On 3/22/2014 9:58 AM, Martin G. McCormick wrote:
Jerry Stuckle writes:
That wouldn't work well. Remember, computers are not the only ones which
use UTC - in fact they are the most imprecise. There are many clocks
around
the world which are synchronized with UTC via radio, i.e. WWV/WWVH in the
United States, CHU in Canada, and other stations around the world.
I once accidentally got a Linux system set up with the
wrong algorithm which I don't really remember what I did, but it
maybe added the leap seconds twice as it was twenty-something
seconds slow which is not a good thing at all. The hardware
clock, being nothing but a crystal oscillator tied to a counter
whose final resolution is 32 bits can be extremely accurate but
unless you are running some sort of specialized setup, you can
expect some jitter due to all the tasks your system is doing at
any given time. I have an old Dell Dimension with a 600-MHZ
processor and use NTP to keep the time synchronized as well as
possible, but I notice that chron jobs seem to have a window of
about a second in which they can fire. Sometimes, it is right on
the dot and the next time it may be a second late but not much
more.
By the way, those "Atomic" wrist watches and clocks you
can buy listen to something called WWVB which transmits at 60
kilohertz from the same location as WWV. There is no voice or
tone for a human to hear, but the signal transmits a 59-bit word
if one wants to use that term, which contains several BCD digits
expressing what the UTC year, month, date, hour and minute will
be at the next 0TH second. It's a leap of faith that
civilization will still be here then.:-) There is at least one
extra bit that sets during DST and clears in Winter. 60 KHZ is
called ELF or Extremely Low Frequency and hugs the Earth. Our
WWVB has been received in New Zealand though this is not a
trivial task.
Martin
Crystal oscillators are stable - but they are not necessarily accurate.
Their frequency needs to be adjusted using accurate frequency counters
(which themselves need to be calibrated, usually to WWV or similar
signal - which is very accurate).
Computer manufacturers do not adjust crystal oscillator frequencies for
accuracy - in fact most MBs don't even have an adjustment. And even if
they did, there is always the problem of temperature drift. Software
running on the system will not affect the hardware clock, but the
software can affect how quickly interrupts are handled.
And yes, WWVB is also used by many people, mainly because since it is a
ground wave, propagation distance (and hence, delay time) can be easily
calculated. However, as you mention, while it works well in North
America, it doesn't work so well on the other side of the world. There
is a limit. WWV and WWVH also transmit BCD time signals and operate in
the MF (Medium Frequency) and HF (High Frequency) bands. Although it is
more difficult to predict propagation delay times due to ionospheric
reflections, knowledge of the current conditions and multiple
measurements of a period of time provides a high degree of accuracy.
I had a Heathkit digital clock about 30 years ago which was synched to
WWV (10 Mhz, IIRC). I loved that clock - always accurate and never had
to be set :) I've thought about about designing something similar for
here, but just too many other projects ahead of it.
Jerry
--
To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Archive: https://lists.debian.org/532d9fc4.7090...@attglobal.net
