Posted on by maxfomitchev

In this article I am going to demonstrate how to set up and use an MCA-PRO16 2CH (which is a member of the ANL family) with a low-energy gamma probe.

Hardware Setup

For low-energy gamma spectroscopy I chose a Bicron A567J detector with a thin 1 mm NaI(Tl) scintillation crystal and a beryllium window. Thanks to the beryllium window this detector will respond to gamma energies as low as 1 keV. However, the small crystal thickness drastically limits the detector efficiency for gamma energies much higher than 100 keV.

The detector comes with a single bias + signal SHV connector and therefore it requires a bias /signal splitter in order to be used with the MCA-PRO. I have connected a 10 MOhm / 10 nF splitter to the detector and ran the splitter signal to the CH1 input and the splitter bias to the HV BIAS output of the MCA-PRO – Fig. 1.

Fig. 1. MCA-PRO16 2CH with a Low Energy Gamma Probe.

PulseCounter Setup

I launched PulseCounter Pro and created a new default single-channel experiment. Then I clicked the NEW button to create a new measurement and changed the MEASUREMENT TYPE to ‘CALIBRATION’.

To set up a proper detector preset, I decided to use a Cs-137 source as I expect it to produce a nice 32 keV peak suitable for calibrating this detector. Because I did not know what sort of signal this detector produces I chose the preset settings as shown on Fig. 2-7.

Fig. 2. Preset MEASUREMENT settings.
Fig. 3. Preset CHANNEL settings.
Fig. 4. Preset PULSE SHAPING settings.
Fig. 5. Preset MCA settings.
Fig. 6. Preset PULSE REJECTION settings.

When I do not know the detector parameters I always choose 5 MHz sampling rate, 750V bias (which is safe for all PMT-based scintillators), DC coupling, signal range of 2 V, assume negative signal polarity and employ trapezoidal pulse shaping with 1 us BASELINE and PULSE.

Then I checked the LOG RAW button to record raw detector signal and captured a 120-second measurement – Fig. 7.

Fig. 7. A Cs-137 calibration measurement.

I switched spectrum display mode to logarithmic view by clicking the LOG button for better view of all channels since the 32 keV peak was dominating the spectrum. Looking at the spectrum I see that the 750V bias and the 2V signal range were a good choice since the CLIPPED pulse rate was only 1%. If the clipped pulse rate was much higher I would have reduced the bias voltage or increased the signal range.

In the ACCEPTED PULSE plot I see that the pulses look like nice negative shelves therefore the choice of the trapezoidal pulse shaping was also appropriate (it almost always is).

Preset Tuning

What I don’t like is a strange peak in the lower spectrum spectrum channels. This peak is clearly artificial and therefore must correspond to noise. There are several ways to remove this peak:

  • Increase the THRESHOLD value on the MCA page of the preset (Fig. 5);
  • Setup the countable CHANNELS range on the PULSE REJECTION page of the preset (Fig. 6);
  • Or setup the countable RISE TIME range on the PULSE REJECTION page of the preset (Fig. 6).

I started by increasing the threshold. Because I captured the measurement with the LOG RAW button checked, PulseCounter Pro simply reanalyzed the measurement using the new threshold value without requiring me to re-capture the counts.

I increased the THRESHOLD to 5 mV, but the artificial peak centered on channel 18 still remained – Fig. 8.

Fig. 8. Artificial peak centered on channel 18.

How do I know that the peak centered on channel 18 is artificial? Because I know what sort of spectrum a Cs-137 source produces on a NaI(Tl) scintillator. This spectrum is characterized by a single 32 keV peak in the lower channels, therefore any peaks below 32 keV must be artificial.

Because this peak did not disappear when I increased the threshold I decided to see what sort of pulses make up this peak. To inspect the pulses I highlighted the peak and clicked the SET FILTER button. Now if I go to the RAW page and click the NEXT PULSE or the PREVIOUS PULSE buttons, PulseCounter Pro displays a detector pulse corresponding to the selected channel range – Fig. 9.

Fig. 9. Raw detector signal.

What I see on the RAW page is that pulses that form the artificial peak centered on channel 18 appear normal and do not look like noise due to electromagnetic interference. Still, these pulses must be artificial due to some unfortunate ‘feature’ of the detector design. Is there a way to exclude them?

So, I try an alternative approach by visiting the RISE TIME page of PulseCounter to inspect the rise time histogram of the counted (i.e. accepted) pulses – Fig. 10.

Fig. 10. Rise time histogram of counted (accepted) pulses.

What I see is that most accepted pulses are characterized by the rise time of 1 or 1.2 microseconds. Therefore I decide to restrict the rise time of the accepted pulses to the range of 1.0 to 1.2 microseconds by clicking on the RISE TIME button on the PULSE REJECTION page of the PRESET dialog (Fig. 6) and clickging the REBUILD button. As a result, the artificial peak centered on channel 18 diminished significantly and only 4.9% of all pulses got rejected based on the rise time criterion.

Now I can proceed to calibrate the spectrum by going to the CALIBRATION page of PulseCounter, clicking on the dominant peak, then clicking the NEW button and selecting a 137Cs | 32.3 keV standard from the list of standards to assign energy values to channel numbers – Fig. 11.

Fig. 11. Spectrum calibration.

The spectrum is now calibrated, but the energy resolution is not great. PulseCounter reports 54.7% FWHM @ 32.3 keV. To improve the energy resolution I can try increasing or decreasing the BASELINE and the PULSE values in the trapezoidal filter settings (Fig. 4), let PulseCounter Pro reanalyze the measurement and then recalibrate the spectrum and see the new FWHM value. Changing the BASELINE and the PULSE values requires changing the RISE TIME filter settings as it is the rise time of the shaped pulses that is used to build the rise time histogram.

By setting the BASELINE and the PULSE values to 2 us and limiting the RISE TIME to the range of 2 to 2.2 us I was not able to increase the FWHM but I was able to get rid of the artificial peak completely by rejecting 8.5% of all pulses – Fig. 12.

Fig. 12. Energy spectrum using the 2 us BASELINE and PULSE values for the trapezoidal filtering.

Therefore the somewhat large FWHM value must be caused by detector quality. To make sure of that I captured another calibration measurement at 10 MHz, repeated the tuning steps discussed above but obtained the same FWHM value of ~54%. Thus, it is sufficient to use the 5 MHz sampling rate or I could try an even lower sampling rate to see how low I can go without affecting the resolution (finding the lowest acceptable sampling rate is a good idea when capturing lots of raw detector signals as recording them at high sampling rate invariably takes a lot of disk space).

Conclusion

PulseCounter Pro provides convenient tools for setting up a detector for use with ANL, including spectrum calibration and energy resolution optimization.