This is the ninth chapter of an eleven part article on Ultrasonics by John Drury, the Author of Ultrasonic Flaw Detection for Technicians. This article was first published in INSIGHT magazine throughout 2004/5. The chapters can be downloaded in PDF for you to build into a complete series.

To access the other chapters please use the navigation at the bottom of this page.

J.C. Drury ' BACK TO BASICS - ULTRASONICS'

Chapter9.
COMPRESSION WAVE TECHNIQUES

The important thing to remember when calibrating the timebase for compression waves is that the left hand end of the timebase (Zero) must exactly correspond to the entry surface of the beam and the right hand end represents a known thickness in the material being tested. The exceptions to this rule are those occasions when you are using delay to expand some distant portion of the material, or when you are using a multiple echo technique and only noting the decay pattern. For single crystal probes, the initial pulse contains two elements; the applied square wave voltage pulse and the ringing of the crystal. The end of the applied voltage pulse represents the top surface at which time the crystal ringing starts. Unfortunately, the amplitude of this part of the initial pulse is so large that it is not possible to identify the point at which the ringing starts, nor is it possible to tell from the timebase line.

There is a similar problem when we calibrate using a twin crystal probe because the initial pulse is at the start of the Perspex delay line and the sound enters the work piece sometime later. In any case, because the amplifier is deliberately isolated from the transmitter crystal, there is no signal to mark the entry surface. Our calibration procedure, whether for a single or twin crystal probe must find some other way to identify the entry surface or true zero. The most common way to achieve this is to use two echoes that are a known distance apart, to set one at timebase zero and the other at the right hand end (10) of the timebase. We do this in the following way: -

-      Use the delay control to position the first backwall echo from the desired range on our calibration block to zero.

-      Then use the depth range controls to position the second backwall echo to 10 on the timebase.

-          This may also move the echo from the zero position and so we need to check and adjust this with delay again.

-          These two controls are used alternately until the two echoes are exactly on 0 and 10. We now know that the timebase is exactly equal to the calibration thickness

-          Once we are happy that we have that exact range on the timebase, we lock the depth controls.

-          We then use delay to move the first backwall echo to the right until it is exactly on 10. If the timebase is exactly a known range and the first backwall echo is on 10; then zero must coincide with the entry surface.

Figures 1 to 3 illustrate the procedure for calibrating the timebase for 100mm of steel on the A2 calibration block.

Fig. 1

Fig. 2

Fig. 3

In this example we know that the first and second backwall echoes represent 100mm and 200mm of steel path time because the A2 block is 100mm wide at this probe position. Therefore figure 2 represents exactly 100mm timebase. This timebase remains constant as long as we do not alter the depth control, so figure 3 represents zero to 100mm exactly.

CHOICE OF COMPRESSION WAVE PROBES

TWIN CRYSTAL PROBES

For conventional techniques twin crystal probes are generally used on thicknesses below 50mm. They are also in general use for high temperature thickness measurement, where a thermal insulating material is used instead of Perspex, to protect the crystals. Because the transmitter section of the probe is isolated from the amplifier, there is no dead zone and so reflections from surfaces close to the probe can be identified. However since the Perspex shoe absorbs some sound, less damping is used and the pulses are longer so that resolution is generally poorer than with single probes.

SINGLE CRYSTAL PROBES.

Single crystal probes are generally used on thicknesses in excess of 50mm. They are also used below 50mm if resolution is an important factor, since single crystal probes usually have shorter pulse lengths than twin crystal probes. However, for conventional techniques they can only be used when the transmission noise (Dead Zone) does not encroach upon the useful part of the timebase for that job. As a guide, you can expect the shortest Dead Zone from high frequency, heavily damped probes.

PROBES FOR MULTIPLE ECHO TECHNIQUE

These are usually single crystal probes, although in some cases twin crystals can be used. When dealing with thin walled material it is possible to get resonance and anti-resonance conditions that will affect the decay pattern and may give false in­dications. This can be avoided if you choose a probe frequency such that the plate thickness is more than 1.5 x the wavelength of the compression wave in the specimen material, and a pulse length that is not more than 3 cycles .

