TEMPO 1D/2D - Performance Update - Bossa Nova Technologies
TEMPO 1D/2D - Performance Update - Bossa Nova Technologies TEMPO 1D/2D - Performance Update - Bossa Nova Technologies
Laser Ultrasonic Receiver TEMPO Serie Performance Update - Nov 2009 - . Bossa Nova Technologies 606 Venice Blvd. Suite B Venice, CA 90291 USA Tel: (310) 577 8113 Fax: (310) 943-3280 www.bossanovatech.com info@bossanovatech.com © 2006 Bossa Nova Technologies, LLC. All rights reserved.
- Page 2 and 3: Table of contents 1. TEMPO Update -
- Page 4 and 5: 2. New Features - Description 2.1.
- Page 6 and 7: photorefractive crystal at frequenc
- Page 8 and 9: TEMPO now integrates a noise reject
- Page 10 and 11: 2.4. Simultaneous In-plane & Out-of
- Page 12 and 13: The collecting optic has a high num
- Page 14 and 15: Figure 12: Description of the exper
- Page 16 and 17: 3.2. Thin sample: Lamb mode resonan
- Page 18 and 19: Figure 18: Description of the measu
- Page 20: Figure 20: B-scan results for throu
Laser Ultrasonic Receiver<br />
<strong>TEMPO</strong> Serie<br />
<strong>Performance</strong> <strong>Update</strong><br />
- Nov 2009 -<br />
.<br />
<strong>Bossa</strong> <strong>Nova</strong> <strong>Technologies</strong><br />
606 Venice Blvd. Suite B<br />
Venice, CA 90291<br />
USA<br />
Tel: (310) 577 8113<br />
Fax: (310) 943-3280<br />
www.bossanovatech.com<br />
info@bossanovatech.com<br />
© 2006 <strong>Bossa</strong> <strong>Nova</strong> <strong>Technologies</strong>, LLC. All rights reserved.
Table of contents<br />
1. <strong>TEMPO</strong> <strong>Update</strong> - Summary 3<br />
2. New Features - Description 4<br />
2.1. Calibrated Output 4<br />
2.2. Background vibration Compensation 5<br />
2.3. Laser Intensity Noise Rejection 7<br />
2.4. Simultaneous In-plane & Out-of-plane detections 10<br />
2.4.1. Principle of operation 10<br />
2.4.2. Figures of Merit 12<br />
3. APPLICATIONS: Example of Results 13<br />
3.1. Thin sample: Pulse detection using <strong>TEMPO</strong>-1GHz 13<br />
3.2. Thin sample: Lamb mode resonance measurement with <strong>TEMPO</strong>-<strong>2D</strong> 16<br />
3.3. Thermoelastic generation in Thick Plate: Pulse detection with <strong>TEMPO</strong>-<strong>2D</strong> 17<br />
2
1. <strong>TEMPO</strong> <strong>Update</strong> - Summary<br />
Over the past year, <strong>TEMPO</strong> has been significantly redesigned, integrating a number of<br />
improvements, including:<br />
• Motorized focusing (25 mm) with Manual or computer (USB connection) control. Computer<br />
controlled auto-focus can be implemented via the USB port.<br />
• Very versatile optical platform, allowing <strong>TEMPO</strong> to easily adapt to different performance<br />
requirement:<br />
• Small detection spot = ∅10m Standard detection spot = ∅75m<br />
• High Frequency Detection: Up to 1GHz<br />
• Simultaneous detection of the Out-of-plane and In-plane (horizontal) displacement:<br />
See <strong>TEMPO</strong>-<strong>2D</strong><br />
• Rejection of the Laser Intensity Noise. Laser Intensity Noise is rejected in order to achieve<br />
shot-noise limited detection below 10MHz,<br />
• CALIBRATED Output. The CALIBRATED Output delivers a signal proportional to the surface<br />
displacement, and it is automatically normalized to 100mV for 1nm displacement. The<br />
CALIBRATED Output operates up to 90MHz.<br />
• Improved compensation of low frequency vibrations. The New compensation loop uses an<br />
internal automatic signal normalization scheme, which allows to optimize the<br />
compensation loop independently of the amount of light collected (signal strength). This<br />
leads to fast response with weak or strong signals and it avoids compensation loop<br />
instability (possible oscillation) that could be caused by very strong signal.<br />
A)<br />
B)<br />
Figure 1: View of the <strong>TEMPO</strong> electronic back panel: A) <strong>TEMPO</strong>-1GHz and B) <strong>TEMPO</strong>-<strong>2D</strong><br />
3
2. New Features - Description<br />
2.1. Calibrated Output<br />
