November 15, 2015 at 10:04 am #3714
I was looking for a bit of a challenge and interesting in trying a board like a Tomo Evo, which has a central double concave, within a single concave (which turns to a quad channel at the fins. e.g.
However I am having a few problems with this proposition in BoardCAD, I have built the channels by adding control points to the bottom section curves, and have exagurated them in the attached pictures for illustration:
1. The nurbs model does not seem to build the bottom detail e.g.
This is the same whether I use all of the following options from the “3D Model” menu: Open Model; Closed Model; Approximate Outline and Rocker.
However when I use the Beizer Patch Method, I do get something reflecting my concaves, however the shape is very distored (and not symetrical): e.g.
The behaviour is the same when the concaves are reasonable values too (i.e. not exaggerated).
Has anyone managed to work in BoardCAD, and if so, are the any tricks or tips they could offer? I’m guessing this is a bug with the 3D generation code given that the resulting surface is asymetrical
Ideally I would like to get the model to include these details, so I can make the CNC machine to the work and improve repeatability (i.e. for changes on future versions). I am sure if I try to do these concaves by hand, they will differ greatly
2. The second problem I am having is how to deal with the channel exits at the tail. At the nose I can blend this to flat, and provided the number of control points remains the same the result looks ok, however at the tail, the software is interpolating narrowing cross sections past the last cross section, which are hence throwing all sorts of weird splines into the channel lines. Is there a better way to deal with channels in BoardCAD? Id like the channels to remain straight and exit the board where ever they meet the rail.
This seems a little counter to how BoardCAD extrapolates the boards, so maybe this is not possible?
Pete.November 17, 2015 at 11:11 am #4741
If anyone else is interested in this subject – I have made a bit of progress by squaring off the back of the board (by adding a non-continuous control point square with the tail).
Every corresponding concave control point is set to the exact same X coordinate in an attempt to keep them perfectly straight.
This combined with some cross sections at 1mm, 5mm and 40mm from the tail reduce the warping of the concaves, but don’t completely remove it. The warp still exists between these new cross sections and the next (and to a lesser extent the whole way up the bottom).
It feels like this result is due to the way BC interpolates intermediate cross sections – it seems to scale the control points against width. Consequently the bigger warps occur in the areas of greater cross section. I think this means it may not be possible (with the current logic) to produce straight channels. Forcing them with excessive number of cross sections will minimise the issue, but never completely fix it, not to mention is a nightmare to work with. This is easy to visualise using the sliding cross section, here you can see the channels deviating as you sweep along the board. Id be really happy if someone could prove me wrong!
I have played with the “set nr of segments” option to try to reduce the number of interpolated cross sections used for surface creation, however neither 1 of 4 made any difference to the wireframe mesh and warping.
None of this has made a difference to the malformed 3D surfaces produced.
Pete.November 21, 2015 at 8:51 am #4742
Been interrogating the code trying to resolve my issues:
Beizer Patch algorithm appears to be designed for either 5 or 4 segments per cross section, however anything other than 5 segments is handled by the 4 segment routine. Having the control point for quad channels (ie inside the concaves) produces 6 segments per cross section, and hence the strange output surfaces.
If I remove the channel quad points and revert to 5 segments per cross section a symmetric model is created, however there still seems to be an issue with the board not being thick enough, I am assuming this is a problem with the 5 segment algorithm which should be able to fix when modifying the algorithm for an additional segment… (if I really what those channels!)
Wavy channels are a function of how BC interpolates the cross sections: intermediate cross sections are interpolated by scaling the closest cross section to achieve the required width and thickness. Concave control points are scaled too. I think I will need a procedure to adjust the concave point X values. Probably need to adjust them to a spline / curve created by all the other control points to permit flaring or curve channels and avoid step changes in the channel directions.
Really not sure how this would be best incorporated into the GUI for future versions, maybe the ability to mark corresponding points on each cross section which should be post processed (ie a group for each of the concave points)?November 21, 2015 at 10:10 am #4743
As a proof of concept, I added a few lines to force the concave control points to specific values (50 & 100mm) and was able to produce perfectly straight concaves on the sliding cross section and patch model:
// Perform Concave Adjustment Here
// Test = Put concave points at 100mm (index 2)
BezierSpline PWtempSpline = i.getBezierSpline();
int PWNumCP = PWtempSpline.getNrOfControlPoints();
BezierKnot PWKnot1 = PWtempSpline.getControlPoint(1);
BezierKnot PWKnot2 = PWtempSpline.getControlPoint(2);
November 25, 2015 at 7:06 pm #4744
hi Pete_w I tried too to design an evo with all the bottom concaves, but I never been able to create something good… at the end of the story I designed the board without the concaves,and finally I made them manually, i’m surfing it since october and I love it. By the way for future Tomo creations like the Hydronaut or something else I would like to try once again to design it totally with boardcad. Till now i was able to design it with MOI 3d.December 6, 2015 at 1:40 am #4745
I’ve added some code to linearly interpolate the x value of the concave control points on the bottom and the results seem pretty good. I have also updated the bezier patch routine to accommodate the additional points which seems to produce pretty good results. I have only tried 1 and 2 additional points.
There still appears to be a tiny little bit of spooning at the nose, but otherwise the concaves are straight. might need to check the logic to ensure the right cross sections are used for interpolation…
December 8, 2015 at 11:44 am #4746
Thanks for sharing that code. You got further than I did with the tail. If I can duplicate what you managed to do when I get back home I’ll upload the fileNovember 24, 2016 at 10:00 am #4822
test?November 24, 2016 at 10:04 am #4825
For anyone interested I finally got around to cutting one of these boards. Been busy, and also had to develop a different NC code script to suit the machine.
The machine had a large (Ø42mm) bullnose cutter so I had to settle for halfway with only 2 channels, but the results were good. No reason to see why the the quad channel couldnt be cut with a smaller tool.
If BoardCAD is being updated at all, I’d be interested in getting this put into the main code.November 24, 2016 at 10:13 am #4832November 24, 2016 at 10:14 am #4834November 24, 2016 at 10:15 am #4835November 24, 2016 at 10:15 am #4836November 24, 2016 at 10:15 am #4837
- You must be logged in to reply to this topic.