[LinuxPPS] ntpd

[Udo van den Heuvel]

Folkert van Heusden wrote:
>>> *) Udo: you see? gps is waaay off compared to other sources on the net and my local DCF77 receiver
>>>    so that is why one should use gps only for PPS
>> Just NMEA will always be off.
>> When correlated to PPS ntpd knows what time it was at the PPS.
>> So your  remark says very little.
>> Who can explain Folkert that NMEA plus PPS is not a bad practice?
>> Folkert did not accept the analogy of the phoned time signal.
>> Better check the config of your Garmin if the PPS is on. See the url of
>> the ntp reference driver page I sent you. It has instructions.
> 
> I verified that.

So the config is OK. Pulse length config is OK.
Does the GPS see sats? Does it know a position? How many sats?

> The device also responded correctly to my commands
> (e.g. when I said "be silent", it became silent).
> Now the testpps thing still outputs nothing.

So no PPS if the kernel was patched OK.

> I saw by the way that the 'on-line' status pulses between off-line and
> on-line when I am in minicom.

Online == Carrier Detect?
So there is PPS.
So the kernel is not OK, assuming ppstest is OK and ntpd can find your
pps source normally.



BTW: I was looking at the time_pps_findpath function but do not yet
really understand it. There is a PPS device defined like "/dev/ttyS2" or
similar but where does that info enter the code? What does the function
look for to know what port/device I want ntpd to use as a timesource?