John Pettitt wrote:
> if there is an inverter in the system (a for example MAX 232 or equivalent)
> you'll need flag 2.

I use a VIA Epia CL6000 board, dunno what they integrated in their
chipset, but I do think they need some level conversion here or there.

Situation is that I need this:

server 127.127.20.0 prefer
fudge 127.127.20.0 flag3 1 flag2 1 time1 0.200

in ntp.conf to get close to the remote ntp servers' times.

I.e.: configure for high to low (CLEAR event) instead of low to high
(ASSERT) but also with a 200 ms (PPS width) compensation.
If I remove the compensation I get a GPS time that is 170 ms or so off
from the outside world.

Because of all this, could it be that the assert and clear events that
LinuxPPS reports are in fact reversed/inversed?

May  7 17:58:12 epia kernel: "pps_core":"kapi"[158]: capture clear seq
#137295 for source 0
May  7 17:58:12 epia kernel: "8250":"8250"[1273]: serial8250: PPS clear
event at 47888857
May  7 17:58:12 epia kernel: "pps_core":"kapi"[147]: capture assert seq
#137293 for source 0
May  7 17:58:12 epia kernel: "8250":"8250"[1269]: serial8250: PPS assert
event at 47889057

So how do I figure out the real situation with my GPS nd the ntp.conf?
How do I know what is the right setup?
Do some low level testing by connecting the 5V to my DCD line over a
resistor and see what LinuxPPS reports?

Any advice?

Udo
_______________________________________________
timekeepers mailing list
[email protected]
https://fortytwo.ch/mailman/cgi-bin/listinfo/timekeepers

Reply via email to