- Email: [email protected]
Near surface defects detectable by ultrasonics
Near surface defects detectable by ultrasonics E. E. Aldridge
It is a mistake to believe that sub-surface flaws cannot be detected using pulse echo due to the presence of a dead zone associated with the surface or interface echo. The dead zone is usually caused by the electronic instrumentation either by the envelope detector or by the overload characteristic of the amplifying chain. By using an accurately controlled acceptance gate situated in the tail of the interface echo, where interference of the flaw echo is most pronounced, this dead zone is avoided and the sub-surface flaws revealed. It requires a narrow or focused ultrasonic beam at the surface and it suffers from the limitations that best results are obtained with smooth surfaces and that the response of flaws tends to be preferentially weighted as they lie further from the surface. However, within these limitations it is an extremely useful technique especially if combined with a suitable scan as is shown by the illustrative results.
There are many people who are interested in detecting subsurface flaws ie flaws which lie close to the surface of the material under test, and who believe that there is a region extending from the surface called a dead zone, in which such flaws cannot be detected using pulse echo techniques. The thickness of the dead zone is commonly related to the length of the ultrasonic pulse and, the shorter the pulse, the smaller the dead zone. Although this belief is mistaken, the electronic instrumentation used to process and display the echo pulses often has limitations which prevent the error being discovered, This is illustrated in Fig. 1. In this figure it is assumed that there is a delay path, eg a water column, between the transducer and the material surface such that the interface echo is clearly distinguished. For clarity only the first few cycles of the interface echo are sketched in and the rest of the echo is indicated by the dashed outline representing its envelope. Overlapping this echo, and delayed slightly, there is a smaller flaw echo of similar shape. As sketched, these echoes reach their peak in about three cycles and decay back to zero again in about seven cycles and so represent a fairly long isonifying pulse in terms of ultrasound wavelengths. Tile most common way of displaying an echo is to show not the carrier pulse itself, but its envelope. This is usually obtained using a peak detector, 1 and the type of output obtained is shown superimposed upon the echo pulses, ie whilst the detector output is following the rising edge of E. E. Aldridge is Head of the Ultrasonics Section of the Nondestructive Testing Centre, Atomic Energy Research Establishment, Harwell, Didcot, Berks, UK.
NON-DESTRUCTIVE TESTING . OCTOBER 1971
Interface
/
echo
/
~ •
\
~ -~.._. Detectoroutput .
/._j
---4"
\
,4 \
\
/
\
/ \
/
Fig.1 Interface echo with overlapping flaw echo.
the interface pulse quite well, it is not following the falling edge. This effect stems from the smoothing action of the detector circuitry, which in practice appears to be a design compromise optimised roughly for the average 4 MHz transducer, ie for this type of transducer the detector output will follow both the rising and falling edges of the pulse fairly well, but at lower frequencies the smoothing will degenerate, and at the higher ones the effect will be as shown. A property of this type of detector is that the output does not respond to any incoming signal unless it exceeds this output, and as shown, this means the detector output will not respond to the presence of the flaw echo unless this echo is sufficiently delayed. This delay corresponds to a
309
dead zone behind the interface echo and its thickness is a function, to some extent, of the amplitude of the flaw echo. Usually the flaw echo is so small that a high gain is required and so the signal amplifying chain is grossly overlQaded by the interface echo. Unfortunately, unless the amplifying chain is appropriately designed, the overload will cause it to behave rather like a peak detector and so contribute a dead zone of itself. A common technique for reducing the dead zone is to use twin probes as separate send and receive, oriented at a small angle on either side of the surface normal and arranged to have their beam cross-over just below the surface. The interface echo is then not picked up by the receiver. Another technique using a single probe in pulse echo is described in the paragraphs following.
I
Reducing the effect of the dead zone in pulse echo From the sketch in Fig. 1, it is clear that the flaw echo will perturb the interface echo and that this perturbation is largest towards the tail of the latter. For pulses which appear to cease abruptly, as shown in this figure, there is a region behind the interface echo where the flaw echo can be seen by itself using a suitably placed acceptance gate (Gate 1). Usually, however, pulses do not cease abruptly and so the flaw echo can only be seen by its interference effect on the interface echo. Thus the flaw can be found by placing an acceptance gate (Gate 2) where this interference is greatest.
b
¢
Fig.2 (a) Undistorted interface echo and acceptance gate pulse. (b) & (c) (echo gain increased by 30 dB) Output of acceptance gate for two positions of transducer.
