One dxf, two cads see different things -hiw to resolve?

[Buntehuhn]

Kindly requesting your help to use QCAD to resolve this issue.

Situation:
I want to lasercut parts for a model glider, from 2mm plywood, at the fablab nearby. The fablab accepts dxf-files as input, the fablab's laser-software is then used to create gcode.
The design is created in a specialized model plane 3D cad program, devwing, and exported as a dxf-file.
With QCAD I've opened the devcad-dxf and saved the parts (two ribs) to be cut from 2mm plywood in a separate dxf-file, for the fablab to check if it is suitable for the laser-software. They check that using CorelDraw, I think version 2007 or 2017; anyways, a very much not recent version, but definitely suitable for their compatabilty-check.
I run QCAD Professional 3.32.3.0 (3.32.3) on W10 Pro.

The issue
The file I supply, the attached Amokka Fin geodesie 5_1_Triplex2mm, contains two ribs, with the top one containing two round circles (that will become holes)
The fablab sees not those round circles, but sees each circle as two overlapping ellipses: see attached image What fablab sees_Amokka Finparts triplex2mm.

What I want the fablab to see, are the two round circles.

My question:
1. How can I make those ellipses in the file that I supply, visible in QCAD?

Kind regards,
Arjan

[Husky]

Hi Buntehuhn - welcome to the Qcad forum.

Your drawing was among other things edited with 3D. To be sure that everything is now 2D, I recommend first converting the drawing in Qcad to 2D.
Load the drawing and run ... "Flatten drawing to 2D": Menu ... Misc ... Modify ... Flatten Drawing to 2D.
... just to be on the safe side

Maybe this will already solve your problem - if not please feel free to ask again. Thank you.

[CVH]

Hi, and welcome to the QCAD forum.

IMHO there are no Z coordinates different from zero.
Flattening a drawing that is already 2D has no further effect.
It won't solve any of your issues.

The fablab sees not those round circles, but sees each circle as two overlapping ellipses:

The QCAD representation is perfect.
Model_Space shows 6 entities in total: 2 ribs outlines, 2 additional 'circles' and 2 text entities.

The top of 2 ribs is a Block Reference based on Block '*R20' what already includes 2 holes.
The leftmost hole has a duplicate: A Block Reference based on Block '*R21'.
The rightmost hole has also a duplicate: A Block Reference based on Block '*R22'.
When selecting entities that are superimposed try selecting with holding down the ALT key.

You should already fix Block names because starting with an asterisk is an invalid name (= Anonymous block).

• QCAD Pro .. Misc .. Block .. Fix Block Names

You could consider to remove the (almost) duplicates.
You could also consider to replace the circular shapes by real circles ...
... If you would be able to locate the correct center (See lower).
Eventually exploded to Polylines or exploded further into 2 Arcs by QCAD if that is required.
I would expect more holes to make the profiles even lighter.


In all 3 Blocks a circle is represented with a polyline with 2 semicircular segments.
'Arc like segments' or better 'Bulging segments' because one can convert them in both ways but never exact.
Bulging segments have end-points, Arcs have end-angles, an Arc has a center, a Polyline segment has none.
(Before anyone disagrees: QCAD explodes polylines internally for snapping to centers of Arc like segments for example)
For a perfect semicircle we expect them to have a bulge factor of 1.000...

Comes from bulge = tan(sweep/4) => tan(180/4) = 1.0

A fully circular 'Arc like segment' is not possible: tan(360/4) = tan(90) = Error (= infinite)
QCAD automatically prevents that on exploding a circle, on converting a circle to a polyline, ... , on saving and on loading.

I suspect that these polylines are not generated by QCAD:
- Would be a logical closed Polyline with 2 vertices, not as open with 3 vertices.
- The 2 vertices would be at 0 and 180 degrees.
- Both the bulge factors would be 1.000 exactly.

Why fablab displays them incorrect I can not tell and is outside the QCAD scope.
It has to do with the bulge factors and the orientation.
The third vertices typically has a bulge factor equal to the second.
So said 'circles' in Block '*R20' are drawn CW and in Block '*R21' and '*R22' they are drawn CCW.

Exploding the 'circular' polylines to Arcs reveals:
- Top Arcs in Block '*R20' have a sweep of -174.85847228
- Bottom Arcs in Block '*R20' have a sweep of -185.14143425
- Top and bottom arc of one so said circle don't have the exact same center, the radius is almost 2.5 but not exact and not equal.
- Both Arcs making up one (almost) circle don't connect tangentially.
Similar for Block '*R21' and Block '*R22' but then with positive sweeps.


The application that generated these polylines from circles did not a great job.

The same is true for the outlines of the ribs what probably were Splines at some point.
Stepping through the vertices of the top profile reveals connecting angles varying between 179.82747097 and 181.29127169 degrees.
Meaning that the transition from one Arc to the next is not smooth, your profiles have dents and kinks.
More pronounced at the front tip where small Arcs are replaced by short but straight segments.
The bottom profile then uses a small Arc at the tip: index 1 > about 185 degrees and that is a visual kink, index 16 > about 176.1 degrees.

