Optical Two-Way Time-Frequency Transfer - PowerPoint Presentation

Slide1

Optical Two-Way Time-Frequency Transfer

Second ISSI Workshop on Spacetime Metrology, Clocks and Relativistic Geodesy

March 25-28, 2019

International Space Science Institute, Bern, Switzerland

Nathan R. Newbury

NIST, Boulder, CO, USA

Slide2

Team

Laura Sinclair, Jean-Daniel Deschenes

*,

Hugo Bergeron, William Swann, Isaac Khader, Martha Bodine, Esther Baumann, Fabrizio Giorgetta, Paritosh Manurkar, Sarah Stevenson, Jennifer Ellis, Emily Hannah, Nate Newbury

*

Octosig

Slide3

Optical

TWTFT

Optical

TWTFT

~

~

Remote

site

~

Master Optical Oscillator/Clock

Femtosecond Clock Network

Some uses:

Time/frequency dissemination

Distributed coherent sensing (active or passive)

Navigation (high precision &

gps

denial)

Secure communications

….

quartz oscillator

or

optical cavity

Goal: enable future femtosecond clock network

Slide4

Outline

Introduction

Turbulence

Comb-Based OTWTFT: basic system and results

Overview of configurations and tests conducted

Conclusion

Slide5

Our Clock

Cavity-stabilized

laser

Self-referenced

comb

Ticks @ 5 ns

 fs

locked to underlying oscillator

Controller:

Count and “label” pulses

Define this pulse arrival at reference plane as

12:00:00.000000000000000

Reference plane

Comb

Quartz/DRO

10 GHz oscillator

Clock

Atoms

rarely

Slide6

Free-space transfer:

Turbulence, platform motion…

Clock

Frequency

Comb

Frequency

Comb

Clock

Amplitude noise & signal loss

From turbulence (scintillation & beam wander)

From obstructions & platform motion

Well-known from free-space optical communications

Phase noise (time-of-flight variations)

From turbulence (“piston effect”)

From platform motion

1st

order Doppler shifts -> Need less than 3 nm/sec to reach

v/c

<

 

Slide7

Atmospheric Turbulence:

“textbook” Kolmogorov scaling

Inertial

“Kolmogorov scaling”

Dissipative

k

< 1/l0

Input

k

< 1/L0

From Andrews, Phillips, and Hopen “Laser Beam Scintillations with Applications”, SPIE Press (2001)

Spatial power spectrum of index fluctuations

wavevector (k)

Outer Scale

Inner Scale

density variations -> index of refraction variations -> random phase mask -> time of flight variations

Inertial range

Slide8

Temporal Power Spectral Density of Timing Jitter

Taylor’s Frozen Turbulence Hypothesis

Predicted Timing Jitter

Outer scale L

0

= 10 m

C

n

2

= 1x10-14 m-2/3 vx = 1 m/s

Standard assumption(Van Karman)

GT spectrum*

Kolmogorov scaling~ f -8/3

wind (

v

x )

~ vx/2pL0

input region

inertial region

*Greenwood and

Tarazano

, 1974

Slide9

Clock

Frequency

Comb

Frequency

Comb

Clock

But the link is still reciprocal

Turbulent Atmosphere is reciprocal

*

For

two-way

single-mode

link, time-of-flight variations are common mode

* J. Shapiro

J. Opt. Soc. Am.

61

492 (1971); J. Shapiro & A.

Puryear

,

J. Opt.

Commun

.

Netw

.

4

947 (2012); R. R.

Parenti

et al.,

Opt. Exp.

20

21635 (2012)

(not true for a multi-mode link)

Slide10

Two-Way Time Transfer: Basic Concept

Timer

Site B

Site A

t

A

=0

t

A→B

=

T

link

+

D

t

AB

t

B→A

=

T

link

-

D

t

AB

t

B

=0

Timer

Requires a “Reciprocal” Link

A single-mode optical link is reciprocal!

