Professional Documents
Culture Documents
DR2 L07
DR2 L07
Content
7.1
7.2
Intrinsic method........................................................................................ 3
7.3
7.4
7.5
7.6
7.7
Compensation principles........................................................................... 10
7.7.1
L 07
7.1
The action principle of digital detector arrays is based on the conversion of the incoming
X-rays into electrical charges that are electronically readable. Amorphous silicon (a-Si) is
taken as an electrically semi-conducting material for this purpose.
Out of three different conversion principles
only the last two are applied practically for NDT. Each method has specific advantages
and disadvantages showing limits for practical applications in imaging systems. In case
of the indirect conversion methods, matrices of photo-diodes are employed which are
able to convert the radiation (either X-ray or light) into electrical charges. In each photo
diode, the charge carrier are integrated over a given time period before they are readout
electronically for each single pixel followed by graphic presentation via a suitable data
acquisition. Each photo diode is linked to an adjacent TFT switch (TFT = thin film transistor see also laptop displays) that activates the readout process of the accumulated
charges at a given point of time.
Fig. 7.1:
The single TFTs are linked together line by line, and column by column, respectively, so
the information of each single photo diode can be passed to the supporting electronics.
L 07
7.2
Intrinsic method
The so-called intrinsic method is based on a direct conversion of the radiation into electric charge carriers inside of a photo diode.
Fig. 7.2:
The X-rays generate so-called electron-hole pairs within a photo diode. The application
of an initial voltage (bias) inhibits a recombination so that the charge carrier can be
readout after a given time. Within the silicon, also thin film transistors (TFTs) are attached to the photo diodes to enable the readout of each single photo diode. They are
controlled electronically by line and column selectors enabling the transfer of a signal
proportional to the radiation intensity to a suitable data acquisition system followed by
the display on a monitor.
Unfortunately, the absorption of X-rays is rather poor in silicon. For a typical thickness of
a photo diode in the range of some m, only X-rays with energies below 20 keV can be
detected with sufficient efficiency.
7.3
In order to gain a higher efficiency as compared to the direct conversion method, a material can be chosen that possesses better absorption characteristics and a higher absorption thickness. This material is attached onto the surface of the electrically conduction layers, and storing the charge carrier is supported by capacitors at these electrodes.
The charge carriers need to be stored outside the actually photo-conducting conversion
layer since otherwise an overflow to adjacent cell might be possible to occur causing unsharp and smeared images.
L 07
Varian comp.
Fig. 7.3:
Photo-conducting method
An initial tension (bias) of some 100 up to 5000 V is applied via a contact plate or layer
to avoid a recombination of the charge carriers and to transport the charge carrier in a
direct line to the electrodes at the same time. Again, the readout of the charge carrier
is accomplished by thin film transistors that are controlled line by line or column by column.
Fig. 7.4:
a-Se detector
Currently, only one digital detector array is in production working with this a-Se conversion principle. Fig. 7.4 shows such a device, the matrix structure on right panel; a single
element is marked with an arrow. A human hair is shown on the left side as a scaling
reference. Selenium serves as the photo-conducting conversion layer that features a 10
times improved efficiency as compared to the direct conversion method, but still remains
too poorly efficient for energies higher than 100 keV. In the meantime, materials with
other conversion properties such as CdTe have entered the market.
L 07
7.4
Scintillator method
By this means, a fluorescent layer (scintillator) is covering the amorphous silicon to convert the X-rays into visible light. Scintillator materials are mostly fluorescent substances
such as gadolinium oxisulfide (product name: Lanex by Kodak or DRZ by Kasei Optonics).
Fig. 7.5:
Section through an a-Si detector with scintillator and a photo diode layer
Within the scintillator, an incident X-ray quantum generates multiple light quanta which
are converted into electrical charges in the photo diode layer underneath. The thicker the
scintillator layer the more X-ray quanta are absorbed and converted. Dissipating of light
quanta arbitrarily into any direction causes an unsharpness: The light generated by a
single X-ray quantum will be distributed onto several elements by numerous light quanta.
