Bad 3x3? or is it just me?

[iRad]

Mike - For testing purposes, an even distribution is fine, but you still may not get optimum results for other reasons. Definitely get rid of the ground to grid connection.

In practice on most standard 10-stage pmts, the resistor between Cathode p14 and GRID p13 (if there is one) should be double the remainder of the resistors in the chain. In your case, a 2M from 14 to 13, then 1M the rest of the way to anode.

In my experience, a 1M base value VD could also be a bit of a load on a GS, especially for a detector/PMT that needs higher voltages (not a problem for most good lab HV power supplies). For use with a GS, I recommend 1.2M minimum as a base resistor, which would put a 2.4M at the K to G connection if using my 1.2M minimum recommendation.

[iRad]

Mike -
Here is how I handle the ground to detector casing on a Bicron 3M3.

[Mike S]

Thanks Tom!

That's a great way to do the grounding, I'll implement something like that if I go with a more permanent solution.

I did two short comparison runs with and without a ground strap (For testing, a 1/4" braid from BNC connector body to detector body held in contact with rubber bands). There was a 20% increase in counts with the grounding, and it appears they mostly were at the low end (good counts, aligned with a peak). That's more evidence that at least some of my problems are noise related.

[Mike S]

Thanks Everyone!

We're getting somewhere now. Here's what's been done, and the much improved spectrum is attached:

1. Added grounding strap so detector case is connected to signal/HV ground.

2. After changing the Grid connection on my voltage divider and getting slightly different, but still weird results, I switched back to the borrowed Teledyne PMT base, which exhibited much lower noise. [It's possible that I wasn't getting good connections to the PMT the first time I tried it since the contacts don't grip that tight and may have been slightly oxidized. Just a theory.]

3. Realizing the GS-USB-PRO has a 1M resistor on the HV output and the Teledyne base is only a few Meg total, I figured I was getting much lower than indicated voltage to the divider, so I cranked it up a bit more to 1100V indicated.

4. I increased the "max slope" and "Max noise" parameters in the "baseline test" used for pulse filtering in Theremino MCA. That, and the increased voltage, helped give some low end response.

The result looks pretty reasonable. I think there's more improvement to be had, but I don't think the problem is a bad detector. I'm still hoping I can get a some more low end response.

The attached spectrum shows one of the "better" previous runs in red. I think the low energy peak on the red trace was probably just noise. Note how even the K40 peak looks better. The red trace was 14 hours, the green trace was 5 hours. Both have the equivalent IIR filtering applied. For reference, this Trinitite sample is about 2300 CPM with a pancake. Shield is 1mm copper, 2mm tin pewter, 15mm lead on bottom and sides, with thin partial lead shielding on top.

Thanks again to everyone that pitched in with ideas, knowledge, and help to get me this far.

--Mike S.

[Mike S]

To start to isolate and understand the factors that contributed to the improvements, I did a comparison run with and without the ground strap to the detector case. The difference is clear to see.

Note the big peak about 8 keV is just noise, but I think it helps illustrate the point.

The red trace is with the ground strap, the green trace is without. I didn't move or change anything other than the strap (I did slightly adjust the energy trimmer to keep the Cs-137 peak aligned due to slight temperature change between the two tests).

[Mike S]

I think I'm ready to say the detector is bad.

After taking a break from this, I'm trying to wrap-up and document the detector issue. I'll share what I think I know, and I'd appreciate any feedback or ideas. The next step will be asking the seller for a refund and see how that goes.

What I've done:
- Bought another voltage divider so I have now tested with 3 voltage divider bases and all give similar results.
- Bought another detector from the same series (a 2"x2" 2M2/2). The specifications on the 2x2 are not as good, plus it's used, while the 3x3" detector is new (surplus). Even then, the 2x2 performs noticeably better than the 3x3 using the same divider base.
- I contacted Saint-Gobain to find out exactly what PMT was used in the 3x3 and 2x2 detectors to make sure the voltage divider was set up correctly. The voltage divider used for the tests below is optimized for the 3x3, where the 2x2 recommendation is slightly different (advantage to 3x3).

Both the 2x2 detector and the 3x3 PMTs should be at nominal gain at 850V, but the 3x3 just gives mushy results. I ran the voltage up to 1300V to see if it improved. It does improve slightly (but still not "good"), but there's little improvement between 1200V and 1300V, so I didn't go any further.

To show that it's not something else in my setup, I have test runs from 3 detectors done with the same hardware. The 3x3 and 2x2" detectors used the same voltage divider, the 1.5x2.25 detector has a built-in divider. All used the same cable, were placed in the same tubular shield (open both ends) and the Cs-137 source was oriented roughly the same (distance adjusted to keep count rates in the same general range). Typically I'm seeing roughly 18% FWHM for the 3x3, 12% for the 2x2, and 7% for the 1.5x2.25 (always a great little performer).

One more comparison: Some may remember my concern with a well-used 5"x4" well detector. That detector works, but has a cloudy and cracked crystal, plus some delamination between the crystal and the light pipe. In addition, as a big well detector, I would expect worse FWHM compared to a solid 3x3" detector. This is more of an informal comparison of spectra of the same Trinitite sample done at various times, but you can see how the marginal 5x4 detector it still outperforms the 3x3. [Note the 2x2" detector is different in this comparison (a Ludlum 44-10)]

Can I call it? It sure seems this 3x3 detector has something wrong. It is supposed to be new old stock, and it looks new, but I think something must have happened to it in storage or shipping.

Mike S.

[iRad]

I just got a shinny brand new looking 2M2 in the mail on Friday... dead as a doornail. Seller claimed it was new, pulled from an unused porotype instrument. Unfortunately these things are very fragile and stuff happens, even without signs of abuse. Although not the only way, thermal shock can kill the performance of a crystal, while still leaving the detector looking great on the outside. As far as my recent 2M2 purchase... it's most likely a cracked PMT, even though there is no rattle inside.