—

Clock Time Offset

D

t

AB

Slide11

Two-Way Time Transfer + Feedback

Synchronization

Timer

t

A→B

t

B→A

Timer

feedback

Site B

Site A

Real-time calculation

Communication link

—

D

t

AB

Slide12

Two-Way Time Transfer: With Combs

Timer

Site B

Site A

t

A→B

=

T

link

+

D

t

AB

t

B→A

=

T

link

-

D

t

AB

Timer

Comb timing is at femtosecond level!

But timing information lost in

photodetection

…still picosecond level timing

Slide13

Comb 2

Comb 1

Pulses from Comb A “downsample” pulse train from Comb B

Digitizer

f

rep,A

Time”

Ampl.

f

rep,B

D

f

r

Interferogram:

Cross-correlation

Downsampled pulse train

Our system:

T

Red

= 10 ns = 1/100 MHz

d

T = 300 fs

d

T/ T

Red

=30,000

(Downsampling factor)

Dual Comb Measurement Technique

Slide14

Interferogram

(cross-corr.)

Comb

Comb

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

time

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

d

t

clock

Small timing shifts are “amplified” in the interferogram spacing by

(

f

r

/

D

f

r

= 100,000)

Detection of Pulse Arrival with Femtosecond Precision

Analogous to high-speed Sampling Oscilloscope

1/

D

f

r

1/

D

f

r

1/f

r1

1/f

r2

Comb 1

Comb 2

1/

D

f

r

+

d

t

clock

(

fr

/

D

f

r

)

Combs phase-locked to clock

with differing repetition rate,

D

f

r

= f

r1

– f

r2

Slide15

Controller

Controller

Comb Pulse Arrival Detection

“Linear Optical Sampling” approach

Time offset at 2kHz update with femtosecond precision but

5

nanosec

ambiguity (which comb pulse?)

Clocks tick at different rates…cannot “synchronize”

Purposefully run clocks at different rates

Site B

Site A

(

f

r

+

D

f

r

)

-1

f

r

-1

Linear sampling of Comb A arrival by Comb B

t

B→A

Linear sampling of Comb A arrival by Comb B

t

A→B

—

Clock Time Offset

D

t

AB

Slide16

Comb-based Synchronization System

Controller

Controller

Coarse Two-way &

Comms

Transfer Comb

CW Laser Coherent TX/RX

CW Laser Coherent TX/RX

Coarse Two-way &

Comms

Synch.

Equation

Comb-based timing measurement

Comb-based timing measurement

Three signals between sites:

Phase modulated

cw

light for “coarse” two-way timing

Comb light for “fine” two-way timing

Coherent communication (to close the loop)

Plus full calibration of each terminal delays for “absolute” time

Timing Exchange

Remote Comb

Master Comb

Clock Output

Clock

Output

10-100 Hz

Feedback BW

Slide17

Comb-based Synchronization System

Timing Exchange

Out-of-Loop

Verification

Clock Output

Clock Output

Slide18

Experimental Setup for 4 km Folded Link Across the NIST Campus

Free-space terminals

FPGA controllers

Frequency comb transceivers

~5

mW

launch

Slide19

Turbulence-induced beam pointing and scintillation

Slide20

One-way (time-of-flight) vs. Two-way (clock timing)

Two-way:

T

AB

 

Two-way (TAB): x2000 expanded view

 

Clock Timing (fs)

Time Variation (psec)

Air Temperature (C)

One-way (DTA)

Suppression of variations in path length to fsec level

Slide21

Residual Frequency Uncertainty

90% Link availability

Shorted-path

Optical Clocks

Also … No Systematic Bias

at the 2.6x10

-19

Level

Slide22

O-TWTFT: Experiments

Carrier-Phase OTWTF

for u

ltra-precise clock comparisons

Sub-fs clock Synchronization with 25 m/s Motion

Coherent 10 GHz – Synched Microwave Osc.

Multi-Node network

Simplified PicosecondSynchronization

Modifications to support ultra-long distance links*

Limits to FutureGround-to-Satellite Links

O-TWTFT

Synchronization Across 12 km of Air

Completed

Under Investigation

Comparison of Atomic Optical Clocks

Applications to Coherent Sensing?

