Re: [time-nuts] Fw: Optical transfer of time and frequency

[Scott McGrath]

I seem to recall that NASA used optical inferiometry at LC39 for time transfer 
during the shuttle program for the DSN the inferiometry was used to normalize 
the delays in the fiber distribution system caused by temperature and gravity 



> On May 5, 2016, at 2:44 PM, Tom Van Baak  wrote:
> 
> Hi Mark,
> 
>> Interesting... what happened on day 4 to send them all bonkers?
> 
> What you see as bonkers starting the 5th day is actually normal behavior for 
> a cesium clock (or any clock, for that matter). I mention this plot because 
> it shows that you cannot get 500 ps over a day, even with a 5071A -- which is 
> why we're all trying to help the OP with alternative methods of time 
> synchronization.
> 
> There are several ways to plot phase data. What people often do is plot 
> residuals for the entire data set. That is, showing how good the clock would 
> have kept time *if you knew ahead of time what the rate would be*. The first 
> 5 days of the plot shows the residuals based on 5 days of rate fitting. At 
> that point the calculated rate was fixed and the clocks continued to tick. So 
> it becomes a 10-day plot of phase residuals based on 5 days of fitting. I 
> like this because it more dramatically shows how hard it is to keep accurate 
> time.
> 
> The spread in phase is expected. In fact this spread is what ADEV measures; 
> variations in frequency accumulate to phase drift. Or vice versa: erratic 
> drift in phase is evidence of frequency instability, which in turn is 
> summarized with ADEV statistics. I know I'm not explaining this well enough, 
> and it's a bit off-topic for this thread, but I'll write it up later and post 
> it.
> 
> /tvb
> 
> 
> - Original Message - 
> From: "Mark Sims" 
> To: 
> Sent: Wednesday, May 04, 2016 4:41 PM
> Subject: [time-nuts] Fw: Optical transfer of time and frequency
> 
> 
>>> See www.leapsecond.com/tmp/5071a-12-run8-5d-10d.gif for a plot of a bunch 
>>> of 5071A Cs clocks.
>> Interesting... what happened on day 4 to send them all bonkers?
> 
> ___
> time-nuts mailing list -- time-nuts@febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.
___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Re: [time-nuts] Fw: Optical transfer of time and frequency

[Tom Van Baak]

Hi Mark,

> Interesting... what happened on day 4 to send them all bonkers?

What you see as bonkers starting the 5th day is actually normal behavior for a 
cesium clock (or any clock, for that matter). I mention this plot because it 
shows that you cannot get 500 ps over a day, even with a 5071A -- which is why 
we're all trying to help the OP with alternative methods of time 
synchronization.

There are several ways to plot phase data. What people often do is plot 
residuals for the entire data set. That is, showing how good the clock would 
have kept time *if you knew ahead of time what the rate would be*. The first 5 
days of the plot shows the residuals based on 5 days of rate fitting. At that 
point the calculated rate was fixed and the clocks continued to tick. So it 
becomes a 10-day plot of phase residuals based on 5 days of fitting. I like 
this because it more dramatically shows how hard it is to keep accurate time.

The spread in phase is expected. In fact this spread is what ADEV measures; 
variations in frequency accumulate to phase drift. Or vice versa: erratic drift 
in phase is evidence of frequency instability, which in turn is summarized with 
ADEV statistics. I know I'm not explaining this well enough, and it's a bit 
off-topic for this thread, but I'll write it up later and post it.

/tvb


- Original Message - 
From: "Mark Sims" 
To: 
Sent: Wednesday, May 04, 2016 4:41 PM
Subject: [time-nuts] Fw: Optical transfer of time and frequency


>> See www.leapsecond.com/tmp/5071a-12-run8-5d-10d.gif for a plot of a bunch of 
>> 5071A Cs clocks.
> Interesting... what happened on day 4 to send them all bonkers?
> 

___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Re: [time-nuts] Fw: Optical transfer of time and frequency

[Magnus Danielson]

Hi,

On 05/05/2016 12:03 AM, Tom Van Baak wrote:

Hal,


How close could you get if you brought two of them together, compared phase,
drove them to the site for a nights work, drove them back to the same
location and compared the phase again.


