HP98035 Real Time clock and AC5954N clock chip

[Chris Elmquist]

On Friday (06/07/2019 at 11:01AM -0700), CuriousMarc via cctalk wrote:
> > 
> > I wondered if it's actually a digtal watch chip (2.5V could have been a couple
> > of mercury cells in series, LED watches were not uncommon back then).
> > 
> > In which case it would not normally have come in a 0.6" wide DIP. Perhaps
> > normally it was a bare chip directly mounted on the watch circuit board or
> > something.
> > 
> > The DIP version would be unusual, which is perhaps why we can't find data
> > on it.
> > 
> > -tony
> 
> I think you are on to something. That would make perfect sense.

FWIW, there was an article in Kilobaud magazine, perhaps 1977 or '78
that described connecting a TI LED digital watch to the SS-30 bus in
the SWTPC 6800.

I built this then, wrecking the TI watch as a watch but it made an
excellent RTC for this machine.  Two AA size NiCADs were used to power
the watch and charged through a simple trickle charge when the machine
was powered up.

The interface to the host was through a 6820 PIA, with just the segment
lines and a couple of output bits.  The output bits indeed "pressed"
the buttons on the watch and the code would set it just like a human
would by stepping through each digit in sequence and "pressing" the
other button to increment that digit.

When reading the clock, the code would pretend to set the watch but never
increment any digit.  It would just press the button to advance to the
next digit, stepping through all of the digit positions until it had
read them all (both time and date) and then return this buffered result.

By pretending to set the clock but not actually changing it, it allowed
the code to know which digit was being displayed and therefore it did not
need to have the digit select lines brought into the host.

Chris
-- 
Chris Elmquist