* over 1-10 km-scale testbeds

Slide23

O-TWTFT: Over strong turbulence

Ultra-precise clock comparisons

~ 1e-17 @ 1 sec

Sub-fs clock Synchronization with 25 m/s Motion

Coherent 10 GHz – Synched Microwave

Osc.

Multi-Node network

Simplified PicosecondSynchronization

Modifications to support ultra-long distance links**

Limits to FutureGround-to-Satellite Links

O-TWTFT

Synchronization Across 12 km of Air

Completed

Under Investigation

Comparison of Atomic Optical Clocks

Applications to Coherent Sensing*

*At white paper level

** over 1-10 km-scale testbeds

Slide24

Synchronization Across 12 km of Strong Turbulence

NIST

Valmont

Butte

5.8 km

Collaborated with JH APL to verify compatible with Adaptive Optics Terminals

Same results for “tip/tilt” terminals

and Adaptive Optics terminals

Path: 12 km highly turbulent path o

ver

a City

Slide25

Turbulence Causes many short Dropouts

Dropout = period during which received power is below ~2.5

nW threshold

Strong turbulence

Moderate turbulence

Most turbulence-induced dropouts are < 10

ms

Increased turbulence increases the number of dropouts

12 kilometer

Slide26

Synchronization Across 12 km of Strong Turbulence

Time Deviation

Averaging Time (s)

100 fs

1 fs

10 fs

100 as

1

10

100

1000

0.1

NIST

Valmont

Butte

5.8 km

L. Sinclair, Applied Physics Letters,

109, 151104 (2016).

Path: 12 km highly turbulent path o

ver

a City

Continuous

Gated

Slide27

Carrier-Phase OTWTF for u

ltra

-precise clock comparisons

Sub-fs clock Synchronization with 25 m/s Motion

Coherent 10 GHz – Synched Microwave Osc.

Multi-Node network

Simplified PicosecondSynchronization

Modifications to support ultra-long distance links*

Limits to FutureGround-to-Satellite Links

O-TWTFT

Synchronization Across 12 km of Air

Completed

Under Investigation

Comparison of Atomic Optical Clocks

Applications to Coherent Sensing?

* over 1-10 km-scale testbeds

O-TWTFT: Comparison of Atomic Optical Clocks

Slide28

On-Going Work: Comparing Atomic Optical Clocks

Comparison of state-of-the-art Sr and Yb Atomic Clocks in the 10-18 Range

Yb

clock

Sr clock

O-TWTFT

fiber

link

NIST

U. of Colorado

Large NIST collaboration

See Dave L. Talk later in workshop

Slide29

Carrier-Phase OTWTF for u

ltra

-precise clock comparisons

Sub-fs clock Synchronization with 25 m/s Motion

Coherent 10 GHz – Synched Microwave Osc.

Multi-Node network

Simplified PicosecondSynchronization

Modifications to support ultra-long distance links*

Limits to FutureGround-to-Satellite Links

O-TWTFT

Synchronization Across 12 km of Air

Completed

Under Investigation

Comparison of Atomic Optical Clocks

Applications to Coherent Sensing?

* over 1-10 km-scale testbeds

O-TWTFT: Carrier-Phase OTWFT

Slide30

Using Phase for Higher Timing Precision

Optical phase is more sensitive (but has 5 fs ambiguity)

1

ps

5 fs

envelope uncertainty

phase fringe uncertainty

Combine for high precision and no ambiguity…

Slide31

t

A

Q

A

t

B

Q

B

Oscillator B

Comb

Oscillator A

cavity

laser

fiber link

Two-way carrier-phase extraction

Relative timing

Carrier-Phase OTWTFT

Slide32

Carrier Phase OTWTFT Allows Frequency Comparisons at 10

-17 level @ 1 sec

Previous OTWTFT (pulse envelope only)

Carrier phase OTWTFT

10X-30X improvement to 10

-17

@ 1 second

L. Sinclair et al., Physical Review Lett, 120, 050801 (2018)

Very near limit set by time-dependent turbulence

Slide33

Timing

(femtosec)

Optical phaseDifference (rad)

Tracking relative timing/phase wander between sites without error

Truth data

Difference (at 0.5

ms

sampling) (at 1-s smoothing)

In phase, unwrapped 300 million radian phase wander without a single

p

error

(post processing but real-time should be possible)

OTWTF phase/timing data

Slide34

O-TWTFT: Sub-fs Clock Synchronization with motion

Ultra-precise clock comparisons

~ 1e-17 @ 1 sec

Sub-fs clock Synchronization with 25 m/s Motion

Coherent 10 GHz – Synched Microwave

Osc.

