Changes in v3.

1a of TurboCNC
Release scheduled for June, 2002
This file last updated 08-17-2002 1900 PDT
====================================================

NEW FEATURES:
--------------

*- User settable jog increments (in ini file). Edit the .ini directly to change
them from the defaults to something more appropriate for your hardware, or use the
setup function labelled "Defaults" from within the program.

*- Use "O" (the letter Oh) to enable/disable limits in jog for safety and for
setting up your machine. One potential use for this, although I don't recommend
it, would be to use a limit switch as a homing stop for setup.

*- A default directory for files and tool offsets, plus the extensions for the same
can be now be set in the ini file. This defaults to the .exe directory on startup
if not specified. Use the "defaults" menu under setup to change these to your
liking.

*- The file load function goes back to the main menu right away for convenience.

*- Shift keys and < > now do 1% changes in feed override. This is cool!

*- Fail-safe gcode file run, not "press any key to start".

NYI- Incorporate RDTSC hardware clock for Pentium users.

WNBR- Save abort position for restarting a file expediently.

WNBR- Absolute position for "home"/toolchange in leiu of home switches.

*- It's now an error to command a spindle to reverse direction if it's

already on without turning it off first. Hasn't happened yet, but it could have
before.

*- A backup ini file is automatically generated as a safeguard against

configuration loss each time the ini file is written through the menu. It will be
named TurboCNC.OLD or similar based on the .ini file name.

WNBR- Add a "one shot" G code input to the jog screen

*- Canned cycle for form and surface digitizing. Use the G32 code like this:

G32 Xbound Ybound Zbound Idiscretization Ffeed #reserved

This is 2 or 3-axis general, so the last axis cycles the most and the first axis
only once through a number of points determined by the discretization distance. In
the future, separate algorithms for this will be supported, so until then don't use
the # code. The digitizing scheme is of the "bed of nails" variety, eg: straight
up and down in the last axis on the line.

WARNING: Rounding error may prevent the last row or column from being scanned. Add
a small amount to the bounding distance to avoid this, eg:

G32 X1 Y1 Z-1 I0.250 F10 ;X1,Y1 may or may not be reached

G32 X1.001 Y1.001 Z-1 I0.250 F10 ;This is better

The output is written to the a scan file in the same directory as the executable,
with a filename of surfscan.dat.

*- Special music for cycle end sound. You can use the old do-ray-mi-ray-do as the
default (STANDARD), or use NONE, WEASEL, or STARWARS in the ini file for sound when
the g code file is done. This plays through the regular PC speaker, so don't
expect the London Philharmonic here.

*- Added skip screen command line parameter (-QUICK). This gets you past the
startup screen ASAP.

WNBR- Add description header to ini files. You can add your own header with the
comment character by manually editing the ini file, and it will be preserved until
you re-save it through the setup menu, eg:

;This is my ini file for the mill

...

*- Added Harald's mini-jog as an option while aborted/toolchanging/between parts.

Use the arrow keys for axis one and two, + and - keys for axis three. The keyboard
layout assumes a three axis mill set up as XYZ. For lathes, set up Z as the first
axis, and X as the second. The XYZ keys zero axes as appropriate, N & D enable and
disable the drives (if configured), A & B toggle the coolants, and the # key zeroes
all of the axes.

*- Add CMM capability for hole location and measurement. For now, this is limited
to holes that can be probed by simple XY motion. Put the probe inside the hole
approximately in the center, and press C while in jog mode, or call G50 from within
a program. This is an orthogonal six hit probing algorithm. In jog mode, the two-
place roundness check results and radius will be reported after you input the probe
diameter.

*- Fixed mmpm/ipm display, added feedrate and spindle speed to status window.
Spindle speed is commanded speed (S word) or the speed read from the machine if
your machine has an index pulse configured. Use the special M50 code to read the
tachometer in a program. The "V" key reads the spindle speed in jog mode, if it's
on. Speed will also be read at the start of a G33 threading pass.

*- Re-enabled CV continuity between moves to avoid lost steps. (Delay parameter for
LG). You can add a short (1-1000ms) delay between axis reversals in case your
machine needs to settle a bit between moves of this type. Edit the ini file
directly to enable this:

ReverseDelay(ms)=0 ;no delay

*- Added diameter mode for lathes on the X axis. Change the Prescale parameter in
the inifile to 0.500000 on the X axis to use traditional diameter programming on a
lathe. A prescaler is available for all axes, in case you're really going hog wild
with this. The prescaler is applied to the input position in absolute mode, not
incremental, and not to IJK, R, or other incremental-type inputs for moves and
canned cycles.