That's essentially asking what the ADEV (or, TDEV) is for tau 1 day. Rb isn't 
near good enough. Neither is Cs, for that matter.


Indeed. In this case the TDEV is of interest, as we have a 500 ps 
requirement. Notice that TDEV is a RMS-type of measure and I think we 
had the 500 ps given as a time error, you can't compare these values 
directly, as the noise will have a different scaling factor depending on 
the confidence interval around it that we want. The engineering factor 
of 3 sigma would give us 99,7 % of the time it would be within those 
limits, which may be good enough, this would mean that in normal 
conditions I would divide my 500 ps with 3 to get the limit value.
The trouble is that we don't follow the normal gaussian noise and hence 
error function but rather have the Chi-squared noise, so we need our 
confidence factor from that. The principle is the same thought.


A bidirectional link should not be too hard to set up, and with some 
care should be able to achieve this target.


Cheers,
Magnus


See www.leapsecond.com/tmp/5071a-12-run8-5d-10d.gif for a plot of a bunch of 
5071A Cs clocks. They are compared together for 5 days to determine their 
relative phase and frequency offsets and then go on a 5-day trip. You can see 
how the phase drifts as random walk does its thing. It's way more than 500 ps 
per day.

That's why the OP cannot use free-running clocks. He needs some method to 
actively keep them in tight phase lock or passively compare them to within 500 
ps in order to adjust the timestamps in post-facto.

/tvb

- Original Message -
From: "Hal Murray" <hmur...@megapathdsl.net>
To: "Tom Van Baak" <t...@leapsecond.com>; "Discussion of precise time and frequency 
measurement" <time-nuts@febo.com>
Cc: <hmur...@megapathdsl.net>
Sent: Wednesday, May 04, 2016 10:30 AM
Subject: Re: [time-nuts] Fw: Optical transfer of time and frequency




t...@leapsecond.com said:

Any of these methods is going to be a challenge, given their 500 ps
requirement and their $2k budget.


How stable are surplus rubidium oscillators?

How close could you get if you brought two of them together, compared phase,
drove them to the site for a nights work, drove them back to the same
location and compared the phase again.


--
These are my opinions.  I hate spam.




___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.


___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Re: [time-nuts] Fw: Optical transfer of time and frequency

[Bruce Griffiths]

One option not yet considered is to use a portable clock (or clocks) 
transported regularly (every 20 minutes?) between stations for frequency 
comparisons. It may be feasible to use a set of rubidium clocks (for the 
station clocks and the portable clocks) in this manner at least for short 
baselines.

