Making a scintillator

[AnomalySmith]

Hi folks,

I would like to have some advice(s) on a project I'm actually working on.
Indeed, my current project is a LYSO scintillator coupled with a R580 Hamamatsu PMT.

To have a linear response, I finally opted for the datasheet solution with the resistance ratio provided.

However, the wiring announced in this datasheet is actuyally for a 2 connector type PMT (Signal and HV).

To obtain a single connector (like BNC) to have both signal and HV, I took for inspiration the famous paintcan DIY scintillator and build the same circuit as shown on this PDF, except for the resistors values, where I choose 1,5MOhm for my based ratio and the same value as in the datasheet for my capacitors.

Not being an expert on electronics, I would like to have some advice generally, but also why this isn't working :D
Indeed, I shortly tried this with a Ludlum Model 12 at 1250V and a scintillator-type sensitivity selected and it seems that the countrate is hugely high (> 1M/min). I naturally think it was not normal and start from asking me if the based resistor value I choose was not too low?
I also checked if there was shortcuts or sparks on the circuit but saw nothing (in total darness this kind of leak is easly visible at this voltage, eh?).

Here is also a picture of this mounting (which is purely temporary, I have to specify :D waiting for a PCB clean version).

What do you think?

Thanks in advance!

Yvan.

[iRad]

Hi Yvan - First I must say I like your breadboard voltage divider approach. Good way to learn what works. Second, I hope you are keeping the entire assembly (entire PMT and crystal) in a light tight container. No light can be allowed to enter anywhere into the PMT, including the sides and socket area. Thirdly, you will need a much higher value voltage divider string to power with a Ludlum meter. I would consider 60M (many Ludlum supplied scintillators use this value or higher) as the minimum value for use with their meters, otherwise you will simply drag down the power supply. Stevens' very nice Gamma Spectacular HV Bias adapters will have no problem powering your tube with the lower value string. But if your using a Theremino PMT Adapter, or other battery powered HV bias supply, then you may need an even higher value, of say 100 MegOhm or better.

Light leakage or an improperly wired VD (or both) may be causing your runaway count rate. Although it's hard to follow your breadboard layout from the picture, it seems way overly complicated for a 2-wire (single HV and signal shared cable), and even a bit more than needed for a 3-wire VD. Send me a post to iradinc at att dot net and I will be happy to send you a proper VD to use. Let me know what you plan to power the tube with for spectroscopy. Also, remember that LYSO has a certain level of activity inherent in the crystal due to the Lu176 content. Cheers, Tom

[Sesselmann]

Yvan,

My first reaction was the same as Tom's, your PMT needs to be completely covered from any light before you attempt to power it up. These things are designed to detect single photons, not broad daylight...enough said.

You need to understand what a capacitive coupling is and why it works. Analogue electronics is intuitive and easy to understand once you get your head around what the components do. The coupling is just a ways to convert a small voltage drop on the high voltage end, to a small voltage drop on the low voltage end. The capacitor isolates the high voltage from the low voltage and still lets the signal through, think of it like a drum skin, you can easily transmit sound from one side of a drum skin to the other, even though there is no air flowing from one side to the other.

The coupling has the same function regardless of weather it sits in the detector or in the rate meter, the only difference is the length of the cables. Even a single wire detector hooked up to a rate meter has two wires, but the second (signal wire) is inside the rate meter so you don't see it.

Here is a simple diagram with a detector, load resistor, coupling and load on the signal end and you will see what's going on.

coupling.png (26.07 KiB) Viewed 18599 times

[AnomalySmith]

Hi iRad and sesselmann.

Thank you for your fast and precious replies!

Sure, I don't want to burn out my PMT and only power it up when covered in darkness, I saw some articles on it and want to avoid the destruction of my dynodes, photocathode and the gaz production inside the PMT (which transform it into a little neon tube) :D

@iRad: I guess I'll mainly use this probe with the Ludlum 12.
Concerning the value resistor when you say 60M as the minimum value, it's the entire voltage divider or the minimum value to adopt between two dynodes?

Indeed I noticed my voltage (when pressing the HV button on the ludlum) was bellow the initial value (like 900V instead of 1100/1200V) but was not sure if it could be a reliable measure during the process.

Now, for the LYSO, I actually choose this kind of cristal due to it's high shock resistance, non-hygroscopic, high energy absorption and low cost which makes him the ideal candidate for outdoor, quick detection and auto-calibration :D (even if the resolution is inferior compared to NaI(Tl) ones).
(Here is also in attachment a Gamma spectroscopy I've done on a Lu176 crystal with a 30*70mm NaI(Tl) in a lead chamber).

