New detector and improved dynamic time-over-threshold calibration

Scintillation crystals, PMTs, voltage dividers etc...
Post Reply
Posts: 19
Joined: 23 Apr 2017, 00:04

New detector and improved dynamic time-over-threshold calibration

Post by miree » 16 Apr 2018, 04:12

Hello everyone,

I have a new detector and some improved calibration formula for the time-over-threshold data acquisition that I would like to share.

The new detector is one of Luuk's 2x2'' NaI detectors from his recent offering in the forum. He sent me a number of reference spectra with different sources and the peak widths
3.3% @ 2505 keV Co-60 sum peak
4.4% @ 1332 keV Co-60
6.0% @ 662 keV Cs-137
I don't have any of these sources, but I used other peaks that I used to get an idea if I can reproduce the performance with my home setup:
3.4% @ 2614 keV (room background)
4.6% @ 1461 keV (room background)
5.9% @ 509 keV (sum of LYSO peaks)
These results make me confident that I'm not loosing resolution in my data acquisition box, which a slightly modified version of the dynamic time-over-threshold detection that I've shown in a previous post in this forum. The modification is that I increased the amplitude of the threshold movement by a factor of 10. Before, the threshold moved by ~30 mV after the leading edge of the pulse was detected, now it moves by ~300mV.
The effect is that the time-over-threshold (ToT) value is almost proportional to the signal amplitude for small pulses ( < 100 keV). The logarithmic behavior is only for larger pulses. With the a smaller amplitude of the threshold movement, the linear region is only for much smaller signals. This makes it easier to take this effect explicitly into account during calibration.

The relation between signal amplitude and ToT is:
a_signal = u_max*( exp(ToT/tau) + u_min/u_max - 1)
where tau is the decay time of the pulse, u_min is the low threshold value, and u_max is the high threshold value. (see attached writeup.pdf file for more details)

I took known energies of 5 peaks in my ToT spectrum to fit the parameters (leaving u_min at zero, because my threshold was just above the noise level)
The curve fits nicely to the data points. The resulting calibration is not perfect, but already very close to linear. For now, to get a really good calibration, I have to apply an additional polynomial calibration afterwards.

Best regards,
(117.89 KiB) Downloaded 103 times
Michael Reese
Darmstadt Germany

User avatar
Steven Sesselmann
Posts: 613
Joined: 27 Apr 2015, 11:40
Location: Sydney

Re: New detector and improved dynamic time-over-threshold calibration

Post by Steven Sesselmann » 19 Apr 2018, 11:10


Thanks for posting, your method seems to be giving you some good results, the spectra look quite acceptable and as long as you have a method or formula to calibrate the energy axis, it's useful.

Steven Sesselmann | Sydney | Australia | | | | |

Post Reply

Who is online

Users browsing this forum: Google [Bot] and 1 guest