310
In practice this simple approach is not enough. The shape of the ultrasonic pulse is rarely gufficiently well defined: there are often small low level echoes generated from the structure of the probe which confuse the picture at the high gains which are usually necessary to find the flaw; also surface conditions can cause perturbations. Hence in order to find a flaw it is necessary to compare the echo shape from one probe position with that from another position. This is illustrated by the wave forms shown in Fig.2. The undistorted interface echo and the acceptance gate pulse is shown in Fig.2 a. The frequency is 11 MHz, the gate width is 1.8/~s and it is delayed on the interface by 0.7/~s. Figs.2 b and c show the result obtained when the gain is increased by 30 dB and for two positions of the ultrasonic transducer as it sweeps over a flaw. In these figures the top trace is the interface echo and the bottom trace those positive half cycles of that part of this echo which is passing through the acceptance gate. As can be seen, there is a very marked change from one position to another. Since this effect can also be produced by the interface pulse moving with respect to the gate, precision gating is required, ie the position of the gate has to be slaved to that of the interface echo. Generally speaking the best results are obtained using this technique in combination with a C-scan. 2 The narrower the ultrasonic beam at the surface, the more sensitive the technique, since a beam which is wholly on a flaw in one position and wholly off in the next will produce a much greater perturbation than a beam which is broad enough to be partially on the flaw for a number of positions. A comparison of this technique with normal pulse echo is of interest in that there the flaw is found by using a narrow pulse whereas here it is found by using a narrow beam, ie there has been an interchange between time and space to achieve the same objective. NON-DESTRUCTIVE TESTING . OCTOBER 1971
to produce a voltage level proportional to the output of the rectifying gate. ~-L;~ , ' .... The ti-ansducer is scanned over an aperture, in a similar manner to a television raster, and the recorder is synchronised to the scan such that the accepted part of the echo is recorded at a point corresponding to the position in the scan at which it was received.
Results Transducer
V
=
~lntegrator
Fig.3 Diagram of equipment.
These are shown in Figs.4 a and b. The object consists of a thin steel sheet, 0.030 in thick, to which is vacuum brazed a circular boss 0.100 in thick and 6 in. in diameter. The brazing material was contained in four concentric shallow channels, about 0.020 in deep, cut in the boss and these channels tie adjacent to the sheet. The view is taken through the sheet. Fig.4a shows the results oStained using an 11 MHz focused transducer with a resolution of about 1 mm in steel. Typical wave forms are shown in Fig.2. The outline of the boss is clearly seen, as are the four channels (marked by
Limitations As mentioned previously, surface conditions can cause perturbations which obscure those due to flaws. Where the pattern of surface variation is not very marked and differs in shape from the flaws the latter can be usually clearly seen depending upon the degree of ultrasonic scatter. On the other hand where the surface variations are very marked and localised, the flaws are usually obscured entirely. Not enough experience has been accumulated to lay down any definite ruling for this condition and at the moment it is largely a matter of cut and try. From Fig. 1 it is clear that the perturbation on the tail of the interface echo becomes greater the further that the flaw is from the surface, up to the point where the flaw echo ceases to overlap the interface echo. This means that the response is weighted against the flaw nearest the surface and, if two flaws are in line, then the one at the greater distance from the surface will tend to mask the other if the degree of scatter is small. Thus there is little depth discrimination.
I
2
3
4
A
I l l u s t r a t i v e results Equipment
The schematic of the equipment used is shown in Fig.3. Ttie transducer is pulsed by discharging a condenser into it, this being initiated by the output of the pulse generator. The echo pulse after being amplified, is then passed through a gate which is opened in synchronism with the initiating pulse. The time delay and width of this gate are chosen such that it will completely discriminate against the transmission pulse and will always accept all the interface echo however its position may vary. The output of this gate is taken to a discriminator which generates a trigger pulse which is phase locked to the interface echo. This trigger pulse then operates the acceptance gate which in this case is a rectifying one which only accepts positive pulses. To ensure that this gate can be positioned appropriately, the output of the first gate is passed through a delay line to this gate. The output o.f the rectifying gate is subsequently passed to an integrator and a sample and hold circuit
NON-DESTRUCTIVE TESTING . OCTOBER 1971
b
Fig.4 Vacuum braze joint between steel sheet 0.03 in thick and circular boss 0.1 in thick. View through the steel sheet: (a) 11 MHz resolution 1 mm approx. (b) 5 MHz resolution 5 mm approx. Arrows 1, 2, 3 and 4 indicate channels in the boss which originally held the brazing material. Arrow A is a central hole. Non-bonds are shown by dark areas on the boss.
