To be precise about the orientation, I always get a polarization which shows a larger in plane component than the out of plane component (which is in contradiction to the theory)..

Thanks

Hi,

please post a sketch of the wedge transducer. May be its doing something weird.

LDV = Laser velocity detector????????

Which plane is in plane and which is OP?

How long a distance away from the transducer did you do the measurement? There is a similarity to optical waves in near field and far field.

Was the measurement region stationary or did you track (and plot polarisation) along the direction of wave travel?

Did you excite with a pulse or with the expected frequency? Or what else?

Did you try to excite the zero order mode or also higher order ones?

What about excitation multiple other waves: compression and or shear waves? And measuring both?

RHABE

Thanks.

Sort of material: does not matter as long as low damping and good elasticity, that is very low internal loss.

Same wave principles in SAW (surface acoustic waves) in quartz or LiNb03 or steel or rocks. Rayleigh waves look similar as waves on water, but are different, both show elliptical movement of one point of material.

But waves on water are kinetic-energy versus gravitational-energy oscillations where

Rayleigh waves are elastic-energy versus kinetic-energy oscillations.

RHABE

Thanks,

Don't the particles near the surface move in ellipses with their long axis vertical (perpendicular to the surface) and the particles deep down move in ellipses with their long axis horizontal (// to the surface)?

What sort of material, and what size of materials are involved here, do you think we're talking about earth quakes (hitting the ground with a lump hammer) or Surface Acoustic Waves in piezoelectric substrates or somewhere in between?

This is a rail steel, and for now I am trying to get a polarization of Rayleigh wave from this material. The geometry of the material should not have any effect in the polarization.

Don't the particles near the surface move in ellipses with their long axis vertical (perpendicular to the surface) and the particles deep down move in ellipses with their long axis horizontal (// to the surface)?

This is the image from the same thesis (Gokhale's) for a better illustration purposes:

where x1 is the surface plane and x3 is the depth of the material. And what I got from my experiment is contradictory to the theory.

Thank you

More questions:

is the Plexi-wedge secured to the surface (glue or else?)

Is the coupling between Plexi and steel good enough - there should be considerable impedance mismatch?

There should be a damping material on the free face of the transducer, coupled by a matching layer of half-wavelength thickness. If not, then the reflections from the Plexi-rail-surface will be reflected from the free surface too and bouncing back and forth according to angles of the wedge. Thus the rail will be excited by all these.

Your spot - where your measurement is done - may be too big, so that at vertical orientation everything is ok, but at inclined orientation I doubt about the result.

Spot size should be near Λ/8 or below.

10 cycles is much too low. This is an oscillator with a Q-factor of likely above 10,000.

What about the length of your rail? And the end-faces? Reflections from these?

How do you discriminate between Rayleigh waves and other modes?

RHABE

is the Plexi-wedge secured to the surface (glue or else?)

Is the coupling between Plexi and steel good enough - there should be considerable impedance mismatch?

High vacuum grease is used as a coupling agent between the transducer and the wedge, and also between the wedge and the test specimen. I even use clamp to tighten the wedge to the specimen to make sure that there's no air bubble in between.

There should be a damping material on the free face of the transducer, coupled by a matching layer of half-wavelength thickness..

What kind of material that can I use for this?

Your spot - where your measurement is done - may be too big

I tried to use the smallest spot size by focusing my laser into the specimen. The size of the spot should be about .5 mm. I even tried to get the out-of-plane displacement measured directly 90 degree from surface and compared it with displacement measured from different angles and used trigonometry to get the out of plane displacement. And they had a very close result. So, spot size should not change my orientation.

10 cycles is much too low. This is an oscillator with a Q-factor of likely above 10,000.

Actually, 1 cycle should be enough. 10 cycles of Rayleigh wave is to make sure that the ellipses have about the same size. I'll show you the image of the result in the next question.

How do you discriminate between Rayleigh waves and other modes?

I used amplifier to increase the magnitude of the wave. the Rayleigh wave is the one with the highest amplitude. For example :

The wave on the left is the 10 cycle Rayleigh wave.

What about the length of your rail? And the end-faces? Reflections from these?

The sample is a 12" long rail steel with cross section of 12.88 in^2. This specimen is coated with "reflective finish" (an aerosol) to increase the reflectivity of the surface.

Hi,

"damping material" : I only know that the producers of medical ultrasonic transducers use these. I would try with any grainy material (sand, balls, also compacted or glued to solidify, grain size near wavelength).

"Spot size": I do not understand why a probe of 500 µm spot size can measure amplitudes of a wave of below 5 µm wavelength? Depending on averaging over spotsize any result may exist.

"Plot of Rayleigh wave": Can you plot a higher horizontal magnification of the plot. This seems to be a mixture of two components.

All other questions solved.

RHABE

About the spot size: The LDV measures the velocity of the particle, and I simply integrate the results to get the displacements. I am not sure if spot size have anything to do with this.

This is the close view of Rayleigh wave. This is not from the same measurements as the plot that I have shown before, but this should give the general idea about the Rayleigh wave. Yes, there might a mixture of two components in the wave. Please let me know if you can see it.

"The LDV measures the velocity of the particle, and I simply integrate the results to get the displacements"

This measurement assumes that the full region of illuminated spot is moving with one and the same velocity. If not, then the mean velocity is measured.

If so in your measurement system, then only at entering of pulse train into the spot-region a signal is generated and at exiting. The part of the pulse train below the sensitive spot region is averaged, so only one half pulse is generating the measurement signal.

Please look at your measurement: if you generate this pulse train of 10 pulses and look the same way as in your diagram and look also 500 (better 800) µm "later" or "earlier" there should be another train measured, two pulse-trains measured if only one pulse-train generated. The 500µm is the estimated spot size, to be translated to time by the wave-velocity of estimated 3Km/s (?).

