Transcript Slide 1
OFS Perth 2008
A multiplexed CW Brillouin system, for
precise interrogation of a sensor array made
from short discrete sections of optical fibre
John Dakin1, Sanghoon Chin2, and Luc Thévenaz2
1
Optoelectronics Research Centre, University of Southampton
[email protected]
2
Ecole Polytechnique Fédérale de Lausanne, Switzerland
[email protected]; [email protected]
Summary
• We present, for the first time, a novel multiplexed sensing architecture
for real-time monitoring of a small array of optical fibres.
• Signal separation, is in the frequency domain, rather than the usual
time domain, and relies on each fibre having a different Brillouin shift.
• The ability to monitor with a 100% duty cycle gives enhanced signal to
noise ratio, allowing precision measurement.
• We will show first the basic feasibility of the method, then show how
the it may be used for precise temperature measurement.
Schematic diagram of basic sensing system
DFB-LD
nc
F3
1W
EDFA
VOA
F2
F1
Chamber
PC
n B3 n B2 n B1
Delay
nc
nc
Beating notes
Det
ESA
The DFB laser pump into the fiber, via a 1W EDFA a circulator, and 3 Brillouin signals
return to be mixed with part of the pump wave on a ~12 GHz photo-receiver .
(The delay line is used to break coherence of the pump beam, to reduce interference between the
pump wave and undesirable pump-wave residues returning from port 3 of circulator)
1
Schematic diagram of basic sensing system
DFB-LD
nc
F3
1W
EDFA
VOA
F2
F1
Chamber
PC
n B3 n B2 n B1
nc
Delay of the cascaded fibers
• Configuration
n
Beating notes
c
Det
diameter: 7 cm
diameter: 7 cm
50 m of DSF
15 m of DCF
2
ESA
50 m of special fiber
with small size of core
OFS Perth 2008
Theory for Brillouin Shift in Optical Fibres
•The Brillouin shift is proportional to the longitudinal acoustic velocity in the glass
material, mainly of the fibre core region in which the majority of energy propagates
nB
2nVa
Va , being acoustic velocity
•Fibres of different composition can have significantly different acoustic velocities, as the
latter is a function of the density and the Young’s modulus of the glass.
•Doping with heavy elements will generally increase the density markedly with respect to
pure silica, and many dopants also reduce the Young’s modulus, both parameters
therefore tending to reduce the acoustic velocity.
It is therefore relatively easy to select fibres having markedly different
Brillouin shifts to suit our multiplexing method!
Measured Brillouin Stokes Spectrum from the 3-fibre array
1.8
fibre-1
Amplitude, a.u.
1.5
1.2
fibre-3
0.9
fibre-2
0.6
0.3
0.0
9.6
9.8
10.0
10.2
10.4
10.6
10.8
11.0
Frequency, GHz
This figure shows three different Brillouin Stokes signals, one from each
fibre, as displayed on the electrical spectrum analyzer (ESA).
3
Temperature dependence of Stokes Frequency.
2.1
1.5
(a)
(b)
fibre-1
fibre-3
1.5
increasing temp.
1.2
fibre-2
0.9
0.6
45 oC
35 oC
1.2
Amplitude, a.u.
Amplitude, a.u.
1.8
25 oC
65 oC
55 oC
85 oC
75 oC
0.9
0.6
0.3
0.3
0.0
0.0
9.6
9.8
10.0
10.2
10.4
10.6
10.8
11.0
10.70
Frequency, GHz
10.72
10.74
10.76
10.78
10.80
Frequency, GHz
Frequency change of the Stokes signal, as the temperature of the chamber
changes, in steps, from 25 oC to 85 oC. It is clearly seen that only fiber-3
shows a variation, whilst the others scatter light at constant frequency.
4
Linear variation of the Stokes on Temp.
10.79
Brillouin Stokes
shift, GHz
Change in Brillouin
shift MHz, relative to
the 250C value
25.1
10.724
0.00
35.2
10.734
9.3
45.2
10.745
21.0
55.3
10.755
30.3
10.74
65.3
10.764
49.7
10.73
75.2
10.776
51.3
10.72
85.2
10.785
60.7
Brillouin shift, GHz
Temp oC
10.78
10.77
10.76
10.75
20
30
40
50
60
70
90
80
Temperature, oC
This display shows the frequency change of the Stokes as the temperature
of the chamber changes from 25 oC to 85 oC. It is clearly seen that only
fiber-3 varies, whilst the others stay at the same frequency.
5
Precision temperature sensor
DFB-LD
n
FBG 1
n
FBG 2
EOM
Probe
Delay
1-km SMF
Chamber
F1
F2
n
n
Pump
EDFA
1W
VOA
F3
Det
BPF
Frequency
Counter
To create a precise temperature sensor, the Stokes scattered light was down-converted,
using a modulation sideband of the pump laser as local oscillator, to give a beat signal of
order 145 MHz.
NOTE: The intermediate-frequency beat signals were then mixed with a 150 MHz local oscillator in
a second electrical (i.e. post-detector) , down-conversion stage, to give ~ 5 MHz signal for
frequency measurement. This 2nd mixing stage will be shown in next slide.
6
Detection system, now showing the second
(electrical) down-conversion mixing stage
Electrical
downconversion
Frequency
Counter
+
Photoreciever:
Bandwidth, 125 MHz
RF L.O
signal
High pass filter:
Cut-off, 150 MHz
Low-pass filter:
Cut-off, 15 MHz
7
Frequency, MHz
Results of heating and cooling cycle
10
9
8
7
6
0
1
2
3
4
5
6
Delay time, hour
The counter frequency was monitored during slow heating and cooling,
with a total temperature excursion of ~ 3.5 C.
The magnified insert shows the short-term frequency fluctuation was only
of ~10 kHz RMS, equivalent to ~ 0.01 C, despite fast (1s) update time.
8
Conclusions
•We have conceived and demonstrated a new frequency-divisionmultiplexed Brillouin sensor system
•We have shown that it is possible to get good separation of Brillouin
signals by simple selection of commercial fibres
•We have shown that the sensor operates with low crosstalk between
sensors (NOTE : WE NEED TO DEMONSTRATE THIS NEXT)
•We have achieved a noise-limited temperature precision of ~ ± 0.01 C0
RMS, with a fast update time of only 1 second
Acknowledgements
Prof John Dakin wishes to thank EPFL for granting him a short visiting
professorship
All the authors wish to thank Andrew Sansom, of Golledge Electronics,
UK, for providing a number of complimentary filter samples at short
notice.