FUNKYWARE.ORG - COM_PORT.TXT - 26.07.1997


-----
FUNKYWARE.ORG - COM_PORT.TXT - 26.07.1997

NOTE: The follow is a series of e-mail exerpts exchanged with
      Ronald Anderson, discussing suitable uses for each port
      on the M-STE and TT. It was included here, because some
      people tell others they better use Modem 2, but fail on
      providing proper reasons as to why one should.

      Ronald has proven very knowledgable on many key aspects
      of Atari hardware and software, so while opinions given
      here are strictly his own, one will notice he knows how
      to support his point with strong evidences.

      About why Atari provided 76800 and 153600 on Modem 2, I
      recall Ronald pointing out that they are respectively 4
      and 8 times 19200. When the M-STE and TT were designed,
      these speeds looked promising because of that aspect.

      One might notice 38400 is twice 19200 (the only related
      multiple that became an established standard) as recent
      developments are all based on 14400 exponentiation.

***
To: q-funk@megacom.net
Subject: STinG DNS resolving on TT & MSTE

STinG was designed for compatibility to the port access methods
of HS-Modem, so unless you have some fully compatible replacement
I strongly recommend using HS-Modem.  The TT users I mentioned
always use 'Modem 2' for STiK/STinG.  'Serial 2' is too primitive
for modem operation, and 'Modem 1' is identical to that on an ST
(slow).

Of course, all serial ports should work with STinG, but a modem
needs control signals beyond what 'Serial 1' and Serial 2' have.

###
To: q-funk@megacom.net
Subject: Re: [2] STinG DNS resolving on TT & MSTE

>> Of course, all serial ports should work with STinG, but a modem
>> needs control signals beyond what 'Serial 1' and Serial 2' have.
>
>Using STiK, Serial 2 has proven both faster and more reliable.

Serial 2 uses an SCC port, just as Modem 2 does.  The speed
capability of those ports is identical, but there are some
differences:

1:  Modem 2 has the RI interface line, used for incoming modem
    calls. This does not matter for any outgoing calls of course.

2:  Serial 2 is actually the very same SCC port as is also used
    by LAN, which has three consequences:

2.1: LAN and Serial 2 activation each excludes the other.

2.2: A hardware selection is required to activate the hardware
     drivers for either LAN or Serial 2.  If that selection is
     in LAN mode, the Serial 2 port will be completely 'dead'
     and vice versa.  Not all terminal programs are capable of
     controlling this.

2.3: The start-up default is for the LAN hardware to be active,
     and so the default speed is also set for LAN use (highest).
     This may be why you have come to regard it as faster, but
     with proper software the Modem 2 port has exactly the same
     speed capability.

[ NOTE: as stated below, it turned out Modem 2 and Serial 2 have
        separate clocks, so changing the software will not solve
        this --making point 2.3 partially wrong--, hence why the
        14 MHz PCLK modification was invented. ]

>However, Serial 1 is indeed primitive and hardly usefull beyond
>driving a serial mouse!

For local STinG use it is equally useful as Modem 1, since it
is in fact an identical MFP port and STinG requires no physical
handshake.  Data loss is ensured against by the acknowledgements
of TCP protocol.

In fact IP itself is not designed to be 'lossless', since all high
level protocols are required to have their own error recovery.  I
myself have used Midi in my experiments with local STinG networks,
and that [ Midi ] has only TXD and RXD without handshaking, just
like Serial 1.

###
To: q-funk@megacom.net
Subject: Re: [4] STinG DNS resolving on TT & MSTE

>***Serial 2 deficiencies***
>
>Well, one mod I intend on doing is Franz Sirl's 12MHz upgrade for
>Modem 2, which make it possible for Modem 2 to acheieve speeds up
>to 230.4 like the LAN port. It is recognized by both HS-Modem and
>Fast_serial, so this should be a noticable improvement! ;-)

Sounds interesting.  I'm not at all well informed on the clocking
system used for the SCCs of Falcon and TT myself.  I guess I need
to read up on it, but I only have one machine with an SCC anyway,
so I cannot use these speeds in my networking.
(My 14k4 USR modem won't swallow anything faster than 38k4 either.)

###
To: dlanor@oden.se
Subject: Re: Modem 2 vs Serial 2

>Note that I have since checked the circuit schematics for the TT
>again and have found that the clock generation for the two ports
>of the SCC is indeed separate.  This means that some modification
>of the hardware may be necessary to achieve full baudrate control.

This is what the modification described in Fast_Serial does, and
HS-Modem can also recognize if a PCLK of 14 MHz is used, instead
of the stock 8 MHz speed.

>I still say that Modem 2 has the same potential for speed as does
>Serial 2/LAN, though I now realize it may need some clock changes.

Indeed! ;-)

###
[ NOTE: The modifications discussed above are described in my
        English translations of texts by Franz Sirl and Harun
        Scheutzow, found elsewhere on this site. ]
-----
NOTE:  those who need to print this may crop HTML tags above and
       below the five dashes and re-save as straight ASCII text.



   HOME


