Re: A crazy idea

Stephen,

Good morning! This idea is not fully baked yet; it only occured to me
about an hour ago. Here's how it would work:

(1) Trip a 2uS one-shot on the 0 to 1 transition of the STEP/DIR line.
(2) Have the STEP/DIR line connected to the conventional DIR input.
(3) Have the one-shot Q-not go to the conventional STEP input.

The one-shot would fire on every STEP/DIR pulse leading edge. It
would time out in 2 microseconds. The Q-not output would then trip
the STEP input.

If the STEP/DIR line is "0" at that time, the DIR would be CCW. If
the STEP/DIR line is "1", then the DIR would be CW.

In other words, if the STEP/DIR line is "1" for 2uS or less, the
direction is CCW. If it is a "1" for more than 2uS, the direction
would be CW.

It is conventional for step pulses to be a constant width, say 1uS.
It is also convetional to invert the pulse polarity. The polarity
inversion would function as the "direction bit" then.

Mariss

--- In CAD_CAM_EDM_DRO@y..., "stephen_stallings" <stevesng@n...>
wrote:

> Hi Mariss,
>
> This sounds a bit like the encoding schemes used
> to record data on magnetic media. All that I am
> aware of have a time sensitive element that
> requires a one-shot or a PLL to recover the data.
>
> Can you describe how you would recover the data
> in the case of one step CW, pause, one step CCW,
> pause, one step CW?
>
> If time is a factor, then the software and the
> motor driver would have to agree and conform
> to a common time scale. This could be troublesome,
> as we already have problems with pulse jitter at
> high step rates.
>
> Regards,
> Steve Stallings
>
>
> --- In CAD_CAM_EDM_DRO@y..., "mariss92705" <mariss92705@y...> wrote:
> > Hi,
> >
> > Parallel port output CNC programs are limited to 3 or 4 axies
> because
> > of the limited number of port bits. Each axis requires a STEP and
> > DIRECTION bit pair. What if it could be done with a single bit

per

> > axis where STEP and DIRECTION were combined on the same line? It
> then
> > would be possible to run 6 to 8 axies.
> >
> > Here's how it could be done:
> >
> > (1) 100000000010000000001000... is CW
> > (2) 011111111101111111110111... is CCW
> >
> > In (1), a narrow positive going pulse would be interpreted as a

CW

> > motion. In (2), a narrow negative going pulse would result in CCW
> > motion at the same speed. One would simply invert (1) to generate
> (2).
> >
> > This scheme would place no additional burden on the software or

on

> > the bandwidth of the now combined STEP/DIRECTION compared to a
> > standard STEP pulse line. Decoding hardware in the drive would be
> > minimal; just a single quad gate package.
> >
> > So, what do you think, a crazy idea?
> >
> > Mariss