Re: [CAD_CAM_EDM_DRO] Re: CNC division

Hi Ian,

I haven't exactly gotten my head around moving the X, Y and A axis all
coordinated to do a cut like you propose, but as long as each discreet
angle is calculated "individually", I.E., not "added" relatively from a
previous move, then you still get to sub hundredths of a degree for
resolution (in my Sherline example). Now if you are writing a program,
just always calculate X, Y and A points as ABS. Relative mode may sound
like an ideal way to do the "loops", but you may loose a little, as you
have observed.

On the "little Italy's", do you really have to use X & Y in combination
with A? I would think just a linear and a rotational axis would
suffice. If you have a .DXF file of your part (or a similar part file),
I and perhaps others would like to see it!

I can see why the loop would be much easier for you to manually code.
I'd like to explore this a little more. I KNOW I don't know much at all
about clock making, and 4 or 5 special cutters CERTAINLY implies to me
that this is a very complex part! But I think I could learn something
from the concept.

Anyway, all I'm saying is that you should be able to position to ANY
point on the work using an X OR Y axis and the table, within the
calculated resolution and backlash limits of the hardware. And that's
all you need. I hope I haven't made any BIG obvious blunders in my
appraisal of your project and spoken TOO far past my current
understanding! I'm sure someone can correct me if I have! (and I'll
learn a little more...). ;>)

Best regards, and good luck on your project!

Alan KM6VV
P.S. I'd love to see a .DXF file!

Ian Wright wrote:

>
> ----- Original Message -----
> From: <rab@...>
>
> > I'd have thought that using any ratio of wormwheel and / or pulleys
> > that will divide up cleanly into 360 should be a good start, so that
> > a specified number of steps will give you exactly one whole degree.
> > I'm testing a 120 tooth wormwheel with a 200 step motor to turn a
>
> Hi,
>
> This is really the point I was making - most divisions DON'T divide cleanly
> into 360. Wheel counts which I often have to divide include 7, 11, 13, 17,
> 54 and 366, none of which divide cleanly into 360. I appreciate that, by
> drawing the wheel out in a CAD program, it should be possible to generate
> g-code which minimises errors but I actually want to do something a little
> more complicated than the norm and was hoping there might be a way to
> separate the division from the cutting. I'll explain - What I am wanting to
> achieve is cutting unusual, specialised tooth forms by generating them. If
> you can imagine a disk with 13 maps of Italy sticking out from the edge this
> will give some idea of the shape of a watch escape wheel. To cut such a
> wheel normally involves buying or making 4 or 5 special cutters at a cost of
> around 40UKP each - not very realistic for a one-off. So, I reckon that,
> with a bit of clever maths, it should be possible to persuade a little cnc
> mill to generate the form by coordinated movements of the A, Z and Y axes
> while running the thin, round edged slitting saw cutter back and forth with
> the X axis. I haven't yet found any software which would seem able to
> produce the g-code for this from a dxf file (unless you can suggest one) and
> so am almost resolved to coding it manually, in which case, having a method
> of producing a 'loop' to do the tooth shape and a separate 'loop' to index
> from one tooth to the next would seen to be easier than coding the whole lot
> as one long file.
>
> Perhaps I am wrong, I have been before - it was Tuesday I
> think.....196x..... ;O) <g> Any sugestions or observations would be very
> welcome.
> Thanks to everyone else who has replied on this thread.
>
> Ian
> --
> Ian W. Wright
> Sheffield UK
> www.iw63.freeserve.co.uk