Re: [CAD_CAM_EDM_DRO] Re: PC based CNC system architecture

Wally K wrote:

>
> Mike, i think the problems with Jons soultion for EMC and stepper
> control is as follows.
>
> Jon please correct me if i am wrong.
>

Somewhat out of order:

> 2) Jon is building what he wants with no input from the group.

No, actually Jon is building what _I_ want, with no input from the
group. Jon had done one other PC board project for me in the past and I
was very pleased with the results, so a while back I called him and told
him I'd like him to build me a clone of the Servo-to-go card with maybe
a few small changes. At the time I thought we could:

1. Cut the cost.
2. Make some improvements to the quadrature decoder circuitry.
3. Come up with a better interconnection scheme for the field wiring.
4. Add something to the board to alleviate the stepper pulse generation
problems.

Well, then we stated talking about ISA vs. PCI and after some research
we decided that a PCI implementation was more expensive than we wanted,
and it didn't do much to improve the wiring problems. At some point we
started researching ways to mount the card external to the PC's case for
several good reasons:

1. Ease of wiring.
2. Ability to use it with laptops and other types of computers.
3. Jon really wanted it to be modular so that it could be configured to
suit almost any type of control application.

I thought we should just use a single board and put enough stuff on it
to suit 80% of all users, but Jon pointed out that it would add cost for
people who didn't need all those features which is another problem with
the STG board. I later came around to his way of thinking on this and
other things as well.

At this point we spent quite some time researching how to connect the
controller to the PC. I read most of the USB spec and concluded that it
wasn't feasible since USB devices can only start data transfers at 1ms
intervals. The data rate is high, but sending a command and receiving a
reply would take 2ms minimum due to the way the bus works. I'm still not
sure whether this is built in to the USB hardware. If anyone knows the
USB interface at the hardware level, I'd like to hear your thoughts on
this. Another problem with USB is that someone will inevitably plug
something else into the bus and goof everything up.

High speed serial was also a candidate, but we eventually settled on the
parallel port using EPP mode. It's true that ISA slots are disappearing,
but PCI will be around a long time and PCI bus parallel port cards will
as well. Other crazy ideas (mine) such as IDE, were thankfully
discarded. The guy who mentioned no-protocol ethernet might be on to
something. Firewire might be good as well, but you have to join some
industry association to get a device address and I think the dues were
like $5000 or $10,000.

Another area of research was what chips to use and I'm getting a little
tired of typing so I'll spare you the details, but Jon wanted to use
PLDs for most of the digital stuff and I disagreed, mainly because I
didn't want any problems with the supply of a single source device like
that, plus I don't fully understand them (at least not the new complex
ones). Now that I've seen all the stuff Jon crammed into these things I
see why he wanted to do it this way. One other decision was to use 16
bit DACs rather than the usual 12 or 13 bit ones.

I guess I'm most responsible for the packaging decisions and I might
post a DXF of the finished card cage one day soon. My two most important
issues were ease of mounting inside a control cabinet, and dirt simple
field wiring without special tools (crimpers, etc.) The whole unit will
snap onto a DIN rail, and all the field wiring (except the parallel port
cable) will be to detachable screw terminal plugs from Phoenix Contact.
No special tools are needed, each cable (like each encoder cable for
example) will have it's own connector, and they can be plugged and
unplugged quickly. The down side is cost, each of the connectors costs
several dollars, but I feel very strongly about ease of installation and
service so I feel it's worth it.

> 1) It is a closed proprietary system.

Not so, it will come with schematics, programming information and a
theory of operation so that you can understand how it works and fix it
if need be. I don't think anyone would want this as a kit, since most of
the ICs are surface mount.

> 3) it will be fairly expensive. From memory $350.00 or so.

My own personal cost goal is that finished units with the capabilities
of a STG card will sell for $800-$1000. This is basically what it costs
to get the STG card going ($888 for the card, 2 ribbon to screw terminal
interface boards at $75/each and $15-$20 for cables). Obviously, a
simpler configuration to run steppers would be less. You might think
this is expensive, but only a few years ago when I first started fooling
around with making CNCs, a commercial quality CNC control, without
motors or drives, cost $15000-$20000. Using the EMC, a PC, and something
like a STG card, or what Jon is designing, this cost is now only $2000
or so. For another comparison, AHHA sells there Artisan software and ISA
card combo for $984. Perhaps volume production will bring this down
some, but CNC controls aren't exactly a household item, so I don't know
how that will play out.

Matt