Page 1 of 2

Re: Still not desired resolution (new scint from GEO)

Posted: Sun Jun 21, 2020 5:07 am
by Richard Hull
Like Carl Willis noted long ago, in this very thread, it is tough to diagnose "at range". I am not there to see touch and adjust.

If you truly have 2048 channels, I see what looks like a fixed, uncontrolled amplifier gain there before hitting the ADC. Some radium or Uranium ore would tell the tale due to the large number of easy peaks. In all gamma spec systems, amplifier gain and windowing play the real role of putting what you really want to see on the screen.

Richard Hull

Re: Still not desired resolution (new scint from GEO)

Posted: Sun Jun 21, 2020 8:15 am
by ChristofferBraestrup
I'm very much aware how big a mess tele-troubleshooting is, That also wasn't my purpose posting in this thread, but I really appreciate the effort!
[+HV]
V
My setup is something like [scint]->[bias tee]->[preamp]->[ortec 572 spec amp]->[Canberra Accuspec MCA]

1kV bias
x6 gain in preamp
3µS shaping time
500x gain in spec amp
2048 bins adc gain
lld=0 uld=2048

What is meant by "fixed, uncontrolled amplifier gain before hitting the ADC"?

Re: Still not desired resolution (new scint from GEO)

Posted: Mon Jun 22, 2020 2:36 am
by Richard Hull
Most ADC signal input amps in spectro work allow windowing. This, for initial convenience, is usually zero on the lower level and the upper limit set to the max height. This allows for a 1kev/channel spead over the 2048 channels. You have done this.

This displays a 0kev to 2048kev range. (1kev per channel) the gain will determine pulse height positions within the spectrum. With the smoothly variable gain control and the window set as above, You can calibrate the system with known isotopes.

Am241 59kev gamma. adjust gain so the huge peak top is centered on channel #59
Cs137 610kev adjust gain so that the peak top is centered over channel #610

I assume you have a movable channel blinking cursor as you seem to have your system setup nicely as above. Move the cursor to the channel peak desired and calibrate as below. The amp gain can't just be set to an even number (500)! You much also have a widely variable gain pot to move the spectrum peaks!

In time, with experience, you can set the LLD and ULD windowing and gain to attempt to separate nearby peaks to within the resolution of the scintillator.
Note, that slight non-linearities can exist such that if you calibrate solely at 59KEV the 610kev peak might be at 605 or 618kev. Often, a balance is needed to allow for total span accuracy that is close. One can fine tune, naturally, for one hyper accurate point anywhere in the spectrum for a specific energy expected or any peaks near it.

A full fledged, all-in-one gamma spec usually has a vertical range selector switch that has ranges like 1k,2K,4k, 8K, 16k and so on counts to the top of the screen. Set to 4k counts, small slow peaks on weak sources can be seen to develop quickly. as the peak grows you can move up to 8k or 16k counts at the top of the screen with much finer peaks displayed.

Calibration demands hot isotope sources so that a 20 second acquisition will produce the desired peak easily seen. The gain can be tweaked and re-run for another 20 seconds as the peak is now moved. I use 10 uCi isotopes and adjust the gain on the fly as the little nipple peak can be actively gain tweaked and the fast rising nipple peak slides along the channel numbers. In short, what works, works.
with weak sources you are lost on calibration. You have to wait a long time to see the result of our calibration efforts.

Richard Hull

Re: Still not desired resolution (new scint from GEO)

Posted: Mon Jun 22, 2020 6:43 pm
by ChristofferBraestrup
What a great summary of MCA setup, thank you!

I will keep at it optimizing the setup, but for now I think I will conclude that the square bicron probe with mods behave decently.

One thing that hinder the performance of these probes are the very long RG174 cables built in as standard. RG174, like RG58 has a capacitance of around 100 pf/m,
so half a metre of cable yielding 50 pf to ground from the detector might be significant!

If there is room I would like to put the SHV and BNC connector directly on the back plate (grounded, of course), and put the bias resistor directly in the probe.

Re: Still not desired resolution (new scint from GEO)

Posted: Tue Jun 23, 2020 10:09 am
by Rich Feldman
Very interesting to read about tapered PMT voltage dividers, I guess related to linearity of the charge multiplication.
Nice idea about embedding the bias tee to eliminate some cable and connectors.

This is a chance to point out (for noobs) a simple thing about transmission lines.
No coincidence that RG174 and RG58 both have 100 pF/m, because they are 50 ohm cables with PE dielectric.
Velocity factor = 66%, so a 1 meter section has delay of 5 ns. C = delay/impedance = 5000 ps / 50 Ω = 100 pF.

When low capacitance is desirable, and cable can't be made shorter, it can be made with higher impedance and/or higher velocity stuff.
50 and 75 ohm are popular for RF and video transmission, but 93 and 125 ohm are available, as discussed in this forum.
A TDR measurement of my BF3 neutron detector tube revealed that the gas-filled section has Z0 of about 320 ohms!

Re: Still not desired resolution (new scint from GEO)

Posted: Sat Jun 27, 2020 5:41 pm
by ChristofferBraestrup
So I've been optimizing my setup quite a bit, and it turns out most of my problems still lie with preamplification.

From different sources (esp. the Hamamatsu PMT handbook) I gather that preamp time constant (R*C of feedback net) should be LONGER than the scintillation time constant, which is about 250 ns for NaI(Tl). The Ortec 113A scintillation preamp uses about 50 µs as decay constant (=50 pf over 1 Mohm), and I'm currently experimenting with 20 µs.

The only worrying things left that can be attributed to the detector are a lot (multi-counts per sec!) very very big pulses, that completely saturates my preamplifier. Thankfully, these 12V clipping pulses are over the detectable range for my MCA so they just disappear but my big worry is if they stem from arcing in the divider circuit. I feel my biggest weakness is never having worked with a functioning gamma spectrometer to know what behavior is normal. These show up regardless of voltages (600-1000V tested)