Re: Re: Re: Re: CNC Test File

Art

Someday I have to try your CNC software and see how it compares. Does that
mean I would have to dig up that MS-Windows I burried in the yard? Or can I
run it in the windows emulator under Linux?<g>

(comments mixed in)

From: "Art Fenerty" <fenerty@...>

>Jon:
>
>Unless I'm mistaked, the BPS of emc will vary more with feedrate than
>interpolation or trajectory cycles. The next block will process when the
>previous block is completed in the trajectory cycle. Upping the trajectory
>cycle would not do much more than increase resolution by decreasing the
>velocity of trajection.

The look ahead on the EMC is a couple hundred blocks. Once you get
beyond that, then the next block is read when the one on the top or
bottom ot the stack is done. One reason that I could get those rates was
that the number of steps per unit rather small. I think that the NIST
hexapod used something like 680 steps per inch.

>Raising the velocity to ridiculous rates will up the BPS to its max. Of
>course Block size would have a lot to do with it.

Block size is right on. I had a discussion with Tom Kramer, the fellow who
wrote the interpreter, and the best way to optimize throughput is to strip
everything out of the g-code that is not absolutely necessary -- no line
numbers, only type modal codes when you need to change a mode, remove
comments, etc -- makes for some pretty unreadable, hard to edit stuff. For
those contouring files with lots of 0.002 linear moves it makes sense.

>( been poking around in the emc code ...stumbling like a greenhorn)
>Art

Most of it is way beyond where I can even dream of reading.

Ray