Long haul time distribution in existing telecommunication networks Jeroen Koelemeij (VU Amsterdam, NL) Tjeerd Pinkert (VU Amsterdam, NL) Erik Dierikx (VSL Delft, NL) Henk Peek (NIKHEF, NL) Rob Smets (SURFnet, NL)

Motivation Scientific motivation: Astronomy (LOFAR, SKA, CTA, VLBI) needs timing over medium-to-long distance VU Amsterdam: high-precision frequency measurements of atoms and molecules for tests of fundamental physics (QED/GR searches for dark matter and higher dimensions, etc.): Need SI second with highest accuracy! GPS 12-13 decimals Stand-alone Cs 13 decimals UTC realization at VSL Delft 14 decimals Fiber-optic WR link VSL-VU (137 km)? Societal/economical motivation GPS back-up facitility needed Integrate accurate time & frequency in fiber-optic telecommunications

Fiber-optic T&F links in the Netherlands Provided by SURFnet 2 x 300 km DWDM unidirectional fiber roundtrip VU-KVI-VU Time transfer for science: 2 x 137 km WR link from VSL (Delft) to NIKHEF (Amsterdam) Amplified fiber link to distribute UTC(VSL) via WR Target performance: <<10-14 frequency stability < 1 ns time uncertainty (0.1 ns) WR link passes three other university campuses, and ESTEC

Roundtrip measurements 1470 nm 1470 nm 1490 nm 1490 nm 1490 nm 1490 nm 1470 nm 1470 nm

Quasi bi-directional amplifier VU Amsterdam / TU Eindhoven / SURFnet SOA MUX OI Commercially available end of 2014 from OPNT B.V. Remote control & monitoring ps-level delay calibration possible Implementation in DWDM systems also possible

WR link VSL- NIKHEF: Results Stability Rb pps in – pps out…

WR link VSL – NIKHEF (Next configuration)

Calibration of delay asymmetries Greg D: thanks for the excellent calibration documentation! WR + Cs ensemble + TWSTFT + GPS equipment @VSL

Dealing with dispersion Use the three-wavelength trick* Exchange one SFP for another with slightly different l: l2 l3 (l1 stays the same) Two Different RTDs (tAC12, tAC13) Measure all li Use formula to find estimated OWD (or, equivalenty, a): < 10ps/100 km if various l only 1-2 nm apart ~ 200 ps/100 km if l1-l2 ~ 20 nm *N. Sotiropoulos et al. Optics Express 21, 32643 (2013)

Delay measurements through fast sampling and cross correlation 40 GS/s DSO 1 Gbps / 10 Gbps pattern generator DUT splitter Differential delay [ns] XCOR High resolution (~150 fs / shot) Works for electrical and optical transmission lines *N. Sotiropoulos et al. Optics Express 21, 32643 (2013)

Outlook Complete methods set for absolute delay calibration of SFPs, WR HW, patches, connectors, etc. (under development, Henk Peek, Tjeerd Pinkert, JK) Use extremely low-noise LO as clean-up oscillator: H-maser stability via WR possible? VLBI through WR possible? (ASTERICS INFRADEV proposal submitted) Fiber-optic dissemination of UTC as a service (part of EMPIR proposal SRT-i32, see also talk by P.-E. Pottie)

Credits €€ STW €€ FOM €€ EMRP NEAT-FT €€ SURFnet Rabbits @CERN and @7S/UGR Ah. Hmkay? Questions? j.c.j.koelemeij@vu.nl