A low cost X-ray imaging device based on BPW-34 Si-PIN photodiode

A low cost X-ray imaging device based on BPW-34 Si-PIN photodiode

Nuclear Instruments and Methods in Physics Research A 819 (2016) 1–5 Contents lists available at ScienceDirect Nuclear Instruments and Methods in Ph...

1MB Sizes 63 Downloads 417 Views

Nuclear Instruments and Methods in Physics Research A 819 (2016) 1–5

Contents lists available at ScienceDirect

Nuclear Instruments and Methods in Physics Research A journal homepage: www.elsevier.com/locate/nima

A low cost X-ray imaging device based on BPW-34 Si-PIN photodiode E. Emirhan, A. Bayrak, E. Barlas Yücel, M. Yücel, C.S. Ozben n İstanbul Technical University, Faculty of Science and Letters, Department of Physics Engineering, 34469 Maslak Sarıyer, Istanbul, Turkey

art ic l e i nf o

a b s t r a c t

Article history: Received 30 December 2015 Received in revised form 5 February 2016 Accepted 7 February 2016 Available online 16 February 2016

A low cost X-ray imaging device based on BPW-34 silicon PIN photodiode was designed and produced. X-rays were produced from a CEI OX/70-P dental tube using a custom made 7 30 kV power supply. A charge sensitive preamplifier and a shaping amplifier were built for the amplification of small signals produced by photons in the depletion layer of Si-PIN photodiode. A two dimensional position control unit was used for moving the detector in small steps to measure the intensity of X-rays absorbed in the object to be imaged. An Aessent AES220B FPGA module was used for transferring the image data to a computer via USB. Images of various samples were obtained with acceptable image quality despite of the low cost of the device. & 2016 Elsevier B.V. All rights reserved.

Keywords: X-ray imaging Si-PIN photodiodes X-ray detectors Low cost

1. Introduction X-ray imaging techniques have been widely used in industrial and medical applications since the discovery of X-rays [1–4]. High resolution X-ray imaging devices are used for medical diagnostic purposes and the most important requirement for these systems is to obtain images as quick as possible to reduce the dose [5,6]. On the other hand, industrial X-ray imaging devices have wide range of use and they have application specific requirements [7]. Scanning time of X-ray imaging is also important for most industrial applications. However, there are some cases where the measurement time may not be an important issue [8,9]. X-ray imaging studies of archaeological samples could be good example to one of these cases [10,11]. For such a case, a low cost imaging device readily available on hand could be very handy. Based on this argument, we have produced a simple and low cost X-ray imaging device based on a BPW-34 silicon PIN photodiode [12]. Designing a device for fast images was not our priority. However, having an acceptable image quality together with reduced cost was our main goal in this work. The total cost of the imaging device presented in this work is under 1 K$.

2. Experimental setup Energy spectrum of the X-ray source has to be determined for setting up a reliable imaging system. Filtering applications based on known energy spectrum are helpful tools to obtain n

Corresponding author. Tel.: þ 90 212 285 6990; fax: þ 90 212 285 6386. E-mail address: [email protected] (C.S. Ozben).

http://dx.doi.org/10.1016/j.nima.2016.02.022 0168-9002/& 2016 Elsevier B.V. All rights reserved.

images with better quality. For that reason, energy spectrum of CEI OX/70-P dental tube was determined in our previous work [13] using Geant-4 simulations. For cross check, we also measured energy spectrum of the tube with a commercial Amptek X123 spectrometer [14]. The geometries of CEI OX/70-P tube and Amptek X123 detector, provided by the manufacturers, were embedded in Geant4 simulations. Then, electrons were driven to anode of the tube and X-rays were produced according to Bremsstrahlung and atomic de-excitation processes. X-rays penetrating through the glass envelope of the tube and reaching the detector were counted. Only the photons with complete energy deposition in the detector material were taken into account to obtain the energy spectra. Simulation results from this study were compared with the measurements for various energies and it is shown in Fig. 1 [15]. 2.1. Power supplies We first concentrated on building a high voltage power supply for the X ray-tube. This device was built from three stages. The first stage was a PWM (Pulse Width Modulator) circuit with adjustable frequency and duty. Two identical step-up transformers were driven by a MOSFET circuit after the PWM, in the second stage. Finally, a well-known Villard-Cascade topology was used to cascade the initial stepped-up potential to the desired value. As known, high voltage power supplies are tricky to work with. Isolation problems, discharges and safety considerations were some of the issues that one had to deal with. A magnitude of 40 kV anode voltage was used for our most imaging studies. Instead of running the tube for our measurements with a single supply (þ40 kV), it was preferred to generate a symmetrical HV (720 kV) with two independent cascade units. This helps to


