Re: [CAD_CAM_EDM_DRO] Re: Running small machine without computer

The hwml looks really promising. There is another product, almost ready,
called the e-leadscrew, that would also work.
But its new, untried, and all the gotchas ...

What fred smith says is one option ... the ncpod, but you wont save
money, imo.
There is a new hw coming for the mach sw, with usb interface, thats
another (a usb stepper something).
Geckodrives has the g100 - it could run all 4 machines in or from one
box, and would work for your use. 500$, and one would be enough.
I believe industrial plc´s could do this. Expensive and reliable.

What I was trying to express was the the truth that every hw attempt to
date on the planet has faced - 90 to 98% of the work is not in making
the machine work and make the first movement. Its all the ancillary stuff.
Please, I applaud the efforts of the various hardware developers, and
fully support them. The only thing I am saying is that unless You
yourself will support the product, and are willing to put the 100+ hours
into it, the project will produce huge headaches.

Many people on this list could do it in hardware, easily. I am not one
of them. the hw costs are 10$. Making it costs 2000$, at hobby-price of
10$/hour, for a very experienced, talented designer.
My expertise has been systems integration, not hardware design.

If you want it cheap, reliable and soon, get a standard pc. It is the
only reliable, available, cheap standardised hw platform on the planet.
Another things all exotic form-factors share is the likelyhood that in 2
years they will not be available, or if they are, they won´t be
identical or compatible. This has generally occurred to every successful
product on the market. And the unsuccessful ones are not on the market
anymore.

Good luck with your project, h.

> Thanks very much...I see your points.
>
> It's for space saving(among other things).. It's a setup
> to do multiple operations on the plastic block. Being
> as high speed machining is out of reach for me, the
> idea is an attempt to have a semi-continuous flow of
> parts being made. No tool changing mostly...
> I make a groove...move the part to another machine
> to round the slot or maybe dovetail it.
> I'm aware there are things I haven't considered
> yet. An experiment mostly.
>
> I haven't been paying attention to the advancements
> with the micro-controllers and thought I'd recalled
> such a thing being available.
>
> A friend sent this link....
> http://www.hwml.com/ <http://www.hwml.com/>
>
> But the tiny computers is a do-able thing as well.
> It just seems there should be some type of flash
> drive controller that'd execute the simplest of
> gcode files.
>
> I appreciate your insights and experience :)
>
> John
>
> --- In CAD_CAM_EDM_DRO@yahoogroups.com
> <mailto:CAD_CAM_EDM_DRO%40yahoogroups.com>, hannu <hvenermo@...> wrote:
> >
> > I have never used the programmable stuff, but have heard about some.
> >
> > Why do You wish to NOT use a std pc ?
> >
> > If your answer is cost, the truth is you will be MUCH better off buying
> > cheap, integrated pc´s for this. They will be easier and cheaper to
> > purchase, install, set-up, troubleshoot and maintain. they may cost
> > less. They are quite small in size.
> >
> > This reminds me of some instances where people just will not believe
> > that they should not try to use a laptop for application "x" ... The
> > situation might be different, but probably not, if you needed to set up
> > 100-200 machines. Economically, your break-even point is somewhere way
> > out there ...
> >
> > The ONLY instance you would want to do this, would be a
> > cost-does-not-matter scenario like aircraft avionics (space),
> > hi-end-boating or submarines (space, wet, power consumption) or
> > spacecraft (mass). I have not worked on space stuff, but extensively on
> > all the others.
> >
> > You can get cheap, integrated pc´s for about 200$. They look like
> stereo
> > boxes, and have std parallel port. This is what you want.
> > You *can* get special-purpose bare-board pc´s for about 100 with an
> > exotic process and packaging, but the likelihood of problems and the
> > guaranteed non-availability of technical support regarding machine
> > controllers makes this a very bad idea. I highly recommend going with
> > packaged, std, bare-bones pc´s, best if you buy nos and a spare machine
> > or 2.
> >
> > IF you really want to do this with a "small" dedicated funky-looking
> > box, you will spend a lot of time trying and testing weird problems
> (98%
> > probability) and have no spares available when something goes wrong.
> You
> > have 99% plus probability that the machines are not available in exact
> > same format when they blow a power supply etc. which will happen, and
> > the support costs & problems will be very much more complex and
> > expensive. You may want that for job security, which in itself is a
> good
> > enough reason, you will know this if so. In that case your extra,
> wasted
> > hours, in this endeavor may well make sense, but I really wish to
> > emphasize, again, that it will not be easier or cheaper. Your hw cost
> > may finally be a very little lower, but you will definitely waste over
> > 10x more in other parts of the job.
> >
> > For me, fwiw, I have about 20 years experience developing,
> successfully,
> > solutions like this, and would never consider anything other than a pc,
> > for 4 machines. IIRC, my first robots where phone answering machines,
> > diskless, with touch-tone decoders we built, booting into dos, via
> > netware spx (cheaper boot roms) off a windows nt 4.0 server, running
> > modified basic and a bulletin board. I got about 3 months/70.000 calls
> > 100% successfully solved on the best robots, this was in about 1987.
> > Good luck anyway, your YMMMV.
> >
> > If you do decide to go the dedicated-micro route, good luck.
> > Also, to not discourage you, most if not all of the problems you will
> > encounter are not insurmountable, and are fairly easy to solve. It can
> > be done, it´s just too much work to make economic sense.
> >
> > As a guesstimate from a long-time contributor to these arts, I would
> > expect about 60-100-200 hours getting non-std hw to run reliably and
> > well, and at least one blown/failed purchase. This includes
> > purchasing/sourcing time, all installation stuff, reliability testing
> > etc. The hours are best-case, median, probable.
> >
> >
> > > Hi to All,
> > >
> > > I'm setting up a row of 4 small desktop
> > > routers. They'll be single purpose machines
> > > to do the same task repeatedly and nothing
> > > else. This will be basically milling grooves
> > > in 1" blocks of plastic that will be no
> > > larger than 5" x 6".
> > >
> > > But I don't need 4 computers or want to
> > > swap wiring...I want them all running at
> > > the same time. Load and let run.
> > >
> > > Where or what would I look at, as far as some
> > > small single purpose controller for
> > > each machine. It doesn't even need a video
> > > monitor....just push go and let it do it's
> > > job. Is there a small controller that could
> > > be programmed easily with a single gcode
> > > file or maybe a flash drive?
> > >
> > > Thanks in advance!
> > > John
> > >
> > > __._,_.__
> >
>