Re: [CAD_CAM_EDM_DRO] Re: EDM retrofit

Graham Stabler wrote:

> [snip]
>
>Thanks for some further info on what EMC can and can't do. Mach3 can
>also do feedrate override, you can also pause and even reverse however
>I think the reverse takes a straight like back to the end of the last
>move before following the g-code.
>
>

That seems like something close to a reasonable compromise :)

>I had the idea of recording step and direction pulses to memory with a
>microcontroller, then either the whole program could be played back
>according to gap conditions (including full reversal if needed).
>

Well, you've made two simplifying assumptions about the machine. First
is that it uses step/dir for motor positioning. Second is that there
would be no change to desired positions based on feedback (which could
be servo position feedback or gap voltage/current/whatever). Since any
solution we'd want to do must work with servo machines, recording steps
isn't an option for EMC2. The next best thing is to pre-process the
G-code into a bunch of expected positions, split into 1/1000-second time
intervals. You'd play back the positions in forward or reverse, and
maybe blend a little if the feed rate override is active (meaning that
you don't land on an exact position/time pair)

> The
>other idea which is probably neater is to just pause the CNC control
>program and use a small buffer for reversal over short distances.
>
>

This can be done in HAL. There was a person who wanted to add a "weave
pattern" to the programmed G-code, for a welding robot. I think the
solution we came up with was to use the signal generator component
(which provides sine/cosine/triangle/square waves), and add the output
of that to the motion controller position outputs. (subtract the same
amount from the feedback loop if you want the display to show you
programmed position instead of actual) This isn't an ideal solution,
since the sine output isn't constrained to the acceleration and velocity
limits. I think there was some discussion about adding offset inputs to
the motion controller, but it hasn't been done at this point.

>However all of this is probably the easy part with a retrofit, you
>have to work out how the machine works, the generator, feedback etc
>etc I think it could be quite difficult without support. And that is
>before you consider 4-axis cutting.
>
>

Indeed. Knowing what to do is pretty easy. Knowing how to do it is
quite a bit harder, and actually making all the hardware is the really
hard part ;)

>Graham
>
>

- Steve