Long-haul optical transmission of non-standard protocols
There was a discussion here about 10MHz transmission across optical
media a few years ago[1] that I missed at the time, but have kept in the
back of my head as something I do want to play with at some point.
Here's a talk from the recent CCC conference about running alternate
protocols across optical transport systems.[2]
https://www.youtube.com/watch?v=3qojgJGtTos
By day I do planet-scale networking and would end up talking to our time
people, who run one of the largest fleets of distributed cesium clocks
on the planet every year or so when they'd ask for dedicated wavelengths
for their time synchronization systems and have to explain to them that
at truly long-haul distances (thousands of kilometers) the distances
aren't constant for a bunch of reasons, and by the time they handle all
of that they might as well just use their existing network paths.
1: Thread subject "Query about List and about 10 MHz Distro" from August
2021
2: There's a few inaccuracies in this talk, some are clearly handwaves
for simplifications, but a few seemed to be actual errors, if you want
to play with this yourself, please find someone who actually does
optical networking in practice before you go too deep into design.
Julien,
Thanks for that. It's a good video (36 min) and worth watching to the end.
If you know the author have him try sending a 1 us wide 1 PPS instead of
audio. That way he will get a direct measurement of toslink latency
without having to pattern match on a 'scope. It would also lower his
baud rate to 1, improving his previous record of wasting fiber bandwidth
by a factor of a million. For extra credit, he could use a time interval
counter instead of a 'scope and then calculate the consistency of
latency of the optical link. This would result in an interesting ADEV
stability plot.
Thread subject "Query about List and about 10 MHz Distro" from August
2021
https://febo.com/pipermail/time-nuts_lists.febo.com/2021-August/thread.html#start
The optical thread itself starts here:
https://febo.com/pipermail/time-nuts_lists.febo.com/2021-August/103830.html
/tvb
For truly long haul (as in solar system scale) time distribution, there are all kinds of interesting things going on.
There's a CCSDS (Consultative Committee for Space Data Systems) working group on Time Management within the Systems Engineering Area. (Disclosure: I'm the newly appointed director of the Systems Engineering area.)
They are working on standards for time distribution. Right now, the existing standards cover time code formats and things like time correlation (the process of reconciling spacecraft on-board time reports with some other time scale like TAI).
There's also a whole effort on establishing a Lunar Time System, at least for US Government purposes. There's a White House directive saying "make it so" and handing the responsibility to US Naval Observatory (who are the "keeper of time for astronomical bodies for the US Government" - they publish the Nautical Almanac, for instance) and NIST (who we all know). Mind you, that could all change in a month with a new administration.
https://public.ccsds.org/about/techOrganization.aspx. - they're in the middle of moving the ccsds website so this may break or change.. https://ccsds.org/ should always work.
https://public.ccsds.org/about/sea.aspx#_SEA-TIME - is the Time Management working group (I might add that if you're in a position to contribute, I think that contribution would be appreciated. There's a LOT going on with PNT in space these days.)
https://www.whitehouse.gov/wp-content/uploads/2024/04/Celestial-Time-Standardization-Policy.pdf
On Mon, 6 Jan 2025 16:38:28 +1100, Julien Goodwin via time-nuts time-nuts@lists.febo.com wrote:
There was a discussion here about 10MHz transmission across optical
media a few years ago[1] that I missed at the time, but have kept in the
back of my head as something I do want to play with at some point.
Here's a talk from the recent CCC conference about running alternate
protocols across optical transport systems.[2]
https://www.youtube.com/watch?v=3qojgJGtTos
By day I do planet-scale networking and would end up talking to our time
people, who run one of the largest fleets of distributed cesium clocks
on the planet every year or so when they'd ask for dedicated wavelengths
for their time synchronization systems and have to explain to them that
at truly long-haul distances (thousands of kilometers) the distances
aren't constant for a bunch of reasons, and by the time they handle all
of that they might as well just use their existing network paths.
1: Thread subject "Query about List and about 10 MHz Distro" from August
2021
2: There's a few inaccuracies in this talk, some are clearly handwaves
for simplifications, but a few seemed to be actual errors, if you want
to play with this yourself, please find someone who actually does
optical networking in practice before you go too deep into design.
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe send an email to time-nuts-leave@lists.febo.com
Hello everybody!
It's always a good day when you work appears on one of the mailing lists
that you subscribe to purely to lurk on!
