Adaptive demodulation of dynamic signals from fiber Bragg gratings

download report

Transcript Adaptive demodulation of dynamic signals from fiber Bragg gratings

Adaptive demodulation of dynamic signals from fiber
Bragg gratings using two-wave mixing technology
Yi Qiao , Yi Zhou and Sridhar Krishnaswamy
Center for Quality Engineering and Failure Prevention
Northwestern University
Illinois
Applied Optics
Vol.45,No 21,Pg 5132-5142
Journal Club Presentation
9/18/06
Presenter: Ashwin Kumar
Outline
 Motivation
Fiber Bragg grating (FBG) sensors
Demodulation Schemes
Two-Wave Mixing Technology
 Theory
Adaptive Two-Wave Mixing
Adaptive Two- Wave Mixing Wavelength Demodulator
 Experiment
Experimental Configuration
Wavelength demodulation
Adaptivity to Quasi-Static Drift
Detection of transient Impact signals
Four-channel Multiplexed Two-Wave Mixing Demodulator
 Conclusions
Fiber Bragg Gratings (FBG) Sensors
 FBG reflects a very narrow bandwidth of
light centered at the Bragg wavelength within the
transmission spectrum
 The reflected wavelength is dependent on the period
of the Bragg grating and the guiding properties
of the fiber
 Proportionally related to environmental variables
(i.e. temperature, stress and pressure )
 Fiber Bragg gratings operate as sensors when changes
in a particular environmental variable are correlated
with shifts in the reflected wavelength of the FBG.
 FBG sensors are insusceptible to EMI
and have no EM emission
 They are intrinsically safe and have unique
optical multiplexing potential
 A large number of FBG sensors can be
addressed and can be read out by a
limited number of lead fibers
 The FBG sensors are potentially lightweight,
small and can be embedded and
integrated in (composite) structures
 Long-term stability and fatigue durability
 Main Application : structural health monitoring
Demodulation Schemes
 Spectral shifts has to be monitored
Demodulation Schemes
Scanning Type
Only one FBG sensor can be investigated
at a given time.
Not Suitable for multiple sensor investigation
Spectrometry based
Interferometry based
Low sensitivity
Ideal for dynamic strain monitoring
Not suitable for dynamic measurements
In case of multiple sensor array
Requires electronic feedback to actively
Compensate for quasi-static drifts to keep
The interferometer at quadrature.
 Need for a cost-effective and parallel demodulation scheme
for arrays of FBG sensors
Cost of Multiplexing is high
Two-Wave Mixing Technology
 A novel two-wave mixing (TWM) wavelength
demodulation scheme for FBG sensors is demostrated
 Ability to compensate for quasi-static drifts without the
need for active stabilization
 Suitable for measuring dynamic and transient strains
induced by vibrations , impact, ultrasound, or acoustic
emissions
 This Scheme can be readily multiplexed by using
wavelength division multiplexing
 Demonstration of a four-channel TWM device to
demodulate dynamic strain signals from four FBG sensors
simultaneously
Adaptive Two-Wave Mixing
Signal Beam
<001>
Transmitted
Signal Beam
Pump Beam
<001>
Diffracted Pump Beam
 Photorefractive TWM is a dynamic holographic process
 Two coherent beams – Pump and Signal – interact within the photorefractive crystal
 Wave-Mixing Process
a) Creation of Intensity gratings due to coherent stationary interference of the beams
b) light is absorbed and free carriers are generated in the bright regions of the intensity pattern
c) Carriers diffuse and/or drift from the bright regions leaving fixed charges of opposite sign
d) Free carriers can be trapped by ionized impurities at other locations ,depositing their charge
there as they recombine
e) Creation of an inhomogeneous space-charge distribution that modulates the refractive index
of the crystal through electro-optic effect.
f) This causes diffraction of the interacting beams and this process of dynamic coupling
is called wave mixing,
Adaptive Two-Wave Mixing
 A part of the pump beam is diffracted by the index grating in
the direction of the transmitted signal beam
 Diffracted pump beam is wavefront matched with the quasi-static signal beam
 Diffracted Pump beam and the transmitted signal beam interfere
with each other to demodulate any phase difference between them
 The TWM interferometer is adaptive as the created index grating is a dynamic
grating, and the crystal can adapt to any phase shift that is slower than the PRC
response time by forming a new index grating.
 Diffracted pump beam will therefore track any quasi-static changes in the signal
beam phase, resulting in no quasi-static phase difference between the diffracted
pump and transmitted signal beam.
 Dynamic phase changes faster than photorefractive response time of the PRC will
not be present in the diffracted pump beam
 This will result in a net phase difference between diffracted pump and the
transmitted signal beam. Which can picked up by the interference of these two
beams.
 By applying a DC field to the PRC, the interference pattern can be kept nearly in
in phase with the created index grating.
 By adjusting the DC field and mixing angle, the TWM interferometer can be made
to work at near quadrature. (the diffracted pump beam, is 90 phase shifted w.r.t
transmitted signal beam.)
Adaptive Two-Wave Mixing
 Adaptivity of the TWM process comes from it’s ability to stay at
quadrature
 PRC can therefore be regarded as a high pass filter that selectively monitors any
high frequency changes in the mixing beams.
 It is ideal for monitoring small dynamic strain signals
 Transmitted Signal Beam
Es  E0 exp[i (t )]
 Diffracted Pump Beam
 (t ) – dynamic phase shift
E0 - Complex amplitude of the signal
L – crystal length in the beam propagation
direction
   ' i " - TWM complex gain
Edp  E0 {exp[ L]  1}
 Interference signal at the photodetector
I  E0 {e 2 ' L  2sin( " L)e ' L (t )}
2
Interference signal varies
linearly with dynamic
Phase shift
 /2
DC level , contributed to Photodetector shot noise
Adaptive TWM Wavelength Demodulator
 Reflected light from a FBG sensor is split into two beams (signal and pump)
 The Two beams are made to travel unequal paths before mixing
 Any wavelength shift would cause a phase shift between the beams due to travel
over unequal path lengths
 Spectral shift induced by strain and temperature
B
B
 {1 
2
neff
2
[ p12  ( p11  p12 )]} z  (    N )T
B – center wavelength of the FBG sensor
B– wavelength shift caused due to strain or temperature
neff
Pij
– effective refractive index of the fiber
– components of strain optic tensor
 – Poisson’s ratio
 Z – strain along the fiber
 – thermal expansion coefficient
 n – Thermo- optic coefficient
 For Bragg sensors at 1550nm, it has been estimated
That 1 microstrain will lead to 1.2 pm change in
Wavelength and 1 C change in temperature will lead
to about 13 pm change in wavelength
Adaptive TWM Wavelength Demodulator
 Phase Shift between the beams
 (t ) 
2 d
B2
B
d is the optical path difference (OPD)
 Greater the OPD , larger is the phase shift and stronger is the interference signal
 Using broadband light sources to illuminate FBG’s, results in the FBG reflection
spectrum having a finite line width of the order of 0.1-0.4 nm
 Implications: coherence of the two interfering beams needs to be taken into account.
 Fringe visibility due to the interference of two beams of finite spectral width k
2 r
k 2 d 2
V
exp{
}
r 1
16 ln 2
r – Intensity ratio of the two beams
 Incorporating the degradation in fringe visibility due to low coherence in the
Interference signal expression , we get
k 2 d 2 d B
S  exp( ' L)sin( " L) exp{
} 2
16ln 2 B
Adaptive TWM Wavelength Demodulator
 Wavelength Demodulation Signal
k 2 d 2 d B
S  exp{
} 2
16 ln 2 B
 L = 1 cm ,  ' (TWM gain) = 0.3 cm-1
 OPD =0 , no wavelength demodulation
 Amplitude increases with OPD to a max
and then starts to decrease due to decreasing
fringe visibility
 Narrower the linewidth, larger the optimum
OPD, larger the demodulated signal
 Trade Offs
Narrower linewidth FBG are longer in length
which decreases the highest frequency to
which the FBG can respond
Larger OPD decreases the dynamic range
the FBG can measure
Experimental Configuration
 Broadband Amplified spontaneous emission
(ASE) source in the C band (1530 to 1570 nm)
 Optical Amplifier : EDFA working
at 500 mW
 Both the beams enter the crystal
by the [-1 1 0] face and the DC
field is applied along the <001>
direction
 Peltier cooler is used to prevent electrical
breakdown due to crystal overheating.