Therefore, the thickness of the scintillator has to be chosen to suit the resolution of the
detector (width of a single element) to balance optimal efficiency (-> low noise) with the
least possible unsharpness (-> perceptibility of details).
A very good compromise between the essential spatial resolution and the light recovery
is achieved by the use of a CsI scintillator. The property of CsI to crystallise in needle
shaped structures entails that the light is dissipated only within the needle like structures
and thus hits focussed just a single element. As a result, higher spatial resolutions can
be achieved herewith.
L 07
Fig. 7.6:
Scintillator structure
The amorphous silicon can be covered with this scintillator substance up to a thickness of
1 mm. This results in a high efficiency in converting X-rays into light. A disadvantage of
this structured CsI layer has become evident storing the information of a previous image
for a longer time (ghost images).
The silicon layer incorporates, in addition to the light sensitive photo diodes, also thin
film transistors (TFT) to enable the readout of each single photo diode. They all are controlled electronically by means of line and column selectors, respectively, making the
transfer of a signal proportional to the radiation intensity to a suitable data acquisition
system followed by the display on a monitor.
A digital detector array consists of a million of light sensitive picture elements (pixels)
arranged squared in a form of a matrix (Figure 7.1). The readout is achieved by the TFT
switching the photo diode to the amplifying electronics followed by measuring the flow of
charges that have been generated by the light.
Figure 7.7 shows a single pixel (dashed box) and the downstream electronics.
L 07
The capacitor of the photo diode is discharging itself even without light. Therefore, this
loss of charge has to be determined individually for each pixel (dark image). This offset image will be subtracted subsequently from each image to neutralise this offset.
The individual degrees of efficiency of all photo diodes and the slight variations of the
readout electronics from line to line and from column to column results in an amplification individual for each pixel. This has to be determined and will be needed for the image
correction (gain image).
Digital detector array are offered with various pixel sizes and dimensions (see Table 7.1).
The weight of a digital detector array amount to some 8 10 kg and thus comes noticeably below that one of an X-ray image amplifier. With respect to the depth of the
casing which is less than 100 mm the digital detector arrays are flatter than the image
amplifiers.
Currently available a-Si detectors are based on materials developed for medical applications. In 2003, two different basic versions have been realised, one with 127 m and
another one with 200 m and. 400 m, respectively. Several providers have assembled
their products incorporating one of these two versions.
Table 7.1:
In general, detectors with a larger area per pixel are less noisy since more quanta per
time period hit a single pixel.
A pixel of the 127 m dedetcor has an area of 0,0161 mm2. A
pixel fo the 400 m Detektor has an area of 0,16 mm2 an thus
receives nearly ten times more quanta in the same time. The
price for this is a geometric resolution that is worse by the
factor of three. To provide the capacity for a larger number of
quanta, higher resolving digital converter are provided in this
type of detectors (16 bit); in comparison, most 127 m detectors only have a 12 bit converter. As a consequence, the
high resolving detectors are used preferably in weld testing
while the 400 m detector mainly is applied for the inspection
of cast components where the contrast range and the short readout times are of bigger
importance.
In spite of several disadvantages, the a-Si detectors utilising the scintillator method represent currently the most prevalent technology for digital detector arrays. Their price is
L 07
significantly higher than that one of the X-ray amplifier what is restricting their area of
application. Enabling even an acceptable price in spite of the sophisticated technology
and manufacturing processes has been made possible only by the mass production in
electronics and of a-Si structures. This also forms the basis for the screens of laptop PCs
(just principally operating the other way round).
Fig. 7.8:
7.5
Fig. 7.9:
L 07
The section through the detector (Figure 7.9) shows large similarities with the a-Si detector. A particularly selected scintillator has to be chosen because of the small pixel size;
in this case, these are CsI:Tl crystals (s. Figure 7.6, right).
