On Mon, 4 Nov 2002 13:42:08 -0000
Richard Urwin <[EMAIL PROTECTED]> wrote:
> The suggestion is not entirely out-to-lunch. Computers certainly used to
> use the line frequency for the real-time clock; the PDP-11 comes to
> mind.
>
> In fact the line frequency is a better solution than crystal because, at
> least here in the UK, the _average_ line frequency is guaranteed to be
> 50Hz to a very high precision - much better than a crystal. This is
> achieved by adjusting the short-term line frequency to correct
> accumulated errors. My old bed-side alarm clock used line frequency and
> it never needed adjusting except for power-cuts and summer time.
AFAIK the same system is used in the Netherlands. Indeed I had a digital
alarm clock once that I'm sure used the line frequency to keep it's time.
(It's microprocessor got it's clock signal from a conventional circuit I
think; the microprocessor just monitored the powerline signal). It's the only
way to keep accurate time if you don't have access to an accurate timing
source via the internet or such to sync with now and then
(or your own atomic clock :) )
> You would, of course transform it down to a lower voltage first, but
> nobody suggested otherwise. It wouldn't blow up a motherboard just
> because it was AC. In fact it would be dead simple to design a PC board
> to take a feed from the AC and provide a Linux driver with a 50/60Hz
> signal. The hardest bit would be getting the AC feed from the
> self-contained power supply.
And the most dangerous to you and your hardware :) How about this: an
external box connected to an AC outlet, inside an optocoupler with
schmitt-trigger(sp?) and a connector for a serial cable? Connect to a spare
serial port on your PC, you could get a nice TTL signal on one of the
readable pins on the port.
And you could pick up the 50 / 60 Hz EMR 'out of the air' (well there should
be an AC power source nearby, and it may not work inside a well shielded
and grounded computer).
Open spec (I hope its not patented...) hardware design and a GPL driver:
now you can use line power frequency as (yet another) timing source
in Linux. Oh well :-)))
> Does the GMT/LOCALTIME mismatch solve the problem? If so the clock
> errors would all be of the form "Damn! The clock is off by exactly an
> hour again!" (or three hours...) And first in one direction (using
> Linux) and then the other (using Windows.) Neither of these actually
> alter the hardware clock unless you tell it to, except for summer time,
> which is topical right now, but each OS should only ever do it once and
> only by an hour.
>
> If it is ever off by some minutes and seconds GMT/LOCALTIME is not the
> answer. It can only make errors of whole numbers of hours.
>
> --
> Richard Urwin, Private
<snip>
I guess I got a bit of topic, sorry :)
-Frans
Want to buy your Pack or Services from MandrakeSoft?
Go to http://www.mandrakestore.com
