If I may chime in with my 2c, I looked at this not long after I
received my 7250 board because the drift was horrible like Don
mentions. Like Jim, I used adjtimex (which BTW was available out of
the box for me, so something must have changed there) to set a
manual drift compensation on boot up with the following line
in '/etc/rc.d/rcS.sysinit' ...
/sbin/adjtimex -t 10002 -f 920000 > /dev/null 2>&1
Those parameters were determined experimentally.
Is it possible that the source of confusion (and I acknowledge that
I don't know too much in this area, just enough to get the above
working) is that adjtimex works enough to be able to set the timing
parameters manually, hence working for Jim and myself. But the issue
that Bob is describing is that, by default, there isn't enough
info/resolution available from the kernel for automatic drift
compensation, by either ntpd or the more advanced functionality of
adjtimex, to function correctly (or at all ??).
And Don, for what it's worth, the above addition to my startup
scripts gives me accuracy in the area of 1 second per week (in a
quite stable environment though). Obviously, this is a per unit
setting, but for my product I will just develop a timing calibration
procedure when setting up each new board. I have to do other setup
tasks, acceptance testing & other hardware calibration anyway,
so this will just be one more step. This may or may not be useful
for your application but it is just an idea... Good luck
Cheers
Phil
--- In Bob Lees <> wrote:
>
> Jim
>
> As we say round here "You got me!" I don't have an explanation:(
I vaguely
> remember some early posts and I thought that you had changed your
divider on
> test for an optimal value? Although on reading the adjtimex man
page
> (followed own advice and rtfmed;)) it may be that the kernel
internals will
> work with whatever the clock gives it and adjust accordingly. The
values
> from adjtimex -p would presumably take account of the lack of
precision in
> the stock gettimeoffset.
>
> Bob
>
> On Tuesday 18 April 2006 23:17, Jim Jackson wrote:
> > Bob,
> >
> > I'm baffled by this. I'm using a ts6 kernel and a rebuilt
busybox with
> > adjtimex. I ran the arm developemnet image with ntpd and used
the results
> > to feed adjtimex. I then ran live tests and fine tuned it to
give me
> > accuracy of 1 sec in 2/3 days in my environment.
> >
> > Something appears adrift here, or am I not understanding this.
> >
Yahoo! Groups Links
<*> To visit your group on the web, go to:
http://groups.yahoo.com/group/ts-7000/
<*> To unsubscribe from this group, send an email to:
<*> Your use of Yahoo! Groups is subject to:
http://docs.yahoo.com/info/terms/
|