Re: PC based CNC system architecture

Mike, i think the problems with Jons soultion for EMC and stepper
control is as follows.

Jon please correct me if i am wrong.

1) It is a closed proprietary system.
2) Jon is building what he wants with no input from the group.
3) it will be fairly expensive. From memory $350.00 or so.

Wally K.




Mike Gann <mikegann@h...> wrote:

> Jeff -
>
> I think what you say below is correct. I also think
> this is exactly the thing Jon Elson is working on.
>
> Here is a quote from a post Jon previously made:
> (Jon - I don't mean to throw you in the middle of this)
> ******************************
> What I am working on is a 4-channel programmable pulse generator
> that would actually send the step and direction pulses to the

stepper

> driver. Each channel would be given a velocity value and a

direction.

> It would also generate quadrature pulses that followed the step and
> direction signals, and could be used to drive old phase-drive type
> stepper drivers, as well as be counted by an encoder counter, for
> open-loop systems. For closed loop, the encoder counters would
> read actual encoders on the motors, or on the machine itself.
>
> I am running EMC with a 100 MHz Pentium, and updating the
> servo loop at a 1 KHz rate, with plenty of CPU power to spare.
> This type of device would look about the same to the CPU,
> encoder values are read in, servo calculations are done, and a new
> velocity command is written out.
>
> A 24-bit counter/comparator circuit running at 10 MHz (100 nS
> resolution) could output up to 5 Million steps / second (although

the

> jump from 3.333 to 5 million is obviously too big a jump).
> But, reasonable step rates would be well in excess of 100,000/sec,
> which might be needed for people using high microstepping
> division ratios. And, the timing resolution there would still be
> 100 nS, giving extremely fine velocity control, no large jumps
> in velocity, and smooth acceleration and running at any sped
> desired.
>
> The low end is one step every 1.6 seconds. The code could either
> stop the counter if no steps are needed, or just let it bump every
> 1.6 seconds, first one way, then the other. This would cause
> some VERY slow and small hunting, and might not be objectionable.
> If it is, the code could shut off the pulses when the commanded
> velocity is below some threshold, like .001" / sec.
>
> Jon
>
> ***************************************