<strong>TEMPO</strong> allows for absolute amplitude measurement. A calibrated 1nm-amplitude<br />
modulation is induced in the path of the internal reference beam. By monitoring the strength<br />
of the interference signal corresponding to this internal 1nm-displacement vibration, we can<br />
directly convert the output voltage of the <strong>TEMPO</strong> signal to nanometer displacement value.<br />
For convenience, in order not to interfere with the measurement, the frequency of the<br />
internal reference vibration is set outside of the <strong>TEMPO</strong> detection bandwidth. Two modes of<br />
operations are possible:<br />
1) With CALIBRATION set in "Auto mode", the Calibrated output signal is automatically<br />
set for a conversion of 100mVnm.<br />
2) With CALIBRATION set in "Free mode", the Signal is not normalized and the output<br />
amplitude varies according to the amount of collected light. The conversion<br />
coefficient (to convert mV to nanometers) is continuously displayed and it is also<br />
available as a DC voltage. The DC voltage corresponds to a 1nm displacement.<br />
During scanning, the amount of collected light will vary from point to point. With<br />
calibration in "Auto mode", the CALIBRATED output of <strong>TEMPO</strong> is always normalized in order to<br />
maintain 100mVnm conversion. Automatic normalization is done with a feedback loop where<br />
an Automatic Gain Control (AGC) amplifier is controlled by the amplitude of the internal<br />
calibration signal. The AGC amplifier limits the CALIBRATED output to frequency below 90MHz.<br />
For measurement above 90MHz, absolute calibration is achieved in "Free mode" by recording<br />
the DC calibration value and by scaling (dividing) the HF signal after digital acquisition by the<br />
DC calibration value.<br />
Figure 2 illustrates the calibration features. Figure 2 shows result for C-scan done by<br />
scanning a thin metal plate glued on a piezo electric transducer. The scan surface is 0.4” x 0.4”.<br />
These are only out-of-plane measurements. The C-Scan is a <strong>2D</strong> representation of the out-ofplane<br />
displacement at a given time. The large point in the center corresponds to the place<br />
where the piezo is glued. The other peaks are different modes propagating at the surface of<br />
the plate. Figure 2 shows a comparison among: (a) HF output signal not normalized, (b) HF<br />
output signal (a) normalized by dividing by the DC calibration value ("Free mode") and (c) the<br />
CALIBRATED Output signal ("Auto mode"). We see than the two normalized C-scans (b) & (c)<br />
are similar. The two normalized scans (b) & (c) are calibrated at 100mVnm.<br />
4
(a)<br />
(b)<br />
(c)<br />
Figure 2: C-Scans of a thin metal plate glued on a piezo electric transducer vibrating. (a) Output without<br />
normalization, (b) Output (a) with a post processing normalization (scaling by the DC calibration value) and (c)<br />
Output with the automatic normalization provided by the <strong>TEMPO</strong>.<br />
2.2. Background vibration Compensation<br />
<strong>TEMPO</strong> takes advantage of the high-efficiency and high-sensitivity of two-wave mixing<br />
in a Bi 12 SiO 20 (BSO) photorefractive crystal. The corresponding response time for the<br />
photorefractive (PR) effect to record the dynamic hologram (to adapt to the change in the<br />
backscatter light) is around 2ms. This response time is given by the nature of the<br />
photorefractive material and the experimental setup (interference pitch, applied High-Voltage<br />
and amount of light intensity).<br />
A plot of the detector output at constant displacement versus the frequency is given in<br />
Figure 3. At high frequencies (above a cutoff frequency f o ) the photodetector signal is linearly<br />
proportional to the displacement. At low frequencies (below f o ) the photodetector signal is<br />
proportional to frequency, or equivalently to velocity. The decrease in the signal below f o<br />
provides the well-known insensitivity of the signal to low-frequency noise sources. The noise<br />
