Nature - USA (2020-05-14)

168 | Nature | Vol 581 | 14 May 2020

Article

shift (see Fig. 2c). We define the chirp nonlinearity as the deviation of
the measured instantaneous frequency from a perfectly symmetric
triangular frequency-modulation scan, estimated with least-squares
fitting, and depict results for the pump and two comb teeth in Fig. 3c
(bottom). Narrow peaks of the chirp nonlinearity are attributed to
single-mode dispersive waves^29. We do not observe intermode breath-
ing of the soliton^32 in the present system. The channel-dependent

root-mean-square nonlinearity is depicted in the inset of Fig. 3e and

remains below 1/500 of the full frequency excursion for all channels at

100-kHz modulation frequency. The frequency-dependent transduc-

tion of the frequency modulation from the pump laser to the DKS teeth

is calculated from the transduced chirps (see Extended Data Fig. 3)

and is plotted in Fig. 3d. We find a lower bound for the 3-dB modula-

tion frequency cutoff of 40 MHz, which corresponds to a maximum

192 192.5 193 193.5 194 194.5 195

Frequency (THz)

0

0.5

1.0

Standard

devaition (cm)

a

bcd

966 lines

per millimetre

228 Hz

v 2 v 14

v 8

Frequency-modulatedmicrocomb

EDFA

DEMUX

90/10

50/50

DSO

COL

DEMUX

ESA

Frequency (THz)

Power

(^102030) Frequency (MHz) (^40506070) (20 dB per div)^184188192196200
Power
(20 db per div)
2.5 7.5 μμss 194.4 THz
2
(^814)
194.4 THz
193.8 THz
193.2 THz
30
50
30
50
10
30
50
10
10
Frequency (MHz)
0102030405060
Time (μs)
0102030405060
Time (μs)
30
50
30
50
10
30
50
10
10
Frequency (MHz)
194.4 THz
193.8 THz
ef193.2 THz
2
8
14
–15
–10
–5
0
5
10
15
Radial velocity (m s
–1)
gh
x 8 = 10.0 m
x 14 = 10.0 m
v 8 = 0.02 m s–1
v 14 = 3.26 m s–1
x 2 = 10.0 m v 2 = –3.40 m s–1
x 8 = 10.0 m
x 14 = 10.0 m
x 2 = 10.0 m
v 8 = 0.0 m s–1
v 14 = 0.0 m s–1
v 2 = 0.0 m s–1
9.98
9.99
10.00
10.01
10.02
Distance (m)
192.0 192.5 193.0 193.5 194.0 194.5 195.0
Frequency (THz)
192.0 192. 51 93.0 193. 51 94.0 194. 51 95.0
Frequency (THz)
9.97
fd fu
RBW 530 kHz RBW 530 kHz
Fig. 4 | Demonstration of massively parallel velocity measurement
using a soliton microcomb. a, Experimental setup. The amplified
frequency-modulated lidar microcomb source is split into signal and local
oscillator pathways. The signal is dispersed with a transmission grating (966
lines per millimetre) over the horizontal circumference of a f lywheel mounted
on a small direct-current motor. The ref lected signals are spectrally isolated
before detection. COL, fibre collimator. b, Radio-frequency spectrum of lidar
backref lection mixed with the local oscillator (sampling length 3.75 μs) around
2.5 μs (upward ramp) and 7.5 μs (downward ramp). c, Optical spectrum of comb
lines after amplification. Blue shading highlights 30 comb lines with sufficient
power (>0 dBm) for lidar detection. d, Schematic illustration of the f lywheel
section irradiated by the frequency-modulated soliton microcomb lines
indicating the projection of the position xμ and velocity vμ of the wheel onto
the comb lines. e, Time–frequency maps of selected microcomb FMCW lidar
channels (sampling length 0.5 μs) for the static f lywheel. f, Same as e, but for
f lywheel rotating at 228 Hz. g, Multichannel distance measurement results
for the static f lywheel. Distance measurement not corrected for fibre path
difference between signal and local oscillator path. h, Multichannel velocity
measurement for the f lywheel rotating at 228 Hz. The accuracy of distance
and velocity measurements in case the rotating f lywheel is affected by
vibrations.