Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!uunet!husc6!bloom-beacon!oberon!cit-vax!ucla-cs!zen!ucbcad!ames!lll-tis!ptsfa!ihnp4!cbosgd!mandrill!neoucom!wtm
From: wtm@neoucom.UUCP (Bill Mayhew)
Newsgroups: comp.sys.amiga
Subject: Re: Insider Clock
Message-ID: <729@neoucom.UUCP>
Date: Mon, 12-Oct-87 13:10:33 EDT
Article-I.D.: neoucom.729
Posted: Mon Oct 12 13:10:33 1987
Date-Received: Fri, 16-Oct-87 00:08:11 EDT
References: <540@inria.UUCP>
Organization: Northeastern Ohio Universities College of Medicine
Lines: 56
Keywords: late french amiga
Summary: 60 Hz Amiga, 50 Hz power

In article <540@inria.UUCP>, rouaix@inria.UUCP (Francois Rouaix) writes:
> 
> I have somebody here on my BBS  who asks this question:
> he has a Franch Amiga (french voltage 220v, 50 Hz)
> He bought an Insider 1Mo expansion RAM with a permanent clock in the States.
> The problem is that the clock is always late.


You are correct in your assertion that your friend's Amiga
gradually looses time becuase european power is 50 Hz, rather
than 60 Hz.  The time-of-day on the Amiga appears to be incremented
by an interrupt service routine that is triggered by a 60 Hz sqare
wave from the power supply.  When the power supply is excited by 50
Hz, fewer interrupts per second are generated, thus causing the
clock to lag behind the proper time.

The battery backed-up Insider clock derives its time base from a
crystal oscillator.  When the Amiga boots, a program called by
s:startup-sequence reads the Insider's clock chip and sets the
proper time.  --so,... the time should be pretty much correct when
the computer is turned on, and then gradually get behind the
correct value.

There are a number of corrective steps that you can take:

The most severe step would be to europeanize the Amiga and outfit
it with 50 Hz video chips and firmware.  Obviously, this would also
entail getting a PAL monitor.  Likely, this is not a cost-efficient
alternative.

A second approach would be to modify the interrupt service routine.
You could normalize the amount the time-of-day register gets
incremented so that 50 interrupts per second would suffice.  You'll
have to consult with the ROM Kernel Manual to figure this one out,
as this is a bit much for my level of expertise.  An alternative
might be to read the Insider for each interrupt, but this would
prbobly use up a little more CPU time.

A third approach would be to generate your own 60 Hz reference and
substitute it for the signal provided by the power supply.  You can
use a National Semicondutor MM5369 17-stage oscillator/divider and
American TV 3.579545 MHz color reference crystal to make a 60 Hz
square wave source.  See page 4-3 in the 1978 Edition of the
National Semiconductor CMOS Databook.  The 60 Hz signal is supplied
to the motherboard on the pin of J-14 that is closest to the rear
of the machine (probably a gray wire).  At least this is true in
fairly old versions of the A-1000.

Since I'm an EE, the third way would be easiest for me.  For
software Gurus, the second choice would probably be best.

I'm sure there are probably better methods of solving the clock
problem, but these are the ones I thought of.

Bill
(wtm@neoucom.UUCP)