THICKNESS MEASUREMENT

One of the most important uses of ultrasonics is that of thickness measurement. It is particularly useful because it is relatively quick, simple and accurate, and access to only one surface of the specimen is required. There are many types of equipment and techniques made ex­clusively for thickness measurement. It is not intended to deal with all of them here. We will only discuss the use of the pulse echo system with an A-scan display.

A-SCAN RECTIFIED DISPLAY

This is the most common display presentation for ultrasonic flaw detection equipment . In the sixth article we described the display for an unrectified trace and various types of rectification.

The basic calibration of the timebase should be carried out to ensure proper positioning of the zero and backwall echo as described above. The calibration block should be made of the same material as the work piece .

For best results the range chosen for calibration should be the shortest range which allows the first back wall echo to be displayed. For example, if the nominal wall thickness of the work piece is 9mm. and your flaw detector is capable of showing 10mm across the full graticule, then the 10mm range should be used. Since the graticule of most flaw detectors can be divided into 100 small units it follows that a timebase calibrated such that those 100 units represent 10mm gives you a reading accuracy of 0.1mm per division. If on the other hand you calibrate such that 100 units represent 25mm, the reading ac­ curacy is 0.25mm per division.

The amplitude of the calibration echo and the amplitudes of thickness echoes made on the work piece should be ad­ justed to the same predetermined amplitude. This is normally between 1/3 and 1/2 full screen heights.

The specimen thickness is determined from the left hand edge of the backwall echo. This is normally a steep sloping line. If a small half cycle appears at the left hand edge of the signal that was not present during calibration, this may be removed by inserting a small amount of suppression or by choosing ‘positive’ or ‘negative’ rectification. (See figure 4).

d. READING THE THICKNESS (MULTIPLE ECHOES)

If the specimen thickness and calibrated range are such that multiple echoes are produced, the most accurate result can be obtained by reading the thickness of the last multiple echo displayed and dividing the answer by the number of backwall echoes. In the example shown in figure 5, the fifth backwall echo shows at 22 mm. so the true thickness is 22 divided by 5 = 4.4 mm. In this case, a single crystal probe has been used and the initial pulse is obscuring the start of the first backwall echo.

Fig. 5

Sometimes the initial pulse obscures the entire first backwall echo and maybe all orpart of other back echoes. Figure 6 showsthe same thickness but with the first two back echoes obscured. Caremust be taken to assess the number of echoes that have been obscured.

Fig. 6

e. USE OF TIMEBASE DELAY

Apart from its use to correct for Perspex path distance in twin crystal compression wave probes, "Delay" can be used as an aid to more accurate thickness measurement. For example, you may want to accurately measure the thickness of a component whose nominal thickness is 80 mm. If you calibrate the timebase so that 100 scale divisions represents 100 mm of that material, each small division represents 1 mm. If instead, you calibrate the timebase so that 100 scale divisions represent 25 mm of the test material each division on the scale represents 0.25 mm instead of 1 mm. The delay control can then be adjusted so that the third backwall echo from the calibration block is set at O, and the fourth backwall echo at 100 scale divisions. The timebase would represent a thickness range of 75mm to 100mm. The first back echo from the work piece (80mm) will appear at approximately 1/5th of the timebase range.

A-SCAN UNRECTIFIED DISPLAY.

There are occasions in thickness measurement, particularly if the scanning surface is rough, when a lot of unwanted signals, "noise " or "grass" appear on the CRT and make it difficult to determine the point at which the back echo starts. If the ultrasonic set allows an unrectified trace to be selected, then measurements can be made using the tip of a particular down going half cycle instead of using the point at which the signal first leaves the timebase.