311
arrows 1,2, 3 and 4). Within the area o f the boss the nonbond areas are shown by the dark patches and it is clear that on the outer periphery there are two large areas of non-bond and extensive areas of non-bond between channels 2 & 3, and 3 & 4. The bright area at A is a hole through the middle. For comparison Fig.4 h shows the results obtained for the same object using a 5 MHz transducer with a beam width of about 5 mm in steel. Although the same features are apparent the reduction in resolution and clarity is very marked. The pulse length in terms o f wavelengths of ultrasound was about the same but the wider beam width made the positioning and width of the acceptance gate more critical than in the other case. In both these cases the adjustment procedure was to let the scan run, observe the tail o f the interface pulse, place the acceptance gate where there was most variation with the scan and adjust its width so it only accepted this part of the pulse, and finally adjust the gain to give maximum output.
Conclusion Sub-surface flaws can be found using pulse echo even though they lie within the minimum range resolution of the ultrasonic pulse. It requires accurately controlled acceptance gating of the tail of the surface echo and a narrow or focused ultrasonic beam at the surface. It suffers from the limitations that best results are obtained with smooth surfaces and the response o f flaws tends to be preferentially weighted as they lie further from the surface. However within these limitations it is an extremely useful technique especially if suitable scanning equipment is available.
References 1 PARKER, P. Electronics. Arnold. 1950. p 499. et seq. 2 Nondestructive Testing Handbook. ed: McMaster, C. vo111. Ronald Press 1959, pp 34-43.
non_
destructive testing Reprints Reprints of all articles published in non_destructive testing are available, in quantities of 100 or more R e p r i n t s are e s s e n t i a l - • • • • •
for the company that wants to distribute impartial comment on its activities to potential customers and clients for the company that wants to up-date its technical staff on new techniques and new technologies for the company that wants to publicize its research and development work for the training course organizer who wants to assemble key reading material for his students for the university or technical college lecturer who wants to distribute the latest information on a topic under study
For full details of prices and availability of reprints, please write to The Reprint Department, non_destructive testing IPC House, "32 High S t r e e t , G u i l d f o r d , Surrey, England
312
NON-DESTRUCTIVE
TESTING
. OCTOBER
1971
It is a mistake to believe that sub-surface flaws cannot be detected using pulse echo due to the presence of a dead zone associated with the surface or interface echo. The dead zone is usually caused by the electronic instrumentation either by the envelope detector or by the overload characteristic of the amplifying chain. By using an accurately controlled acceptance gate situated in the tail of the interface echo, where interference of the flaw echo is most pronounced, this dead zone is avoided and the sub-surface flaws revealed. It requires a narrow or focused ultrasonic beam at the surface and it suffers from the limitations that best results are obtained with smooth surfaces and that the response of flaws tends to be preferentially weighted as they lie further from the surface. However, within these limitations it is an extremely useful technique especially if combined with a suitable scan as is shown by the illustrative results.
There are many people who are interested in detecting subsurface flaws ie flaws which lie close to the surface of the material under test, and who believe that there is a region extending from the surface called a dead zone, in which such flaws cannot be detected using pulse echo techniques. The thickness of the dead zone is commonly related to the length of the ultrasonic pulse and, the shorter the pulse, the smaller the dead zone. Although this belief is mistaken, the electronic instrumentation used to process and display the echo pulses often has limitations which prevent the error being discovered, This is illustrated in Fig. 1. In this figure it is assumed that there is a delay path, eg a water column, between the transducer and the material surface such that the interface echo is clearly distinguished. For clarity only the first few cycles of the interface echo are sketched in and the rest of the echo is indicated by the dashed outline representing its envelope. Overlapping this echo, and delayed slightly, there is a smaller flaw echo of similar shape. As sketched, these echoes reach their peak in about three cycles and decay back to zero again in about seven cycles and so represent a fairly long isonifying pulse in terms of ultrasound wavelengths. Tile most common way of displaying an echo is to show not the carrier pulse itself, but its envelope. This is usually obtained using a peak detector, 1 and the type of output obtained is shown superimposed upon the echo pulses, ie whilst the detector output is following the rising edge of E. E. Aldridge is Head of the Ultrasonics Section of the Nondestructive Testing Centre, Atomic Energy Research Establishment, Harwell, Didcot, Berks, UK.
