Re: Dual port serial card?

[david@wraith.cs.uow.edu.au (David E A Wilson)]

gtolar@ghgcorp.com (Glynne Tolar) writes:

>In article <50pljr$rpg@flood.weeg.uiowa.edu>, snelson@news.uiowa.edu (Steven Nelson) wrote:
>> (Scott Barker):
>>> Has anyone ever heard of a dual-port serial card? I've got an apple // card
>>> which claims to be one. I've got no docs, though. Any ideas how it might
>> work?

>>I've never seen one, but one could be done.  Most 6850 and 6551 based
>>serial cards used only 4 addresses in the Cn8x space (6850 used 2, 6551
>>used 4).  Since 16 addresses are available in that space, a quad-port
>>serial card would not have to be too exotic.  Each port would be controlled
>>by a different set of addresses.  If phantom slots were implemented, even
>>more could be controlled.  

That describes a dual port card I built many years ago - 2x 6850 (one at the
standard offset and one 4 or 8 bytes away) plus a 74LS259 8 bit addressable
latch to select buad rates (2 bits for port 1 plus the /16 and /64 options in
the 6850 gave me 300, 600, 1200, 1200/75, 75/1200, 2400, 4800 in my original
design [port 2 was 1200 and 4800 only] and 1200, 2400, 4800, 9600 for both
ports when I modified the card to run my v32 modem) and provide DTR outputs. 

>If you used some of the space that a card's ROM uses you could put even
>more then 4 ports.  Or use the same address the 6551 does but another
>address to "bank switch" which port you want to use.  I would think a better
>card would be one with a microcontroller that could catch incomming data
>or offload outgoing data to take the load off the CPU and allow it to process
>other requests.  Like reading the hard drive (DMA stuff would get in the way
>of an IRQ).  But with a uC you could have plenty of time to play catch up. 

Are people still interested in such a card? Cirrus Logic has a nice 4 port
chip with builtin PPP support - just the ticket for the following:

>Maybe if TCP/IP for GNO/ME ever comes out the demand for such a card
>of the type I just described would be justify its creation.
-- 
David Wilson	Dept CompSci Uni Wollongong Australia	david@cs.uow.edu.au