[ntp:questions] Very large jitter and offsets on GPS ref clock after upgrade to "p5"

David J Taylor david-taylor at blueyonder.co.uk.invalid
Wed Jan 11 19:46:59 UTC 2012


"unruh" <unruh at invalid.ca> wrote in message 
news:yHjPq.70127$8O1.15459 at newsfe07.iad...
[]
> For timing it is only the rise time that is important. The length of the
> pulse is irrelevant, unless your Hardware/software requires a given
> length pulse in order to recognize and interrupt ( and if it did , it
> would be broken since it had better interrupt on the rising edge if it
> is to be useful to timing. )
> Certainly if you are using TTl-RS232 converters their times are
> important. However most rs232 cards will handle ttl signals fine anyway.

Distinguish, though, between internal-level signals where a 100 ns pulse 
may be quite enough, and signals sent down cables where the pulse length 
needs to be sufficient to allow the signal to be recognised at the other 
end.  I've seen RS-232 inputs with quite strong filtering, and taking 
115,200 baud as a typical maximum for older units, I would not want a 
pulse less than 10 microseconds long.  First guess estimate.

How well the rise-time needs to be preserved depends on the precision 
required - I would take more care where sub-microsecond precision was 
needed.

The 100 ms pulse from the GPS 18/x and the Sure board is nice because you 
can put an LED (or scope) on that line and see the pulse, giving visual 
confirmation of PPS data.

Cheers,
David 



More information about the questions mailing list