PMT Philips XP2020 damaged

[ChrisRadium2023]

Hello,

I've started to make my onwn gamma spectrometer ...
I have a Philips XP2020 pmt found on ebay (new old stock) and of course when I started to make the voltage divider I knew nothing about Pmt physics (so i read all thats available about it).
I had to buy an oscilloscope and after a few days was able to see the pulses on the screen : to get the signal i read it trough a 10M resistor after the coupling capacitor and put the probe a the "terminals of the resistance" (sorry , dont know the english tech term for it). I didnt have a preamp yet and am still working on it.

The HV sourve is a cheap HV module found on Aliexpress ; the kind of that can be found in a flly killer racket that is annouced to reach 1800V @ 3.9V power supply ....

So far so good as it works....but .....

The HV has high ripple up to 200mV and the signal read up to 12V!!!!

(That was in october 2022) 2days ago I realised that I got it all wrong about my HV probe to read the real voltage with a meter...and that i've been in fact using the PMT under 2000 to 3400V instead the 1800V I thought!!

The max rated in the datasheet is 3000V .....

My voltage divider is devised to limit the current up to 10uA so I guess what saved the pmt as it still works fine now that I apply the proper voltage. And of course the pulses are now much way down than 12V.

But still something is puzzling me. The dynamic range of the Pmt that seems not to be able to reach around 600 keV. The crystal used is perfectly clean and transparent , almost never used (70mm dim and height).
The Am 241 give a nice peak but the FWHM is still quite wide.
I dont have Cs137 so when I try with Ra226 its not very nice and the Bi214 peak @ 609 kev is almost not visible if not at all.

Is my Pmt damaged ? Is the voltage ripple responsable for its limitation ? (i thought that the ripple would eat the low energies depending how much ripple and the amplitude of the pulses.

Yesterday at last I was able to add a 8 stages RC filter to reduce the ripple from 200mv down to 250uV and I intend to add some more stages to reduce it down to some 10-20uV.
But when I tryed last night a capacitor unsoldered itself and the noice produced by the arcing produced some voltage spike that burned the usb sound card or worse the audio chip on the mother board...

What could be the cause of the reducing of the dynamics? (i also tried different voltages and chain divider but nothing resulted appart from shifting the spectrum).

Thanks for help
Chris

[ChrisRadium2023]

Here are some spectrum I got so far...

and the pmt XP 2020 schematic

[Sesselmann]

Chris,

I assume you already know this, but a warning "Never power up a PMT in daylight" it will strip the cathode clean in a few seconds. After such an accident the brown coating on the front of the PMT becomes clear. This is something to look out for if you are buying on eBay. Talking from experience 😉

Assuming your PMT is fine, you need to filter the HV from your HV power supply to reduce the ripple, a couple of low pass filters in series -- 1M--100K--- resistors with 10nF HV capacitors to ground.

Use a simpler voltage divider scheme, with cathode to ground and HV to the anode via a 1M load resistor.

I am not familiar with the Photonis tube you are using, but the voltage seems extremely high, most PMT's used with NaI crystals have sufficient gain between 600V to 800V, but may be rated for higher voltages.

Best of luck..

Steven

[ChrisRadium2023]

Hello and thanks for your fast answer
I didnt know at first that pmt shouldnt be run in full daylight and at the very begining i ran my pmt il very dim light. I could see many pulses or hear then if i linked the output to the speakers i usually use for my geiger counter (DP66 that i reparaired).
The photocathode seemed still ok and amber gold color (i saw post on this forum of the pmt ripped of its coating a few days ago )..

THis morning i just bought a new usb sound card and could check that everything was ok. And with a 8 stage RC filter the ripples have gone down to 250uV wich seem to be even lower than most of the pmt HV supply...so I think this issue is solved so far ( at last!! and indeed on the oscillo the pulses arent made of wavelets like they used to before the filtering).

Now i have to built my CRRC filter to give the pulse the lengh and gaussian shape that the MCA should be able to use ....and then maybe I will be able to see if the dynamic has improved.

The datasheet of my pmt XP2020 gives 2000 V typical and its cousin from EMI THORN that i have too is rated 2000 to 3000V.

I tried as you were saying earlier to power it up at 600-800V but the pulses are very very weak. The load resistance recommended of 50 Ohms gives very short and fast pulses and my oscillo bandwidth cant cope with them... if i dont lenghten them with the 1 to 10M that I use to read the pulse voltage with the oscillo.

