gnu not unix wrote:
> In message <[EMAIL PROTECTED]> you write:
>> Udo van den Heuvel wrote:
>>> If the commandline at http://time.qnan.org/ for shmpps really configures
>>> for low-to-high (i.e.: rising edge) then why do I see -180+ ms offset
>>> for my Garmin GPS 18 with rising edge configured?
> (...)
>>> So what is the correct way to tell ntpd what's going on?
>>> Is there some peculiarity going on?
>>> Or am I missing something obvious?
> 
> You are using a TTL (well, CMOS) level signal to drive the
> RS-232 input of your serial port. The expected level conversion
> that should be done is thus missing. This level conversion also
> does a signal inversion. 

Ah, this helps explain some of the stuff.

> The only thing to care about is that one uses the correct side
> of the PPS pulse. The "rising edge" (as seen in the CMOS level
> signal) is the proper edge to trigger on. The falling edge 
> will exhibit the "200ms offset" symptom, along with jitter.

So with

server 127.127.20.0 prefer
fudge 127.127.20.0 flag3 1 flag2 1 time1 0.200

I am indeed using the (inverted) rising edge of the original PPS signal
in its TTL form but then I should not need the 200 ms offset time?

Thanks, this helps a lot but not 100%.
What am I missing further?

The only issues I have left are the possibility for compensating the DCD
interrupt time of my system, but it looks like this time is below 1 ms?
Also there is the situation that ntpd chooses an external clock as the
clock to use and marks my GPS clock as a falseticker. (!)
How can I fix that?
_______________________________________________
timekeepers mailing list
[email protected]
https://fortytwo.ch/mailman/cgi-bin/listinfo/timekeepers

Reply via email to