make a hyperbolic shape in solidworks 4

[supernova1387]

Hi
I need to draw feet of my robot using a curve which follows this equation in 3D space:

z=sqrt(Rf^2-x^2)-Rf+sqrt(Rs^2-y^2)-Rs

where Rf and Rs are radius of curvature of the feet in frontal and sagittal plane(take them as 0.5 and 0.38 m respectively). I have drawn the curve in Matlab and you can see it in the 1st link below. I put the actual picture of the robot as well for you(second 1).

http://i54.tinypic.com/sywl52.jpg

http://i56.tinypic.com/2z8zpjt.jpg

the 2 feet should be made as 1 piece (we will then cut the middle to get 2 feet). Please explain how can I make this in solidworks? I am not an expert so if you use any unfamiliar term please explain that.

Regards

 

[rstupplebeen]

SolidWorks can not create a equation driven surface unfortunately. I would try to export your geometry from Matlab into a IGS or STEP file and import into SolidWorks. There are several free programs that can do this directly and can be found with some Google searching I hope this helps.

Rob Stupplebeen

https://sites.google.com/site/robertkstupplebeen/

 

[rollupswx]

What is your manufacturing tolerance. Looks like it should be a simple Loft would get you well within an manufacturing tolerance.

 

[supernova1387]

We can use CNC machining to make this. I have seen a video on youtube that can import file from matlab to solidworks but I am using matlab and solidworks 2010 and the commands in both of them are not working and solidworks doesn't support that file format anymore or at least I can't find it. Plz have a look

http://www.youtube.com/watch?v=-R-TpWUSaX8

 

[supernova1387]

Oh this is sweet
I found the problem. Watch the video I just put, then export the file as *.txt rather than *.xyz in Matlab. when you go to solidworks, and when you want to open the txt file, if you don't see the option: point cloud files, go to tools/addins and turn on the click for scanTo3D for both of them(left hand side and right hand side). This will activate the point cloud files in your open menu .

This is really cool

 

[rgrayclamps]

Hi, supernova1387:

This geometry is doable in Solidworks. If you section the geometry, along planes parallel to either YZ or ZX direction, you will get ellipses. Solidworks has "Partial Ellipse" or "Ellipse" command which allow you to create ellipses. You can then create your surface through lofting the profiles with an ellipse guide curve.

Good luck!

Alex

 

[Updraft]

SWX also has a parabola sketch element. Though neither the ellipse nor the parabola are mathematically the same as the hyperbola perhaps one of these will be close enough. You could also make a cone and cut it parallel to the axis to get a true hyperbolic shape. Adjusting any of these elements to fit your function could take a little work.

- - -Updraft

 

[lumenharold]

I used a boundary surface with formula driven sketches in each axis. The scale isn't correct but this is what I whipped up.


Harold
SW2011 SP1.0 OPW2011 SP1.0 Win 7 Ultimate
BOXX 8550 Xtreme Dual Xeon 5680 @4.2Ghz
nVidia Quadro 4000
www.lumenflow.com

 

[lumenharold]

This is the formula by the way:

(x^2/.38)/(1+(1-(1+(-1.1))*(x/.38)^2)^.5)

And if you don't have SW2011 loaded this is hwat it looks like.

Harold
SW2011 SP1.0 OPW2011 SP1.0 Win 7 Ultimate
BOXX 8550 Xtreme Dual Xeon 5680 @4.2Ghz
nVidia Quadro 4000
www.lumenflow.com

 

[rstupplebeen]

I would not trust boundary surface, lofts or sweeps if you want accurate geometry. In 2010 I created a partial ellipse and revolved it, lofted it to an identical ellipse with and without a circle as a guide curve and did the same for a boundary surface and sweep. None of them were close enough for my tolerances to the revolve. Try similar tests with your equations.

Rob Stupplebeen

https://sites.google.com/site/robertkstupplebeen/

 

[lumenharold]

Hey Rob. I've certainly been there with the surface accuracy issues but mostly with optical surfaces. With a machined radius of .5M do you think it will matter? I wouldn't use it (again) for a parabolic reflector for imaging but I've used it for non-imaging systems and it seems to work well enough.

Harold
SW2011 SP2.0 OPW2011 SP2.0 Win 7 Ultimate
BOXX 8550 Xtreme Dual Xeon 5680 @4.2Ghz
nVidia Quadro 4000

http://www.lumenflow.com

 

[rstupplebeen]

Harold,

Attached is a model showing identical unit quarter arcs that are revolved or lofted and how the geometry varies.

Rob Stupplebeen

https://sites.google.com/site/robertkstupplebeen/

 

[lumenharold]

Whoa! I didn't go as far as to try all the different constructions you did Rob but I did make a quick comparison between a surface created by two formula driven sketchs (one in each axis) and a revolved sketch. I changed the surface from a hyperbolic to a parabolic since SW doesn't have a hyperbolic in the pulldown and there still isn't a conic sketch tool in SW.

I knew there would be a difference in the area. For a disk that's .9mm diameter with a parabolic profile of radius .5mm the difference is .0077mm^2. I'm certain that if I used the surface to image a point source niether would be accurate.

Still, pretty close for machined foot. Perhaps the error scales up as well?



Harold
SW2011 SP2.0 OPW2011 SP2.0 Win 7 Ultimate
BOXX 8550 Xtreme Dual Xeon 5680 @4.2Ghz
nVidia Quadro 4000
www.lumenflow.com