Re: [CAD_CAM_EDM_DRO] Re: Smart box / Dumb box

Mariss Freimanis wrote:

> I will take the plunge on the axis pulse generator cards. It will
> cost about $2,500 to get 1,000 of them. I believe in it enough to do
> it.

I've been thinking about your idea and I had two comments:

1. I really like your pulse generator circuit because of the equal
frequency divisions throughout its range.

2. I have some reservations about the velocity/time/distance
relationship that you are going to rely on to determine your position. I
know you've used this before with microprocessor controlled equipment,
but I think there could be problems using it with PC control in "Dumb
Box" mode. Are you going to interrupt the PC through the parallel port's
IRQ line? If so, that's usually IRQ 5 or 7, and they're prioritized from
the lowest to highest numbers:

IRQ0 - timer tick
IRQ1 - keyboard
IRQ2 - daisy chain to second interrupt controller (RTC, hard drive, FPU,
etc.)
IRQ 3 & 4 - serial ports
IRQ 5 or 7 - you
IRQ6 - floppy controller

What could happen is:

a. Some device with higher priority than you requests service and you're
pushed on the stack for longer than your interrupt period and you miss
the chance to update your position data.

b. Windows (or another OS) puts out the "Do Not Disturb" sign by
executing a CLI instruction and your interrupt is ignored.

Some possible fixes are:

1. An external computer (which gets you closer to the "smart box" crowd
;) ).
2. Accumulate your pulse count in hardware out on your pulse generator
boards (my personal favorite).
3. Flag an error if your INT is not ACKed.
4. Return a count of interrupt attempts so the software on the PC can
adjust the position data to account for missed interrupts.
5. Use a RTOS like RTLinux that gives you (rather than the PC) control
over interrupt priority.
6. Use NMI and hook its handler for your position update & fifo stuffing
code (this will be like your dedicated microprocessor systems).

Matt