NON-DESTRUCTIVE TESTING . OCTOBER 1971
Interface
/
echo
/
~ •
\
~ -~.._. Detectoroutput .
/._j
---4"
\
,4 \
\
/
\
/ \
/
Fig.1 Interface echo with overlapping flaw echo.
the interface pulse quite well, it is not following the falling edge. This effect stems from the smoothing action of the detector circuitry, which in practice appears to be a design compromise optimised roughly for the average 4 MHz transducer, ie for this type of transducer the detector output will follow both the rising and falling edges of the pulse fairly well, but at lower frequencies the smoothing will degenerate, and at the higher ones the effect will be as shown. A property of this type of detector is that the output does not respond to any incoming signal unless it exceeds this output, and as shown, this means the detector output will not respond to the presence of the flaw echo unless this echo is sufficiently delayed. This delay corresponds to a
309
dead zone behind the interface echo and its thickness is a function, to some extent, of the amplitude of the flaw echo. Usually the flaw echo is so small that a high gain is required and so the signal amplifying chain is grossly overlQaded by the interface echo. Unfortunately, unless the amplifying chain is appropriately designed, the overload will cause it to behave rather like a peak detector and so contribute a dead zone of itself. A common technique for reducing the dead zone is to use twin probes as separate send and receive, oriented at a small angle on either side of the surface normal and arranged to have their beam cross-over just below the surface. The interface echo is then not picked up by the receiver. Another technique using a single probe in pulse echo is described in the paragraphs following.
I
Reducing the effect of the dead zone in pulse echo From the sketch in Fig. 1, it is clear that the flaw echo will perturb the interface echo and that this perturbation is largest towards the tail of the latter. For pulses which appear to cease abruptly, as shown in this figure, there is a region behind the interface echo where the flaw echo can be seen by itself using a suitably placed acceptance gate (Gate 1). Usually, however, pulses do not cease abruptly and so the flaw echo can only be seen by its interference effect on the interface echo. Thus the flaw can be found by placing an acceptance gate (Gate 2) where this interference is greatest.
b
¢
Fig.2 (a) Undistorted interface echo and acceptance gate pulse. (b) & (c) (echo gain increased by 30 dB) Output of acceptance gate for two positions of transducer.
310
In practice this simple approach is not enough. The shape of the ultrasonic pulse is rarely gufficiently well defined: there are often small low level echoes generated from the structure of the probe which confuse the picture at the high gains which are usually necessary to find the flaw; also surface conditions can cause perturbations. Hence in order to find a flaw it is necessary to compare the echo shape from one probe position with that from another position. This is illustrated by the wave forms shown in Fig.2. The undistorted interface echo and the acceptance gate pulse is shown in Fig.2 a. The frequency is 11 MHz, the gate width is 1.8/~s and it is delayed on the interface by 0.7/~s. Figs.2 b and c show the result obtained when the gain is increased by 30 dB and for two positions of the ultrasonic transducer as it sweeps over a flaw. In these figures the top trace is the interface echo and the bottom trace those positive half cycles of that part of this echo which is passing through the acceptance gate. As can be seen, there is a very marked change from one position to another. Since this effect can also be produced by the interface pulse moving with respect to the gate, precision gating is required, ie the position of the gate has to be slaved to that of the interface echo. Generally speaking the best results are obtained using this technique in combination with a C-scan. 2 The narrower the ultrasonic beam at the surface, the more sensitive the technique, since a beam which is wholly on a flaw in one position and wholly off in the next will produce a much greater perturbation than a beam which is broad enough to be partially on the flaw for a number of positions. A comparison of this technique with normal pulse echo is of interest in that there the flaw is found by using a narrow pulse whereas here it is found by using a narrow beam, ie there has been an interchange between time and space to achieve the same objective. NON-DESTRUCTIVE TESTING . OCTOBER 1971
to produce a voltage level proportional to the output of the rectifying gate. ~-L;~ , ' .... The ti-ansducer is scanned over an aperture, in a similar manner to a television raster, and the recorder is synchronised to the scan such that the accepted part of the echo is recorded at a point corresponding to the position in the scan at which it was received.
Results Transducer
V
=
~lntegrator
Fig.3 Diagram of equipment.
