RE: [CAD_CAM_EDM_DRO] EMC and Rounded Corners

Les, Thanks for the quantitative analysis. The 3/8" rounding was a wild
guess, not an actual measured radius. It could well be more like 1". Here's
the qualitative analysis: My parts look like marshmallows.

It is a stepper system.

Tonight I will try to find my max accel values. Yes, plasma and routers need
high accel values to run at high feedrates. Keeping the gantry mass down
helps with plasma but the routers need the rigidity.

I think the G61(exact path mode) will eliminate the tool path variation and
maximizing acceleration will minimize the slow motion around the corners.
The EMC apparently defaults to G64(continuous mode)where tool path accuracy
is compromised to optimize velocity. That's probably a great feature for a
mill cutting a mold cavity at 10 or 20 IPM, and making slight changes in
direction, not so great for a plasma machine running 200 IPM trying to cut
square corners. Actually G64 may be useful with the plasma for curves
generated from small line segments as long as you go back to G61 to make
sharp corners.

Terry

>
> Hmmm... is this stepper or servo?
>
> If you figure using the centripedal acceleration formula
> a=v^2/r to relate acceleration to minimum cutting radius
> that gives (solving for r)
>
> r=v^2/a= (3.33 in/s)^2/8 in/s^2= 1.39 in
>
> -much more rounded than you are getting.
> If two coordinated axes were set to the same acceleration
> and a corner was commanded it would have a parabolic
> countour- just like a projectile path fired at 45 degrees.
>
> This brings up the fact that some of us need quite high
> acceleration values in high speed routers and plasma.
>
> I think acceleration for a vertical mill might be more like 5
> or so but that is not enough for us.
>
> I have been playing around with MAX_ACCELERATION
> and DEFAULT_ACCELERATION in the .ini. I am running
> about 20 for MAX... but I don't really know when one
> or the other kicks in or whether they are numerically
> accurate. Perhaps
> someone could enlighten us.
>
> I think you need to play around with those as well. And feed.
> That v^2 term would seem to give rapidly tighter corners as
> you slow down.
>
> For example for your 8 in/s^2 accel,
>
> 200 ipm gives 1.39 " radius as mentioned
>
> 100 ipm gives 0.34"
>
> 50 ipm gives 0.083"
>
> The last thing to note... High accel like I am using requires
> very careful pid tuning to avoid instability as the rapid starts
> and stops excite system resonances. Before I did a careful
> Zeigler-Nichols closed loop tuning analysis I could not run
> with ACCEL more than about 10. This is for servos of course.
>
> It also puts a load on the bearings etc- but my machine is
> designed for 0.5g
> safely. That's a real sledgehammer blow
> start and stop.
>