Multi-Node network

Simplified PicosecondSynchronization

Modifications to support ultra-long distance links**

Limits to FutureGround-to-Satellite Links

O-TWTFT

Synchronization Across 12 km of Air

Completed

Under Investigation

Comparison of Atomic Optical Clocks

Applications to Coherent Sensing*

*At white paper level

** over 1-10 km-scale testbeds

Slide35

Clock

Frequency

Comb

Frequency

Comb

Clock

What About Moving Platforms?

Relative velocity of

3 nm/sec →

v

/c

<

Relative motion of 300 nm

→

1 femtosecond

 

Turbulence still present -> signal fades & piston noise

Platform motion worse: mm/sec → 10’s of m/sec

Qualitatively different!

Time-of-flight varies over millisecond measurement time

Time-of-flight varies over time-of-flight!

Doppler shifts lead to false time shifts from dispersion

Calibration much more complicated

moving

v

Slide36

Solution to Moving platforms

All effects can be “calculated” or compensated for at the Sub-Femtosecond level… at least for terrestrial velocities of ~ 30 m/s (~ 50 mph)

New-generation FPGA

Implemented new Algorithms in Real-Time Processing Platform

Slide37

Delay Doppler coupling:

System Dispersion + Doppler -> systematic timing shifts

“High” Dispersion(meters of optical fiber)

Low Dispersion

Slide38

How to Induce Doppler Effect

Out-of-Loop

Verification

Moving Clock

Master Site (Site A)

Remote Site (Site B)

Do not yet have:

moving optical clock

tracking free-space terminals

Slide39

Testing OTWTFT with Motion: Two methods

“Doppler Simulator”

Multi-passed retroreflector on a rail in series with open-path linkMimics moving clock (without time dilation)25 m/s maximum effective velocity12 m maximum displacement

Connect clocks via retroreflector mounted on a quadcopterTwo-way polarization-multiplexed link20 m/s maximum effective velocity250 m maximum displacement

Because no moving clock/transceiver yet…

A

B

A

B

Via a Quadcopter

Slide40

Synchronization Off of a Quadcopter

Out-of-Loop

Verification

Master Site (Site A)

Remote Site (Site B)

Quadcopter with reflector

Slide41

Synchronization off of a quadcopter:Telescope pointing servo using image analysis

Region of Interest

850 nm LED reflected from retroreflector

850 nm LED

Si camera + 500 mm lens

Azimuth-Elevation gimbal

Slide42

Synchronization at the femtosecond level off of a Quadcopter

Slide43

Previous Results:

Synchronization Off of a Quadcopter

No velocity-dependent bias observed (statistically) down to ~ 300 attosecond level

Slide44

Equal Times at Both Sites(from tests with Doppler rail at different velocities)

Time deviation < 1 fs

24 m/s @ 0, 2, 4 km

0 m/s @ 0, 2, 4 km

Slide45

Equal Frequencies at both Site:Despite +/- 24 m/s motion:

24 m/s motion across:

0 km

2 km 4 kmNew static measurement across: 0 km

Slide46

O-TWTFT: Multi-Node Network

Ultra-precise clock comparisons

~ 1e-17 @ 1 sec

Sub-fs clock Synchronization with 25 m/s Motion

Coherent 10 GHz – Synched Microwave

Osc.

