Circuit Testing

The first test with the electronics, regarding discrimination techniques, was to de-termine how well the electronics/crystals captured the energy spectrum of the 511 keV peak from ²²Na. The results of this are shown in Fig. 26. This was important to ensure that there is a linear response of the crystal/electronics.

The results of the initial tests with the BGO and the LYSO are as expected.

Because of the quicker decay time and the higher light output of the LYSO, it was expected that the peak would have higher resolution. The BGO on the other hand, has a moderate light output but much slower decay time. The expected result is that the BGO would exhibit a lower resolution. The decay time for a BGO crystal is about 300 ns and the decay time for LYSO is 41 ns [21]. The results give good indications that the energy resolution is high enough for pulse-height or pulse-shape discrimination. A second trial run was accomplished to determine how well digital and/or post-analysis filters work to shape the output signal. The initial results of this testing are shown in Fig. 27.

Figure 27 shows both the Fourier transform of the spectrum and also the results of applying a Butterworth filter to the LYSO signal. The filter parameters are shown on the figure: first order, low-cutoff of 10 MHz, and a high-cutoff of 100 MHz. The 10 MHz cutoff reduces any slow-rising component. This would be effective if the goal was to separate two particles that interact differently with the crystal and have well-separated timing properties. The high-cutoff of 100 MHz is effectively to filter out the high frequency noise and increase the signal to noise ratio (SNR). Using the two test crystals, BGO and LYSO, the electronics and crystals configuration demonstrated that it is capable of resolving energy peaks with varying timing properties. This is an important capability for utilizing pulse-shape analysis (which will be discussed later in the paper).

Figure 26. The output from the fabricated circuit board being evaluated with both BGO and LYSO crystals. The BGO was used because it has a slower decay signal with high light output (similar to LiCAF) and the LYSO was used to evaluate the timing performance of the circuit since it has a very fast decay time.

Once it was decided that the wavelength shifting fibers would be used to transport the photons to the SiPMs to created a current, and the appropriate circuit was de-signed and fabricated, the next step was testing it. At this stage, the Eu:LiCAF had not yet been acquired so the initial testing was performed using both BGO and LYSO crystals. Fig. 28 shows the setup that was used for the testing of the circuits. These crystals work well for the electronics testing because of their opposite characteristics.

BGO has low light output and a long decay time, while LYSO emits a bright, fast, pulse of light. A comparison of the signals from the two crystals is shown in Figures 29 and 30.

A schematic of the pulse counter circuit was previously shown in Fig. 22. The timing and gain of each component was carefully chosen to ensure that proper pulse-shape filtering and amplification can be achieved. The first essential component is the Sensl C-Series SiPM, which has a microcell size of 35 µm and a peak sensitivity of 425 nm. The stage 1 is a simple npn transistor used to buffer the current (unity gain) from the SiPMs and the stage 2 amplifier is an Analog Devices AD8007 (ultralow distortion high-speed amplifier, 650 MHz, 1000 V/µs slew rate) with a gain of +2.

Initial testing was conducted with the SensL evaluation board

(MicroFC-SMA-300xx-Figure 27. The fourier decomposition of the BGO waveform (bottom) and the origi-nal/filtered output (top) of the BGO signal. The Butterworth filtering accomplished here was done using MATLAB as a post-process. The MATLAB code used to filter the spectra is included in Appendix B.

Figure 28. The fabricated circuit board being evaluated with both BGO and LYSO crystals. The BGO was used because it has a slower decay signal with high light output (similar to LiCAF) and the LYSO was used to evaluate the timing performance of the circuit since it has a very fast decay time.

Figure 29. The analog test signal from a 4 mm × 4 mm BGO crystal after the circuit shown in Fig 22. The BGO signal has oscillations in the falling edge, likely resulting from the relaxing time of the photocells internal to the SiPMs.

Figure 30. The analog test signal from a 4 mm × 4 mm LYSO crystal using both the SensL evaluation board, and the custom electronics.

35u) to ensure that the light emitted from the Eu:LiCAF would result in a sufficiently high signal-to-noise ratio (SNR) after the inefficiencies from both the WSFs and the photon detection efficiency of the SiPMs (maximum of 42%). Fig. 29 shows the results from comparing the SensL evaluation board to the custom circuit using a

BGO crystal and ⁶⁸Ge gamma source. A BGO scintillator was used for the early electronic testing because it has a well-documented light output from the 511 keV annihilation gammas that could be used for comparison in simulations. The original signal from the evaluation board (green trace, 5 mV/div) had a peak amplitude of about 10 mV, whereas the custom circuit using the AD8007 amplifier (red trace, 50 mV/div) had a peak amplitude of approximately 250 mV and a faster slew rate with the same BGO crystal.

The signal shown in Fig. 31 is collected at the output of the stage 2 amplifier; it then must go through a filter, comparator, and finally a counter. Fig. 31 shows the persistent oscilloscope traces of the output of the stage 2 amplifier (left column of scope image), and the right column of the scope image shows the comparator output due to the gamma interactions in the BGO. The digital output of Fig. 31 (right) is not in persistence mode and only shows a pulse each time the comparator threshold voltage is exceeded. The purpose of the filter after the amplifiers was to suppress the faster pulses in the rubberized Eu:LiCAF wafer and WSFs. The gamma/WSF pulses have a larger high-frequency component than the neutrons, thus filtering the faster pulses in conjunction with pulse-height discrimination allowed most of the gamma/WSF pulses to be rejected.

An active low-pass filter was developed using the Analog Devices ADA4857-1 Operational Amplifier. This amplifier was chosen because of its desirable properties:

ultralow distortion, low power, low noise, and high speed. The next step after the filter was to perform pulse-height discrimination. This was done with the Maxim Integrated MAX995, high speed, low voltage comparator. The comparator outputs a digital pulse anytime the user-defined threshold was exceeded. Using a micro-controller or field programmable gate array, the rising edges of the comparator output can be counted to determine the number of neutrons that interacted with the Eu:LiCAF. The

Figure 31. Persistent oscilloscope traces of electronic circuit mated to a 4mm × 4mm BGO crystal with a⁶⁸Ge gamma source, and the output digital pulse from the MAX995 comparator.

comparator was setup in burst guard mode, which limits pulse pile-up. Counts were only recorded when the pulse signal amplitude exceeded the user-defined threshold.

After testing with the SiPMs, it was evident from the oscilloscope trace that there were oscillations in the signal from the long decay time of the light in the BGO crystal.

This oscillatory nature was not seen in the LYSO crystals. The oscillations are likely from the size of the photocell in the SiPM. The recovery time, or decay time of the pulse, is primarily determined by the microcell reset period, given by the product of the effective capacitance of the microcell and the value of the quench resistor. Since the capacitance of the microcell will depend on its area, the reset time will vary for different microcell sizes. An additional factor that can affect the recovery time is the series resistance from the rest of the sensor that will be more significant in larger sensors [27]. These effects may be insignificant in lower flux neutron fields but would need to be considered in high flux environments.