@sesselmann: Thanks for your explanation, it's true that part was still a bit blur to me!

Finally, I let you another picture, slightly better than the first one about the breadboard mounting.

Now the most difficult part will probably be to found those resistance values without using 3/5 of them at each dynodes stages (And at reasonable costs!).

EDIT: I already own a GS1100A, do you think I sould give it a try? :D

Yvan.

[iRad]

Hi Yvan - For use with a Ludlum meter, your entire VD string should be ~=>60M. I would use at minimum a 4M7 for R (Dy1 thru P), and a 10M for R2 (between K and Dy1). That only totals 57M, but it's probably close enough. You could use R=10M/R2=20M, which would be even better for use with the Ludlum meter. Your GS-1100 should have no problem with even your current value resistors, once you get your divider wired correctly that is.

For starters, you may wish to simplify your VD by getting rid of all the capacitors but one .01 uf HV cap (103/2kV) between the K and Dy10. This is sufficient for detection of even moderate count rates. Once you get that going, then try adding additional caps to the last two or three dynodes. Do not put a cap between the last dynode and anode (Dy10 to P). Then your positive HV will go to the anode (P), and your ground lead will go to cathode (K). Cheers, Tom

[Sesselmann]

Yvan,

Your GS-1100A will work with any divider over 6M Ohm. Remember when you are running a single wire detector, the load resistor is in the rate meter is technically forms part of your divider, so when you start hooking up low impedance dividers the actual voltage on the dynodes is less than the rate meter is set to. Confused?

The voltages on a rate meter or GS assumes infinite impedance, so when you put a load across the load resistor, the voltage is pulled down. Simple enough to calculate. V_f = V_i * (R_d/(R_d+R_l) where V_f is the final voltage, V_i is the initial voltage without load, R_d is the divider resistance and R_l is the load resistor.

Detectors like geiger counters and gas proportional counters have almost infinite impedance, so this really applies to PMT's only.

[AnomalySmith]

Thanks iRad, I followed your advice and use now 10MegOhm at each stage, except between K and Dy1 where 20MegOhm is put.

Now my countrate is less than the last time, but still pretty high (about 100 000cpm). The difference is I'm able to detect a "slight" difference when I put a radioactive source next to the detector (up to to 150 000cpm with Am241).

Here are screenshots from the jack output of the GS1100A with my bitscope, it seems to be a reliable signal, but do you think it's normal to have this kind of countrate with 16 LYSO crytals (I probably bought from you :D)? Where could be the problem?


@sesselmann: Thank you again for this clarification, I guess it's actually the "signal splitter" part sometimes quoted in some rate meter devices?

Yvan.

[iRad]

Hi Yvan - Your HV could be set too high, try starting at about 650 to 700 volts. Your post above said that you had voltage set at about 1100/1200 volts, so make sure to drop that quite a bit. You should not see a large voltage drop now that you have a higher value VD.
Also, if you are using 16 LYSO crystals coupled together in an array, then your count rates could be very high. When using more than a few LYSO crystals, you will probably need to subtract the background generated by the Lu176. Even a single LYSO crystal has enough Lu176 to make a nice little check source for a scintillator. Several crystals together are not enough energy to register on a GM tube, but any good scintillator can see the energy in even a single crystal easily enough. For gamma spectroscopy, a single crystal attached face-on (the clear 4mm x 4mm end) is all you need, with FWHM in the 6 to 7% range very possible. Some have used a 2X2 array (4 pieces) with good resolution results, but more than that could be a problem without using a multi-anode PMT, diode array, and or very sophisticated electronics. For general survey meter use, LYSO may not be the best choice due to the additional background. Cheers, Tom

[valentin]

Hi Yvan,

100.000cpm is pretty much. I would suggest testing your PMT without a crystal attached. If you still see these pulses, then something is wrong with your setup. For example, the voltage may be too high, causing a lot of random discharges inside the PMT.

Valentin

[AnomalySmith]

Hi iRad and valentin,

I will test these advices as soon as come back to home: next weekend..

But when looking my oscillo output, the pulse seems to be correct and when calculate manually the coutrate from one of my last screeshots, I obtained about 84 000 cpm which is close to the reading I actually have. If it's right, I really not expected this level!

It seems also difficult to use this kind of configuration for survey purpose, even with substract.

Keep in touch :)

Thank you all for your help!

Yvan.