Spectrum shifts to the left between run

[AnomalySmith]

Hello there,

I decided to build a graded Z-shield with Lead, Cadmium and Copper long time ago and I'm finally about to finish !
So I started longer-than-usual test runs with my LaBr3 detector and noticed I had my Cs137 peak left shifted between each 24h-long run (especially in the medium-high energies).

I thought first it could be related to the MCA soundcard and/or crystal temperature changes but it'll be more than 1 week that everything is plugged and slight deviation still occurs with almost no temperature change (+/-1°C).

Here are the raw Cs137 (662keV) bins obtained in PRA:
24h -> 1959
48h -> 1945
72h -> 1937
168h -> 1927

Do you know where it could come from ?

Thanks !


Soft:

• PRA 25.0.2.1 (384kHz sampling rate settings)
  Windows 11 (USB Soundcard maxed at 48kHz sampling)

Hardware:

• 10x10LABR w/ Hamamatsu R980 PMT @675V
  GS-USB-PRO (Discontinued)

Note: In attachment, background spectrometry: the white spectra line is taken 4 days after the red one. Red with lead shielding and white one with Cadmium / Copper shielding added.
Note2: The spectra isn't calibrated :)

[jneilson]

That still sounds like temperature drift to me. I've previously worked with a LaBr3 detector as part of a pan-and-tilt gamma imaging system, and we regularly saw drift from one spectrum to the next, even with relatively small temperature changes of a degree or so. We had some degree of active gain stabilisation based on tracking the position of the La-138 intrinsic peaks, but it was never perfect, and with spectra taken at various points in time as it panned round over the course of a day, I'd often need to tweak the gain in post processing to get it to correctly match our predefined ROIs so that it could get the correct peak areas in the ROIs for quantifying activity.

[AnomalySmith]

Thanks for your feedback jneilson. I've already read some papers on the light yield versus temperature of LaBr3:Ce detectors and it seemed to me that the deviation was pretty small at first. But given your feedback and this source [broken link removed - Steven ] I'll have to do more tests.

What was weird is that none of my tests showed a right shift from the previous one, as it should be when global temperature is decreasing.

In the meantime, I'll try cooling down a bit the room (easier now as the temperature drops here) to see the effect.

EDIT: I launched a Cs137 acquisition yesterday evening and the still-going spectra is already clear enough to tell that the Cs137 gamma bin is at 1933 (finally a significant right-shift).
This is a bit annoying, as if it's the right cause, it would suggest I have to find a way to have a room temperature change very very low.
But that could also explain why my very weak Cs137 source shows up with a resolution of ~3.5% in a long-run spectro instead of the 3% known specification !

[Rob Tayloe]

I also think that the peak shift is likely due to temperature effects. Even if one carefully controls the room temperature where the measurement occurs, the electronics can introduce some thermal effects. Early in my working career I made a large number of measurements of uranium throughout a gaseous diffusion uranium enrichment facility. As the name suggests the uranium would ideally be in a gaseous phase (as UF6) which requires operating temperatures greater than 100 C. Even with insulated housings the process buildings were very warm in some places and more tolerable in others. One device that we used was a NaI detector, in a lead shield, with an Eberline SAM-2. The SAM stands for "stabilized assay monitor" and included Am-241 with the NaI crystal to help stabilize temperature effects. We were monitoring the 186 keV photopeak from U-235.

Links follow for the SAM-2 technical manual and a couple of papers related to temperature stabilization. Alas, the full papers require a costly subscription (or payment) or one having access to a facility that does maintain a subscription (e.g. university library).

https://ia801707.us.archive.org/3/items ... l_text.pdf
https://www.sciencedirect.com/science/a ... 0286902251
https://www.sciencedirect.com/science/a ... 8712001692

In a separate thread I show the results from SrI(Eu) detectors. A couple of the detectors were coupled to an MCA with included temperature stabilization and one detector did not use any stabilization. One can see some differences in the peaks of spectra that are likely due to thermal effects.
viewtopic.php?f=14&t=885&start=20

[Sesselmann]

Hmm, small proportional drift over a long time.

Hard to guess the exact cause, maybe try logging the temperature along side the spectrum.

The new GS-MAX-8000 comes with a temperature probe and can be temperature stabilised.

https://www.gammaspectacular.com/blue/g ... s-max-8000


Steven

[jneilson]

Alas, the full papers require a costly subscription (or payment) or one having access to a facility that does maintain a subscription (e.g. university library).

I find the "Unpaywall" browser extension very useful in a lot of these cases - it helps find copies elsewhere that are not paywalled. In this case, the second paper in your list can be found at
https://digital.library.unt.edu/ark:/67 ... 844841.pdf

[AnomalySmith]

Thanks all for your answers and docs, there is indeed a lot to learn of and it is very interesting !

In the meantime, I tried other runs at stabilized and monitored room temperature this time (the best I can).

Drift still occurs (but maybe a bit less) even with measured 0.1°C / 0.18°F average difference between the two acquisitions.

I don't know if it's really only related to temperature but as the average T goes slighty lower, the actual drift goes in the right side direction, which stays coherent..

[Sesselmann]

I did some tests here on my bench as well and over a period of 3 days the voltage on the GS-PRO remained unchanged, but the output voltage as measured by my bench meter drifted by around 1 volt in 1000 with a temperature change of 1 C˚.

The GS-PRO is voltage stabilised with a precision voltage source on the high side of the 1M load resistor and when I connect the SHV connector on the GS to the Bench meter I use a 1 G ohm divider, so it is not surprising to see some drift.

Essentially the same with your detector it has a voltage divider with 12 resistors and these are affected by temperature.

It should be possible to make a temperature probe which sends a frequency modulated signal to the right audio channel and then add some code to convert frequency into temperature compensation but not sure if the juice is worth the squeeze 🤔