Bruce 

On Thursday, May 05, 2016 11:14:37 AM Bruce Griffiths wrote:
> In the same vein if it takes 1000 seconds to measure the relative phase of a
> pair of clocks to within 500ps then the relative ADEV of the clock pair at
> 1000 sec needs to be somewhat less than 5E-13.
> For 100 s averaging the relative ADEV of a clock pair needs to be better
> than 5E-12 @ 100sec.
> For 10s averaging the relative ADEV of the clock pair needs to be better
> than 5E-11 @ 10s.
> Thus if the measurement takes too long the cost of the local clocks becomes
> unaffordable.
> Comparison techniques that don't require more than 10-100 sec of averaging
> are preferable to keep the cost of the local clocks sufficiently low.
> 
> Bruce
> 
> On Wednesday, May 04, 2016 03:03:59 PM Tom Van Baak wrote:
> > Hal,
> > 
> > > How close could you get if you brought two of them together, compared
> > > phase, drove them to the site for a nights work, drove them back to the
> > > same location and compared the phase again.
> > 
> > That's essentially asking what the ADEV (or, TDEV) is for tau 1 day. Rb
> > isn't near good enough. Neither is Cs, for that matter.
> > 
> > See www.leapsecond.com/tmp/5071a-12-run8-5d-10d.gif for a plot of a bunch
> > of 5071A Cs clocks. They are compared together for 5 days to determine
> > their relative phase and frequency offsets and then go on a 5-day trip.
> > You can see how the phase drifts as random walk does its thing. It's way
> > more than 500 ps per day.
> > 
> > That's why the OP cannot use free-running clocks. He needs some method to
> > actively keep them in tight phase lock or passively compare them to within
> > 500 ps in order to adjust the timestamps in post-facto.
> > 
> > /tvb
> > 
> > - Original Message -
> > From: "Hal Murray" <hmur...@megapathdsl.net>
> > To: "Tom Van Baak" <t...@leapsecond.com>; "Discussion of precise time and
> > frequency measurement" <time-nuts@febo.com> Cc: <hmur...@megapathdsl.net>
> > Sent: Wednesday, May 04, 2016 10:30 AM
> > Subject: Re: [time-nuts] Fw: Optical transfer of time and frequency
> > 
> > > t...@leapsecond.com said:
> > >> Any of these methods is going to be a challenge, given their 500 ps
> > >> requirement and their $2k budget.
> > > 
> > > How stable are surplus rubidium oscillators?
> > > 
> > > How close could you get if you brought two of them together, compared
> > > phase, drove them to the site for a nights work, drove them back to the
> > > same location and compared the phase again.
> > 
> > ___
> > time-nuts mailing list -- time-nuts@febo.com
> > To unsubscribe, go to
> > https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the
> > instructions there.
> 
> ___
> time-nuts mailing list -- time-nuts@febo.com
> To unsubscribe, go to
> https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the
> instructions there.

___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Re: [time-nuts] Fw: Optical transfer of time and frequency

[Bruce Griffiths]

In the same vein if it takes 1000 seconds to measure the relative phase of a 
pair of clocks to within 500ps then the relative ADEV of the clock pair at 
1000 sec needs to be somewhat less than 5E-13.
For 100 s averaging the relative ADEV of a clock pair needs to be better than 
5E-12 @ 100sec.
For 10s averaging the relative ADEV of the clock pair needs to be better than 
5E-11 @ 10s.
Thus if the measurement takes too long the cost of the local clocks becomes 
unaffordable.
Comparison techniques that don't require more than 10-100 sec of averaging are 
preferable to keep the cost of the local clocks sufficiently low.

Bruce

On Wednesday, May 04, 2016 03:03:59 PM Tom Van Baak wrote:
> Hal,
> 
> > How close could you get if you brought two of them together, compared
> > phase, drove them to the site for a nights work, drove them back to the
> > same location and compared the phase again.
> 
> That's essentially asking what the ADEV (or, TDEV) is for tau 1 day. Rb
> isn't near good enough. Neither is Cs, for that matter.
> 
> See www.leapsecond.com/tmp/5071a-12-run8-5d-10d.gif for a plot of a bunch of
> 5071A Cs clocks. They are compared together for 5 days to determine their
> relative phase and frequency offsets and then go on a 5-day trip. You can
> see how the phase drifts as random walk does its thing. It's way more than
> 500 ps per day.
> 
> That's why the OP cannot use free-running clocks. He needs some method to
> actively keep them in tight phase lock or passively compare them to within
> 500 ps in order to adjust the timestamps in post-facto.
> 
> /tvb
> 
> - Original Message -
> From: "Hal Murray" <hmur...@megapathdsl.net>
> To: "Tom Van Baak" <t...@leapsecond.com>; "Discussion of precise time and
> frequency measurement" <time-nuts@febo.com> Cc: <hmur...@megapathdsl.net>
> Sent: Wednesday, May 04, 2016 10:30 AM
> Subject: Re: [time-nuts] Fw: Optical transfer of time and frequency
> 
> > t...@leapsecond.com said:
> >> Any of these methods is going to be a challenge, given their 500 ps
> >> requirement and their $2k budget.
> > 
> > How stable are surplus rubidium oscillators?
> > 
> > How close could you get if you brought two of them together, compared
> > phase, drove them to the site for a nights work, drove them back to the
> > same location and compared the phase again.
> 
> ___
> time-nuts mailing list -- time-nuts@febo.com
> To unsubscribe, go to
> https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the
> instructions there.

