes to get a
feeling for what kind of clearance you need.  I'm not quite sure what
you're trying to make, but it looks sort of like an oversized 1/2-13 bolt,
so if a 1/2-13 would suffice you could also try screw("1/2-13,1/2") which
might get you off to a better start.  If you set the tolerance to the
loosest option (1A for bolts, 1B for nuts) and a $slop suitable for your
printer the parts should fit easily.  (The loose tolerance is in fact loose
enough that I believe on my MK3S, the parts fit with a $slop of zero.)

Note that clearance and slop are meant to be two different concepts:
clearance is the actual space between parts, whereas slop is an adjustment
for inaccuracy of the printer.  Correct setting of $slop is intended to
produce parts that are the exactly specified size.  But you can abuse $slop
to introduce clearance so the parts fit.  If you want to do that, probably
trying 0.2 is a reasonable starting point.

On Wed, Apr 5, 2023 at 3:12 AM Karl Exler karl.exler@meinklang.cc wrote:

Dear Adrian
By now I am not making anything. I learn and try and so I have my
troubles. But many thanks to your help.

1. I would prefer to use the screw function.. but this ends at M12.. And
   I would like to make larger threads and nuts

2. so, due my humble understanding I have to use the "threaded_rod"
   function :-( About that I am very sad, as e.g. the tolerance, the head
   and some other parameters do not exist for this function. Please peg me,
   if I use wrong termini technici...

3. I will use my tools only for 3D-Printing and there I only use PLA or
   PETG. No massive force, only bla bla to connect quick and easy. High
   pitch, big diameter, sometimes with a hole, sometimes with an head -
   very very less tolerance.

What does the community thing about this demands?
Thanks
Karl

Am 05.04.23 um 15:20 schrieb Adrian Mariano: