Jones and shipley cylindrical grinder CNC conversion, grinding camshafts, cranks

I am getting ready to embark on converting a very nice jones and shiply grinder that I recently aquired for a steal $400 USD, to linux cnc. Now the place where I could use some help from the gurus off mathmatics here. 

I want to be able to use it to grind camshafts, and crankshafts, but I think eccentric turning hal components can probably be tweaked for crankshafts. I am curious how I would 

1. generate g code for the camshaft but I suppose I could use a 2d tool paths with a end mill to generate tool tip positions, around the Z axis with x & y coordinates, the only issue is the grinder will only have z and X like a lathe. 

2. design considerations during the retrofit fit. I am thinking a servo for the z axis to control position along z in a rotary C axis fashion for grinding camshafts as indexing will be important. 

I have 3 good yaskawa AC servos packs and servos out of a hurco I tore down for parts, balls screws and mechanical tidbits for days, the hardpart is trying to figure out how to grind the camshaft lobes.  

open to suggestions if anyone has done anything similar before. 

Looks like i can use A axis rotary moves with the grinder spindle on Z with A being aligned on Y to create Gcodr 

Seems more of a cam problem than a problem for the machine controller?

yeah, the big issue is generating tool paths. I have been working on it and I feel like I might have gotten somewhere with it.  It would be nice if there was a way to upload the lobe geometery and then pick the camshaft lobe arrangement by engine type and then hit grind. That's more of a UI side issue i suspect. 

It seems a lot of effort to program any meaningful thing from the start.
It is quite expensive to "rent" a ready programmed solution.

Boils down to the question is it worth it? Pay someone to program it, or (provided one is able to) spend all the time to do it.

I have a guy I know working on it. the rented solutions kinda suck. Most of the packages are aimed at very large scale operations. Mostly this will be used for protyping and R&D work on a small scale. 

What I saw in hobby forums are mostly "copy" machines using a master(template)cam and some mechanical system to follow the template. (this boils down to 2d operation.

What do you mean by scale? Does it matter what size the parts are?

Very cool project. Please post photos of the grinder when you get it.

timo post=315275 userid=29371What I saw in hobby forums are mostly "copy" machines using a master(template)cam and some mechanical system to follow the template. (this boils down to 2d operation.

What do you mean by scale? Does it matter what size the parts are?

lobe shapes/profiles are not directly related to the base circle. That is to say that, the size of a lobe is correlated to the base diameter of the camshaft journal, but not always. I can upload some 3d models of camshafts and screen shots to illustrate this. 

Maybe an electronic 'copy' device to build up a library of lobe shapes?

I'm thinking of a motorized fixture to rotate the cam, with a high-res rotary encoder and a roller follower with a high-res linear encoder.  Rotate the cam and you can now plot the toolpath based on the follower location to rotational position (and follower wheel diameter)

Once in CAM, the plot should allow reverse engineering the lobe max lift and duration @0 or 0.050" lift since it should be obvious when the follower is riding on the cylindrical portion of the cam.

If the fixture were built with the follower on a track that's aligned with the camshaft long axis, it'd be easy to just work your way down each lobe without much setup time.

As for generating toolpaths, this looks like a good use of LCNC's external offsets.  Like everything, the details are probably fussy... but I think you could essentially load the plot back in to an external offset table and the 'X" axis would follow the rotary axis's rotation.  Multiple step-downs, of course, and I'm not sure how to avoid air cutting during the roughing phase.

Cool project.  Can't wait to see how it works out.