___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Re: [time-nuts] Fw: Optical transfer of time and frequency

[Tom Van Baak]

Hal,

> How close could you get if you brought two of them together, compared phase, 
> drove them to the site for a nights work, drove them back to the same 
> location and compared the phase again.

That's essentially asking what the ADEV (or, TDEV) is for tau 1 day. Rb isn't 
near good enough. Neither is Cs, for that matter.

See www.leapsecond.com/tmp/5071a-12-run8-5d-10d.gif for a plot of a bunch of 
5071A Cs clocks. They are compared together for 5 days to determine their 
relative phase and frequency offsets and then go on a 5-day trip. You can see 
how the phase drifts as random walk does its thing. It's way more than 500 ps 
per day.

That's why the OP cannot use free-running clocks. He needs some method to 
actively keep them in tight phase lock or passively compare them to within 500 
ps in order to adjust the timestamps in post-facto.

/tvb

- Original Message - 
From: "Hal Murray" <hmur...@megapathdsl.net>
To: "Tom Van Baak" <t...@leapsecond.com>; "Discussion of precise time and 
frequency measurement" <time-nuts@febo.com>
Cc: <hmur...@megapathdsl.net>
Sent: Wednesday, May 04, 2016 10:30 AM
Subject: Re: [time-nuts] Fw: Optical transfer of time and frequency


> 
> t...@leapsecond.com said:
>> Any of these methods is going to be a challenge, given their 500 ps
>> requirement and their $2k budget. 
> 
> How stable are surplus rubidium oscillators?
> 
> How close could you get if you brought two of them together, compared phase, 
> drove them to the site for a nights work, drove them back to the same 
> location and compared the phase again.
> 
> 
> -- 
> These are my opinions.  I hate spam.
> 
> 
>
___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Re: [time-nuts] Fw: Optical transfer of time and frequency

[Magnus Danielson]

Tom,

L1 code and carrier phase GPS will be challenging.
Tracking it as such is relatively straightforward.
For ionspheric and tropospheric shifts there will be a significant 
common mode, thus cancels fairly well as the time-difference.


What is a concern is the multipath, which will shift around differently 
for the sites as constellation shifts and reflections moves around.
Getting such antennas cheap to keep it within the budget is however 
troublesome. With luck you can find them.


I believe more in a direct link to lock the nodes together.

Cheers,
Magnus

On 05/04/2016 07:06 PM, Tom Van Baak wrote:

If every station has its own GPSDO, what is the purpose of the optical transfer?


The purpose of the optical transfer is to keep the LO at each site in sync at 
all times to within 500 ps. GPSDO are not good enough for this level of timing. 
That's why some sort of optical transfer is being discussed.

A optical transfer would allow them to either:
1) measure the phase difference of each LO and phase lock them to within 500 
ps, or
2) measure the phase drift amongst all the LO and then back out the drift 
during post-processing.

Another proposal is using L1 code and carrier phase common view GPS techniques 
at each site and back out the observed Rb phase drift during post-processing. 
The question is if this gets you down to 500 ps.

Any of these methods is going to be a challenge, given their 500 ps requirement 
and their $2k budget.

/tvb

- Original Message -
From: "Azelio Boriani" <azelio.bori...@gmail.com>
To: "Discussion of precise time and frequency measurement" <time-nuts@febo.com>
Sent: Monday, May 02, 2016 5:20 AM
Subject: Re: [time-nuts] Fw: Optical transfer of time and frequency



If every station has its own GPSDO, what is the purpose of the optical transfer?

On Mon, May 2, 2016 at 11:14 AM, Michael Wouters
<michaeljwout...@gmail.com> wrote:

One other possibility occurs to me that might be doable with surplus
gear and sticks to the  budget. Instead of using WR, give up on
getting time of day and just send a 1 kHz pulse stream in each
direction. Each station then measures against its own GPSDO clock
using a standard/homebrew TIC and records the difference. This is
ambiguous modulo 1 ms but this is trivially resolved using GPS. You
also probably know the distance between the stations to much better
than 1 ms = 300 km :-) . You then post-process but this can be done
with very little latency if you're keen.

