[ntp:questions] Oddities in termination of cable from gps18.

David J Taylor david-taylor at blueyonder.co.uk.invalid
Thu Feb 23 06:35:43 UTC 2012

> I could not get good result using cat-5 either.  My "fix" was to move
> the GPS and computer closer so I would not need a long cable.
> I sort of understand coax but cat-5 is a bundle of pairs.  If you use
> the cat-5 for both RS232 and PPS then the twisted pairs are
> un-balanced because the return current path is shared by the RX, TX
> and PPS signals.   I think that might have been my problem
> I was thinking about use balance line driver and receiver chips but
> then figured moving the equipment would be better
> Chris Albertson
> Redondo Beach, California

It's all guess work unless you actually /measure/ the signals.  For the 
sort of cable lengths being discussed, the cable acts mostly as a 
capacitive load on the output signal, and terminating in its 
characteristic impedance is /not/ the correct thing to do unless you add 
line drivers at the sending end.  The GPS 18x LVC signal is TTL level, and 
not line driver level.

Again, for the sorts of distance being discussed, cat-5 cable should be 
fine for the GPS 18x LVC, using the four pairs as:

- ground and +5V
- ground and RX
- ground and TX
- ground and PPS/DCD

unless you are very unfortunate in having the cable pass near some 
considerable interference source (where mains-level voltages and currents 
are being switched).  The signal level currents involved are negligible, 
but you /might/ want to add a small decoupling capacitor on the +5V at the 
GPS 18x end of the cable run.

If Bill is really unhappy with the pulse edge shape, and it appears to 
have a step (photo please!), he could try adding a small resistor /in 
series/ with the PPS line from the GPS to the cable, so that any 
reflections travelling back up the cable have a better chance of being 
absorbed.  Considering that these devices are specified at the microsecond 
level, and likely there is filtering on the DCD line inside the computer 
or chip, looking at reflections at the tens of nanoseconds level may be 


More information about the questions mailing list