Toolpath sequence

Discussions around the CAM Add-On of QCAD.

Moderator: andrew

Arctic_Eddie
Active Member
Posts: 28
Joined: Sat Dec 12, 2020 5:23 pm
Location: St. Petersburg, Florida

Toolpath sequence

Post by Arctic_Eddie » Wed Dec 16, 2020 7:08 pm

I have multiple holes to laser cut in several pieces in one session. I select them in the sequence I want for a particular tool path but the simulator shows an entirely different pattern. Where is the optimizer or sequence generated and how do I turn it off?

User avatar
andrew
Site Admin
Posts: 7731
Joined: Fri Mar 30, 2007 6:07 am

Re: Toolpath sequence

Post by andrew » Wed Dec 16, 2020 9:05 pm

The selection order or selection procedure is not relevant for the order in which contours are processed. All contours that are processed in one single toolpath are automatically processed, optimised for shortest path. Contours inside other contours (islands) are processed before the outer contours.

If you need more fine tuned control over the order, you'd have to create a separate toolpath for each contour. You can then re-order toolpaths as desired.

Arctic_Eddie
Active Member
Posts: 28
Joined: Sat Dec 12, 2020 5:23 pm
Location: St. Petersburg, Florida

Re: Toolpath sequence

Post by Arctic_Eddie » Wed Dec 16, 2020 9:18 pm

That's what I was afraid of. In some cases, that's going to be rather cumbersome. For my present project with multiple ribs on one sheet, it's not a problem as I can do all the inside holes first then all the outside perimeters last.

I was hoping for a way of turning off the optimizing function.

Thanks for clarifying the situation.

User avatar
andrew
Site Admin
Posts: 7731
Joined: Fri Mar 30, 2007 6:07 am

Re: Toolpath sequence

Post by andrew » Thu Dec 17, 2020 9:26 am

Can you provide more details, a screenshot or a drawing file?

QCAD/CAM does cut inner contours before outer ones automatically, so this should be taken care of.

Arctic_Eddie
Active Member
Posts: 28
Joined: Sat Dec 12, 2020 5:23 pm
Location: St. Petersburg, Florida

Re: Toolpath sequence

Post by Arctic_Eddie » Thu Dec 17, 2020 3:01 pm

In this particular PNG, the sequence would probably be the same, inside holes then outer perimeter. I don't have a particular case where I would want to select a different sequence. My objects are usually airplane parts cut with a laser. In most cases, the laser cut width is very small so cutting "On" is OK. When selecting otherwise, the long arc problem shows up with this piece even after exploding all splines.
Attachments
s3021-rib-3028-balsa4c.png
s3021-rib-3028-balsa4c.png (58.74 KiB) Viewed 3020 times

Arctic_Eddie
Active Member
Posts: 28
Joined: Sat Dec 12, 2020 5:23 pm
Location: St. Petersburg, Florida

Re: Toolpath sequence

Post by Arctic_Eddie » Thu Dec 17, 2020 3:27 pm

Here's the DXF for the companion piece. It does not work for an outside cut even though the splines are exploded.

PS
This should have gone in a previous post. I tried to delete here and repost elsewhere but it won't let me.
Attachments
s3021-rib-3028-foam4cX.dxf
(855.17 KiB) Downloaded 94 times

CVH
Premier Member
Posts: 1369
Joined: Wed Sep 27, 2017 4:17 pm

Re: Toolpath sequence

Post by CVH » Thu Dec 17, 2020 6:08 pm

Hi,
shouldn't the outer and the rectangle be a closed polyline?
My 3dparty CAM needs that ... When it fails to find the contour ...

Regards,
CVH

Arctic_Eddie
Active Member
Posts: 28
Joined: Sat Dec 12, 2020 5:23 pm
Location: St. Petersburg, Florida

Re: Toolpath sequence

Post by Arctic_Eddie » Thu Dec 17, 2020 6:42 pm

The rectangle is just four individual lines. The balsa needs a narrow slot for the spar but the foam version has the slot widened because the spar passes through at a 35° angle and the balsa on one side is offset horizontally by 2.8mm from the other side. It's easier for me to do that with a four line rectangular box and maintain the same center position.

The perimeter starts as a single closed spline. The end cutouts result from overlaying the correct shape then trimming the unwanted parts. I had been using DXF2GCODE and it handles the splines OK. In one case, exploding the splines to polylines in QCAD solved the problem but in the attached DXF, that did not work. The laser beam is so small that specifying "On" is OK. If not, I can still use DXF2GCODE. That program adds many lines of gcode that confuse an Anilam Crusader 2 controller so I prefer to do everything in QCAD.

Note:
On the underside near the tail there is often a small section of reversed curvature with very large arc radii. This is common in glider airfoils. One of the arcs had a radius of 360,000mm for an arc length of 10mm. This is an angle of 0.00159 degrees. Computer trig functions are usually polynomial series expansions so I can understand why the calculations are stressed. I also found that if the tool path entered a spline at it's flatter end, the calculation failed while the other end was OK. Maybe that's a clue.