Cheers
Michael

On Mon, May 2, 2016 at 1:44 AM, Tom Van Baak <t...@leapsecond.com> wrote:

Has anybody experienced with free-space optical gigabit Ethernet
links? I am curious about whether the transceivers have a fixed
latency or at least a latency one can easily quantify online. This is
the trickiest part for adding WR support on top of a given physical
layer.


Hi Javier,

When searching this topic I ran across a commercial laser solution:

http://www.laseroptronics.com/products.cfm/product/27-0-0.htm
http://www.laseroptronics.com/index.cfm/id/57-66.htm
http://www.laseroptronics.com/index.cfm/id/57-69.htm
etc.

But, according to /57-67.htm it "starts" at $15k per node. Plus there's the cost of 
all the WR pieces, assuming the two are even compatible. So this is vastly above the ~$2k 
budget mentioned by OP. I also assume OP is not ready to embark on a one-off, multi-man-year 
R project.

This particular issue -- how to synchronize (or, at least phase compare) 
multiple oscillators by a two-way laser link over a few km to within 500 ps -- 
is really quite interesting. It would, for example, allow me to do live 
monitoring of 5071A Cs time dilation on my next mountain-valley relativity 
experiment.

/tvb
___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.


___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Re: [time-nuts] Fw: Optical transfer of time and frequency

[Hal Murray]

t...@leapsecond.com said:
> Any of these methods is going to be a challenge, given their 500 ps
> requirement and their $2k budget. 

How stable are surplus rubidium oscillators?

How close could you get if you brought two of them together, compared phase, 
drove them to the site for a nights work, drove them back to the same 
location and compared the phase again.


-- 
These are my opinions.  I hate spam.



___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Re: [time-nuts] Fw: Optical transfer of time and frequency

[Tom Van Baak]

> If every station has its own GPSDO, what is the purpose of the optical 
> transfer?

The purpose of the optical transfer is to keep the LO at each site in sync at 
all times to within 500 ps. GPSDO are not good enough for this level of timing. 
That's why some sort of optical transfer is being discussed.

A optical transfer would allow them to either:
1) measure the phase difference of each LO and phase lock them to within 500 
ps, or
2) measure the phase drift amongst all the LO and then back out the drift 
during post-processing.

Another proposal is using L1 code and carrier phase common view GPS techniques 
at each site and back out the observed Rb phase drift during post-processing. 
The question is if this gets you down to 500 ps.

Any of these methods is going to be a challenge, given their 500 ps requirement 
and their $2k budget.

/tvb

- Original Message - 
From: "Azelio Boriani" <azelio.bori...@gmail.com>
To: "Discussion of precise time and frequency measurement" <time-nuts@febo.com>
Sent: Monday, May 02, 2016 5:20 AM
Subject: Re: [time-nuts] Fw: Optical transfer of time and frequency


