Re: [CAD_CAM_EDM_DRO] Re: Black box again

Alan Marconett KM6VV wrote:

> >From what I've seen, the problem is more that DOS can't see an LPT at
> C400h. The port doesn't appear to be there at all, it simply doesn't
> show up!
>

Well, DOS sure won't look for it at that address. Many earlier PCs
didn't even decode more than 10 I/O address bits, so the same I/O
addresses repeated from 0 to 3FF, then 400 to 7FF, etc.

> So you're doing full 1284 xfers (handshake managed by hardware)?

Umm, sort of. The 1284 spec is both hardware and software protocol.
We didn't want the heavy protocol overhead, with a CPU in the device
to identify its make and model, etc. But, the single I/O instruction to
handshake 1 to 4 bytes of data across the port is a HUGE help in keeping
the port from becoming the bottleneck. I can do the servo cycle for
3 axes in 50 uS on a 333 MHz Pentium II, and I haven't made the best
use of the 1284 features, yet! (This includes the PID calculation, etc.
to compute the next velocity to send to the motors.)

So, I just grab the port registers and start throwing bytes at them.
The real time section of EMC is pretty devoid of driver software,
anyway. Using single byte transfer cycles, it does a transfer every
1.2 - 1.6 uS. Using the multi-byte transfers, it should go a bit faster.
I will eventually get set up on the logic analyzer again and make new
measurements.

Jon