Multi-Node network

Simplified PicosecondSynchronization

Modifications to support ultra-long distance links**

Limits to FutureGround-to-Satellite Links

O-TWTFT

Synchronization Across 12 km of Air

Completed

Under Investigation

Comparison of Atomic Optical Clocks

Applications to Coherent Sensing*

*At white paper level

** over 1-10 km-scale testbeds

Slide47

Turbulent air

A

B

X

cavity stabilized laser

NIST

Table mountain

A

B

X

28-km roundtrip

Clock offset

(should be zero)

Goal: compare clocks A & B over 28 km roundtrip by way of “transfer” site X

comb

transceiver

First Deployment of Multi-Node Network

Slide48

First Deployment of Multi-Node Network

NIST

Table mountain

A

B

X

28-km roundtrip

Established 3-node network between NIST and Table Mountain Antenna Sites

Slide49

Initial Measurements over three-node network

Site X terminals—view from Table Mtn

Slide50

Strong

Turbulence

NIST

Table mountain

A

B

X

28-km roundtrip

NIST telescopes

B

A

Movie of incoming beam spread & turbulence at Telescope A

Slide51

First results on three-node network

Turbulent air

A

B

X

cavity stabilized laser

Clock offset

A-B clock offset, averaged over 1 s

Fractional Frequency

Instability (Mod. Allan)

10

-12

10

-14

10

-16

10

-18

1

10

100

1000

0.1

0.01

0.001

Averaging Time (s)

3-node network

28-km roundtrip

2-node

over 4 km

26% dropouts

2-node

over 4 km

1% dropouts

(should be zero)

Slide52

O-TWTFT: Limits to Future Ground-to-Satellite Links

Ultra-precise clock comparisons

~ 1e-17 @ 1 sec

Sub-fs clock Synchronization with 25 m/s Motion

Coherent 10 GHz – Synched Microwave

Osc.

Multi-Node network

Simplified PicosecondSynchronization

Modifications to support ultra-long distance links**

Limits to FutureGround-to-Satellite Links

O-TWTFT

Synchronization Across 12 km of Air

Completed

Under Investigation

Comparison of Atomic Optical Clocks

Applications to Coherent Sensing*

*At white paper level

** over 1-10 km-scale testbeds

Point ahead effects on O-TWTFT

Slide53

Transfer to orbiting satellite:Motion + Finite Speed of Light + Long Distance  Uplink & downlink paths separated!

What if the Link Isn’t Truly Reciprocal?

Pseudo wind

5 to 190 m / s

V ~ 5.3 - 7.5 km / s

Point-Ahead Angle

V/c

Typ. ~ 35 – 50

uRad

C. Robert, J.-M. Conan, and P. Wolf,

Phys. Rev. A

93

,

033860

(

2016

).

A. Belmonte, M. T. Taylor, L. Hollberg, and J. M. Kahn,

Opt. Express

25

,

15676

(

2017

).

Slide54

Experimental Setup to Test Breakdown of Link Reciprocity

30 cm

Slide55

Results

W.C Swann et al, to be published in PRA

d

minutes

d = 30 cm

d = 110 cm

non-reciprocal time-of-flight (fs)

Slide56

Results: Residual Time-of-Flight Noise at different terminal separations

Time-of-flight PSD

(s2/Hz)

Fourier frequency (Hz)

Slide57

ResultsTwo-way Measured vs. Model

Time-of-flight PSD

(s2/Hz)

Fourier frequency (Hz)

Model for

d = 0.5 m separation

V = 1.m/s wind speed

Adapted from A. Belmonte, Opt. Exp.

25

, 15676 (2017).

Slide58

Time Deviation

MEO satellite model

Our data (1 meter separation

W.C Swann et al, to be published in PRA

One-way

With O-TWTFT

Slide59

Conclusion

Comb-based OTWTF can successfully synchronize distant clocks:

In real time

With millisecond update rates

To below 10

-18

in frequency

To femtoseconds in time

Across turbulent links

Despite motion of up to 25 m/s & displacements of 250 m (with quadcopter)

Current system shows:

No velocity-dependent bias at sub-fs level

No range limitation from turbulence (other than SNR)

More advanced algorithms & protocols could extend range and velocity for comb-based OTWTFT

Networks possible