where expertise comes together - since 1996

Web's Largest Portal of Nondestructive Testing (NDT)
Open Access Database (Conference Proceedings, Articles, News), Exhibition, Forum, Network

All Forum Boards
Technical Discussions >
wavelength/resolution
Career Discussions
Job Offers
Job Seeks
Classified Ads
About NDT.net
Articles & News

5063 views
08:34 Oct-20-2001

Karen Ellis

Consultant
Self employed,
United Kingdom,
Joined Oct 2001
3
wavelength/resolution

Could anyone explain from basics the link between ultrasonic wavelength and resolution of an inspection.

Regards

Karen


 
08:59 Oct-22-2001

Rolf Diederichs

Director, Editor, Publisher, Internet, PHP MySQL
NDT.net,
Germany,
Joined Nov 1998
602
Re: wavelength/resolution It is quite difficult to give an answer to your interesting question.
Nevertheless I hope that some experts will try it.
I think that it is to simple to say "one wavelengh is the minimum resolution"

A related subject you can study in the discussion of our former transducer workshop in Sept '96.
"Smallest detectable flaw size, especially for low frequency test (f= 0.5 MHz) and in general."
http://www.ndt.net/wshop/wshop_tr/messages/72.htm

Best regards
Rolf Diederichs

-----
: Could anyone explain from basics the link between ultrasonic wavelength and resolution of an inspection.
.
: Regards
.
: Karen
.



 
01:36 Oct-23-2001

A. Asundi

R & D, -
Nanyang Technological University,
Singapore,
Joined Feb 2001
10
Re: wavelength/resolution Does not Rayleigh's criterion of resolution apply here.
Resolution = 1.22 lambda * L/D

Anand

: : Could anyone explain from basics the link between ultrasonic wavelength and resolution of an inspection.
: .
: : Regards
: .
: : Karen
: .
.



 
01:09 Oct-23-2001

Manfred Johannes

R & D
Technology Service International (TSI),
South Africa,
Joined Mar 2000
2
Re: wavelength/resolution : Does not Rayleigh's criterion of resolution apply here.
: Resolution = 1.22 lambda * L/D
.
: Anand
.
: : : Could anyone explain from basics the link between ultrasonic wavelength and resolution of an inspection.
: : .


: : : Regards
: : .
: : : Karen
: : .
: .
.
Hi Karen,

The way that I explained these issues to students and technicians is as follows:

1. Sensitivity is the ability to find small indications.
2. Resolution is the ability to distinguish between two reflectors, which are close to one another in a body.
3. Sensitivity is a function of wavelength, in that reflectors which are smaller than half a wavelength (lamda/2) will not necessarily reflect energy back to the source (receiver). They may reflect energy back, but this is not a certainty. Hence it is postulated that the smallest reflector one can pick up with a certain wavelength is lamda/2. (All UT probes have a bandwidth of frequencies, which they emit - typical 1 to 4 MHz for a 2 MHz probe. This info can be found on the probe data sheet.)
4. From the above one could conclude that where two reflectors are closer than half a wave length (lamda/2) from one another, one should not be able to distinguish between them, and visa versa. In reality (experience) the distance between two reflectors, to be distinguished as separate entities, is usually larger than lamda/2, if one uses the central frequency of the probe as reference. This is due to the diffraction patterns (Airy discs) which are associated with waves and slits/edges in general. In optics (microscopy) it is assumed that one can still distinguish two reflectors close to one another if the sum of the intensities (amplitudes) of the airy discs is 20% less than the maximum amplitude of one of the discs.

I hope this helps you along.




 
04:00 Oct-23-2001

Carlos Valdecantos

Engineering
MTORRES, Ing.,
Spain,
Joined May 2001
9
Re: wavelength/resolution : : Does not Rayleigh's criterion of resolution apply here.
: : Resolution = 1.22 lambda * L/D
: .
: : Anand
: .
: : : : Could anyone explain from basics the link between ultrasonic wavelength and resolution of an inspection.
: : : .
.
.
:
: : : : Regards
: : : .
: : : : Karen
: : : .
: : .
: .
: Hi Karen,
.
: The way that I explained these issues to students and technicians is as follows:
.
: 1. Sensitivity is the ability to find small indications.
: 2. Resolution is the ability to distinguish between two reflectors, which are close to one another in a body.
: 3. Sensitivity is a function of wavelength, in that reflectors which are smaller than half a wavelength (lamda/2) will not necessarily reflect energy back to the source (receiver). They may reflect energy back, but this is not a certainty. Hence it is postulated that the smallest reflector one can pick up with a certain wavelength is lamda/2. (All UT probes have a bandwidth of frequencies, which they emit - typical 1 to 4 MHz for a 2 MHz probe. This info can be found on the probe data sheet.)
: 4. From the above one could conclude that where two reflectors are closer than half a wave length (lamda/2) from one another, one should not be able to distinguish between them, and visa versa. In reality (experience) the distance between two reflectors, to be distinguished as separate entities, is usually larger than lamda/2, if one uses the central frequency of the probe as reference. This is due to the diffraction patterns (Airy discs) which are associated with waves and slits/edges in general. In optics (microscopy) it is assumed that one can still distinguish two reflectors close to one another if the sum of the intensities (amplitudes) of the airy discs is 20% less than the maximum amplitude of one of the discs.
.
: I hope this helps you along.
.

