4th axis cam machining

[kvom]

I know that other CAM programs have ops for machining cam lobe profiles.  An idea I had was to draw the lobe in CAD, import to CB, and divide the polyline into multiple short segments.  I could then extract the points and write a Java program to move such that each segment between two points is horizontal with the tool touching it.  I haven't needed to machine a cam yet I haven't moved on this idea.

But then it occured to me that a plugin could be written to take as input a pointlist created by dividing the lobe polyline.  The output would be a straight line in Y with Z values equal to the distance from the center for each point and Y values equal to the spacing of the point.  Then an engrave Mop will follow the Z values as it moves in Y.  The output gcode would then be converted to A moves via the wrapper plugin.

Thoughts?

[dh42]

Hello

Can you get a 3D of the lobe ?

If this is the case then you can maybe use my unroll plugin ?

http://www.atelier-des-fougeres.fr/Cambam/Aide/Plugins/Unroll3D.html

++
David

[Bubba]

If this is the case then you can maybe use my unroll plugin ?
*******************
Once you get solid model, Davids plugin definitely should do this easily. Good luck.

[kvom]

Cams have a flat surface so the unwound surface would be a ribbon.  This would need to be machined with a flat, not a ball mill.

Does unrolling on the Y axis mean machining needs to be oriented the same way, since a camshaft would almost always be setup on the X axis?

[dh42]

Does unrolling on the Y axis mean machining needs to be oriented the same way, since a camshaft would almost always be setup on the X axis?

No, the plugin just create a surface object from the source object, you can rotate it as you want.

have a look on the doc
http://www.atelier-des-fougeres.fr/Cambam/Aide/Plugins/Unroll3D_doc_EN.html

++
David

[Bubba]

This would need to be machined with a flat, not a ball mill.
************************
Yes you can machine with ball endmill as long as the stepover is minimal. Little tiny groves as oil groves, would help with oil lubrication. 

Of course I'm joking here not knowing what loads the cam will see and how it gonna be used.

[EddyCurrent]

I didn't fully understand the requirements from kvom's first post but having read through the thread again plus some other web pages I can now see how it works.

[kvom]

Over in the model engineering websites I frequent, those who build internal combustion engines need to machine camshafts for the valves.  The "tried and true" method is to calculate a series of flats around the perimeter and mill across, using a rotary table to position the shaft.  The cams are obviously oriented at different angles, and one slip in positioning will ruin the entire shaft. Then file to produce a smooth surface.  The single cylinder hit&miss engines use a single cam that can have it's profile milled normally.

I've just confirmed that the X and Y coords of points can be copied from CB and pasted into a text file, so I think I'll write the program in Java to generate the G1 moves.

[EddyCurrent]

This is just a test, try the attached plugin and cb file. CamBam 1.0 only

It gives the rotation angle and height (currently in Y but easily changed)

1. The user draws a cam profile centered on 0,0
2. Select profile (as polyline)
3. run plugin ( called BamCAM in the Plugins menu)
4. enter a value for each rotation angle step, e.g. 2 degrees.
5. values are shown in the bottom message log window


It seems a pity to have to use the Wrapper plugin when it's possilbe to find angle and Z height easily.
I've never written g-code out from a plugin before so that's going to be a challenge.

[kvom]

The height above the center is irrelevant.  The tool must move so that it's tangent to the line.  So the lobe is rotated so that segment is horizontal in the YZ plane with a positive Z value.  The amount of the rotation is the A axis.  The Z value is the height of the section after rotation.  The g-code for that move is just G1 Axxx Zxxx.

If you take two consecutive deltas of the angle specified intersected with the lobe, you get two points in YZ.  The tangent angle is calculated from a line connecting the two points, and that angle then tells how to rotate to bring it horizontal.  That's one reason I suggest creating a pointlist via the divide geometry tool and using that as input.   The tangent angle between the two points becomes a simple calculation.

Using an angle increment from the center can cause an issue as it make make the division of the top of the lobe, where the radius is smallest, too large.

[lloydsp]

Kvom wrote:
"The tangent angle between the two points becomes a simple calculation."
-------------

Kvom, I'm not "preaching to the choir".  I KNOW you know this, but some prior posters might not have been aware of how easy it is to do.

Indeed, it's easy! 

Mnemonic for Americans: "Indian Chief Soh Cah Toa". Once you've learned it, you never forget it!

Sin=opposite/hypotenuse, Cosine=adjacent/hypotenuse, Tangent=opposite/adjacent

You might have to come up with your own Mnemonic for other countries/languages. 

Simply calculate the one that's most-convenient for the point-pair;  The tangent would seem to be the easiest and most-logical one. 

Once you've determined the angle of the point-to-point line, calculate the perpendicular for the actual milling tangent angle by just adding or subtracting 90 degrees.

Lloyd

[EddyCurrent]

The Z value is the height of the section after rotation

Height from where ? what is the datum ?

Using an angle increment from the center can cause an issue as it make make the division of the top of the lobe, where the radius is smallest, too large.

Yes, I see that but if a small enough angle is used it should be okay.


If you draw a line between points it makes a facet that is inside the lobe outline so the lobe will be smaller but if you use the method I'm suggesting the Z height is on the outside of the lobe outline.

@Lloyd, I don't think any of the prior posters have a problem with the maths.

[lloydsp]

"@Lloyd, I don't think any of the prior posters have a problem with the maths."
0-------0

Good!  I wasn't meaning to be didactic, but merely to show a tool I use nearly every day, and most folks need frequently.

Lloyd

[kvom]

Once rotated, the Z height of the facet is from the center of rotation, which is 0,0 in the YZ plane.  It's true that using the facets would make the output slightly smaller than the lobe as drawn.  Unless you have a math formula for the lone calculating the true tangent is difficult.  A closer approach is to use 3 consecutive points:  two outside points for the rotation and the center point for height.  Here the points should be the same distance apart.

For model-scale cams any differences are likely to be insignificant.

[EddyCurrent]

Once rotated, the Z height of the facet is from the center of rotation, which is 0,0 in the YZ plane.

Exactly, which is why the test plugin gets the height from centre. The fact that it is currently in the 'wrong' plane does not matter at this stage.