a. Firstly, let us identify our measuring point. Figure 7 shows a back echo from the 25mm range on the V1 block with the timebase calibrated for 50mm using the conventional rectified display. The presentation has then been changed to ‘unrectified’ and the vertical or 'Y' shift used to raise the timebase to a level between 1/3 and 175th full screen height. Gain has been adjusted so that the peak of the longest down going half cycle just meets the graticule. In this case, it is the second half cycle that is the longest, and we will use the second half cycle as our measuring point. (Note that sometimes a back echo from the work piece may show the 1st or 3rd half cycle as the longest - despite this, if you calibrate on the second half cycle you then always measure from the second half cycle even if it is not still the longest,)

 

b. Having identified the half cycle that you are going to use, you calibrate the timebase so that this part of the signal coincides with the correct point of the graticule. In the case shown in figure 7, if we wished to calibrate for 50 mm we would use "delay" to move the second half cycle from 5.15 to 5.0 divisions and check that the second half cycle of the second back echo coincides with 10.0 divisions (see figure 8).

 

Fig 8

c. Calibration for other timebase ranges would be done in the conventional way but using the second half cycle instead of the left hand edge of the pulse, as your measuring point.

LAMINATION TESTING

Lamination testing of plates and pipes that are to be welded or machined is a very common NDT task. It is also a simple application of compression waves in ultrasonic flaw detection, but one that might give some problems when examining thinner samples.

STANDARD PROCEDURE

a) Calibrate the timebase to allow at least two backwall echoes to be displayed.

b) Place probe on the work piece and adjust the gain controls so that the second backwall echo is at full screen height.

c) Scan the work piece looking for lamination indications that will generally show up near half specimen thickness together with a reduction in back echo amplitude. In some cases, a reduction in the amplitude of the second back echo may be noticed without a lamination signal being present. Care must be taken to ensure that this reduction in amplitude is not due to poor couplant or surface conditions.

MULTIPLE ECHO TECHNIQUE

Lamination testing of plate or pipe less than 10 mm. in wall thickness may be difficult using the standard procedure because multiple echoes are so close together that it becomes im­possible to pick out lamination signals between backwall echoes. In such cases, we can use a technique called the "multiple echo technique" using a single crystal compression wave probe. The procedure is as follows: -

a) Place the probe on a lamination free portion of the work or calibration piece.

b) Adjust the timebase and gain controls to obtain a considerable number of multiple echoes in a decay pattern over the first half of the time base. A typical example is shown in figure 9.

Unlaminated plate
Fig 9

c) Scan the work piece, the presence of laminations will be indicated by a collapse of the decay pattern such as the one shown in figure 10. The collapse oc­curs because each of the many multiple echoes is closer to its neighbour in the presence of a lamination.

EXAMINATION OF BRAZED AND BONDED JOINTS

Com­ pression waves can also be used for the detection of areas of non-adhesion in brazed or bonded (glued) joints,

If the wall thicknesses permit clear separation bet­ ween back wall echoes, brazed joints can be examined using the standard procedure for lamination checking. However, since the braze metal separating the two brazed walls will have a slightly different acoustic impedance to that of the parent metal, a small interface echo may be present for a good braze. The technique, therefore, is to look for an increase in this interface echo amplitude. F igure 11 shows the type of indication when the unbrazed portion is smaller than the beam diameter. If the two brazed walls are too thin to permit clear back echoes, a multiple echo technique can be used as described above.

Fig. 11

b. BONDED JOINTS

These may include metal-to-metal glued joints and metal to non-metal glued joints (e.g. rubber blocks bonded to steel plates). The technique used is a multiple echo technique. Each time the pulse reaches a bon­ ded interface; a portion of the energy will be transmitted into the bonded layer and absorbed. Each time a pulse reaches an unbonded layer, all the energy will be reflected. If we look at the multiple echo pattern for a good bond, the decay will be relatively short because of the energy loss at each multiple echo into the bond. However, for an unbonded layer each multiple echo will be slightly bigger because there is no interface loss, and the decay pattern will be significantly longer. Figure 12 shows a good bond (probe 1) and poor bond (probe 2).

Fig. 12

Reference: -

‘Ultrasonic Flaw Detection for Technicians’ - Third Edition, June 2004 by J. C. Drury