Though this technology has been realised for large series (digital cameras) there are only
few systems found in radioscopic applications. The size of the pixel in such systems is
50-100 m, the readout time for a single image with full resolution amounts to ~ s
and the data are processed with a 12-bit converter. The active area ranges from 50*50
mm2 to 220*180 mm2. Due to the optimisation of the production of semiconductors filling factors (ratio of the areas of the photo diodes over those of the other electronics per
pixel) of 80% are achieved even with 50m detectors. The CMOS systems currently
available produce a considerable electronic noise. The superb resolution makes these
detectors being used particularly in specific applications where structures of <100 m
need to be visualised without employing a -focus instrument.
Fig. 7.10:
L 07
10
7.6
The whole system consist of the X-ray sensitive detector itself, a connecting cable, the
power supply and the readout electronics with the data acquisition to produce the X-ray
image on a monitor. For this purpose, an industry PC is taken normally nowadays (s. a.
V8).
computer
Graphic
board
Fig. 7.11:
The sensitivity of the individual pixels varies arising from the production of the digital
detector arrays. In addition, changes in the temperature also have an influence on the
detector performance. By this reason, the detector needs to be calibrated periodically
(gain and offset adjustment). This is accomplished automatically for each single image
by most of the programs.
Moreover, it is unavoidable in the production of photo-diode-semiconductor plates that
some pixels are left producing no relevant signal (Bad Pixel). When calibrating the detector prior to a series of measurements, these pixels are identified and their response is
replaced by the mean of the adjacent ones.
7.7
Compensation principles
Contrast-to-noise (CNR) management is the key element for image quality control in
digital radiology. Modern DDA calibration techniques allow an extraordinary increase of
CNR in comparison to all other digital detectors and film. Contrast reduction due to increase of X-ray energy can be overcompensated by noise reduction. This enables the
reduction of acquisition time and increase of the thickness range per radiograph. Even
limitations in the spatial resolution, constrained by the individual picture element (pixel)
size of the detector, can be compensated with an increased CNR. Bad pixel management
of DDAs in combination with CNR enhancement enables the safe application of this technology for film replacement. Considering these points, three compensation principles
have been formulated for the implementation of DDAs as a viable film replacement technology:
1st Compensation of reduced contrast (eff) by increased signal-to-noise ratio (SNR): If
optimization of contrast cannot be achieved, the noise must be reduced (i.e., in-
L 07
11
creased SNR). If contrast can be increased, there is more tolerance for higher noise
(moderate or lower SNR can be used).
2nd Compensation of insufficient detector sharpness (high unsharpness) by increased
SNR.
3rd Compensation of interpolation unsharpness, due to bad pixel correction, by increased
SNR.
(1)
Typical IQIs as plate holes (ASTM E 1025) change the hole diameters with its thickness.
Therefore, its hole visibility depends on the image unsharpness and the achieved CNR. If
the hole diameter is much larger than the unsharpness, the equivalent IQI sensitivity
(EPS, defined by IQI thickness in % of the penetrated material thickness for 2T hole visibility) changes proportional to about 1 /
SNR eff .
L 07
12
b)
Contrast
Signal
(base material)
Intensity
Intensity
a)
Contrast
Signal
(base material)
Length
Length
Notch visible!
Contrast/Noise is high
Signal/Noise is high
Contrast/Noise is low
Signal/Noise is low
Fuji IX25
SNRnorm~ 265
DDA exposure
Fig. 7.13: High Contrast Sensitivity Technique: Better detail visibility of flaws by
increased SNR of DDA image in comparison to digitized film image of test weld BAM 5.
L 07
13
L 07
14
smallest thickness are even radiographed with 2-1T quality. Here, the tube voltage increase yields a higher efficiency and an increased thickness range based on the digital
high CNR technique.
L 07
15
Fig. 7.15: Comparison of visibility of wire type IQIs according to EN 462-1 for film
(left) and DDA (right) at 8mm wall thickness (images high pass filtered for better visualization). The improved SNR of the DDA allows to detect the wire W19 (50 m diameter) at a detector pixel size of 200m without magnification technique!
