Gamma Spectrometry Forum

Hi all,

Since recently I was only working with negative HV supply for the PMTs of scintillation detectors. Reason: lack of a positive HV source. I do have detectors with "housing at cathode potential, use positive HV only" written on them... and I did operate them with my negative supply and it works great (I get good resolution), but they are annoying to use because I cannot touch them while HV is applied. (For those who are concerned about safety: This is not dangerous if the power supply is a weak (not more than around half a mA) and as long as there are no large capacitors involved! )

Recently I bought a power supply that can do both, +HV and -HV, and I immediately started building a positive divider. But I quickly realized that It is not as easy as expected. When I saw the first signals on the oscilloscope, there was huge noise ~10mVpp present. First I thought my new power supply was bad, but I measured the ripple voltage of it and it is comparable to the one of my negative supply.

I realized that the reason is the topology of the voltage divider for +HV. For negative HV, the HV-ripple is far away from the anode and the anode is not connected to the rest of the resistor network at all, while for the positive HV, there is a direct path from HV supply to the readout capacitor.
First attempt to get clean signals: Put a HUGE filter between supply and detector to remove the supply ripple to a negligible level. I tested this and it works. But as I've written above, I would like to avoid large capacitors in connection with HV, even if the HV parts cannot be touched.
Second attempt to get clean signals: Put a high pass filter between detector and oscilloscope, I just tried and it was a fail (maybe the filter needs a bit more tuning)... but why invest the time if I could just use negative HV??

So I would like to ask the community: What are the pros and cons of positive/negative HV for PMT bias supply? This is my list so far, but I'm sure you can add to the list

negative HV:
+ signal doesn't need to be going through a capacitor (no baseline shift at high count rates) -> allows to use time-over-threshold technique
+ high voltage ripple is far away from the signal (makes it easy to get low noise levels on signal)
+ need fewer parts to build a voltage divider
- cannot safely operate detectors labeled "housing at cathode potential, use positive HV only" (Serious question: why detectors are built like this?)

positive HV:
? feel free to fill the list :)

Cheers
Michael

There is a good explanation from Hamamatsu. There is a very similar text by EMI. Peter

Michael,

Detectors with PMT's wired for negative HV are relatively uncommon. Hamamatsu seems to be the one who promotes this scheme, but for all the thousands of GS unit's I have made, only 2-3 customers ever asked me for negative bias. When I was making the original GS-1100A it was easy to flip the diodes the other way and turn it into a GS-1100A-N but unfortunately the new GS-USB-PRO can't be flipped around easily.

When it comes to making the detector, it's simply a matter of applying a conductive coating all around the PMT which in turn floats at cathode potential, and then another layer of insulation so the housing remains at ground potential.

I can see benefits by being able to connect the anode directly to a preamplifier without an RC coupling, especially if you want fast counting, but in our case we are sampling at relatively slow rates, so the pulse would in any case need to be stretched.

I haven't seen any direct comparison experiments, but would love to see someone do this.

Steven

Thanks Peter and Stephen for the information

I didn't know about glass scintillation.

I did a ltspice simulation of my positive divider circuit an found out that the reason for the noise that I could see was probably the high input impedance of the oscilloscope I used to look at the pulses. I think if I would put the signals into a preamplifier with ~10 kOhm input impedance, the noise should disappear (I didn't have time to test it in real life yet).

A direct comparison of positive and negative supply with the same detector would indeed be interesting. Maybe I can do this once my positive base is working fine.

Michael