Re: [CAD_CAM_EDM_DRO] PC based CNC system architecture/Black Box

Chris Paine wrote:

> I wont be offended if you tell me this wont work:-
> Supposing you control a voltage controlled oscilator as if it was a servo,
> Count the output pulses as if they were encoder pulses so you can start
> and finish a move at a controlled accelerate/decelerate and have nice even
> pulses for your stepper drives. I think the VCO would be followed with a
> flip flop or divider to give a nice square wave out.

The only reason this might be messy is that you need a VCO with a step-less
range of 10,000 : 1 or so. I can build a digital version of this quite easily with
a step-less range of 1 million to 1.

> The limit to step
> speed would be only how fast you can count output.
> The software would have to switch on one paralell port pin to raise VCO
> frequecy and another pin to lower so an input (serial port ?) counts
> oscilator output . so the program commands a move, the Vco starts from 0
> accelerating at a rate set by a resistor and capacitor untill the pulses
> comming back exceed or equal the desired rate, the port pin is swithed off.
> remembering the software is counting pulses, when it reaches a point maybe
> derived from the acceleration time it switches on the decelerate pin slowing
> smoothly to around about the right position (havent thought this out quite
> right yet) but the pulse count knows where we realy are.

How do you manage different rates for 3 axes? How do you create precision
rates, so that the direction of travel is as intended? Can a one-bit Delta-Sigma
DAC perform that well with only a few thousand samples / second?
I'm a little dubious. With that RC, there is only one rate for acceleration, but
the several axes moving simultaneously need to have their accelerations
proportional to their final velocities. I think you'll have either large
jumps in velocity, or a severe limit in acceleration from too much smoothing.

> Can a PC manage to
> keep a good count while deriving a speed figure and dead reckoning positon,
> also may be re calculating while decelerating? I shold mention In am
> thinking of a high resistance and capacitor on the VCO input which controls
> the maximum rise and fall time to suit stepper hardware.
> If it works It would be very cheap apart from programmers time......
> I suppose you will say that phase jitter on the VCO would spoil the nice
> step pulses but I think it could work ok.
> This idea is cheap and old technology so I could afford it and understand
> it!

You're planning on using a serial port to count the step pulses? That takes
one serial port / axis, I guess. Yes, it is simple to understand, but it is also
fairly simple to see the severe problems such a simple scheme causes.
You really would need to change the RC for every acceleration ramp
needed - messy. I think any scheme to sample the step pulses and not
ever miss one is bound to limit the speed severely. I think a dedicated
counter will be much better.

This idea is so cheap that it could be worse than what is being used now.

Jon