If the first half-wave of the pulse-train has entered the sensitive region of the transducer this is generating a signal of 1half wavelength divided by transducer sensitivity length (spot size in direction of wave-travel). So this will be near 2µm/500µm or 1/250. So your 30nm (0.03µm) Amplitude shall in reality be near 7µm.

If the first full wave has entered the sensitive region of the transducer then no measurement signal will be generated as half is moving up and half is moving down.

If the transducer has no sharp transition (assumed above) between non-sensitive and sensitive area then the convolution integral of sensitivity shape by signal is giving the response of the transducer.

If you look at an inclination you will have another artifact: The peaks will be seen by the optical eye of the sensor at full amplitude but the valleys will be partially shielded by the peaks and by the aperture (lens diameter to distance ratio). So the inclined look will have a much better sensitivity to the regions that are near maximum. This is valid for both directions of motion - these to be resolved (projected) onto the sensitive direction.

I am not totally sure that this is a complete and true description of the situation.

So to test: is the second measured pulse train existent?

Is the measurement (nearly) the same if the amplitude of excitation is to 0.1 and 0.01 and at the noise floor of the instrument?

What is the lower and upper frequency limits of the instrument? Which rolloff of the high/low-pass filter?

I never experienced anything similar. I think that you can imagine what happens if you attach a dime onto your fingertip and try to detect an oscillation with 1mm wavelength.

RHABE

Hi,

my statement about the amplitude cannot be correct, at this amplitude the steel surface would spew bits of steel.

So what I estimate is that the averaging of the partially oscillating surface where a small part of the illuminated spot only is oscillating is done in a way that includes a high pass filtering: the instrument (LDV) is measuring above a lower frequency limit only and near and below this frequency limit according to filter rolloff with ~f or ~f².

So the non-oscillating parts will not have a considerable effect and the amplitude measurement may be nearly true.

This may affect my statement about inclined view - I have to think about.

RHABE

Many thanks to RHABE and Randall for the replies. I really appreciate the thoughts from you guys. I will try some other experiments to this, and I will update it again if I am able to solve this problem.

please post a sketch of the wedge transducer

This is the sketch that I got from Shailesh Gokhale's thesis (txspace.tamu.edu),

(and this is the experiment that I am trying to do) :

There is an angle (theta w) needed to generate Rayleigh wave, and I used the same angle as Gokhale's.

LDV = Laser velocity detector????????

Laser Doppler Vibrometer

Which plane is in plane and which is OP?

In plane is the plane of the surface and out of plane is the plane perpendicular to the surface

How long a distance away from the transducer did you do the measurement?

I tried various of distances, but none of them agrees with the theory

Was the measurement region stationary or did you track (and plot polarisation) along the direction of wave travel?

The measurement is only at a single point

Did you excite with a pulse or with the expected frequency? Or what else?

I used function generator at 1MHz with 10 cycle excitation

What about excitation multiple other waves: compression and or shear waves? And measuring both?

There might be longitudinal waves or shear waves detected, but the Rayleigh wave can be easily distinguished in this experiment.

Thank you

I had better emphasize here that I know nothing about this subject, but, for what it's worth I just created this simple simulation in phun: a simple 2D physics simulator. Although it's essentially a toy, the physics is accurate (within the 2D constraints) so if you can set up the correct starting position it can give you a lot of insight into what's happening.

The orange wheel just rotates so that the "lump" keeps bashing the wedge. In this simulation the movement of the individual balls is almost entirely horizontal. The exact position of the wedge and consequent shear "fracture line" may be critical.

If you're interested I'll send the model file for the above: send me your EM in a private message.

Hi,

if you add flexible interconnections between any balls and its neighbours, and these flexibilities to have longitudinal and shear stiffness.

And the exciter driven from the vertical side and coupled by viscous or friction forces.

RHABE

if you add flexible interconnections between any balls and its neighbours, and these flexibilities to have longitudinal and shear stiffness.

Will springs do?

I might try it if I get time. Would the balls be better in a square or hexagonal matrix?

Then I'll get you to explain what

And the exciter driven from the vertical side and coupled by viscous or friction forces.

means

Hi,

balls or any other shape is not important as long as only the springs generate forces.

The lengthwise stiffness of the springs is simulating the elastic modulus, the crosswise stiffness the shear-modulus.

Elastic modulus = lengthwise-stress/lengthwise-strain

shear-modulus = shear-stress/(angle of shear deformation)

(stressing an elastic material longitudinally by 1% of the elastic modulus will yield 1% elastic longitudinal deformation,

stressing by 1% of shear-modulus vertical to length direction will result in 0.01rad of shear deformation, without bending deformation, so valid for short beams only .)

Exciter drive: the triangle made from Plexiglas has a horizontal part that is coupled by glue or viscosity to the steel, has a vertical part where the lengthwise oscillating exciter is firmly attached with prestress and a free hypothenuse that should be covered by an absorbing material in order to avoid reflections.

RHABE

For what it's worth I've got the "material" nearly behaving realistically (the balls near the surface go round generally anticlockwise and those further down go round clockwise, but neither motions are circular or elliptical).

I haven't messed with the exciter yet. Not sure if I'll get time (damned work).

Interesting!

Is the motion of the exciter small, so that the balls don't touch each other?

Is there any friction between the balls? If so set friction coefficient to 10^-4 or 0.

Try to excite pure horizontal movements.

Look at some distance of the exciter (to the right), usually these effects need 3 to 10 wavelengths to purify (any near field of an exciter or antenna is distorted).

Looks good, I will try the program for other applications!

RHABE