RE: [CAD_CAM_EDM_DRO] STEPSTER: how many use it and are you aware of any particular problems

HI Daniel,

Very interesting! I never would have thought one could implement a CNC
controller on a stamp!

You might consider moving to a PIC, and a BASIC interpreter. Better yet,
write it in C. You can even get a free "LITE" C compiler from Hi-Tech.
Seems like we always need more speed in CNC controllers, and compiled
(compiled BASIC?) code runs much faster.

Sure hope your 9V batteries don't die in the middle of a long machining
operation! ;>)

I found it interesting that you're using ULN2803A drivers to run your
steppers, I just designed a PIC board with ULN2803A drivers to drive my
Geckos (boards should be in today!)!

I'd forgotten how much had been done in Kevin Carroll's STEPSTER. I recall
studying it probably 5 years or so ago.


Alan KM6VV

> -----Original Message-----
> From: CAD_CAM_EDM_DRO@yahoogroups.com
> [mailto:CAD_CAM_EDM_DRO@yahoogroups.com] On Behalf Of D. Daniel McGlothin
> Sent: Thursday, July 21, 2005 8:39 AM
> To: CAD_CAM_EDM_DRO@yahoogroups.com
> Subject: RE: [CAD_CAM_EDM_DRO] STEPSTER: how many use it and are you aware
> of any particular problems
>
> Brian,
>
> Thanks for the information.
>
> Alan,
>
> Thanks for the encouragement. I've made a modest start on the project.
> If
> so inclined, you can read about it at
> http://mcglothin.us/RobotScrapbook/CommunityPcbMill2005/BasicStampGCodeInt
> er
> preter. It is just a start, and has not been tested on any 3-axis
> hardware
> yet.
>
> Daniel
>
> > -----Original Message-----
> > From: CAD_CAM_EDM_DRO@yahoogroups.com
> > [mailto:CAD_CAM_EDM_DRO@yahoogroups.com]On Behalf Of Alan Marconett
> > Sent: Tuesday, 19 July 2005 12:49
> > To: CAD_CAM_EDM_DRO@yahoogroups.com
> > Subject: Junk E-Mail RE: [CAD_CAM_EDM_DRO] STEPSTER: how many use it and
> > are you aware of any particular problems
> >
> >
> > HI Daniel,
> >
> > Ambitious undertaking!
> >
> > If you don't mind Pascal, I'd look at TurboCNC. Minimal
> > contribution needed
> > for source access, well worth it.
> >
> > Stepster is no longer supported to my knowledge, TurboCNC has a
> > development/support team.
> >
> > Controllers that process Gcode don't normally process Gerber or Excellon
> > files, at least not hobby programs.
> >
> > Are you really going to move an aperture, and expose film? Or do you
> just
> > want to extract the paths?
> >
> > I added limited capability to my STEP4 CNC controller program to
> > read Gerber
> > files, and generate Gcode. A tough job! And I'm limited on how
> > big a board
> > I can handle. Are you reading RS-274D (hope I have that right)? Then
> > you'll also have to use the aperture data to build a "pad table" of some
> > sort. Gerber files are also "scaled", i.e., no decimal points
> > are used. So
> > the values might be 1020 for a 10.20" coordinate with a particular
> scale.
> >
> > There is also a header that tells (on RS-274D?) what the scale is, board
> > size and a few other things I forget.
> >
> > Assuming you are generating ISOLATION traces (big assumption!),
> > then you'll
> > need to transverse the Gerber data, and build a b-tree or similar
> > structure
> > to hold nodes. The nodes would be the pads, vias or endpoints of your
> > traces. Each node would have the pad size, maybe hole size (mind
> didn't),
> > and pointers to the parent and children (traces to other pads).
> > You'll also
> > want the trace sizes (apertures) in this node.
> >
> > Once you have the tree built (an array), then you can discover the entry
> > points for all of the nets.
> >
> > To generate Gcode for a trace (actually, I just generated PLC
> > code to import
> > into Vector CAD/CAM) you just "walk" each net going around one side of
> all
> > the traces and pads. You will arrive back at the start of the
> > net. You'll
> > want to add an offset to the Gerber path as you go around a net, such
> that
> > 1/2 the trace width and the radius of each pad you encounter is
> > accommodated. Don't forget, you can have round or square pads, and
> > intersections, and angled traces, and ....
> >
> > Go through all the nets, and the PLC (point, line, circle) or Gcode you
> > directly generate describes the board.
> >
> > The Excellon file I think, is just a list of holes and sizes, with the
> > coordinates for them. Not hard to generate a Gcode file from
> > them, or do a
> > simple conversion to PLC and import them into Vector CAD/CAM.
> >
> > With the PLC imported into Vector, Any fixes can be done, and then you
> can
> > let Vector generate the Gcode. Each net generates a tool down, a
> > transverse
> > of the net (go around the net and come back to the start), and
> > then tool up.
> >
> >
> > You'll want to arrange the start of each net in the most efficient order
> > (traveling salesman algorithm), of course!
> >
> > I found it an interesting project! I wish you luck on yours!
> >
> > Alan KM6VV
> >