First of all, yes there are certainly parts of this talk that are not as
accurate as they should be, however they were simplified for the
audience and the general goal of the topic (ultimately this was a talk
that was basically half comedy, half education)
It did come across my mind that it would be very useful to attach and 10
mhz input into the setup, but i haven't really put enough thinking into
the implications of doing that. Maybe if others can put a "this is going
to go wrong because {X}" I'm all ears.
Happy to take feedback on it, but do keep in mind i was not trying to
the a one hundred percent accurate talk, as reality presents a lot of
nuances which are not necessarily very entertaining for the average
person who just wants to see something silly happen and lot a little bit
about the gist of optical transmission stuff.
Cheers
Ben
On 1/6/25 05:38, Julien Goodwin via time-nuts wrote:
There was a discussion here about 10MHz transmission across optical
media a few years ago[1] that I missed at the time, but have kept in the
back of my head as something I do want to play with at some point.
Here's a talk from the recent CCC conference about running alternate
protocols across optical transport systems.[2]
https://www.youtube.com/watch?v=3qojgJGtTos
By day I do planet-scale networking and would end up talking to our time
people, who run one of the largest fleets of distributed cesium clocks
on the planet every year or so when they'd ask for dedicated wavelengths
for their time synchronization systems and have to explain to them that
at truly long-haul distances (thousands of kilometers) the distances
aren't constant for a bunch of reasons, and by the time they handle all
of that they might as well just use their existing network paths.
1: Thread subject "Query about List and about 10 MHz Distro" from August
2021
2: There's a few inaccuracies in this talk, some are clearly handwaves
for simplifications, but a few seemed to be actual errors, if you want
to play with this yourself, please find someone who actually does
optical networking in practice before you go too deep into design.
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe send an email to time-nuts-leave@lists.febo.com
Hi Ben,
Even if it was a talk designed for entertainment it was highly
inspirational and confirmed real possibilities.
I had seen that page a week before on this same list about the HH3 clock
at cern https://dvaluch.web.cern.ch/hh3/
And the last images were very appealing to the nuts side of me,
confirming that it was indeed used in quite a serious manner.
I have completed my own design of such a board using components
available to me at farnell. I believe it would work for transport of a
10 MHz reference, as SFPs are made to transport fast signals, so they
will have fast rise times, and the LVPECL transceivers also have fast
rise times.
It should be ordered soon.
I'll open source everything after validation.
I also have a frequency counter that I will be able to use for allan
variance measurements, to see if fiber transport has any influence on
clock accuracy.
I'm new to time-nuts but not to hacking :-)
Sebastien
On 17/01/2025 17:22, Ben Cartwright-Cox via time-nuts wrote:
Hello everybody!
It's always a good day when you work appears on one of the mailing
lists that you subscribe to purely to lurk on!
First of all, yes there are certainly parts of this talk that are not
as accurate as they should be, however they were simplified for the
audience and the general goal of the topic (ultimately this was a talk
that was basically half comedy, half education)
It did come across my mind that it would be very useful to attach and
10 mhz input into the setup, but i haven't really put enough thinking
into the implications of doing that. Maybe if others can put a "this
is going to go wrong because {X}" I'm all ears.
Happy to take feedback on it, but do keep in mind i was not trying to
the a one hundred percent accurate talk, as reality presents a lot of
nuances which are not necessarily very entertaining for the average
person who just wants to see something silly happen and lot a little
bit about the gist of optical transmission stuff.
Cheers
Ben
On 1/6/25 05:38, Julien Goodwin via time-nuts wrote:
There was a discussion here about 10MHz transmission across optical
media a few years ago[1] that I missed at the time, but have kept in
the back of my head as something I do want to play with at some point.
Here's a talk from the recent CCC conference about running alternate
protocols across optical transport systems.[2]
https://www.youtube.com/watch?v=3qojgJGtTos
By day I do planet-scale networking and would end up talking to our
time people, who run one of the largest fleets of distributed cesium
clocks on the planet every year or so when they'd ask for dedicated
wavelengths for their time synchronization systems and have to
explain to them that at truly long-haul distances (thousands of
kilometers) the distances aren't constant for a bunch of reasons, and
by the time they handle all of that they might as well just use their
existing network paths.
1: Thread subject "Query about List and about 10 MHz Distro" from
August 2021
2: There's a few inaccuracies in this talk, some are clearly
handwaves for simplifications, but a few seemed to be actual errors,
if you want to play with this yourself, please find someone who
actually does optical networking in practice before you go too deep
into design.
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe send an email to time-nuts-leave@lists.febo.com
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe send an email to time-nuts-leave@lists.febo.com
