Photomultiplier Tube Destroying Amplifiers - Suggestions?

[C. Law]

Hi everyone, new poster here. I'm a chemist who's in a little over my head with an electrical engineering problem, and I'm hoping someone here might have some advice.

The short version (details to follow) is that my experimental setup has a photomultiplier tube (PMT) detector whose output has to be amplified, and our amplifiers keep burning out. We think we know what the problem is, and we think replacing the (very old) PMT will solve the problem, but that will take a couple months, and I'd love some ideas for temporary fixes that could help extend the life of the amplifiers and let me continue taking data.

---

So, here's the setup. The detector is a microchannel plate PMT with a -3kV DC power supply (the PMT's channel plates need to be held at large negative voltages to function), but its normal output is pretty weak; pulses of ~ -5mV amplitude, produced at random intervals of ~100us, with each pulse having a width on the order of tens of picoseconds. (Strictly speaking it's an electron source and not just a voltage source, but we have an in-line attenuator on the output to deal with that). This goes into a commercial inverting amplifier (20dB gain, 13dBm max input power) that boosts it to the level we need in order to trigger some timing electronics.

This setup worked fine for years (maybe decades? Since before I joined the project, anyway). Then one day an amplifier burned out. Okay, no big deal. Then they started burning out more frequently. Now we're lucky to get more than a day of use out of an amplifier before it dies, and the instrument that uses this setup is out of operation.

We think that the PMT, which is very old, is breaking down and allowing high voltage to intermittently arc to the signal output, dumping a huge voltage into the amplifier and killing it (if we're right about this, it isn't something we can repair, and long-term it means replacing the PMT, which unfortunately has a multi-month lead time from the manufacturer). We think this is the problem in part because it has gotten progressively worse over the course of a year, suggesting something is progressively deteriorating. We also think this because one of our attempts at protecting the amplifier involving attaching a splitter to the output, and after a couple days we discovered that two of the 50Ω resistors in the splitter had been burned out. It seems to us (and again, we're chemists and not electrical engineers) that if something as simple and robust as a 50Ω resistor is getting burned, the high voltage must be involved.

---

Does anyone have any insight into this problem? First of all, does our conclusion that this has to be crossover from the high voltage seem reasonable, or could something else cause this kind of damage? And secondly, does anyone know of anything that could be used to protect the amplifiers while still letting the small, short pulses that we want to detect through unperturbed? Some sort of protection circuit, or a really fast fuse, maybe?

We really only need a temporary fix, something that would allow us to use the instrument for the next couple months without spending thousands of dollars on amplifiers. I haven't been able to find a commercial component that seems like it would help, and none of the homemade protection circuits I've tried to kludge together with diodes have worked, but maybe someone here has experience with something similar?

If anyone can help, my praise will be unreserved and effusive. Embarrassing, even. Thanks!

 

[Bobbywhy]

C.Law, Welcome to Physics Forums!

In the Hammatsu PMT-MCP technical manual, on page 229, Section 8. 9. 3 Reliability, Part (1) Failure mode, there is this:

“Failure mode for photomultiplier tubes is roughly classified into gradual failure and breakdown failure. The main failure mode is gradual failure, which includes cathode sensitivity degradation, a loss of gain, an increase in dark current and a decrease in dielectric resistance. Breakdown failure includes cracks in the
faceplate, bulb envelope and stem portion, and also air leakage through microscopic cracks. Breakdown failure fatally damages the photomultiplier tube, making it permanently unusable.”

Also, on page 106, Section 4. 1. 3, Voltage -divider circuit, there is a schematic diagram of a typical MCP circuit. Shows 3 high voltage capacitors, anyone that has deteriorated over time may intermittently short and send a high voltage spike into the MCP, and then to the output, which may cause your amplifiers to fail.
http://www.coseti.org/www.boseti.org/pdf/hamamatsu pmt handbook ver2.pdf

One suggestion: you might open up the HV section and perform a complete visual inspection. Look for cracks in HV wire insulation, and for any deformed or discolored components. A simple remove/replace maintenance effort in this area may stop the damage to the output amplifiers.

Bobbywhy

 

[C. Law]

Thanks so much for the link, Bobbywhy, I hadn't seen that document before; it's very useful! Figure 7-28 on page 183 even has a diagram for a protection circuit to deal with exactly the problem I described.

What I've discovered, though, is that putting the MCP-PMT's output pulse through a resistor is killing the output. I can see pulses from photon events on a scope when the signal just goes through a BNC cable, but once I solder a resistor into the middle of the cable (even just 50 ohms) I don't see anything. Does this make sense to anyone? Is there some reason a fast electron pulse should be getting wiped out just by going through a resistor? It seems like this shouldn't be happening, since even aforementioned Figure 7-28 in the Hamamatsu calls for putting a PMT output through a resistor, but for some reason it's not working for me.

Thoughts?

 

[Mike_In_Plano]

I recall PMTs being (1) very picky about supply voltage, and (2) very sensitive to the voltage divider accuracy.

The key testing with confidence is a high impedance voltmeter. Common volt-ohm meters load the readings, and can result in your power supply voltage being adjusted too high.

Check for too high a supply voltage, and test the divider taps to be sure their divided equally. It's not uncommon for the resistors to increase in value or open over time. Don't flex or prod the resistors during test, or you may cover up the problem temporarily.

Cleanliness is essential in high voltage circuits. If your machine has been sitting downstream of acid vapors, you make have a built up contaminates that will throw off the divider voltage when the room is more humid. I've had great luck with breathing onto the suspect board with a drinking straw to prompt failure.

Best Luck,

Mike