due to vibrations or turbulence is compensated by the adaptive two-wave mixing in the<br />
5
photorefractive crystal at frequencies below f o . Thus, f o is often called the compensation<br />
frequency or compensation bandwidth. This parameter is dependant of the material properties<br />
and of the incident optical power. For <strong>TEMPO</strong> with a laser wavelength of 532nm, the<br />
compensation bandwidth is around 75 Hz (Figure 6) corresponding to a crystal response time<br />
of 2.1 ms (Fit of the experimental data).<br />
1.2<br />
Normalized amplitude<br />
1<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
0<br />
Response time = 2.1 ms<br />
0 500 1000 1500 2000<br />
f (Hz)<br />
Figure 3: Frequency response of the photorefractive effect: - Theoretical fit; • Experimental.<br />
For measurement on thin samples, background vibrations can easily be picked-up by<br />
the thin sample (resonance of a membrane), leading to reduction in the system performance if<br />
the PR crystal can not compensate fast enough. To overcome this limitation, <strong>TEMPO</strong> includes<br />
an internal stabilization loop, which improves the compensation speed (the response time) of<br />
the crystal. This stabilization loop has been recently highly improved, showing much faster<br />
response time, Typically 1s (a X2 improvement is response speed compared to the<br />
initial measurement of the photorefractive effect).<br />
Data shown Figures 4-A and 4-B illustrates the improvement in response time that is<br />
achieved with "Compensation ON". These measurements were recorded for our <strong>TEMPO</strong>-<br />
120MHz. A step displacement corresponding to the maximum displacement measurable by<br />
the interferometer (8) is applied to the sample and the DC output signal is displayed,<br />
showing how fast the interferometer adapts to this displacement (response time of the<br />
interferometer). With "Compensation OFF", the response is given by the response time of the<br />
photorefractive effect. It must be pointed out that the response time of the photorefractive<br />
effect is slightly dependent on the direction of the displacement. This is a well-known effect for<br />
photorefractive crystal with an applied DC high-voltage [Refxx]. With "Compensation ON", the<br />
response time is highly improved, with only about 100S to recover from a 66.5nm<br />
displacement step. In the measurements shown Figure-4, the Photorefractive two-wave<br />
mixing effect exhibited a "natural" response time of 3ms to ms.<br />
6
Figure 4-A: Response (measured at DC output) to a negative displacement step of 66.5nm<br />
Figure 4-B: Response (measured at DC output) to a positive displacement step of 66.5nm<br />
2.3. Laser Intensity Noise Rejection<br />
In theory the signal-to-noise ratio (SNR) of a laser-ultrasound interferometric detection<br />
system should increase with the square-root of the amount of light power received on the<br />
detector. This statement assumes that the interferometer is shot-noise limited and all the<br />
other noise sources (electronic noise, laser intensity noise and laser phase noise) are negligible.<br />
However in practice, as the power on detector increase, measurements become more and<br />
more sensitive to the laser intensity noise. The use of interferometers with high collection<br />
efficiency and the use of detection lasers with higher power are advantageous only if the<br />
intensity noise is kept below the shot noise level.<br />
7
<strong>TEMPO</strong> now integrates a noise rejection circuitry [patent Pending] which rejects the<br />
Laser Intensity Noise (LIN) for frequencies below 10MHz. With single-frequency CW Lasers used<br />
for interferometric detection (and for ~1mW of light at 532nm on the detector):<br />
- Intensity Noise is dominant at lower frequencies: < 1MHz<br />
- Shot Noise dominates over LIN at Higher frequencies: > 10MHz<br />
- Electronic noise dominates for low power on detector at high frequency.<br />