> If every station has its own GPSDO, what is the purpose of the optical 
> transfer?
> 
> On Mon, May 2, 2016 at 11:14 AM, Michael Wouters
> <michaeljwout...@gmail.com> wrote:
>> One other possibility occurs to me that might be doable with surplus
>> gear and sticks to the  budget. Instead of using WR, give up on
>> getting time of day and just send a 1 kHz pulse stream in each
>> direction. Each station then measures against its own GPSDO clock
>> using a standard/homebrew TIC and records the difference. This is
>> ambiguous modulo 1 ms but this is trivially resolved using GPS. You
>> also probably know the distance between the stations to much better
>> than 1 ms = 300 km :-) . You then post-process but this can be done
>> with very little latency if you're keen.
>>
>> Cheers
>> Michael
>>
>> On Mon, May 2, 2016 at 1:44 AM, Tom Van Baak <t...@leapsecond.com> wrote:
>>>> Has anybody experienced with free-space optical gigabit Ethernet
>>>> links? I am curious about whether the transceivers have a fixed
>>>> latency or at least a latency one can easily quantify online. This is
>>>> the trickiest part for adding WR support on top of a given physical
>>>> layer.
>>>
>>> Hi Javier,
>>>
>>> When searching this topic I ran across a commercial laser solution:
>>>
>>> http://www.laseroptronics.com/products.cfm/product/27-0-0.htm
>>> http://www.laseroptronics.com/index.cfm/id/57-66.htm
>>> http://www.laseroptronics.com/index.cfm/id/57-69.htm
>>> etc.
>>>
>>> But, according to /57-67.htm it "starts" at $15k per node. Plus there's the 
>>> cost of all the WR pieces, assuming the two are even compatible. So this is 
>>> vastly above the ~$2k budget mentioned by OP. I also assume OP is not ready 
>>> to embark on a one-off, multi-man-year R project.
>>>
>>> This particular issue -- how to synchronize (or, at least phase compare) 
>>> multiple oscillators by a two-way laser link over a few km to within 500 ps 
>>> -- is really quite interesting. It would, for example, allow me to do live 
>>> monitoring of 5071A Cs time dilation on my next mountain-valley relativity 
>>> experiment.
>>>
>>> /tvb
>>> ___
>>> time-nuts mailing list -- time-nuts@febo.com
>>> To unsubscribe, go to 
>>> https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
>>> and follow the instructions there.
>> ___
>> time-nuts mailing list -- time-nuts@febo.com
>> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
>> and follow the instructions there.
> ___
> time-nuts mailing list -- time-nuts@febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.
___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Re: [time-nuts] Fw: Optical transfer of time and frequency

[Magnus Danielson]

If the dynamics is working for you, yes, you can use something like 
that. If you have noise, you would like a spreading code such that the 
correlation in the receiver suppress the noise.


MVH
Magnus

On 05/02/2016 11:14 AM, Michael Wouters wrote:

One other possibility occurs to me that might be doable with surplus
gear and sticks to the  budget. Instead of using WR, give up on
getting time of day and just send a 1 kHz pulse stream in each
direction. Each station then measures against its own GPSDO clock
using a standard/homebrew TIC and records the difference. This is
ambiguous modulo 1 ms but this is trivially resolved using GPS. You
also probably know the distance between the stations to much better
than 1 ms = 300 km :-) . You then post-process but this can be done
with very little latency if you're keen.

Cheers
Michael

On Mon, May 2, 2016 at 1:44 AM, Tom Van Baak  wrote:

Has anybody experienced with free-space optical gigabit Ethernet
links? I am curious about whether the transceivers have a fixed
latency or at least a latency one can easily quantify online. This is
the trickiest part for adding WR support on top of a given physical
layer.


Hi Javier,

When searching this topic I ran across a commercial laser solution:

http://www.laseroptronics.com/products.cfm/product/27-0-0.htm
http://www.laseroptronics.com/index.cfm/id/57-66.htm
http://www.laseroptronics.com/index.cfm/id/57-69.htm
etc.

But, according to /57-67.htm it "starts" at $15k per node. Plus there's the cost of 
all the WR pieces, assuming the two are even compatible. So this is vastly above the ~$2k 
budget mentioned by OP. I also assume OP is not ready to embark on a one-off, multi-man-year 
R project.

This particular issue -- how to synchronize (or, at least phase compare) 
multiple oscillators by a two-way laser link over a few km to within 500 ps -- 
is really quite interesting. It would, for example, allow me to do live 
monitoring of 5071A Cs time dilation on my next mountain-valley relativity 
experiment.

/tvb
___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.


___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Re: [time-nuts] Fw: Optical transfer of time and frequency

[Michael Wouters]

One other possibility occurs to me that might be doable with surplus
gear and sticks to the  budget. Instead of using WR, give up on
getting time of day and just send a 1 kHz pulse stream in each
direction. Each station then measures against its own GPSDO clock
using a standard/homebrew TIC and records the difference. This is
ambiguous modulo 1 ms but this is trivially resolved using GPS. You
also probably know the distance between the stations to much better
than 1 ms = 300 km :-) . You then post-process but this can be done
with very little latency if you're keen.

