[LinuxPPS] Experience with current linuxPPS kernel patch.

[clemens at dwf.com]

Im happy to see this thread back in action. 
First to answer some of the questions that came up.

> What is your PPS source?

I have ONCORE VP, M12 and M12+T receivers.
The timing spec on the M12+T receiver, using the 'clock granularity message', 
which
the ONCORE receiver does, is 2ns.  So any problems being seen are not due to 
the
receiver.  Now interrupt response time in the Kernel is a whole different 
story, and
the lp interface being cleaner than the tty interface is the reason to prefer 
this
interface.  Unfortunately the sign of the pulse needed to trigger this 
interface is
the opposite of the tty interface, so one needs at least one more transistor to
make that work...

> Other question: How close to (a half sphere's) 2Pi angle does your GPS antenna 
> see the sky?

Mine is up on a flat roof, sitting (well it has a magnetic 'sticker') on top 
of a
Furnace/AC unit, so it has a nice large 'backplane' of metal around it.

While the crystal in the ONCORE receiver is not OXCO, its an order of magnitude
better than the 25c crystals used on motherboards.

> For last 120000 samples I'm seeing this:
> http://fedorapeople.org/~mlichvar/tmp/18xlvc_errdist.png

Thats an interesting plot, I should try doing something like that.

One thing it brings to mind tho.
Your plot is symmetric on the + and - sides.  
I do an offset vs time plot, and it shows the same symmetric fuzz around a 
central
value.  Now the central value can move up and down, especially at a reboot, but
also due to room temperature (I assume).  I guess the question with your plot 
is,
are we seeing fuzz around a stable central value, or are we seeing the central 
value moving back and forth?

The interesting thing here, is that when I do the same plot on data from ntp
on FreeBSD, the fuzz is ONE SIDED.  If you think about it, the one-sided 
'answer' is
the correct one if you believe that what you are seeing is delays in answering 
interrupts.  There is no way to answer an interrupt early, but plenty of ways 
to
answer it late.  So, the symmetry that you and I see on Linux puzzles me.

Just the one other thing I alluded to.
Someone mabe a year ago, on this list or the ntp list, was discussing problems
with the conversion of ns <-> us in the Linux kernel and libraries.  Seems 
that it
was to have been fixed by now, but I dont remember the details.  Also, there 
were
questions about the Linux internal clock, and which of 3 or 4 sources were 
being
used on a given machine and  given kernel.  Also whether the interpolation of
fractions of a second was being done correctly/well (thee was an 
initialization
problem).  I guess I would really like to know the status of all of those 
problems
today, because when the accuracy seems to be pegged at 1us, and the GPS unit
is good to 10ns,  and Machine interrupts are probably in the 100ns range, it 
just feels
like something is wrong somewhere...  And that symmetry.  Mumph.

-- 
                                        Reg.Clemens
                                        reg at dwf.com