The LIN rejection circuit is designed to reject the common intensity noise at<br />
frequencies up to about 10MHz. LIN Rejection is now included as a standard feature for the<br />
<strong>TEMPO</strong> Serie:<br />
- <strong>TEMPO</strong>-<strong>2D</strong>:<br />
- Out-of-plane output includes LIN rejection.<br />
- For In-plane detection, LIN is naturally rejected because of the differential<br />
process of the in-plane detection principle.<br />
- <strong>TEMPO</strong>-1GHz:<br />
- The LIN rejection is available at the CALIBRATED OUTPUT (Frequency<br />
bandwidth up to 90MHz)<br />
- The FULL-BANDWITH Output [1MHz - 1GHz] does not include the LIN<br />
rejection.<br />
- <strong>TEMPO</strong>-12MHz:<br />
- The LIN rejection is available at both the CALIBRATED Output [90MHz] and<br />
The FULL-BANDWITH Output [120MHz].<br />
Examples of LIN rejection are shown Figure 5 and Figure 6. The laser from Figure 6<br />
exhibited a strong noise frequency peak at 350kHz. With the LIN rejection, this noise frequency<br />
was reduced by more than 30dB.<br />
8
Figure 5: Rejection of the Laser Intensity Noise (<strong>TEMPO</strong>-120MHz detection bandwidth). Comparison among the<br />
noise spectrums: 1) Electronic noise from the rejection circuit only, 2) Electronic noise from the photodetector<br />
(newFocus-12MHz), 3) output noise for 1mW without LIN rejection and 4) with LIN rejection.<br />
Figure 6: Rejection of the Laser Intensity Noise (<strong>TEMPO</strong>-20MHz detection bandwidth): Comparison between with &<br />
without Noise Rejection.<br />
9
2.4. Simultaneous In-plane & Out-of-plane detections<br />
The <strong>TEMPO</strong> measures a phase shift that is proportional to a surface displacement along<br />
the direction of the probe beam (Z-axis). For most measurement, the probe beam is at (or<br />
near) normal incidence of the inspected surface. Thus, It is often referred as Out-of-plane<br />
measurement. Out-of-plane displacement is easily measured by interferometer, specifically if<br />
the surface is highly reflective (strong specular reflection). However, for measuring in-plane<br />
displacement the light backscattered at large angle (away from the specular reflection) must be<br />
used instead (Figure 7) because the directly backscattered light (specular reflection) doesn’t<br />
carry any in-plane displacement information.<br />
Figure 7: Schematic Principle for detecting In-plane and out-plane displacement<br />
2.4.1. Principle of operation<br />
Adaptive interferometers based on two-wave mixing (TWM) in photorefractive crystal<br />
(PRC) are optimized to collect many speckles. High collection efficiency is achieved using a high<br />
numerical aperture collecting optic. <strong>TEMPO</strong> with its large aperture optical system is thus<br />
potentially well suited for simultaneous in-plane & out-of-plane detections. Beams<br />
backscattered at large angle carry more information about the in-plane, than beams in the<br />
center that are carrying only out-of-plane information (Figure 8).<br />
An important feature of an adaptive interferometer using TWM in PRC, is that multiple<br />
beams can be independently processed inside the same crystal without cross-talk issue. This<br />
feature is used to realize a multiplexed interferometer for simultaneous detection of multiple<br />
beams corresponding to the observation at different viewing angles of the same illuminated<br />
point. For detection of the two components, we use a similar layout than our standard <strong>TEMPO</strong>.<br />
The optical setup was adjusted in order to ensure that the entrance pupil (the collecting optic)<br />
is imaged on the detector and the single-element detector was replaced with a detector array<br />
(Figure 9).<br />
10
Figure 8: Detection principle using a single, large aperture collecting lens.<br />
Figure 9: <strong>TEMPO</strong>-<strong>2D</strong> Optical Set for in-plane and out-of-plane detection.<br />
Each element of the detector array corresponds to a small area of the entrance pupil<br />