These are shown in Figs.4 a and b. The object consists of a thin steel sheet, 0.030 in thick, to which is vacuum brazed a circular boss 0.100 in thick and 6 in. in diameter. The brazing material was contained in four concentric shallow channels, about 0.020 in deep, cut in the boss and these channels tie adjacent to the sheet. The view is taken through the sheet. Fig.4a shows the results oStained using an 11 MHz focused transducer with a resolution of about 1 mm in steel. Typical wave forms are shown in Fig.2. The outline of the boss is clearly seen, as are the four channels (marked by
Limitations As mentioned previously, surface conditions can cause perturbations which obscure those due to flaws. Where the pattern of surface variation is not very marked and differs in shape from the flaws the latter can be usually clearly seen depending upon the degree of ultrasonic scatter. On the other hand where the surface variations are very marked and localised, the flaws are usually obscured entirely. Not enough experience has been accumulated to lay down any definite ruling for this condition and at the moment it is largely a matter of cut and try. From Fig. 1 it is clear that the perturbation on the tail of the interface echo becomes greater the further that the flaw is from the surface, up to the point where the flaw echo ceases to overlap the interface echo. This means that the response is weighted against the flaw nearest the surface and, if two flaws are in line, then the one at the greater distance from the surface will tend to mask the other if the degree of scatter is small. Thus there is little depth discrimination.
I
2
3
4
A
I l l u s t r a t i v e results Equipment
The schematic of the equipment used is shown in Fig.3. Ttie transducer is pulsed by discharging a condenser into it, this being initiated by the output of the pulse generator. The echo pulse after being amplified, is then passed through a gate which is opened in synchronism with the initiating pulse. The time delay and width of this gate are chosen such that it will completely discriminate against the transmission pulse and will always accept all the interface echo however its position may vary. The output of this gate is taken to a discriminator which generates a trigger pulse which is phase locked to the interface echo. This trigger pulse then operates the acceptance gate which in this case is a rectifying one which only accepts positive pulses. To ensure that this gate can be positioned appropriately, the output of the first gate is passed through a delay line to this gate. The output o.f the rectifying gate is subsequently passed to an integrator and a sample and hold circuit
NON-DESTRUCTIVE TESTING . OCTOBER 1971
b
Fig.4 Vacuum braze joint between steel sheet 0.03 in thick and circular boss 0.1 in thick. View through the steel sheet: (a) 11 MHz resolution 1 mm approx. (b) 5 MHz resolution 5 mm approx. Arrows 1, 2, 3 and 4 indicate channels in the boss which originally held the brazing material. Arrow A is a central hole. Non-bonds are shown by dark areas on the boss.
311
arrows 1,2, 3 and 4). Within the area o f the boss the nonbond areas are shown by the dark patches and it is clear that on the outer periphery there are two large areas of non-bond and extensive areas of non-bond between channels 2 & 3, and 3 & 4. The bright area at A is a hole through the middle. For comparison Fig.4 h shows the results obtained for the same object using a 5 MHz transducer with a beam width of about 5 mm in steel. Although the same features are apparent the reduction in resolution and clarity is very marked. The pulse length in terms o f wavelengths of ultrasound was about the same but the wider beam width made the positioning and width of the acceptance gate more critical than in the other case. In both these cases the adjustment procedure was to let the scan run, observe the tail o f the interface pulse, place the acceptance gate where there was most variation with the scan and adjust its width so it only accepted this part of the pulse, and finally adjust the gain to give maximum output.
Conclusion Sub-surface flaws can be found using pulse echo even though they lie within the minimum range resolution of the ultrasonic pulse. It requires accurately controlled acceptance gating of the tail of the surface echo and a narrow or focused ultrasonic beam at the surface. It suffers from the limitations that best results are obtained with smooth surfaces and the response o f flaws tends to be preferentially weighted as they lie further from the surface. However within these limitations it is an extremely useful technique especially if suitable scanning equipment is available.
References 1 PARKER, P. Electronics. Arnold. 1950. p 499. et seq. 2 Nondestructive Testing Handbook. ed: McMaster, C. vo111. Ronald Press 1959, pp 34-43.
non_
destructive testing Reprints Reprints of all articles published in non_destructive testing are available, in quantities of 100 or more R e p r i n t s are e s s e n t i a l - • • • • •
for the company that wants to distribute impartial comment on its activities to potential customers and clients for the company that wants to up-date its technical staff on new techniques and new technologies for the company that wants to publicize its research and development work for the training course organizer who wants to assemble key reading material for his students for the university or technical college lecturer who wants to distribute the latest information on a topic under study
For full details of prices and availability of reprints, please write to The Reprint Department, non_destructive testing IPC House, "32 High S t r e e t , G u i l d f o r d , Surrey, England
312
NON-DESTRUCTIVE
TESTING
. OCTOBER
1971