Gamma Spectrometry Forum

Sorry for the title but I feel it's adequate. :)

Still testing my "portable" gamma spec system. I use the same setup (sound capture at 192KHz/24bit, PRA22, GS-2000 Pro, 1" NaI(Tl) probe with R6095), the only change is that I bumped the voltage up from 700V to 750V. I get better resolution at low energies this way.

So in this experiment I ran a 1h background, and then a 1h test with potassium. The sample was a plastic container with 480g pure KCl, which according to my calculations contains 30mg of K-40.

The results are below, one raw and the other with background subtracted. The system isn't calibrated yet (running some Cs-137 and then Co-60 as I type this) and I was hoping to use the K-40 peak for calibration purposes too.

Only... there are too many peaks where the K-40 should be.

So, which is the real K-40? Will the real K-40 please stand up? And what are the other 2 peaks?

Silviu,

Your volume is too high, it needs to be set lower so your spectrum falls in the range 0 -100 arb.u.

What you are observing above is a pileup in the last bin.

A good rule of thumb is to get the k40 peak showing at 50 arb.u (middle of the screen). For the most part you can ignore the peak in the last bin, alternatively cut off your spectrum at 100 arb.u using the advanced filter.

Good luck..

Steven

PS: When switching to a new computer, it's always good to check the pulse direction, as it's not necessarily the same. I have seen a lot of people confused about why one computer has a perfect spectrum and another doesn't, and it has turned out that the sound card on one inverts the signal. Not a problem, you just change the polarity of the height threshold.

Thank you Steven, I am re-running this with a different volume setting (volume was at 80%, have decreased it to 25% to get the K-40 peak around 50au). It's little things like this that ruin your day.
As for polarity, I always check. Every single computer and sound card I have give me negative pulses. So the minus is a must. :)

And the real K-40 stood up.

Silviu,

Remember that the volume setting is not the only way to change gain, you also have voltage and I recommend that you find optimum voltage first by doing a quick linearity check.

The most convenient source for checking the linearity is Thorium because it has so many peaks. The attached chart can be plotted using any spreadsheet like Excel, in this case I used Apple Numbers.


1 - Concave - Voltage too low
2 - Linear - Voltage correct
3 - Convex - Voltage too high

BTW, do you know that the Chinese words for "concave" and "convex" are ? Easy to remember isn't it?

Steven

That convex word is giving me the finger. :D

I have been adjusting my voltages to suit the part of the spectrum I'm most interested in but mostly by trial and error; actually I've played with that a lot just this afternoon as I have built a new detector with a CsI crystal which I will probably post about in a few days in the "detectors" section. As to the detector used above, it has a large linearity shift between 700 and 750V. I will do some plots with thorium time permitting.

But yes, my observations are consistent with that; I sometimes use a higher voltage if I want to spread the lower energies farther apart. That would be consistent with a convex curve.

At 750V, convex as expected. :)
(I skipped higher energies because with this small detector I don't get great peaks)

725V, better but still convex to some degree.

And even better at 700V (my default voltage for this setup). I had to increase the volume to 30% to keep the K-40 put.
TBH the nonlinearity at the lower end is not unexpected. The Lu-176 peak at 88keV is much less defined than at higher voltages so there's an amount of uncertainty as to its exact location.

Silviu,

Good analysis above, it seems your detector will become linear at around 670 Volts, so as long as you can get enough volume there shouldn't be a problem running at 670.

Once you have the linearity right you then need to check the resolution at that voltage. Sometimes you can be fooled by higher than expected resolution by pushing the voltage too high.

Steven