DIY CNC Camgrinder

I saw this video on youtube of a CNC cam grinder that looks like it works very well.
I have a manual cam grinder that follows a mechanical master.
It would be nice to have a CNC but I had assumed that because the very hi-end cam grinders are super expensive and still have surface finish problems that a DIY project would not be practical.

I am wondering if anyone has any ideas for what kind of hardware would be good to control the axis that moves the wheel in and out?
Would it be best to drive the rotation of the cam with a rotary table or would an encoder be good enough?
Would it make sense to re-purpose hardware off a a milling machines of are there better options for new parts?

Good day,
What is the problem with the cnc grinders , is it form shape or surface finish.
cheers john

I have been told that the surface looks faceted as one might expect if the motion was not smooth.
This is supposedly on machines that cost between $400K and a $million.

I suppose it could be wheel balancing or dressing but I don't know.

I'm looking into this as well. Have you made any progress?
Maybe can compare notes on going ahead.

marsheng wrote:

I'm looking into this as well. Have you made any progress?
Maybe can compare notes on going ahead.

You should become familiar with external offsets. There are a number of simulators using them included with V2.8 and V2.9. Andy has done a video doing a mockup of cam grinding.

i think these machine are a lot like a lathe in the respect that the work is being done between 2 centers. where things get different is how the spindle gets controlled. so rather than having a spindle that works like a lathe it would be more like a set up you would find on a milling axis of control. you need precise control over the spindle like a rotary axis in a nut shell.

things that would need to be addressed is backlash on the spindle and cross slide for the grinding wheel. if we are talking about grinding cams in motors you are getting into some close tolerance work being done so things better be spot on if you want to do this kind of work.

so as far as how to control it i don't think it would be difficult to program, i think it's more about knowing how the machine is configured and choosing the right plane to set your programs to based on how the machine is configured in respect to the axis layout of the machine.

the hard part is working all the backlash out of the mechanics of each axis and not being stingy with quality when it comes to parts.

basically any cylindrical grinder would be a good starting point for a retrofit of this sort i would think.

I found Andy's video


Basicallly you use external offsets which plasmac uses for torch height control and the cam grinder will follow the profile based on the angular position of the chuck. Supper simple if you know the maths behind a cam!

with a cam it's basically like any other contour you do just in a different plane in the program. all your work with plasma is done in the XY plane. with a application like this the contour of the cam is going to be done in a X/variable plane based on the configuration of the machine. so if the axis of the spindle is set up as a B axis the program would be generating the code in a XB plane rather than what you use in plasma ( XY plane)

an application like this would be a servo based machine with a fully closed loop control. this would be a good application for dual loop feedback with a 1 micron encoder or better since it is a grinding application with very close tolerances. the mechanics of the machine would have to be very tight and as close to zero backlash as possible. this would be a good example of where any kind of position comp that is end point in nature will not work at all.

probably have to synchronize the 2 axis's based on speed of the rotary axis somehow like the video you posted too. i would think it might be something linked to the arcs in the cam. probably would have to figure out a feed per revolution based on the geometry taking a guess.

cam software might have this built in if it is specialized software for this application but i don't know honestly.

interesting regardless. i don't think half a million or better for a machine build for this is in reach for most of us