Timing calibration using WR

P.Kooijman

Fibre network

• Had a PRR of the network• Comments reasonably positive• Complicated system• Build a full testbench

• Will work as a fibre network This is what the network will look like

• Two systems• DU control bidirectional• DOM control/DATA different

returnpath

DU control path goes form

ABRDCToBDUFromCDUTo DDU

Several contributions to the path

ABRDC CirculatorCirculator cable terminationCable termination cable termination (MEOC)Cable termination Junction Box in (IL cable)JB in JB out (Amplification, splitting)Jbout DUBASEin (IL cable)DUBASEin BDU

CirculatorDDUcable termination Circulator Cable termination cable termination (MEOC)Junction Box in (IL cable) Cable termination JB out JB in(Amplification, splitting)DUBASEin Jbout (IL cable)BDU DUBASEin

Different paths up and downneed to be calibrated

MEOC and Interlink cables have different fibre type

  CP Toulon

L 100 km 45 km

1576 1575

B 0.574 1.075

C -52 -30

0.999995(8 2) (-2.13 ns) 0.999993(2 4) (-1.48 ns)

0.999989(8 2) (-5.16 ns) 0.999978(2 3) (-4.82 ns)

Effect of dispersion in MEOC (have been measured)

𝑡=𝐴 ¿

  SMF-28 Bend-Brite NZDSF (It) NZDSF (Fr)

 

Different characteristics for different fibres

Dispersion relatively small effect. Corrections maximally of the order of 5 nsWe know the correction to 2-4% Systematic shifts of 100-200 ps are possible

Amplifiers are doped lengths of fiber (30-100m long)Lengths are not very critical for the amplificationBut Very high amplification is also longer length

Can expect differences of a few metres up versus down~10 ns

This has to be measured.

Device for measuring time delay TCD (time calibration device)

Was used to measure the MEOCs

Asymmetry  on shore

Measure somewhere beyond the circulator

ABRDC  CT1

And

CT1  DDU

Is not critical where.

Can use the TCD

Effect of not knowing exactly how long the interlinks are

Dispersion is quite large in interlink cablesSignal l

(nm)t  for 2x500 m interlink(ns)

Dt (ns) 

t for 100 km(ns)

Dt (ns) 

T total

Asym.(ns)

Asymmetry

SC Downlink 1530.33 4904.08 0.00 490012.55 0.00 494916.62 - -Uplink DU 1 1535.82

4904.28 0.21 490010.42 -2.13 494914.70 -1.93 0.9999961149Uplink DU 2 1536.61

4904.31 0.24 490010.12 -2.42 494914.44 -2.19 0.9999955848Uplink DU 3 1537.40

4904.34 0.27 490009.84 -2.71 494914.18 -2.45 0.9999950621Uplink DU 4 1538.19

4904.37 0.30 490009.55 -3.00 494913.92 -2.70 0.9999945467Uplink DU 5 1538.98

4904.40 0.33 490009.27 -3.28 494913.67 -2.96 0.9999940386Uplink DU 6 1539.77

4904.43 0.36 490008.99 -3.56 494913.43 -3.21 0.9999935378Uplink DU 7 1540.56

4904.47 0.39 490008.72 -3.83 494913.18 -3.45 0.9999930443Uplink DU 8 1541.35

4904.50 0.42 490008.44 -4.10 494912.94 -3.69 0.9999925581Uplink DU 9 1542.14

4904.53 0.45 490008.18 -4.37 494912.70 -3.93 0.9999920792Uplink DU 10 1542.94

4904.56 0.48 490007.91 -4.64 494912.47 -4.17 0.9999916017Uplink DU 11 1543.73

4904.59 0.51 490007.65 -4.90 494912.24 -4.40 0.9999911375Uplink DU 12 1544.53

4904.62 0.54 490007.39 -5.16 494912.01 -4.63 0.9999906749

Ignoring dispersion difference of Interlinks gives a maximum error of  500 ps.

Giving an approximate ratio of 1%  (1 km of interlink  100 km MEOC)

Which is about what we have corrects timing.

