[time-nuts] SLIP vs Ethernet for NTP
magnus at rubidium.dyndns.org
Sun Oct 23 14:09:29 EDT 2011
On 10/23/2011 07:15 PM, Chris Albertson wrote:
> On Sun, Oct 23, 2011 at 2:28 AM, Iain Young<iain at g7iii.net> wrote:
>> Hi Guys,
>> I have often heard it said that since RS-232 is more "deterministic",
>> and suffers from less jitter, and uncertainties, than ethernet, that
>> it makes a better medium for time distribution (no CDMA for a start).
> I think this is only true if the Ethernet is using a shared media.
> Like the old coax 10BaseT cables. Modern system use switched
> 100BaseT that works much like a point to point link. In fact if you
> can do point to point RS232 why not point to point 100BaseT. Just use
> a crossover cable between two computers.
If you do a crossover cable and do the home-work in both ends on keeping
timing and delays where they should be, then it should be a good
solution. However, it is much easier to break it by myriads of reasons
than RS-232. Point-to-point and care in packet timing and you have a
good stable solution.
> The "best" why to distribute time over an area that is small enough to
> run cable is to distribute PPS. Most cat-5 cable has unused pairs,
> makes an easy way to send PPS.
You can lock up the 25 MHz oscillator and provide PPS and clock. Sync-E
locks up the baud-rate like that, but doesn't provide any solution for
time (it is not intended to).
There is also the AES packet solution which distributes the word-clock
on the remaining pairs such that audio over wordclock + Fast Ethernet
comes in the same connector. Cool solution which nobody seems to be using.
More information about the time-nuts