Re: Turbo232 UART Cart HDL

On 12/17/2013 10:04 PM, hwarin@neuf.fr wrote:
> Welcome, Spiro !
>
> At my point of view, there's really few interest in the use of any kind
> of other chip than a real 6551 (8250, 16x50, 6850) for Commodore 8 bits.
> It's available, cheap, well known, and it makes the job...

With one little exception. The NMOS version has a nasty bug in the way 
CTS is handled. If the receiver pulls CTS, the 6551 will stop sending 
immediatly, even if that means destroying the byte it's currently 
sending. This is mentioned in the datasheet for the 6551 from Rockwell 
in Figure 8. If I understand this right, it means that the Byte will be 
send again once CTS goes LOW again. The only way the CPU has to detect 
that state is that it gets an IRQ but the status register doesn't say 
that the Transmit  Register is empty. This is fixed in the 65C51, even 
if you pull CTS, it will finish sending the current byte.

If you take a look at the schematics for the Plus/4, you will see that
the CTS pin of the 6551A/8551 is tied to GND while one of the portbits 
of the 6529 handles the job from the software side.

The MOS datasheet for the 6551 doesn't mention any of this in clear 
words but when you know about it, you see it CAN be read that way.



> From
> memory, 6551 is (was) also in Apple's super serial card (2 ports,
> integrated into //c) and CBM128/P500 (1 port only - is it supported in
> kernel/basic ?).
>
> Long time ago, I had developped a 4 serial expansion board for
> A500/A2000/(A1000 & CDTV never tested mecanically) to setup some
> multinode Minitel server using four 6850, over one of the 8520s and
> inspired from an Atari Mega ST board.

There was a 7port serial card for the Amiga. It used a 65CE02 and 7 
8551. I once suggested to replace the crystal to get at least one port 
up to 38400 for the fast modems at the  time.

  Gerrit



       Message was sent through the cbm-hackers mailing list