and thus corresponds to light backscattered along well defined incidences. Processing of the<br />
interference signals, as a function of the back-scattered angle, provides simultaneously inplane<br />
and out-of-plane displacements. The schematic of the electronic signal processing is<br />
shown Figure 10.<br />
11
The collecting optic has a high numerical aperture (F0.75), with a maximum collection<br />
angle of 31 o . The multi-detector is a linear-array with 16 elements. The collected light is<br />
focused on the linear array using a cylindrical lens. In this configuration, we detect the in-plane<br />
component along the orientation of the linear array.<br />
The light backscattered by the sample may not be uniformly distributed and each<br />
channel must be normalized before calculating the in-plane and out-of-plane. Normalization is<br />
achieved using automatic gain controlled (AGC) amplifiers monitoring the low frequency signal<br />
generated by an internal piezo-mirror in the path of the reference beam. After amplitude<br />
normalization, the signals are processed by pairs, with same incidence angles. For each pair,<br />
the two normalized signals are added to generate the elementary out-of-plane component and<br />
their subtraction gives the elementary in-plane component (Figure 10).<br />
Figure 10: Principle of the electronic demodulation circuitry for simultaneous in-plane and out-of-plane detection.<br />
2.4.2. Figures of Merit<br />
A critical receiver parameter is the surface displacement sensitivity or minimum<br />
detectable displacement called the Noise-Equivalent Surface Displacement (NESD). NESD is the<br />
RMS surface displacement amplitude that can be detected at a signal-to-noise of unity for 1 W<br />
incident power and 1 Hz bandwidth. The units of NESD are nm (WHz) 12 .<br />
To determine the minimum detectable displacement, the NESD is multiplied by the<br />
square root of the bandwidth and divided by the square root of the power on the detector.For<br />
12
Out-of-plane displacement, <strong>TEMPO</strong> (out-of-plane) has NESD better than 2x10 -7 nm (W/Hz) 1/2<br />
for f > 1 MHz. The noise spectrums for both the out-of-plane and in-plane outputs are shown<br />
Figure 11. For frequency above 1MHz we measured the NESD for Out-of-plane and In-plane :<br />
NESD OUT = 1.7 x10 -7 nm (W/Hz) 1/2<br />
NESD IN = 11x10 -7 nm (W/Hz) 1/2<br />
For the In-plane displacement, the sensitivity strongly depends on the spatial distribution of<br />
the backscattered light. Optimal conditions are achieved when uniform scattering occurs. The<br />
example of NESD measured in Figure 10 corresponds to near-optimal conditions. If the light<br />
scattering is not uniform, the noise level of the weakest signals will be amplified through the<br />
normalization process, resulting into a higher noise level for the in-plane output.<br />
Figure 11: Noise spectrum for the Calibrated In-plane Output and the Calibrated Out-of-plane output.<br />
3. APPLICATIONS: Example of Results<br />
3.1. Thin sample: Pulse detection using <strong>TEMPO</strong>-1GHz<br />
Examples of high-frequency measurement using the <strong>TEMPO</strong>-1GHz are shown below. In<br />
this experiment, generation was carried out with a MicroChip laser from Teem Photonics<br />
(PowerChip PNG), with a 300ps pulse duration. The experiment is described in Figure 12 and<br />
results are shown Figures 13, 14 and 15.<br />
13
Figure 12: Description of the experiment using the <strong>TEMPO</strong>-1GHz.<br />
Figure 13: Result recorded on an AlTi plate with 230m thickness.<br />
14
Figure 14: Result recorded on an AlTi plate of 265m thickness.<br />
Figure 15: Result recorded on a Copper plate of 400m thickness.<br />
15
3.2. Thin sample: Lamb mode resonance measurement with <strong>TEMPO</strong>-<strong>2D</strong><br />
Figure 16 shows an example of high-frequency Lamb wave detected with <strong>TEMPO</strong>-<strong>2D</strong>, in<br />