Some practical points of view to help to resolution-sensitivity discussion:

To talk about sensitivity it is interesting to classify the potential reflectors as:

Big reflector: larger than the beam diameter, at least in one direction.
Small reflector: smaller than the beam diameter but well above the wave length.
Very small reflector: with size of the same order as wave length and smaller.

- From big to very small, reflection goes from purely specular (mirror), clearly directional and high energy (amplitude), to pure scatter with low amplitude, diffuse (any direction) response.
- Unfortunately, there are not well defined limits defining these three ranges in terms of wave length.
- It is basically true that a given reflector may behave as scatter for a given frequency and as a mirror for higher frequencies. But, at the same time, other small reflectors in the material may produce its own signals that may bury or hidden the signal in the A-scan. This is because it is also useful to take into account signal-to-noise ratio (SNR) to define whether a reflector is or not detectable. As a general rule, the attempt to increase sensitivity through testing at higher frequency (beyond some practical limit) may lead to worsen SNR.
- A compromise is always to be found between increasing frequency to improve sensitivity and the loss of SNR because of scattering.

From the resolution side it may be useful to make differences between:

- Axial resolution: measured in the direction of beam propagation. It is the parameter people refer to when they say resolution. Ability to detect reflectors at distances slightly different from the transducer. Axial resolution is more directly connected to the pulse length, rather than to the wave length.
- Lateral resolution: measured 90º to the previous one. The leading parameter in this case is the beam diameter.

Pulse length and beam diameter are both connected to the wave length but perhaps the transducer diameter, piezoelectric material, backing, front and coupling layers, focus and other details of the probe design/construction are of primary importance.

Regards,

C. Valdecantos





 
06:15 Jun-03-2014

edwardnxs4

Oil&Gas Industry,
Indonesia,
Joined Dec 2013
20
Re: wavelength/resolution In Reply to Rolf Diederichs at 08:59 Oct-22-2001 .

So If I use 5 Mhz transducer for pipe stainless steel measurement,
my resolution is 1.16 mm ???
I think is so bad resolution,,

cosasco product claim their resolution 0.0025mm , are they using Gigahertz frequency of transducer ?

Please explain to me, thank you

-------------
I think that it is to simple to say "one wavelengh is the minimum resolution"

 
11:23 Jun-03-2014

Frank Lund

R & D,
United Kingdom,
Joined Apr 2005
219
Re: wavelength/resolution In Reply to edwardnxs4 at 06:15 Jun-03-2014 .

Edwardnsx4,

It looks as if you've strayed into the ignoble field of Specmanship.

Resolution of 0.1 mil (0.0001 inch) is just an expression of the number of digits to which the answer is displayed AFTER PROCESSING.

For a velocity of sound ~6 x 10^3 metres / second and a claimed resolution of 2.5 x 10^-6 metres

2.5 x 10^-6
-------------- = appromately 0.4 x 10^-9
6 x 10^3

gives a time resolution of 0.4 NANOSECONDS

Allow for timing the to and from transits, you still need a clock rate of 1.25GHz

There are fast DSP devices out there, but this looks like a thickness monitoring device that is put in place then just watches one spot so that it can take the average of very many readings to get the number of displayed digits.

Cheers,
Frank

 
04:08 Jun-04-2014

edwardnxs4

Oil&Gas Industry,
Indonesia,
Joined Dec 2013
20
Re: wavelength/resolution In Reply to Frank Lund at 11:23 Jun-03-2014 .

So the key is in Receiver ?

and for your calculation, 0.4 nanosecond resolution , only can reach with 2.5Ghz transducer..

so if you mean , that resolution is result from Digital Signal Processing, isn't that a pseudo resolution ? or fake resolution ?

I mean, if I using 5Mhz transducer for stainless steel pipe inspection, the true resolution is 1.16mm but with a Digital Signal Processing I able to get 0.0025mm resolution , but how precision of the result ?

regards,
Edward

 
12:31 Jun-04-2014

Frank Lund

R & D,
United Kingdom,
Joined Apr 2005
219
Re: wavelength/resolution In Reply to edwardnxs4 at 04:08 Jun-04-2014 .