I'll post some photos of my very crude setting ....


radioactive regards
Chris

[ChrisRadium2023]

Philips XP2020.pdf

(4.89 MiB) Downloaded 313 times

here are datasheet

[Sesselmann]

Chris,

Thanks for posting the data sheet, I see the recommended voltage for the 12 stage PMT is quite high and so is the gain 2 x 10^6, so your circuit I assume is designed accordingly.

Most detectors I make have 8 or 10 stage PMT's, voltages in the 600-800V range and a gain in the high 10^5, and pulses are in the 300-450 mV peak voltage, which is good for audio sampling.
The dynamic range of sound card ADC is in the range +-1.2V so for me this works well.

Ultimately it doesn't matter if you are getting good spectra 😉

Steven

[iRad]

Chris - More stages usually means more gain, but also more noise. Most 12 stage and higher PMTs I've seen and used can be good for counting applications, but in my experience not so great for gamma spec (because of high noise). As Steven says, most PMTs we work with are 8 to 10 stages. Some of the best PMTs I've seen for gamma spec use only 8 stages.

I'm guessing 70mm crystal you're using with your PMT is a NaI(Tl) crystal. A rule of thumb with building a good gamma spec detector is to use a PMT with a larger active area than the crystal face. For best resolution and performance (if your crystal is clear and white), you should be looking for a good quality 3" PMT.

Also, those "fly swatter" and similar power supplies are notoriously noisy and very low power. But as long as you're having fun...

Welcome to the forum, your experience and comments are very much appreciated. Cheers, Tom

[ChrisRadium2023]

Thank you iRad^^
I took the 70mm because it was the cheapest i could find at the time in my aeria (in France) on ebay and it was as new compared to may I saw on ebay. But yes, you're right , a 50mm would habe been better.
Still , I made several experiments last night : I wired my EMI THORNE 9813KB and tried both pmt for dark current and then with the NaiTl. I took advantage of the fact that I could try my XP2020 without the crystal again but this time very carefully : I made a pin hole in the black plastic cover (provided with it when I bought it to protect the photocathode end) and watched the pulses on the oscillo. Then I lit my blue led torch (quite far away) and I could see bigger pulses. Pulses that were bigger than the biggest I could see with the crystal but still in the range of a few volts. That means that the pmt isnt damaged and indeed can translate the amount of light into a proper current pulse.

With the EMI THORNE pmt and NaITl I could obtain the same quality gamma spectrum than with the XP2020 so I first concluded that the crystal could be faulty in the end. But I remember I saw this post on this forum where someone showed the different spectrum He got with a good and a bad NaITl crystal : Even with the faulty one the dynamic was still ok ; only the resolution and amplitude were affected.

I have a LYSO crystal and I tried it on the pmt : also the spectrum obtained shows the same lack of dynamic range as with the NaITl. That means that in the end the problem comes from the electronics : preamp, USB sound card.

Enfin , I checked the Theremino PDF about how to use the USB sound card and it says that there are modifications to be made to make the sound card suitable for gamma spectrometry. Such modifications as to "change" the coupling 1uF for a 33nF. I only have a 10uF at hand so I tried to see on the oscilloscope what would happen if I swap the 10nF coupling capacitor of my pmt for the 10uF.

And guess what : most of the pulses are bloqued up to 98% My guess is that the impedance is too high with the 10uF. Could it be the reason why I only get short range energie gamma spectrum ???

In the Thermino file they dont explain why they this modification is required . They speak about adding another big cap but that because they want to use the 5V bias from the USB port to power up the PMT adapter ...

SO I'm going to see what happens if I change the cap....

You're right too when you say that the Fly swatter power supplies is noisy but all my power supplies are (even the one from rHelectronics thats meant for pmt). My recent try with RC filters seems to be working as I was able to lower the voltage ripple down to 250uV (so it seems).

To ber continued

[ChrisRadium2023]

YouCut_20230127_214256609.mp4

With 10nF

(1.05 MiB) Downloaded 347 times

With 10nF everything passes through!

YouCut_20230127_214146458.mp4

With 4.7uF

(4 MiB) Downloaded 316 times

With 4.7uF , pulses are so attenuated...no wonder my spectrum misses most of the energie

[Sesselmann]

Chris,

Are you referring to the coupling capacitor?

If so I recommend a 1M load with 1 nF coupling.

Steven