a 1mm thin aluminum sheet after propagation along 20mm. Generation was carried out with a<br />
line-focused pulsed laser beam. For this measurement, the measured in-plane direction is<br />
along the propagation direction. The signals were high-pass filtered with cut-off frequency set<br />
at 2MHz, in order to only visualize the higher frequencies. The high frequency vibration of the<br />
plate experiences low attenuation and the plate vibration is visible on both in-plane and outof-plane<br />
displacements for a long time, much longer than the time window showed Figure 16.<br />
We can see that the In-plane component carries higher frequency resonances compared to the<br />
out-of-plane component.<br />
When the detection and the generation are alignedsuperimposed then strong<br />
resonances are detected. Some Lamb modes exhibit an anomalous behavior at frequencies<br />
where the group velocity vanishes while the phase velocity remains finite. The zero group<br />
velocity (ZGV) leads to Sharp cw resonance and ringing effects. Figure 17 shows the fast Fourier<br />
transform computed on the first 50s of the signals. The spectrum of the out-of-plane signal<br />
(Figure 17-A) clearly shows the resonance of the S 1 mode and of the A 2 mode as described by<br />
Clorennec et al [*]. The resonance of the A 2 Lamb mode corresponds to the thickness shear<br />
resonance ( F ⋅ d = 3⋅V<br />
2 ), where V S is the shear wave velocity and d is the thickness.<br />
2 S<br />
20mm.<br />
Figure 16: High frequency Lamb waves detected in a 1mm thin aluminum sheet, after propagation along<br />
16
Figure 17: Time signal and corresponding Spectrum of the high-frequency Lamb wave, with detection at epicenter.<br />
A) out-of-plane displacement and B) In-plane displacement<br />
On the spectrum of the in-plane signal (Figure 17-B) the S 1 and A 2 resonances are also<br />
visible. We also see many other resonances at higher frequencies, which were not visible on<br />
the out-of-plane displacement. These modes correspond to the higher order Lamb modes,<br />
demonstrating the wealth of information available with the simultaneous measurement of inplane<br />
and out-of-plane.<br />
[*] D.Clorennec, C. prada, D. Royer and T. Murray , “laser impulse generation and interferometer detection of<br />
Zero Group Velocity Lamb mode resonance”, Appl. Phys. Lett. 89, 2006<br />
3.3. Thermoelastic generation in Thick Plate: Pulse detection with <strong>TEMPO</strong>-<br />
<strong>2D</strong><br />
An other example of In-plane measurement is shown below. Figure 18 describes the<br />
through transmission experiment. Generation was achieved in thermoelastic regime, with the<br />
laser beam focused along a line and detection was carried out on the other side. The sample<br />
was a thick aluminum plate: 12.7mm thickness. In a first experiment, we use a sample free of<br />
defects (no blind holes). The detection was scanned along a 50mm line. The B-scan results for<br />
both In-plane and out-of-plane displacements are shown Figure 19. The strong shear wave is<br />
clearly visible. Many reflected and converted wave arrivals are visible. After 12s, some<br />
reflections from the sample edges are visible. For these measurements, the calibration<br />
coefficient is 100mVnm for both In-plane and out-of-plane outputs.<br />
17
Figure 18: Description of the measurement setup for the sample with Blind-holes.<br />
Two defects (blind holes as described on Figure 18) were introduced and the<br />
measurements were repeated. The result comparing the B-scan, without and with, defects are<br />
shown Figure 20. Reflection diffraction from the two defects are clearly identifiable on the Inplane<br />
B-scan.<br />
18
Figure 19: B-scan results for through-transmission In-plane and Out-of-plane measurement on a 12.7mm thick<br />
aluminum plate (no defect) with laser line generation.<br />
19
Figure 20: B-scan results for through-transmission In-plane and Out-of-plane measurement on a 12.7mm thick<br />
aluminum plate: A) without defect and B) with Defect (2 blind holes).<br />
20