Programming docs for the ISDN-Surfer

This document is freely distributable as long as it is not changed, and one
of the sources www.jschoenfeld.com, www.jschonfeld.de, www.siliconsonic.com
is mentioned. 

last update: August 2nd, 2000

Vendor ID: 4626 ($1212)
Product ID: 5   ($05)
Serial ID: 0 for Siemens Chipset. Other IDs reserved. If you write
           your software, make sure that the serial ID is checked properly!

The autoconfiguration is optimized to fit in a single MACH110. One drawback
had to be taken in order to make the design fit: The ISDN surfer can only
be mapped to addresses in the regular IO area. That means, if the IO area is
filled with other cards, the card will not work. However, this only happens
if the ISDN surfer is the 8th card in your system that uses IO address space
at $ex0000. This hardly happens, and if it does, a simple change of the board
order will make the system work. It's a real advantage over the old ISDN
Blaster board by VMC, because that one only worked properly as first, second
or third card in the system.
The data bus is buffered, meaning, the board should work in all known Zorro-
setups. The bus design has been optimized for working with bad or even
missing clock signals, so it should even work in RBM-towered 1200s. Another
enhancement over the ISDN Blaster is the S0 bus part, that one is now
symmetrical (as recommended by Siemens), and it uses a VAC ISDN transciever.
VAC is a Siemens daughter, they guarantee to meet all the requirements of the
used chipset.

Memory map:

All addresses are byte-addresses. The whole board has only got even
addresses. Do not use word-accesses to the board, because they may harm
future options.

$0000-$00fa  Configuration nibbles
             The board does not only say that it supports shutup_forever,
             it really supports this feature. Although it is not used in
             normal configurations, it is implemented here. Usually, you
             don't use the airbag of your car every day, but you want to
             know that it's there!
$00fc        IRQ lock register. A write to this register locks the IRQ
             selection. Once the IRQ is locked, the write-accesses to $00fe
             and $0240 do not have any effect. The lock can only be removed
             with a hardware reset. It is recommended to use the lock option,
             because there has been a lot of software that plays around with
             hardware registers. Intentionally or not - it could stop the
             product from working properly. Locking makes the hardware more
             stable, and more convenient for the customer (because there are
             no jumpers on the baord).
$00fe        IRQ select register. A read from this register tells you which
             IRQ has been selected: If the MSB of this byte is set, the ISDN
             surfer uses IRQ6. If the MSB is 0, IRQ2 is used. This is useful
             for cross-checking if the lock-access has been successful.
             A write to this register selects IRQ2, no matter what value you
             write. 
$0100-$01fe  Siemens ISAC-S
$0200-$02fe  IRQ control, continued
             after a reset, IRQ6 is selected. If IRQ2 has been selected by
             writing to register $00fe, it can be switched back to IRQ6 with
             a write to $0240.
             $0200 read: Test reset access  (hardware testmode only)
             $0200 write: Test set access 0 (hardware testmode only)
             $0220 write: Test set access 1 (hardware testmode only)
$0300-$03fe  Siemens HSCX
$0400-$7ffe  mirrors - do not acces, reserved for future use.
$8000-$9ffe  clock-port (spare_CS only)
$a000-$bffe  26-pin expansion port

The hardware testmode is not implemented in all boards, because it does not
fit in a MACH 110 chip (a minimum of 210 is required). The testmode can only
be activated in our facilities, or with the special equipment that has been
handed out to our betatesters. Table of testmodes:

mode bit 0    mode bit 1        chosen mode
0             0                 2189/3
1             0                 5000/1
0             1                 2092/64
1             1                 5001/1

The testmode bits are NOT reset with a hard-reset, so the desired mode
stays in the MACH chip after a warm start. Since the only reset to these
registers is the read from $0200, the combination is undefined after a cold
start. The modebits are locked with the IRQlock access, so once the IRQ is
locked, the mode bits cannot be changed until the next reboot.

the 10-pin port for the telephone module can only be programmed via the
Siemens Chipset, because it controls the IOM bus. The IOM bus is not
directly mapped into the Amiga's address space. It wouldn't make any sense,
because you wouldn't want a phone module that loads the CPU because it
cannot talk to the ISDN chipset directly.

If you have any questions, feel free to ask me via e-mail:
jens@siliconsonic.de