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

David Lord snews at lordynet.org
Thu Feb 23 01:30:20 UTC 2012


unruh wrote:
> On 2012-02-22, David Lord <snews at lordynet.org> wrote:
>> unruh wrote:
>>> I have a garmin gps 18 connected to my computer with a long (maybe 15m)
>>> cat 5e cable, with the PPS carried on one of the twisted pair. I figured
>>> it would be a good idea to terminate the cable with a 100 ohm resistor. 
>>> One testing this the other day, I notices that the signal level was down
>>> at 1V, with a staggered risetime of the pulse. -ie it would
>>> exponentially rise, to a little plateau, then rise a bit more. Thinking
>>> that I had underestimated the cable impedance ( it is after all a single
>>> sided pulse, not a balanced signal) I upped it to 200 ohm. Now the pulse
>>> rose to 2V but with a very similar shape to the rise. I finally removed
>>> the resistor entirely, and now got a 4.5 V pulse, but the shape of the
>>> rising edge remained much the same. I would have expected a much sharper
>>> rise, with ringing , but no ringing in evidence. I do not understand
>>> this. Clearly the 100 ohm was overdriving the output of the gps, but the
>>> cable should have looked like 100 ohm to the pulse anyway (at least at
>>> first). The open termination of the line should surely have resulted in
>>> much more structure to the pulse. (The scope's input impedance should
>>> not have altered things much since that is more like a Mohm.)
>>> From this it seems that trying to terminate the line is a mistake, and I
>>> do not understand why. 
>> PPS from my Garmin is at TTL level.
> 
> Yes, that is what I get if I leave the cable unterminated.
> 
>> There are scope traces in "The ART of Electronics" in
>> section on RS422 where the traces are bumpy ramps with
>> several plateaus due to the reflections.
> 
> That is what it looks like, no matter what the termination is. 
> As the termination resistance goes up, the reflected pulse should be
> negative, giving ringing. But no ringing is seen.
> 
> 
>> For RS232 the conditioning is by limiting the rise time by
>> a capacitor across the output and from MAX232 datasheet
>> about 330pF would seem ok for limiting the slew rate.
> 
> Slowing down the risetime is not what I want. That pulse edge of the
> PPS  gives the time. 

300pF gives a slew rate of 30V/us which isn't too much
in the way of slowing down. The slew reduction leads to
a more consistant timing at expense of slight added delay.

Using balanced RS422 would give less delay and still be
reliable with the receiver switching at the crossing point
of the two signals.

I was doubtful of RS232 being useful over 15 metres so I
added a 17 m length to my cable run to check. I've not
noticed any difference and except for the two 30us blips
from cron jobs the offset is mostly < 5 us. This from the
Sure GPS with PPS modified to use the spare rs232 port.

typical line from peerstats:
        ident  cnt   mean   rms   max delay  dist  disp
127.127.22.2 5383 -0.000 0.004 0.032 0.000 0.473 0.236

David



More information about the questions mailing list