Re: Black Box Goals

Ballendo,

I got to start somewhere and that somewhere is to tackle the
difficult part first. Kind of like a roller coaster; a steep initial
climb followed by a fun downhill ride the rest of the way.

The process involved is an initial overview of the whole project so
it can be defined (Black Box Goals) and what resources will be
required (hardware and software). The project then breaks down into
individual tasks. These are the equivalent of building the individual
pieces of a jigsaw puzzle. While each piece is being built, there is
no thought of the whole puzzle, just the faith all of the pieces will
come together properly because the initial planning was well thought
out.

I am working on the velocity manager which is the very heart of the
project. It requires the following parameters:

For linear interpolation:

VECTOR LENGHT
VECTOR ANGLE
or:
NEW X COORDINATE
NEW Y COORDINATE

For circular interpolation:

START ANGLE
END ANGLE
RADIUS

For all motion:

VECTOR VELOCITY
BASE VELOCITY
RATE OF ACCEL/DECEL
ACCEL/DECEL PROFILE (linear ramp, S-shape, parabolic, user-defined?)
SEGMENT OR CONTINUOUS?

The "velocity manager" uses those parameters to generate all motion
within the 1,024 Hz interrupt service routine. The background program
either generates or looks up those parameters and feeds them in a
timely fashion to the velocity manager. Depending on how it presents
those parameters to or from the velocity manager, it will also do:

PAUSE (end of segment or right now, fixed time or switch input)
GEAR RATIO
HOME
ABSOLUTE AXIS POSITION (on the fly)
HARDWARE HANDSHAKE ( moving/stopped output, go/stop input)
JOG (joystick, quadrature, whatever)
READ HARDWARE INPUTS (1,024 times per second)
WRITE HARDWARE OUTPUTS (ditto)
RUN AUTONOMOUSLY (once programmed, runs without a PC attached)

The program will be open architecture. Anyone that wants to can get
the Rabbit Development system with everything they need to change the
program and flash it to the Black Box ROM for $129. If you want
a "dumb box", remove and replace the software so your PC can run the
all axies directly. If you don't like how it does things but want to
keep the velocity manager, you will be able to do that. Like Burger
King, "you can have it your way".

Right now I've got linear interpolation running just fine according
to the memory dump results. Circular interpolation is a cinch and
will be running in a week. Right now I don't think about or much care
how the data will get into the box, G-Code or other, just that the
motion routines work. I will care when this task is done. Next is to
spring for prototype axis cards ($1,000) so I can solder them to the
Rabbit prototype board and actually generate pulses instead of bytes
for the output.

Mariss

--- In CAD_CAM_EDM_DRO@y..., "ballendo" <ballendo@y...> wrote:
> Mariss,
>
> First four?
>
> So we have:
> 0) rapid linear interp (or non-interp to position)
> 1)Linear interp w/feedrate
> 2 and 3) arc/circle interp at feedrate
> Do we get 4) delay, or pause?
>
> How is plane selection for arcs/linear handled?
> Homing to switches?
> Will you have "gearing" for threading, or will this be a
modification
> of the linear interp?
>
> Looking at the Centent unit, we will send setup parms to the b-box?
> But these won't include SPI? And motor accel/decel parms, like the
> centent unit?
>
> Why are you calling this gcode in? It really isn't, or am I missing
> something?
>
> Thank you in advance.
>
> Ballendo
>
> P.S. This will be an impressive product.