Re: multiple block MDI entry

[Jon Elson <jmelson-at-artsci.wustl.edu>]

Ray Henry wrote:

>
> I like your mods to tkemc.  I still have a question about what emc is
> really using for feedrate because of what Jon E. said a while back.  I've
> seen it do some goofy stuff here, like running f3 when it says f50.

This is a bug.  It may well be in lower levels of EMC, not in the UI,
because behavior seems the same in xemc, and same from Oct-1999
through Dec-20-1999.  What triggers it is a change from MDI to manual.

When you jog, it uses the last feed rate set in MDI (or sometimes, from
recently running a program in auto) after you've switched to manual.
If you hit the 9 and 0 keys, it will speed up.  I don't know whether
that is from having a feed rate override set to 10 or 20% left over
from MDI or auto, or whether the feed override keys force it to
fetch the manual feed rate so it can multiply by the feed override
to get the rate required for this mode.  I think that is it, however.
So, I think that the motion routines are not aware of the mode change
(maybe only sometimes, or under certain sequences) and continue
using the auto or MDI feedrates previously programmed.

I haven't seen this during the last couple times I've used TkEmc,
but then, maybe I'm so used to it I don't notice it anymore, and
hit the 9 0 without thinking.

Now, the other thing is feedrate display in auto.  The feedrate comes
from the interpreter, and the interpreter runs ahead, as far as available
buffers allow it to.  that usually means to the end of the program,
where there is usually a rapid retract or move to home position,
with an F word, like F50.  As soon as you hit run, you will see the
feedrate blip through whatever values are set in your program, and
then show the last commanded feedrate throughout the rest of
program execution.  This is a fundamental problem in the design
of the EMC program structure.  The real-time feedrate could be
recomputed from commanded velocities, and in fact, these individual
velocities are available, I think, in a shared memory structure.
So, TkEmc could just compute the vector sum of velocities and
rescale to FPM (for imperial measure) or m/M for metric.
This gets messy as additional, or non-cartesian axes are added,
of course.

Jon