Two HWP are used to rotate the beam polarization to be
S- polarized. (along the <110> direction)
 Under the applied DC field , index grating is in phase with the interference pattern, with
the TWM kept at quadrature, provides optimal demodulation of phase/ wavelength changes.
Wavelength Demodulation
 Test Parameters: FBG sensor centered at 1552 nm with a line
width of 0.1 nm, length of 10 mm, and reflectivity of 50%
 Glued to a PZT stretcher
 10kHz , 10 strain onto the FBG sensor
 Intermittent DC field appplied
from 1 to 6 ms
 Photorefractive grating builds up
 strain applied to FBG from
2 to 6 ms as a tone burst
 OPD=0, TWM gain =max, zero
demodulated signal detected
 AS OPD increases, slowly the
demodulated signal starts to
appear
Wavelength Demodulation
 Optimum OPD - 0.1 nm line width FBG
= 8mm
 For a OPD =8mm, the wavelength to
phase shift conversion sensitivity comes
about 21 radians/nm wavelength shift at
1550 nm. Implies 0.0252 radian/microstrain
These changes are easily detected using
TWM
 TWM gain ’ experimentally is verified
to be a small number (0.47 to 0.1 cm-1)
for OPD (0-12mm)
 For a 20KHz dynamic strain amplitude
the TWM demodulator output was found
to vary linearly with spectral shift (strain)
 Minimum detectable strain with current setup
~ 0.25 microstrain corresponds to 0.3 pm
spectral shift
 Limited by ASE and EDFA intensity noise
 Improved through balanced photodetection to
cancel the intensity noise.
Adaptivity to Quasi-Static Drift
 To demonstrate the adaptivity of TWM
setup to quasi-static drift
(low frequency strain or temperature drift)
 A frequency sweep signal from 10 Hz to
1.2 kHz with a constant amplitude of 10
microstrains was applied
 Demodulator ignores the low frequency
strain applied in the beginning.
 Starts to respond to frequencies above
600 Hz
 Acts like a high pass filter with a cut off
frequency of 600 Hz
 Cut-off frequency is directly proportional
to the response time of the PRC. Faster the
response, higher is the cut-off frequency.
Adaptivity to Quasi-Static Drift
 In principle, there is no upper limit to detectable
frequency range of FBG spectral shifts from TWM
process
 Limit occurs only from FBG sensor response and
electronics bandwidth of the photodetector
 A high frequency dynamic strain (580kHz ,
3microstrain) is successfully demodulated.
 To Demonstrate adaptability to large
temperature drift
 Temperature drift introduced to FBG
through TEC module connected to
temperature controller.
 Temperature is monitored using a thermistor
 0.1 Hz sine input was supplied to controller
and a temperature drift introduced was 10 C
drift within 5secs (2 C /sec)
 An optical spectrum analyzer was used to
record the reflection spectrum of FBG as the
temperature drifts.
 10 C change causes a wavelength change of 110 pm
Close to 130pm (13pm/C)
Adaptivity to Quasi-Static Drift
 NO variation due to the 0.1Hz 10 C temperature drift
 110 pm wavelength shift would correspond to a 2.3 radians for this
configuration
 This phase shift, if picked up would cause the amplitude to vary a lot
 No such shift demonstrates the system ability to
compensate large temperature drifts.
Detection of Transient Impact Signals
 FBG sensor was covered by a 1mm
thick dry couplant made from silicon
epoxy
 A 3mm metal plate was placed on top of
the epoxy
 A 3mm ball was dropped from a height
of 5cm above the metal plate
 An oscilloscope was used to capture the
demodulated signal
 Demodulated signal shows multiple
bouncing of the ball bearing
 Frequency of the impact signal was
about 5 KHz
Four-Channel TWM wavelength demodulator
 Multiplexing possible
without significant increase
in cost
 Channels are separated
using band drop filters
 Multiple FBG sensors
with distinct spectral
reflectivities are selected
 Center wavelength separation
is chosen to be large enough to
avoid stationary optical
interference between the
multiple channels
Four-Channel TWM wavelength demodulator
 Inside the PRC, each channel creates it’s own index grating with
different index grating pitches
C
 