QCAD explodes Spline to Polylines with tangentially connected Arc like segments.
Perhaps it is an idea to do all the conversion to polylines in QCAD.
Then you could avoid to explode Circles to Polylines, see: Application Preferences .. Modify .. Explode.
Remind that the stricter the exploding tolerance for Splines, the more Arc like segments are created.

Regards,
CVH

[CVH]

PS:
The QCAD book is not really the manual of QCAD.

The title is "An Introduction to Computer-Aided Design" and that at the hand of QCAD.
It does not cover all QCAD drawing methods.

It was issued in 2008.
QCAD is a living entity, updated and enhanced on regular base.

Best option that covers (almost) all QCAD methods is the Reference Manual: Online or offline included in QCAD under menu Help.

Regards,
CVH

[Buntehuhn]

Thank you for your quick replies!
I'll follow your advices and come back with updates of success or with new questions.

[CVH]

Thanks for this particular drawing file.
It is a good example of what I call a 'distant Block origin' where other applications may fail.
In essence: 'How it should be'.

Example:
Block '*R21' content is drawn around let's say (17.99790211, 362.20526181) what is the center of the bounding box.
Thus far from the drawing origin.
Inserted in Model_Space at (19.76712033, 16.2308315) and the content is also rendered around this point.
Seems logical but it isn't in all drawings we might encounter.

Instead the usual (0.0, 0.0) the Block origin is listed as (17.997902, 362.205262) ... Fairly a match with the above box.
If the Block origin was (0.0, 0.0) or not correctly included and reverted to (0.0, 0.0) ...
... Then the content would be displayed near (19.76712033+17.99790211, 16.2308315+362.20526181) or near (37.76502244, 378.43609331).

The rendered content would then be far form the insertion point.
An issue we see in many drawing files created by yet other applications.

Regards,
CVH

[Buntehuhn]

i've read the book and succesfully applied your advice in the previous posts.
However, one thing I can't accomplish: turn the red rib R-A-1a into a closed polyline.
I could do that to the white rib R-A-4a, by exploding its block and the command polylineautoclose | oe.
What do I have to do to close the polyline of red rib R-4-1a?

Amokka Fin geodesie 5_1_Triplex2mm_nodupls_lyr0_fonts_crcls_halfpolyclose.dxf

(110.13 KiB) Downloaded 42 times

[CVH]

Hi,

Please open a new topic for unrelated or new questions.
That keeps the forum a bit searchable.

The white rib in Block *R0 is already a Polyline.
As far as I can tell at glance that is not logically closed (Closed property) but it is geometrically closed (end = start).
OE - OJ converts between the two closed natures.

You don't have to explode an existent Block Reference first.
Now the white rib 'R-A-4a' in Model_Space is unrelated to the rib in Block *R0.

You can edit the content of Block *R0 and apply the conversion on the Block content.
Select the Block in the Block List followed by BE (Shortcut or pencil button)
Or simply double click on the Block name or even on the Block Reference.

Editing a Block by a double-click on a reference is by preference.
Block content can also be edited in-place, other things are then grayed.
See menu Edit .. Application Preferences .. Graphics View .. Behavior

The red shape in Model_Space are loose segments.
Double-click near one of them to select all connected segments. (Also by preference)
Called a contour but double-clicking selects all connected entities at endpoints.
The selection is not per definition a closed contour, more than 2 entities may connect in one point.

Then you can convert the selection into (a) Polyline(s) with 'Polyline from Selection' (OC) (HA) was wrong
If that is geometrically closed, the Polyline will automatically be constructed as logically closed.

If not resulted in logically closed or in more than one Polyline then there are gaps or more 'shapes' were selected.

Regards,
CVH

PS: *R0 is an invalid block name in QCAD what prevents inserting a Block Reference based on that: Use menu Misc .. Block .. Fix Block Names
PSS: I would avoid 'Double-click' = 'Edit block reference in-place' in print preview mode because the Options Toolbar doesn't seems to be complete.

[Buntehuhn]

Thank you for calling my attention to posting follow-up questions as separate posts.

I've managed to close all ribs (using command OC rather than HA), and will present the resulting file to the Fablab to have them check if it is laserable by their laser.
I'll report back in this thread.
Resulting new questions I'll post as new threads.

[CVH]

Sorry, my bad , (HA) was wrong, OC is indeed the correct shortcut.

The reason is that OC is a also a fine check for testing if a contour can be hatched.
When the resulting polyline is logically closed then HA won't throw an error with the perfect closed nature.
Otherwise there are gaps that may prevent hatching.

The associative mind jumped a notch in providing the shortcut .

Regards,
CVH

[Buntehuhn]