If we now vary the MEOC length from 80 to 120 km the variation in timing is 

-60 ps to 60 ps.  (ie we don’t know the real length to 20 %)

If we vary the Interlink length by 10% the variation is 30 ps.

Roughly knowing the lengths of the cables is sufficient

Conclusion for DU control

We need to measure the delays in the Amplifiers 100 ps (up and down) We need to measure the delays on shore 100 ps (100 ps up and down)We need to know the ratio of Interlink length to MEOC to 10% 50 ps

The time offset uses difference between up and down The round trip time uses the sum

So both have identical errors

Error on the timing will be of the order of 200 ps

The DOM control

We have a different fibre back from DU than to DU for this system.Fibre lengths through the MEOC are variable although DOMs of 4 DUs have the same return fibre

Clock at the bottom of the DU is synchronized so if the delay time from DU clock to DOM clock can be determined then the offset of the DOM clock can be determined

Need the delay from hereTo here

RTT measurement from entrance to DUBASE

Delay time from same point to BOB exit

RTT measurement from entrance to DOM

  

  

  

 

When the string goes into the sea the pressure on and the temperature of the fibres will change timing will change

Dispersion variation due to temperature has been measured as

Dt = 40 ps/km/K

Dispersion variation due to pressure

Dt = 3.8 ps/km/bar

Plab = 1 bar Tlab = 20 (C)First DOM: P = 315 bar T=14 (C) (314*3.8 -6*40)* 0.1 = 95 psLast DOM:P=285 bar T= 14 (C) (285*3.8 -6*40)* 0.8 = 675 ps

Putting together the RTT to the DU and the Dt to the DOM Can derive clock offset Putting together the RTT to the DU and the Dt to the DOM and RTT to DOM - derive clock offset within WR determine asymmetry

Time error to DU Base 200psTime error in string (100 ps)RTTDU +(100 ps)DelayVEOC +(100 ps)RTTDOMcalib + (10-60 ps) pressure and temperature correction = 180 psTotal error   270 ps

This is also true for RTT DOM measurement as asymmetry and return delay Are chosen to reproduce this downward delay.

Only errors that can occur are temperature variations on shore. If fibres have same temperature coherent variation RTT for DU will change by the same amount as RTT to DOM The RTT variation will be split between both up and down going Offset will not change

Only if temperatures vary wildly over long distances will an error occur.25 deg difference between DU control fibre and DOM fibre over 2 km gives 1 ns shift of timing.

This would mean Cable is bare in sunlight over 2 km we should do something about it !!

Systematics:

Measuring the DOM RTT gives 18 measurements of the return delay that pass through the same fibre. After correction for Dispersion and delay in the VEOC these should give consistent delay due to length of fibre. Gives extra check on all corrections and delays in the system.The signals of 4 DUs go through the same fibre Gives any differences in the seafloor lengths as the MEOC length is the same

Other DUs go through different fibres Measurement of the day night variation due to possibly different fibre

temperatures Gives an overall measurement of the stability and possible error on the timing

measurements

TT measurement at same time as RTT measurement in “Green” Box.

No need for full string dark room calibration

Conclusion

WR is useable in the DOM control system by measuring the RTT and down going time

Coherent variations of delays in different fibres leave the Offset of clocks unchanged

RTT measurement allows for a stringent consistency check of the timing system Typical accuracy of the system is about 270 ps for the DOM synchronisation TT measurement as part of DOM acceptance test RTT DOM as part of DOM acceptance test All calibration based on WR switch or CLB The system is needed NOW if Line 1 is to be calibrated

All of this requires a process running in DOM that can on request give the RTT back to a shore process, that calculates the relevant asymmetries and can perform the regular systematic checks

WR switches can only have one WR slave.

WR switches can have more than one Ethernet uplink  (already shown by CTA)

Configuration  18 to 2 or 15 to 3 seem reasonable(highest rate in PPMDOM  run of 10 mins average = 1.2MHz = 60 Mbit*1.2(overhead) = 75 Mbit.  For 18:2 multiplexing  675 Mbit  factor 2 below maximum  reasonable  start with 9:1 if problems can always buy a few more switches to go to 5:1

Conclusions II