It is proposed, to permit the application of unsharp systems, if the visibility of the required wire or step hole IQI is increased by compensation of missing duplex wire resolution through SNR enhancement (see EN 462-5, ASTM E 2002 and requirements of EN
14784-2). Several new standards define minimum duplex wire values for specific applications (e.g. ISO/DIS 10893-7, prEN 1435-2). Typically, one higher (smaller diameter, see
EN 462-1) single wire (resulting in higher contrast sensitivity) shall be seen through adjustment of parameters that increase the SNR if an additional duplex wire of spatial resolution is required in the system qualification for a given material thickness and application. It was proposed in CEN TC 138 WG 1 that the compensation should allow maximum
2 wires vs. wire pair compensations. The compensation should be applicable to plate hole
IQIs too. This is still under discussion.
This effect has been proven with Perkin Elmers XRD 1620 detector in combination with
YXLONs Image.3500 software. Even at a magnification of 1 and a basic spatial resolution of 200m (pixel size), the significantly increased SNR of the DDA allows the detection of crack indications which are hidden by noise in the film image with its much better
basic spatial resolution SRb of 40m. Fig. 7.15 shows the radiograph of an # 13 wire IQI
on a 8 mm steel plate. The radiographs were high pass filtered for better graphical presentation. The digitized film shows wire number 16 and the DDA image shows wire number 19 being visible, which has a diameter of 50 m. Therefore, the detector shows the
wire 19 indication with a sub-pixel resolution.
L 07
16
FIG. 7.16: Different Types of Bad Pixel (D) Groups: Cluster, Relevant Cluster, and Bad
Line
L 07
17
guished: Best practice zone, zone with risky detection, and zone of negative detection. It
shall be noted that there are no exact CNR values agreed on for the limiting lines in Fig.
7.18 yet.
10100
10000
Bad Pixel
Fig. 7.17: Effect of bad pixel interpolation. The interpolation reduces the contrast of the
final indication. The increase of the SNR compensates the CNR to the value of the
intact detector image.
The required SNR for the best practice can be recommended as follows: For example, if
film and digital detector are applied at the same kV and the specific contrast (eff) is
comparable, it is sufficient to require a minimum SNR (see eq. 1). Typical SNR values for
best practice in film radiography are 150 for fine grained films (at net density =2), and
75 for coarse grained films. Since DDAs have typically higher unsharpness than films, an
adequate increase of SNR is recommended. Here, the concept of normalized SNRN can be
applied, and the recommended SNR shall be increased as follows:
SNRrecommended = SNRN
SRb
88.6m
(2)
with SRb basic spatial resolution (definition see e.g. ASTM E 2597)
If the film application (coarse grained film) provides SNRN > 75, the adequate DDA radiograph taken by a detector with 200m effective pixel size (SRb=200 m), should exhibit a SNR > 170. This should apply for the last two areas in fig. 7.18 (right side of fig.
7.18 beginning with: isolated bad pixels and irrelevant clusters); best practice line. The
SNR selected will depend on the available time allotted to the inspection, and can so be
improved using DDAs if time is available to do so. The required minimum SNR depends
on the X-ray energy and may vary if compensation principle I and II are applied. The
final proof of image quality requires always the visibility of the agreed IQIs as in classical
radiography.
L 07
18
Bad pixel
considerations
Limited
number of
bad pixels allowed
in area of interest
Detectability: positive (best
practice)
Nyquist
Limit
Contrast-to-Noise
Ratio
Bad pixel
considerations
Detectabilit
y: risky
Detectability: negative
Fig. 7.18: Recommended praxis for selection of minimum SNR for DDA applications
with bad pixel management.
7.7.5 References:
[1]
[2]
C. Bueno, A. D. Matula, "Digital Radiography for Gas Turbine Components", Proceedings from ASM Gas Turbine Materials Conference, 12-15 October 1998, Rosemont, IL. P. J. Maziasz; I. G. Wright; W. J. Brindley; J. Stringer; C. O'Brien, editors. ASM International, Materials Park, OH, pp. 119-122.
L 07
19