[LinuxPPS] parallel PPS

[Remco den Besten]

Hello Paul,

I saw the large jitters too but I do not see them anymore after using time1 
1.0
The PPS (driver 22) and NMEA (driver 20) driver(s) stick to each other here.

Just wait and see ;-)
I promise, when the system goes out of lock, I post it here.

Regards,
Remco

> Remco
>
> The Rockwell Jupiter module has an interesting history. I think the early 
> ones
> outputted 'NMEA' data in a binary mode. (After all, nobody expected them 
> to
> be used as we do - the manufacturers assumed they would be built into car
> GPSs etc.) Later ones output the NMEA data as a standard serial stream. 
> The
> problem is that they only output the NMEA data when the processor is not 
> too
> busy. Hence the one second offset. Unfortunately they sometimes output the
> data two seconds late, and may spend hours doing this. As far as I could 
> see
> this is not related to satellite visibility. If you don't actually see it
> happening, the obvious effect is a surprisingly high jitter on the NMEA.
>
> Paul
>
> On Friday 29 February 2008 20:55, Remco den Besten wrote:
>> Hello Paul,
>>
>> Indeed several people tell me to 'forget' the NMEA from the Jupiter.
>> When I use time1 0.0 or 2.0 I have the 'second or two seconds' 
>> phenomenon.
>> But,  I did not google intensively on this matter.
>>
>> However, as far as I can ascertain I did (and do? ;-) not have this 
>> problem
>> using a 1.0 second
>> time1 which I determined empirically (to be honest, I also tried time1 
>> 0.0
>> and time1 2.0 and
>> saw strange things happen ;;-) .
>>
>> Unfortunately I was not able to use the 31 driver because I received
>> errors, see:
>> http://rembl.org/index.php/2008/02/16/rockwell-jupiter-and-linuxpps/
>>
>> I am experimenting with several combinations now:
>> 'Time' from DCF, internet and GPS, in combinatin with GPPS (PPS derived
>> from GPS) or Rb87
>> for the precision. The combination Rb87 (PPS) and internet time (from PTB
>> Germany) worked well.
>> (see http://remco.org/rb87)
>>
>> This afternoon I recompiled my 2.6.24 kernel with CONFIG_HZ_100=y and
>> now have a contrapsion where Jupiter NMEA delivers 'the time' and the
>> Jupiter PPS
>> the precision.
>>
>> We'll see ....
>>
>> Cheers,
>> Remco