Re: [CAD_CAM_EDM_DRO] Re: UCMI Interface

Art Fenerty wrote:

> Hi Matt:
>
> My understanding is that the APIC was disabled on earlier motherboards, but
> is now universally enabled on all future and present chipsets. Some
> motherboards do disable it, but only via software.It can be turned on very
> easily by an application. What makes it attractive to me is that nothing
> else seems to use it. Master5 has problems with some motherboards due to
> their doing funny things with the int8 interface, but an APIC should be able
> to be programmed to be very fast and asyncronous for all axis. That should
> give resolutions down to uS levels. I think I'll have to write some test
> routines to see how it works. Maybe I can upload them and ask people to try
> the test applications just to see what resolution would be available and how
> many people could use them.

I'm not that familiar with this APIC, I'm guessing it is sort of a combination
of a counter/timer and an interrupt controller?

I made up a gadget (in an FPGA) to produce step pulses with high resolution,
no glitches when being rapidly updated, and logic to enforce the correct
setup and hold times between the step pulse and the direction signal.
This does a lot more than the programmable interrupts, but these extra
things could mostly be done in software. The tough thing, however is
to always know how many pulses went out. Now, if the APIC is actually
interrupting the CPU, and only the program can make the step pulses,
then maybe that solves the problem of knowing exactly how many steps
were made. Since my logic produced the pulses completely autonomously,
I had to also have a counter to count up the steps. This counter was
set up to actually count quadrature phase signals made from the step/dir
signals, so that it could be hooked to an encoder instead.

But, anyway, you need to take into account the odd requirements for setup
and hold of the direction signal. I assumed most stepper drivers needed
the direction line setup prior to the step, and maybe held a little while
after.
The Gecko drives need the direction line held stable for 200 (or is it 500)
microseconds after the step pulse.

Jon