I haven't used that library, but a 2.5mm pitch sounds pretty fine for a 20mm diameter thread. I'm not sure what thread angle the library uses, but a coarser thread pitch helps to mitigate the overhang angle created by a steep thread angle. That might be part of your issue with fit on the printed parts.
On Feb 20, 2021, 12:02 -0800, Roel Vanhout roel.vanhout@gmail.com, wrote:

On 2021-02-20 21:01, Roel Vanhout wrote:

Things like this also depends on the slicer settings and the printing
process, so I don't think there is a universal answer other than you
must combine the software parameters with your printing slicer/printing
process to make it work. If you print with a large extrusion with you
will need a larger tolerance between the parts. With less extrusion
with/layer thickness the tolerance requirement may change.

Years ago I made a test model to learn about making threaded nuts and
bolts fit each other
https://www.thingiverse.com/thing:1445033

Carsten Arnholm

I have had good results using https://dkprojects.net/openscad-threads/ https://dkprojects.net/openscad-threads/  for the bezels on my instrument dial gauges (think SLR camera filters)

First time out it took a few tries to get the right tolerance then I ran into two simple problems- 1- there was too much play in one of my carriages resulting in a slightly eccentric thread and 2 - PLA on PLA has a high friction coefficient - so spray the threads with ptfe lubricant before trying to thread them together -

First time using them it may take a few back and forths to knock off any blobs of PLA, and ease the pointy curves ( unless you have arc support enabled in both Marlin and your slicer (or post process with arcwelder) then you will be printing circles using a lot of small straight lines with pointed joints).

Unfortunately threads are precision bits and so you need to re-calibrate every time you change filament or modify your slicer print settings.

Good luck- it is very satisfying when you get a good print.

cheers
Bob

Roel Vanhout wrote

You got this right. But the required slop depends on your thread, the
filament, the printer and the print settings. Instead of increasing the
inner diameter of the nut, you can increase the slop (default value is 0.2
[1]).
The size of of the inner diameter of the nut is calculated as follows:

[1] https://github.com/revarbat/BOSL/blob/master/constants.scad#L41
[2a] https://github.com/revarbat/BOSL/blob/master/threading.scad#L286
[2b] https://github.com/revarbat/BOSL/blob/master/threading.scad#L99

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

Thanks all. I did realize that exact fit depends on slicer settings &
printer calibration/setup, but in my experience I can download pretty much
any properly designed stl with thread from Thingiverse and I can print
working thread, at any reasonable layer thickness, even if some are a bit
more loose or tight than others.

Anyway with some test designs and cut throughs in OpenSCAD, and a range of
overnight test prints at various diameters and slop settings, I can offer
the following experimental conclusions/guidelines to anyone stumbling upon
this thread at some point in the future:

• Indeed the d of the rod and the id of the nut should be the same and fit
  can be calibrated using the nuts 'slop' parameter, thanks to fe60 for
  confirming
• Slop setting changes of 0.05 have noticeable effects ('measured' by hand
  twisting the nut onto the rod, and for prints at layer heights 1.2 and 2.4
  using PLA+); for a rod d=20, 0.3 is rather too tight, 0.4 works but is
  quite loose and 0.35 gives a snug fit. So any trial prints should go down
  to that level of detail at least.
• For rods d=20, the default slop of 0.2 is not enough (it's a bit much for
  diameters < 12, ok up to 16, not enough above). A good rule of thumb seems
  to be to take 15% of the diameter and round up for 'large' diameters and
  down for 'smaller' (sometimes requiring some adjustment still like my d=20
  -> 0.35). I didn't do an exhaustive study on this, but it seemed to work
  more often than not for the trial prints I did. Maybe it'll save someone a
  few test prints some time.

cheers,

roel

On Sun, Feb 21, 2021 at 1:36 PM fe60 fe_60@web.de wrote:

With my Prusa MK3 I find that in general, $slop values need to be around 0.05
for things to fit well.  The default of 0.2 is much too loose.  I use BOSL2
and am not sure how things might have changed with the threaded rod code
from BOSL, but there is new screw code in BOSL2 that supports standard
machine screws with standard screw tolerances, and in that case I have
printed parts with  $slop=0 that will fit onto standard metal parts or onto
each other, depending on the tolerance settings.  However my testing with
threading has been mostly with small sized parts like 1/4 inch or M6.
However, I would not expect this tolerance to change with a larger part.
If you print a square part 100x100 and measure with calipers, how accurate
is it?

I did recently print pipe threads at 3/4 inch and for that part I found
$slop=0.1 gave me a female part that my (mass produced, not printed) pipe
part threaded in very far (more than is probably normal), and $slop=0.05
gave a possibly normal fit, and $slop=0 a very tight fit (but still possible
to start engagement).

Roel Vanhout wrote

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

How do you print a go/no-go gauge when your printer might be overextruding
and printing inaccurately?  It seems like you need to get such a gauge from
another source with guaranteed precision, and then you can use it to test
printed parts.

I personally tested printed screws by seeing if the mated well with metal
parts of the same dimension.

fouroh-llc wrote

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