E. Emirhan et al. / Nuclear Instruments and Methods in Physics Research A 819 (2016) 1–5

overcome unwanted discharges and to ease the isolation problems. Both positive and negative cascade units were embedded in epoxy-resin. Schematics of the anode HV power supply circuit are shown in Fig. 2. A custom made filament transformer was wounded and it was also embedded in epoxy-resin. An isolation transformer was used between the mains and the filament power supply running at 2 VAC and 1.5 A current in regular operating conditions. Most industrial type power supplies used in X-ray tubes are designed with feedback circuitry to stabilize the voltage/current output. However, our design does not include such a feedback control for simplicity. Obviously the drifts in the filament current cause X-ray intensity to change and this can degrade the image quality. This problem was solved by correcting the image data when necessary by using the data obtained from an X-ray intensity monitor. 2.2. X-ray detector The detector section is considered as three stages. First stage is an ultra low cost BPW-34 silicon PIN photodiode which was used as the active detector element. BPW-34 has 4 mm2 active area and low terminal capacity (15 pF at 12 V reverse bias). A 1.0 mm thick, 1.0 mm diameter aluminum collimator was directly mounted on the photodiode and all measurements were performed with this

collimator. Photodiode was directly mounted on the PCB of charge sensitive preamplifier, the second stage of the detector, with the shortest signal path to the FET as shown in the Fig. 3. The charge sensitive preamplifier is very similar to the one produced at BNL (Brookhaven National Laboratory) Instrumentation Division nearly two decades ago with the code of IO-446. There is no appropriate reference in the literature about the details of this preamplifier. However an updated version (IO-446-2) of it was used at AGS and RHIC [16] during early 2000's. PCB of the charge sensitive preamplifier was redesigned using on-shelf components and geometry of it was modified to match it with our OPAMP based shaping amplifier. An ENC measurement for the preamplifier took place and it was found to be RMS 300 electrons. Shaping time of the preamplifier was determined as 200 ns for minimum noise of the preamplifier. Noise level after the shaping stage was about 20 mV RMS. Energy calibration using 22 keV of 109Cd and 59 keV of 241Am showed that 10 keV detection limit was safely above the peak value of the noise level. Additionally, the simulations showed that X-rays below 10 keV did not pass through the glass window of the tube. This can be seen from Fig. 1. The shaping amplifier was the last stage of the detector. A single bit comparator was also used in the last stage of the detector for obtaining a digital signal where the threshold was set slightly above the noise level. Measurements showed that the

Measurement (20 keV) Simulation (20 keV)


Measurement (25 keV) Simulation (25 keV)

Counts (A.U.)

Measurement (30keV) Simulation (30 keV)


Measurement (35 keV) Simulation (35 keV)



0 0










Fig. 1. Comparison of measured and simulated spectra for various energies [13].

Fig. 2. Schematics of high voltage power supply unit.

Fig. 3. X-ray detector.

E. Emirhan et al. / Nuclear Instruments and Methods in Physics Research A 819 (2016) 1–5


Fig. 4. Schematics of the charge sensitive preamplifier.

Fig. 5. Schematics of the shaping amplifier.

noise level of the detector electronics was around 5 keV X-ray energy. The schematics of the charge sensitive preamplifier and shaping amplifiers are shown in Figs. 4 and 5 respectively. The detector was placed in an aluminum enclosure having 3 mm opening where the photodiode (with its collimator) was centered through the opening that was covered with a black tape to prevent leakage of light. A 2D position control unit was used for moving the detector in small steps and count rate in each step was registered and transferred to a computer using an Aessent AES220B FPGA [17]. 2.3. Control-logic 2D position control unit has parallel port interface and a control pulse obtained from this interface was used for enabling the pulse counter of FPGA. The counting of pulses was done when the detector was stationary between the scan steps. This procedure was performed with the logic circuit given in Fig. 6.

3. Measurements It is known that drifts in the filament and anode voltages cause intensity, therefore the count rate, to fluctuate larger than statistical expectations. Since there were no feedback controls for precise regulation of the filament current and the high voltage of the tube, X-ray intensity was continuously monitored with another BPW-34 based detector as mentioned above. In order to investigate the magnitude of these fluctuations which may deteriorate the image quality, we have made a measurement performed with CEI OX/70-P dental tube that was powered with our custom made HV and filament power supplies. As shown in Fig. 7, X-ray intensity fluctuates due to non-regulated power supplies. The tube was run mostly with low anode current (  50 μA) and the temperature variation on the tube during the measurement was smooth and obviously could not explain this drift. In all measurements, the X-ray tube was placed in a composite lead cylindrical shield with three cm circular opening. The objects


E. Emirhan et al. / Nuclear Instruments and Methods in Physics Research A 819 (2016) 1–5

Fig. 6. Block diagram of the control logic.

IM counts

150 125 100 75 50













Current (A)

Time (min) 1.6 1.55


Temperature (C )






Time (min)

50 40 30 20 0






Fig. 9. X-ray image of a watch. The step size is 0.25 mm and there are 229 horizontal and 200 vertical steps.

Time (min) Fig. 7. Intensity monitor counts (top) and time stability of the filament current (middle). Temperature vs time of the tube wall (bottom).