Edward,

I had a quick look at the Cosasco website that said 0.0025mm (0.1mil) resolution.

The instrument appears to be static thckness measurement device that sends a "ping" then times the return signal. So the repetion rate of that "ping" does not need to be high. It will wait for the echo from one "ping" before it sends the next.

The resolution of the measurement of the transit time is what would give the resolution of the distance. As the echo time is taken from the time taken for the "ping" to travel to the target and then return, a variation of 0.0025mm would make a difference of 0.005mm in the Tx-Rx path length or ~0.8nS in the time.

The high frequency required is not in the Ultrasonic signal but in the "ticks" of the clock that measures the time of the echo.

With the luxury of a permanently placed sensor, many "pings" and echoes can be timed and the average taken. Natural noise (even a little added noise) would make the time measurement dither between at least two time readings, so averaging of the time can use the proportions of the time at T = x and at T = x+1 to give an output which lies at some point between x and x+1. This makes the resolution appear greater than the resution of the actual timer involved.

There is also the matter of temperature compensation, I haven't done the calculation, but I supect that applying temperature corrections will make very small differences. I.E the "resolution" in thickness when transferred from the actual resolution of the temperature measurement will give very small changes distances shown on the readout.

Cheers,
Frank

 
13:32 Jun-04-2014

Ed Ginzel

R & D, -
Materials Research Institute,
Canada,
Joined Nov 1998
1208
Re: wavelength/resolution In Reply to edwardnxs4 at 04:08 Jun-04-2014 .

Edward, Frank's explanation is a good one. You seem to be assuming that the transducer pulse is the resolving limit. In fact it is the clock and the instrument's ability to identify the same point on a waveform. This makes the digitising process the resolving limit for the thickness measurement. Although the digitising clock rate will determine the smallest interval measured the tolerance noted by Frank (i.e. plus or minus one time-interval sample) is a good practical limit to account for noise and jitter.

 
01:41 Jun-06-2014

Wieslaw Bicz

Engineering,
PBP Optel sp. z o.o.,
Poland,
Joined Feb 2009
239
Re: wavelength/resolution In Reply to A. Asundi at 01:36 Oct-23-2001 .

"Does not Rayleigh's criterion of resolution apply here.
Resolution = 1.22 lambda * L/D"

Abbe criterion would be better suitable here, since its applicable to objects, that are emitting the wave.

But bot criteria are not necessary correct in this case, since ultrasonic waves are first: coherent (with unlimited coherence time) and it is possible not only to measure the phase, but also the time of flight with much higher accuracy as the wave length. Additionally the numerical aperture is not easy to define.

Simple example could be the consideration, what happens with signal reflected from two close points having the distance of 1/4 wavelength, situated almost on the same line as the sender (and receiver). The reflection of the distant point will come after 1/2 wavelength and it is naturally very easy to distinguish. Additional example: small object composed from two balls, much smaller than the wavelength will create signal with different time of flight if both balls are placed on the same line as sender/receiver and off line.

The resolution criterion for sound imaging must be different than classical Abbe criterion.

Sampling should be at least as frequent, as required by Nyquist criterion, but if the noise is not significant, it must be not much better.

 
01:49 Jun-06-2014

Wieslaw Bicz

Engineering,
PBP Optel sp. z o.o.,
Poland,
Joined Feb 2009
239
Re: wavelength/resolution In Reply to Ed Ginzel at 13:32 Jun-04-2014 .

Ed.

It is not the sampling clock, that sets the limit of time resolution, but the amplitude resolution and eventual its stability. In the case of measurement between two points in the same signal the noise can be also critical. In the case of the measurement of absolute time of flight the synchronization between the pulser and the sampling clock.

Quite simple calculation is showing, that 8bit ADC and few MHz signal can allow to achieve resolution of time of flight in the region of few ps. Small noise level and averaging can easily increase this accuracy. If you like, I can send you the description of such calculation.

The people, that are building ultrasonic flow meters are using this fact: they are mostly working with low frequencies and quite simple electronics and achieve very good measurement of time of flight

 
06:05 Jun-12-2014

edwardnxs4

Oil&Gas Industry,
Indonesia,
Joined Dec 2013
20
Re: wavelength/resolution In Reply to Wieslaw Bicz at 01:49 Jun-06-2014 .

"8bit ADC and few MHz signal can allow to achieve resolution of time of flight in the region of few ps"

Why you focus on the Bit of ADC (vertical resolution), isn't the time of flight on Horizontal resolution (sample rate)?