Cheers
Michael

On Mon, May 2, 2016 at 1:44 AM, Tom Van Baak  wrote:
>> Has anybody experienced with free-space optical gigabit Ethernet
>> links? I am curious about whether the transceivers have a fixed
>> latency or at least a latency one can easily quantify online. This is
>> the trickiest part for adding WR support on top of a given physical
>> layer.
>
> Hi Javier,
>
> When searching this topic I ran across a commercial laser solution:
>
> http://www.laseroptronics.com/products.cfm/product/27-0-0.htm
> http://www.laseroptronics.com/index.cfm/id/57-66.htm
> http://www.laseroptronics.com/index.cfm/id/57-69.htm
> etc.
>
> But, according to /57-67.htm it "starts" at $15k per node. Plus there's the 
> cost of all the WR pieces, assuming the two are even compatible. So this is 
> vastly above the ~$2k budget mentioned by OP. I also assume OP is not ready 
> to embark on a one-off, multi-man-year R project.
>
> This particular issue -- how to synchronize (or, at least phase compare) 
> multiple oscillators by a two-way laser link over a few km to within 500 ps 
> -- is really quite interesting. It would, for example, allow me to do live 
> monitoring of 5071A Cs time dilation on my next mountain-valley relativity 
> experiment.
>
> /tvb
> ___
> time-nuts mailing list -- time-nuts@febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.
___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Re: [time-nuts] Fw: Optical transfer of time and frequency

[Magnus Danielson]

Javier,

On 05/01/2016 02:54 PM, Javier Serrano wrote:

On Sun, May 1, 2016 at 9:50 AM, Magnus Danielson
 wrote:

If I recall correctly, there where some White Rabbit stuff available from
vendors, was it only Ethernet switches or also cards?


Yes, the switch is available from two vendors that I know of. They are
referenced at [1]. For nodes, we have a page describing what hardware
support is needed [2]. Then there are various boards available
commercially which implement that hardware support, like [3].

For the particular case of WR-enabled TDCs, one could use the SPEC
PCIe carrier [3] with a TDC FMC [4] or a simple DIO FMC [5],
delegating then the TDC function to the FPGA on the carrier, using an
HDL core like [6]. Or roll your own, of course.


Thanks for this listing!


Has anybody experienced with free-space optical gigabit Ethernet
links? I am curious about whether the transceivers have a fixed
latency or at least a latency one can easily quantify online. This is
the trickiest part for adding WR support on top of a given physical
layer.


There exists optical links that is used to solve the "last mile" 
problems. I have not used any of those myself.


I also know that there is microwave links which essentially just 
converts the optical GE encoding onto a microwave carrier and back.

It could be an interesting option to consider.

Most of the microwave links that is in regular use however have modes 
that re-encode things and will break White Rabbit. It also breaks my 
stuff every once in a while, so I know more about these systems than I 
should know. My EFTF-2014 presentation and paper give some comments on it.


Cheers,
Magnus


Cheers,

Javier

[1] http://www.ohwr.org/projects/white-rabbit/wiki/Switch#Commercial-producers
[2] http://www.ohwr.org/projects/white-rabbit/wiki/WRReferenceDesign
[3] http://www.ohwr.org/projects/spec/wiki
[4] http://www.ohwr.org/projects/fmc-tdc/wiki
[5] http://www.ohwr.org/projects/fmc-dio-5chttla/wiki
[6] http://www.ohwr.org/projects/tdc-core/wiki


___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Re: [time-nuts] Fw: Optical transfer of time and frequency

[Javier Serrano]

On Sun, May 1, 2016 at 5:44 PM, Tom Van Baak  wrote:
> This particular issue -- how to synchronize (or, at least phase compare) 
> multiple oscillators by a two-way laser link over a few km to within 500 ps 
> -- is really quite interesting. It would, for example, allow me to do live 
> monitoring of 5071A Cs time dilation on my next mountain-valley relativity 
> experiment.