2sin( / 2)
Center wavelength separation
Angle between signal and pump beams
Change in index grating pitch
 A channel separation of 4nm was chosen which gave a index grating pitch shift of 76 nm
at a beam angle of 3 degs
 4 nm channel spacing allows 10 channels in the C band (1530nm – 1570nm)
 Multiple gratings can be written in a PRC with negligible cross talk if the pitches differ
by 0.03 nm.
 Four 0.1 nm line width FBG sensors were connected in series and are centered at 1548,
1552,1556,and 1560 nm respectively.
 FBG Sensor 1 (1548nm) : 10kHz, 5 microstrains
 FBG Sensor 2 (1552nm) : 5kHz ,5 microstrains
 FBG Sensor 3 (1556nm) : 2kHz ,5 microstrains
 FBG Sensor 4 (1560nm) : 20kHz, 5 microstrains
Four-Channel TWM wavelength demodulator
a) Demodulated signal amplitude for each channel is slightly different despite same input amplitude
b) Each channel has different optical intensities (non uniform EDFA gain)
c) Can be corrected by precalibration or gain-flattened EDFA
d) Fourier spectra confirms there is no cross talk between the channels
Conclusions
 First Adaptive wavelength demodulator for spectrally
encoded FBG sensors based on two-wave mixing
interferometry
 Optimum value of OPD for a 0.1 nm line width FBG was
found out to be 8mm
 Spectral resolution of the TWM wavelength demodulator
is of the order of 0.3 pm
 Resolution limited by EDFA and source intensity noise
 TWM wavelength demodulator is adaptive to quasi-static
drifts and temperature drifts.
 Well suited for detecting dynamic strains.
 TWM was also used to study transient impact signals
 Demonstrated a four channel system investigation
through wavelength multiplexing.