Re: [CAD_CAM_EDM_DRO] Re: Shared Direction pins vs dedicated Direction pins

"Bernard R " wrote:

> Doug,
>
> While there is no technical difficulty doing time division multiplex
> with muliple controllers, I don't really see the need.

If you don't have the requirement, then you wouldn't see the need!

For example, a hexapod with material on a rotary table needs 7
motors, or 14 total S&D output pins (this exceeds the 12 available
on one parallel port).

> The current limitation is using a single parallel port as the
> computer interface.

TurboCNC (http://www.dakeng.com/turbo.html ) for example can use up to
3 parallel port cards simultaneously (I think). But thats not the issue.

> From what little I've seen the G2002 project is well in hand so the
> limited output pins should soon be history.

While the G2002 project is fabulous, that is also beside the point
(besides, the parallel port is practically free, whereas the G2002 is not).

> From the practical side I could see a lot of people having problems
> with the concept and trouble shooting such a system would be a nightmare

I disagree with you on that point, in fact it is simpler which is why
I am suggesting it. Assuming functional software, a VOM or
2 wires & 470 ohm resistor and an LED is all that is required for debugging.

----------------------------------------------------------------
Mariss wrote:

>Doug,
>
>I have a circuit worked out that is very simple that shares step and
>direction on a single line. If the duty cycle of the step pulse is
>less than 50% it's CW, if more than 50% it's CCW.
>
>CW = 100010001000...
>CCW = 11101110111...
>

That too would be useful (N output pins for N channels), but you'd need
drivers that recognized that encoding, and the software wouldn't be useable
for standard S&D drives.

Whereas with the shared Direction & dedicated Step pins (ie N+1 output pins
for N channels), standard S&D drives from many vendors would still be usable,
and therefore makes it an attractive project for the software authors.

Anyway Mariss, is it possible to shorten the Direction pin hold time?

In my naivete' I am assuming it is a simple matter to assert the Direction
pin for perhaps a setup time of 250 to 500 ns, and latch it on the Step
pulse (freeing the Direction pin to change any time after that).

Of course the other drivers sharing the Direction pin who are not Stepping
do not see a Step pulse (because the Step pulses are dedicated) and hence
pretty much ignore the current state of the Direction signal!

It seems so simple to me, but then I'm not privy to
the internal workings of those fabulous Gecko drives.

----------------------------------------------------------------------

In regards to general other Gecko drive improvements, I agree with some
of the other guys in these respects:

- 80VDC max upped to 200VDC range (90VDC & 180VDC are
standard servo motor voltages).

- custom, higher (than 20) amperage drives would be nice, to
drive mill and lathe main motors up to the 5HP (or even 10 HP) range.

- the base could be bigger, moving the mounting holes out from
underneath the wire connect interface.

- separate connectors for the 470uF capacitor, or an internal 470uF cap

- provision for remote fault LED (it would be handy to install
remote fault LEDs to be displayed outside the Control Box), whereas
the Geckos are usually burried deep inside the Control Box. Merely
having two solderable holes for a pair of external wires is sufficient
(although a built in 470ohm current limiting resistor would be great).

- find some way to boost the 50ma current source (to power the encoders)
up to 100 or 125ma. Many encoders are 65ma (exceeds the 50ma), encoders
with index channel are up around 90ma, and differential encoders require
up to 125 ma.


OK, thats my 2 cents worth, now I'll be quiet.

However in summary, I'd like to say I'm extremely happy (no, make that
EXTREMELY happy) that Mariss is making available his great drives to our
cnc community. My 'improvement' comments only result from his direct
request a few days ago.


Doug Fortune
http://www.cncKITS.com











.