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

Wally K wrote:

> > Matt wrote:
> > The above configuration would require 5 boards of approximately this
> > cost (the cost of the DAC boards must be higher since the Burr-Brown D/A
> > chips are not exactly cheap, and the digital I/O board hasn't been
> > invented yet, but I expect it will probably cost about what the
> > quadrature decoder boards cost), plus a motherboard/card cage assembly.
> > That's where that number of $800-$1000 comes from.
> >
>
> Wally replys:
>
> This is were i loose you. When you say the above configuration...ect
> i am not sure what you are talking about. At one point you talk
> about 8 channels then later you talk about 5 boards. I do not
> understand what these boards are for or what boards are required for
> servos and what boards are required for steppers.

A Servo-to-Go card usually has 8 channels of D/A, 8 channels of
quadrature input for encoders, and 32 bits of digital I/O (it also has 8
multiplexed channels of 8 bit A/D, and an interval timer that can
generate interrupts to the PC, but I've never used either of these
features). The system Jon is working on is modular and consists of an 8
slot motherboard with a DB-25 connector near one end which connects to
the PC through a cable to the parallel port. There are (will be) a
number of different plug in cards that each perform a different
function. The list is (so far):

4 channel 16 bit D/A

4 channel quadrature decoder

general purpose digital I/O (probably 24 bits, Opto 22 compatible, not
invented yet)

machine tool I/O board with estop control, home & limit inputs, spindle
control relays (to pick the coils of larger contactors), etc. (not
invented yet either)

4 channel stepper motor control (possibly the same board as the
quadrature decoder board with different pld programming)

Using a combination of up to eight of these boards you could configure a
system to control just about any machine tool. Using the example of the
Servo-to-Go card you would need 2 D/A cards (4 channels each for a total
of 8), 2 quadrature decoder cards (4 channels each for a total of 8),
and a digital I/O card. Obviously, if you had only 3 or 4 axes to
control, you could use 3 cards instead of 5.

With respect to stepper motor control, if Jon is successful in putting
all the logic necessary into 1 card, then you'd have a motherboard and
1 stepper card (up to 4 axis) (this ignores the digital I/O issue all
together).

> You seem to have a billion dollar card cage in here someware with
> million dollar connectors. What is wrong with simple cheap header
> connectors.

1. Unless they are equipped with a latching tab they pull apart rather
easily, and you can only imagine what happens if you accidentally
disconnect the encoder cable while the axis drives are enabled...

2. Most connectors of this type made by Molex, Amp, Panduit, etc.
require special tooling to apply the plug end , usually a special
crimper for the contacts. The connector I'm interested in using combines
the ease of screw terminal termination with the ability to unplug from
the mating connector on the PCB. One factor that made these expensive is
that we needed a miniature version of the connector because there was a
limited amount of room for (4) eight pin connectors mounted end to end
on the front of the quadrature decoder card.

> And are you physically building a mechanical card cage.

Yes, it's actually cheaper than buying a pre made one.

Matt