Motor drivers



watch. No correction due to intensity changes during the measurement was applied for this image. Detector was moved in 0.25 mm steps horizontally and vertically to construct this two dimensional image. Similar procedure was applied for Fig. 10 that shows the X-ray image of a screw with and without the intensity correction. The intensity correction was made based on the following steps. Monitor counts (Mi) were normalized to its minimum value (Mmin) as

Intensity monitor

X-ray source

M ni ¼ Detector

_+ HV Unit Filament power supply

Mi M min


and normalized monitor counts were obtained (Min). Then measurements at each scan point (Ni) were scaled with the normalized monitor counts as follows:

Fig. 8. Measurement setup.

were positioned 15 cm away from the tube. The object and the source remained fixed and the detector was moved with the 2D position scanner. An intensity monitor was positioned close to the edge of the X-ray beam profile to prevent the detector from counting the scattered photons. The setup is shown in Fig. 8. To minimize the image reconstruction artifacts, the object-detector distance was chosen to be smallest possible and the mentioned single hole collimator was used to minimize the scattered X-rays from large angles. The image data containing the position information and the count rate were constructed as gray-scale images. 4. Results and discussions Following images were taken with 717 kV with the experimental setup shown in Fig. 8. Fig. 9 shows an X-ray image of a

Nci ¼

Ni M ni


where Nic were the intensity corrected counts. Fig. 10(b) shows the image from these corrected counts. As a result, X-ray images of some objects were obtained with reasonable image quality. Depending on the desired resolution, images may take up to several hours even for small sized objects. However, when time is not number one priority, this device can provide a good quality X-ray images at low cost. Of course, our measurements with a single detector are time consuming. However, identical detectors can be produced and a detector array can be constructed. This could dramatically decrease the time required for having images. However the total cost will only change by a small fraction.

E. Emirhan et al. / Nuclear Instruments and Methods in Physics Research A 819 (2016) 1–5


Fig. 10. X-ray image of a screw with no intensity correction (a) and with intensity correction (b). The step size is 0.25 mm and there are 90 horizontal and 107 vertical steps.

Acknowledgments We would like to thank to the Scientific and Technological Council of Turkey (TUBITAK) for supporting this work (113F104).

References [1] S. Kelkar, C.J. Boushey, M. Okos, Journal of Food Engineering 159 (2015) 36. [2] R.P. Haff, N. Toyofuku, Sensing and Instrumentation for Food Quality and Safety 2 (2008) 262. [3] J.M. Davis, D.E. Newbury, P.R. Rangaraju, S. Soundrapanian, C. Giebson, Cement and Concrete Composites 31 (2009) 171. [4] S.D. Auweter, J. Herzen, M. Willner, S. Grandl, K. Scherer, F. Bamberg, et al., British Journal of Radiology 87 (2014) 20130606. [5] P. Zhu, K. Zhang, Z. Wang, Y. Liu, X. Liu, Z. Wu, et al., Proceedings of the National Academy of Sciences of the United States of America 107 (2010) 13576. [6] A.J. Gislason-Lee, C. McMillan, A.R. Cowen, A.G. Davies, Medical Physics 40 (2013) 091911. [7] E. Lombi, M.D. de Jonge, E. Donner, P.M. Kopittke, D.L. Howard, R. Kirkham, et al., PLoS One 6 (2011) 4.

[8] J. Jakubek, T. Holy, M. Jakubek, D. Vavrik, Z. Vykydal, Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment 563 (2006) 278. [9] M. Eveno, A. Duran, J. Castaing, Applied Physics A: Materials Science & Processing 100 (2010) 577. [10] H. Matthiesen, E. Salomonsen, B. Sorensen, Journal of Archaeological Science 31 (2004) 1451. [11] L. Jacobson, F.C. de Beer, R. Nshimirimana, Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment 651 (2011) 240. [12] BPW34 Silicon Pin Photodiode, Vishay semiconductors, available online 〈http://www.vishay.com/docs/81521/bpw34.pdf〉(accessed 17.12.05). [13] CEI OX/70-P dental tube, C.E.I. Compagnia Elettronica Italiana, available online 〈http://www.ceixray.com/ox70-p-2/index3dab.html ¼ eng〉(accessed 16.12.15). [14] X-123 Complete X-Ray Spectrometer with Si-PIN Detector, Amptek, Inc. retrieved from 〈http://www.amptek.com/products/x-123-complete-x-rayspectrometer-wth-si-pin-detector〉(accessed 16.12.15). [15] M. Yucel, M.E. Emirhan, A. Bayrak, C.S. Ozben, E.B. Yucel, Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment 799 (2015) 50. [16] H. Huang, K. Kurita, Fiddling carbon strings with polarized proton beams, In: AIP Conference Proceedings, vol. 868, 2006, pp. 3–21. [17] Aes220 High Speed USB FPGA Mini-module, Aessent Technology, available online 〈http://www.aessent.com/products/aes220-minimodule.html#!/  /pro duct/id ¼ 52901862〉(accessed 16.12.15).