User avatar
andrew
Site Admin
Posts: 7731
Joined: Fri Mar 30, 2007 6:07 am

Re: Toolpath sequence

Post by andrew » Fri Dec 18, 2020 8:48 am

Arctic_Eddie wrote:
Thu Dec 17, 2020 6:42 pm
I had been using DXF2GCODE and it handles the splines OK. [...] That program adds many lines of gcode
This sounds like DXF2GCODE might explode splines into tiny line segments rather than arcs. You can achieve the same behavior in QCAD/CAM by setting:
Edit > Application Preferences > Modify > Explode > Explode splines to polylines with line segments

This does tend to generate more segments than the bi-arcs but it is more reliable for offsetting / radius compensation.

Another method would be to do the offsetting "manually" and correct where necessary:
- Select all
- Modify > Offset (with distance = half of the laser width, preferably on a separate layer called "offset")
- Correct the offset path where necessary
- Create a toolpath from the entities on layer "offset" using Side: "On"

Arctic_Eddie
Active Member
Posts: 28
Joined: Sat Dec 12, 2020 5:23 pm
Location: St. Petersburg, Florida

Re: Toolpath sequence

Post by Arctic_Eddie » Fri Dec 18, 2020 1:14 pm

The DXF2GCODE program explodes splines into a combination of arcs and lines. The arcs never have a large radius because I limit the value to 750mm. In one of the successful runs, it was 175mm. Maybe that's the secret, limited radii. When it had no limit to radii then it produced the large circle. The file size for the drawing above was 28K and contained 731 lines.

User avatar
andrew
Site Admin
Posts: 7731
Joined: Fri Mar 30, 2007 6:07 am

Re: Toolpath sequence

Post by andrew » Fri Dec 18, 2020 1:48 pm

Yes, that seems like a good idea. I'll look into that, thanks.

Arctic_Eddie
Active Member
Posts: 28
Joined: Sat Dec 12, 2020 5:23 pm
Location: St. Petersburg, Florida

Re: Toolpath sequence

Post by Arctic_Eddie » Fri Dec 18, 2020 3:14 pm

In addition to arc radius too high, arc angle might need a lower limit. It depends on whether the calculation is done in float or double.

CVH
Premier Member
Posts: 1369
Joined: Wed Sep 27, 2017 4:17 pm

Re: Toolpath sequence

Post by CVH » Sat Dec 19, 2020 9:13 am

PM Summary:
Also or more related to:
viewtopic.php?t=7901

The airfoil originates from a point cloud:
- A Polyline with line segments is under size when bulging out and over size when bulging inwards.
- A controlpoint Spline doesn't come near the points of the point cloud and even fails a smooth shape at some points.
- It should be obvious that the fitpoint Spline is the way to go. :wink:

s3021c-100mm-Fits.dxf
(196.47 KiB) Downloaded 95 times
s3021c-300mm-Fits.dxf
(460.28 KiB) Downloaded 96 times

Trimming the fitpoint Spline for the slots will turn it most correctly in a controlpoint Spline to keep start and end vector at the trimmed point.
To avoid a misunderstanding here: The controlpoints of such are NOT equal to the original fitpoints !!
(Check: Select both the spline and the overlay)

Exploding a controlpoint Spline to a Polyline can induce NaN values for some local Widths, usually when bulge = 0 :shock: :!:
(Many more in the 100mm drawing than in the 300mm drawing)
:!: Arctic_Eddie reports failing CAM offset at exactly these points :!: (300mm version)

Work around:
The NaN values are solved with saving the file and reverting back to it. :wink:

Arctic_Eddie reports a faulty arc with radius of over 360,000mm.
I can not reproduce this under QCAD working further on the Polys with NaN.

Arctic_Eddie reports arc radii of over 10,000mm. (300mm version)
These don't seem to be faulty.
Per offset there are:
- 11 Arcs with R >10,000
- 4 Arcs with R>15,000
- 1 Arc with R≈28,000
- 6 line pieces (or R = infinite)

The trailing edge slot seems a little off.
s3021-rib-3028-CVH.dxf
(142.96 KiB) Downloaded 95 times

In short:
Point cloud > Fitpoint Spline > Trim > =Control Point Spline > Explode to Poly (> NaN?!) >
Save > Revert >
Construct a closed Polyline as a single closed contour to use for CAM.
Clean up: Only keep the perfectly closed Polyline, remove anything hiding under it.
Explode this poly and check for duplicates and zero entities.
Recombine with OC what also can merge some segments reducing the segment count.
Optionally use the Poly exploded to segments for CAM but only after the clean up.

Regards,
CVH

Arctic_Eddie
Active Member
Posts: 28
Joined: Sat Dec 12, 2020 5:23 pm
Location: St. Petersburg, Florida

Toolpath sequence

Post by Arctic_Eddie » Sat Dec 19, 2020 2:21 pm

The 1/8" x 3/4" balsa at the tail passes through the slot tilted at 35°. That's why the slot is about 1" deep. That piece of balsa is slotted along the back edge for the elevon hinges. The elevons and trailing edge pieces are 3/8" x 2" symmetrically tapered aileron stock. The leading and trailing edge angles are both 35°. The shape is similar to the Horten HO9 but using only straight lines. The new B-21 development bomber is closer to the shape.

CVH
Premier Member
Posts: 1369
Joined: Wed Sep 27, 2017 4:17 pm

Re: Toolpath sequence

Post by CVH » Wed Dec 23, 2020 5:27 am

Entered a bug report on the NaN Widths:
https://qcad.org/bugtracker/index.php?d ... sk_id=2157

The faulty offset with radius of over 360,000 is at this spot. :wink:

Regards,
CVH

Post Reply

Return to “QCAD/CAM”