 
17:06 Jun-12-2014

Frank Lund

R & D,
United Kingdom,
Joined Apr 2005
219
Re: wavelength/resolution In Reply to Wieslaw Bicz at 01:41 Jun-06-2014 .

"
Simple example could be the consideration, what happens with signal reflected from two close points having the distance of 1/4 wavelength, situated almost on the same line as the sender (and receiver). The reflection of the distant point will come after 1/2 wavelength and it is naturally very easy to distinguish. Additional example: small object composed from two balls, much smaller than the wavelength will create signal with different time of flight if both balls are placed on the same line as sender/receiver and off line.
"

Wieslaw,

Are you describing a technique based upon interferometry in which a sample of the Tx wave is mixed with the Rx wave so that the beats can be counted as the transit distance and therefore transit time change?

I could see how the beats could be counted as the far wall is slowly eroded in an installation with a thickness monitor fixed in place, but this is a differential measurement.

I would very much like to have details of the system using a few MHz and the 8 bit ADC.

Thanks and Regards,

Frank Lund

 
19:38 Jun-12-2014

Wieslaw Bicz

Engineering,
PBP Optel sp. z o.o.,
Poland,
Joined Feb 2009
239
Re: wavelength/resolution In Reply to Frank Lund at 17:06 Jun-12-2014 .

Download

My examples was valid for a simple system, using pulse-echo technique. It is also easy to demonstrate with a simple rectangular or oval rod with small diameter. But the fact, that I am describing is also known in the case of measurement of layer thickness - it is possible to measure layers with the thickness smaller than wavelength (even significantly).

In the case of time of flight measurement of signals, that are not changing, but only shifting in time, the accuracy of measurement is depending only on the stability of sampling clock (in the case, if the distance between two signals in one shot is measured) and synchronization of pulser with sampling clock (in the case of absolute time of flight).

The accuracy will depend on the accuracy of ADC and is the function of frequency and very easy to calculate. I have attached the paper, describing this exactly.
 
13:40 Jun-13-2014

Frank Lund

R & D,
United Kingdom,
Joined Apr 2005
219
Re: wavelength/resolution In Reply to Wieslaw Bicz at 19:38 Jun-12-2014 .

Wiewslaw,

Thanks for that. Now I see what I had guesed, that you use an ADC to sample some points at fixed times then do an interpolation calculation to find the point at which the signal would have passed through the defined threshold value in the slope between amplitude Vn at time = Tn and Vn+1 at time = Tn+1. This gives you the time Tt at which the signal was at Vt, where Vt = Threshold level and Tt = time at which V passes through Vt.

Cheers,
Frank

 
02:47 Jun-16-2014

Hi Frank,

Consultant, owner of business
NTS Ultrasonics Pty Ltd,
Australia,
Joined Oct 2012
108
Re: wavelength/resolution In Reply to Frank Lund at 13:40 Jun-13-2014 .

From what I have read, the accuracy of interpolation depends on what mathematical function you use to estimate the missing data points, sinusoidal, Gaussian, etc. There are papers that you can find online that discuss systematic errors in interpolation, although all that mathematics is a bit too much for me.

Getting back to Edward's enquiry, I think he has been reading up on advanced corrosion monitoring systems like Ultracorr from Rohrback Cosasco Syetems, and Ultramonit and Pipemonit from Sensolink. These systems claim very high accuracy for measurements, 0.0001" in the case of Ultracorr. They can be used for determining the rate of corrosion if the sensors are installed for periods and ultrasonic signals monitored regularly. There is not much documentation about the workings of Ultracorr available online, but Ultramonit is discussed in a number of papers that can be found online. For example "Wall thickness measurement of new and existing subsea pipelines using ultrasound", from the NACE Corrosion 2007 conference is a discussion on the workings of Ultramonit. This system apparently uses a 50 MHz A/D and then interpolation to obtain higher resolution. They then plot the change in time of particular echoes from earlier thickness measurements with those taken at later times and work out the rate of change of position, which then becomes the rate of corrosion. It is pretty clever work.

The point for Edward is, it may not necessarily be the initial resolution (i.e. hardware) of your measurement that is important, but how good your readings are (stability, noise, etc, as Ed and Wieslaw point out), and how you process the data you have collected (interpolation, dither (for multishot measurements), etc).

If Edward wants good resolution based on hardware he needs to look into design topics like equivalent time sampling, dither, digital Vernier techniques, dual slope integration and the like and choose the one he likes.

Regards
John Norman

 


© NDT.net - The Web's Largest Portal of Nondestructive Testing (NDT) ISSN 1435-4934

Open Access Database, |Conference Proceedings| |Articles| |News| |Exhibition| |Forum| |Professional Network|