[ntp:questions] Very large jitter and offsets on GPS ref clock after upgrade to "p5"
David Lord
snews at lordynet.org
Tue Jan 10 11:24:45 UTC 2012
David J Taylor wrote:
> "David Woolley" <david at ex.djwhome.demon.invalid> wrote in message
> news:jefokf$9m0$1 at dont-email.me...
> []
>> To the extent those constraints don't apply and both ends are
>> terminated well above the characteristic impedance, the output voltage
>> actually goes up in a staircase, with the steps being the round trip
>> time.
>>
>> Some other mistermination conditions cause a ringing approach to the
>> final value, which is why it is better to operate with a high
>> impedance load, and therefore capacitive characteristics.
>
> I wish Chris would just look at the remote signal with a 'scope! A wide
> pulse over the length of line he talks about should be no problem at all
> (but a microsecond-wide pulse might). The timing PPS I've seen are in
> the tens of milliseconds wide. If there is overshoot, perhaps a
> capacitor to slow things down might help, or on the other hand, if the
> edge is too slow providing a better match might help.
>
> Of course, "clean" transmission does rely as I said before on each
> signal being provided with its own ground in the twisted pair, a point
> also made by Bill Unruh. For PPS a screened 50-ohm coax cable is not
> needed except in extra ordinary circumstances (exceptional noise or
> sub-microsecond accuracy). Even level converters may not be needed -
> even though the specification says so. Many RS-232 ports work just fine
> with TTL levels.
My Oncore docs give a suggested circuit with a 74HC132:
GPS-6_1PPS --+-----------------+ +--- MAX232_T2in
| | |
| 4 +-- |
1 +-- 3 & o--+
| & o-- R=10k --+-- 6
2 +-- 5 |
| +-- C=470p -- +5V
|
9 +-- 8
| & o-- R=820 -- D=PPS LED -- +5V
10 +-- K A
I think that the generated pulse is only about 5 usec and was
doubtful it would work for me over a 15-20m cable run. I've had
regular rs232 work without problem over the same distance in
the past.
I've not started on this yet but will test with a short cable
run first and increase the pulse width as needed.
David
>
> Cheers,
> David
More information about the questions
mailing list