*- Added G28 home Gcode. This does the same thing as Home All Axes in jog mode.

*- Added automated job logging. When you run a file, the time it took, filename,
and date are written to a file (jobjog.txt). This is so you can track billable
machine hours, takt time performance, etc....
*- The backlash sense is now written to the ini file when you exit the program, and
restored when you start up again. This avoids position loss, as the backlash was
applied for the first move since startup. When you start with the -nopos switch,
the position and backlash sense will default to 0.

*- You are prompted on exiting whether you want to switch back to G53 mode to avoid
position loss if you left a tool offset active as you exited the program. Hit
enter to switch to T0 and save.

*- The current feedrate and modal Gcode are displayed in the STATUS window, right
next to the ABS/INC indicator. This should save you some grief when you're working
through your files.

*- Added alphanumeric menu selection as an adjunct to the numerical menus. Some

users prefer this sort of arrangement. Kudos to Harald Geier and Jerry Jankura for
programming this one.

BUGS:
-----

*- Backlash was not always stored to the ini file. Inifile logic has been beefed
up.

*- Abort only continues on 'N', to prevent trouble with a double ESC in a real
crash. You have a separate panic button, don't you?

*- Angular axes couldn't be set up easily in metric mode - fixed division by 25.4.

*- Arc feedrate was a bit strange, changed to strict Newtonian acceleration. Arcs
are also feed override cognizant, although you can't hit the override in the middle
of an arc and have it take effect.

*- Swapped G18/G19 definition - G02 & G03 were reversed in these planes. Sorry
about that.

*- An error indication will now pop up when the jogging increment is smaller than
the step increment instead of ignoring you.

*- When you had no .CNC files in the list directory, it was reported as a bad
directory input. Now it reports "no files found" if there are no files, and
"directory not found" as appropriate. Further, the filespec is now .* instead
of .CNC, since hardly anyone uses the .CNC extension anyway. You can enter just
the base filename now, eg: instead of c:\mycnc\part01.cnc , just enter part01, if
the default directory is c:\mycnc\.

*- G02/03 now reports an error if there are too many or too few axes, instead of
just skipping the line.

*- Fix MDI mode logic with the PGUP/PGDN controls, some actions produced array
overflows with the accompanying wierd output.

*- I beefed up the dry verify time estimator to account for acceleration and proper
feedrates on rapids. It should be a lot more accurate now. It also takes the
current override setting into account.

*- The timing algorithm is a bit more robust to avoid "sputtering" steps on some
systems. I can't test this here, so let me know if it occurs for you and what kind
of computer, OS, etc, you have.
*- Some trouble has been seen when using hard drive for Gcode instead of floppy;
timing collision? I think this might be improved in this rev based on the timing
refinements. But I can't test this one directly either.

*- Fixed arc problem. When either endpoint was at 6 o'clock exactly, the arc would
go around twice or do other strange things.

*- G16-19 now updates the status window immediately. Before, it wouldn't occur
until the next move was executed.

*- The G31 code has been revised a bit to make false misses less likely.

*- Feed override was a little wierd, it didn't work on arcs and affected rapids in
a peculiar way if it was over 100%. It no longer affects rapids of any stripe, but
overrides "on the fly" don't work while cutting arcs or during rapid moves. Sorry.
I'm not totally pleased with this arrangement either - this code needs to be
revised somewhat.

*** WARNING *** An important consequence of this change is that G00 moves are no
longer fully interpolated.

*- Added more logic for the special case of a move that only requires one step.

*- IPR logic was somewhat strange if the spindle speed was changed or the feed was
called out on every line.

*- Angular moves in incremental kept adding together.

DOCUMENTATION:
--------------

*- I went through the docs thoroughly, and fixed all the inconsistencies in the
notation and examples.

WNBR- A glossary of terms has been added.

*- A pdf manual is included with the download as well as the usual text version.
The text version has built-in line breaks now so it isn't so painful on a non-
wordwrapped screen.

*- Some source code docs are included so you can figure out what's going on a bit
more easily (registered users only). This is a "tip 'o the iceberg" thing, not a
full expository.

SOURCE CODE:
------------

*- The source code has been run through an auto-formatter to make the indentation
and casing consistent (Thanks Jerry). About 18000 lines of it now...