Machsupport Forum
Mach Discussion => General Mach Discussion => Topic started by: Keith on January 07, 2006, 10:33:12 AM
-
Hi,any hints/short cuts on milling a 2 D carving onto a spherical surface without too much distortion and without too much depth variance(the sphere is hollow)?I have an image I'm rasterizing and would like to just make that impression(negative) on a small area of a plastic sphere.The sphere is about 4" diameter and the area with the image is negotiable but around 2", centered. I actually once had this done many years ago before I did any milling and I believe the guy that did it for me actually calculated the different heights and offset Z with these numbers but there's got to be a better way(curve interpolation?).Thanks,Keith
-
HI keith If you go to the yahoo mach support group and look up starting at
Message 43730 that issue is discussed. Thanks Terry
-
Well yes that seems pretty straight forward and I'm familiar with axis swapping to accomplish this but my brain is having a hard time extropolating this idea to a sphere(the 'message you mention deals with a cylinder).Thanks Terry,maybe I'll have an 'Eureka' moment after reading those posts ,over.-Keith
-
Keith what your are wanting to do is reverse ortho rectification from a flat plane to a sphere. I don't know the answer but hopefully the discussions will get you going in the right direction. (:~)= Terry
-
Hi Keith,
to do this properly you need a 5 axis machine, if you try it on a 3 or 4 axis machine you will get overcutting as you move down the sphere. The correct way is to keep the cutting edge at 90 degrees to the work surface at all times. The nearest you can get with 3/4 axis is to use a lollypop shaped tool so the back edge of the tool is not cutting into the work surface.
Most CAM systems will wrap a design round a ball using surface project, to get a negative you pocket out the background.
Hope this helps
Graham Waterworth.
-
Thanks Graham,I wasn't thinking so much about the overcutting so much since I was going to use the tiniest of tools and the image I want is very 'high contrast' so I'm really looking more for what on paper would look either black or white/on or off,all cuts being the same depth, but I can imagine this still to be a concern.What I was concerned with is doing it without a CAM.I was pretty sure a CAM package could do it.I was more wondering about the feasibility of roughly mapping the Z offset of a sphere since the image would easily be rasterized.I'm going to contact the guy who once did this for me,maybe I'm wrong but this guy was so 'old school' ,I don't even think he ever used a CAM package.Thanks,Keith
-
Keith I have seen it done with 5 axis many times. I also have seen it done with 4 axis. Here is a thought.
If you use a 4axis setup with A: being polar and angle A: to 1/2 the Rad off of vertical. That will lessen the correction needed, then use a cone shape section of your picture to cut as a linear cut around the sphere, it would be fairly close. Not perfect but close. It will depend on the thickness of the sphere surface. Do you have a 4 axis or 3 axis machine.
There are cam packages out there that can do the sphere cut with no problem but they are a little pricy. What you might want to try is a demo package that will write a limited GCODE. Try BITCAM.COM
I'll ask my son they work in 5 axis 4d all the time.
Might think of using a spiralled helix cut as a cone if you only have 3 axis. Or just bite the bullet and sit down and map out the Z corrections from a flat plane as your friend did.
Please let us know how it ends up, Interesting project.
(:~)= Terry
-
Oh right,you brought me a little closer with that one...think polar.
I have a 4-axis.I'm going to chew on this for a bit as a back of the brain distraction.When I do something mentionable,I will.Also,if that guy tells me what he did to achieve it,that did it for me in the past,I'll pass it on to the group.Thanks for your thoughts Terry.-Keith
-
Keith do yo have a digitizing probe???? and an example of the object. Nah that would be too easy. (:~)= Terry
-
Hi Keith,
the way we did it before CAM systems was to work in concentric circles from the centre of the logo/image on a flat drawing and machine one circle at a time working out the drop in Z on the sphere at that diameter, then replace all the Zs in the code with the original Z+the drop.
Hope that makes sense.
Graham
-
Now that makes sense.I guess then the trick is how to index the information contained in each circle of interest.Maybe one circle per layer?
-
This is an example of the old way we used to do things.
See picture attached.
1. Original image centered about X and Y zero.
2. Circles drawn at a known pitch.
3. Pick 2 circles to work between.
4. Remove all the bits inside inner circle and remove all outside outer circle.
5. The bit we machine and generate Gcode for.
6. work out the drop using trig and add this to the Z created in the Gcode.
This is not the perfect way to do it but it works well enough for most jobs.
Regards
Graham
-
I was thinking this could be done with a wizard :) But the more I think about it a cam system would be the best.... I think you would need to have this in C++ as VB would be a little slow. I am going to have to think about this, is this something that many of you would use?
-
I gottcha ,Graham.I like this way and since what I want is high contrast ,any discontinuity will just be part of the randomness.What I think I would do is outline the high contrasted image in Rhino.Draw the circles and then 'trim' the first area of interest and save it as a layer and then continue to isolate each circle and save as another layer this way,generate the gcode from importing it as a DXF and then do the trig to insert into each layer's z-cut.Thanks a million.-Keith
-
Hi Keith,
Glad to be of help.
Graham.
-
There you go, Good job guys. Ifyou keep the z and circles into perspective it will look very good. (:~)= Terry
-
Good going Terry,I didn't see your earlier post,I needed an excuse to buy that probe I've wanted,for a while.Merry Xmas to me! Seriously that is so cheating but I WANT THAT PROBE!!!-Keith
-
I have an IMservices TP-100 combo probe. It is an interesting economical unit. Works very well for the low cost. A touch probe and tool setup probe rolled into 1.
(:~)= Terry
-
Yeah that's the unit I was drooling over.I just went over board this year with all sorts of new equipment but it's high on the 'next to buy' list.I think that tool setup extra makes it a bargain ,too-Keith
-
Brian I don't know about the sphere milling, but I do need a wizard for making cookie cutter dies. Now DON'T LAUGH. Let me explain.
I take a picture of the object you want a cutter/die made to look like. I then bring that picture into AUTOCAD as a base layer. Then I create layers and outline the basic shape and then outline the highlights. I then assign each layer a z height and DXFout the drawing.
What I need is a wizard to cut out all the flat relief areas. Now I have to sit and program out each section or just manually cut it out when the basic shape outlines are finished.
Just getting old and lazy thats all. I really am getting to like conversational programming such as your wizzards.
(:~)= Terry