Re: Thunderclock and UCSD

[mjmahon@aol.com (Michael J. Mahon)]

MouseBoi 2K asks:

>I have a question.  Does anyone know the actual machine calls and
>memory-mapped
>I/O locations of clock cards?  (thunderclock, timemaster, any other popular
>clock cards)?  It's impossible to find any documentation at all for the
>cards,
>and near impossible to find the drivers (and then extraction of useful
>information is near impossible).
>
>I only ask because I'm writing some clock device drivers for the Apple II to
>support an operating system.

You will find that there are really only two useful sources of information
on setting and reading these clock chips:  reverse-engineering the
existing drivers, and reading the chip manufacturer's data sheet.

There are several "flavors":

1) parallel (byte/digit at a time) and serial (bit at a time)  Serial is
   more common because the chips have fewer pins and are cheaper
   to buy and integrate.

2) slot-based cards vs. ROM-socket "no slot" devices.

The slot-based serial cards are most common, and most use a clock
chip that does not store the year, and so the driver inherits the task
of doing a calendar congruence to infer the year within a five or six year
range from the month, day, and day of week.

To understand these cards, disassemble the drivers, both in the on-
board ROM and in the (mostly Apple-provided) drivers in ProDOS.
There's some fairly simple bit-banging going on, and the day of week
congruence calculation is a little tricky (watch the carries), but it's
all there.

No slot devices are also serial, but must be enabled with a peculiar
pattern of address accesses to enable it to replace ROM data with its
own when a particular address is read.  After enabling, other accesses
control the chip and its shifting out of the date.  Again. disassembly of
the driver(s) and study of the data sheet(s) is the only way to really
understand these beasts.

-michael

 Email:  mjmahon@aol.com
 Home page:  http://members.aol.com/MJMahon/