Maybe Koruza [1] could be a good starting point for such a
development. It does not meet the distance spec in its current state
but it is not too expensive and it's all open source hardware (and
software of course). I am pretty sure the Koruza team would be happy
to collaborate.

Cheers,

Javier

[1] http://koruza.net/
___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Re: [time-nuts] Fw: Optical transfer of time and frequency

[Tom Van Baak]

> Has anybody experienced with free-space optical gigabit Ethernet
> links? I am curious about whether the transceivers have a fixed
> latency or at least a latency one can easily quantify online. This is
> the trickiest part for adding WR support on top of a given physical
> layer.

Hi Javier,

When searching this topic I ran across a commercial laser solution:

http://www.laseroptronics.com/products.cfm/product/27-0-0.htm
http://www.laseroptronics.com/index.cfm/id/57-66.htm
http://www.laseroptronics.com/index.cfm/id/57-69.htm
etc.

But, according to /57-67.htm it "starts" at $15k per node. Plus there's the 
cost of all the WR pieces, assuming the two are even compatible. So this is 
vastly above the ~$2k budget mentioned by OP. I also assume OP is not ready to 
embark on a one-off, multi-man-year R project.

This particular issue -- how to synchronize (or, at least phase compare) 
multiple oscillators by a two-way laser link over a few km to within 500 ps -- 
is really quite interesting. It would, for example, allow me to do live 
monitoring of 5071A Cs time dilation on my next mountain-valley relativity 
experiment.

/tvb
___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Re: [time-nuts] Fw: Optical transfer of time and frequency

[Javier Serrano]

On Sun, May 1, 2016 at 9:50 AM, Magnus Danielson
 wrote:
> If I recall correctly, there where some White Rabbit stuff available from
> vendors, was it only Ethernet switches or also cards?

Yes, the switch is available from two vendors that I know of. They are
referenced at [1]. For nodes, we have a page describing what hardware
support is needed [2]. Then there are various boards available
commercially which implement that hardware support, like [3].

For the particular case of WR-enabled TDCs, one could use the SPEC
PCIe carrier [3] with a TDC FMC [4] or a simple DIO FMC [5],
delegating then the TDC function to the FPGA on the carrier, using an
HDL core like [6]. Or roll your own, of course.

Has anybody experienced with free-space optical gigabit Ethernet
links? I am curious about whether the transceivers have a fixed
latency or at least a latency one can easily quantify online. This is
the trickiest part for adding WR support on top of a given physical
layer.

Cheers,

Javier

[1] http://www.ohwr.org/projects/white-rabbit/wiki/Switch#Commercial-producers
[2] http://www.ohwr.org/projects/white-rabbit/wiki/WRReferenceDesign
[3] http://www.ohwr.org/projects/spec/wiki
[4] http://www.ohwr.org/projects/fmc-tdc/wiki
[5] http://www.ohwr.org/projects/fmc-dio-5chttla/wiki
[6] http://www.ohwr.org/projects/tdc-core/wiki
___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Re: [time-nuts] Fw: Optical transfer of time and frequency

[Magnus Danielson]

Javier,

If I recall correctly, there where some White Rabbit stuff available 
from vendors, was it only Ethernet switches or also cards?


Cheers,
Magnus

On 05/01/2016 01:14 AM, Bruce Griffiths wrote:



  On Sunday, 1 May 2016 10:52 AM, Bruce Griffiths 
 wrote:


  White Rabbit is open hardware, you are free to build it yourself should you 
want to do so. All the relevant VHDL etc is available.There will also be 
suitable TDC designs available on the CERN site.You can also integrate these 
into your system if you want.
White Rabbit can handle thousands of nodes thus 3 or more won't be a problem.

Bruce

 On Sunday, 1 May 2016 10:35 AM, Ilia Platone  wrote:


   I found only preliminary data about these transceivers. I was meaning for a 
<2000€ overall solution, does a White Rabbit implementation fill this requisite 
( I couldn't find much information about its costs)? Also consider that nodes 
could be more than three also.
   Ilia.



___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.


___
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.