Page 2 of 2

Re: Criticize my neutron detection setup

Posted: Mon Oct 23, 2017 2:32 pm
by Finn Hammer

Re: Criticize my neutron detection setup

Posted: Mon Oct 23, 2017 3:32 pm
by Silviu Tamasdan
Oh. I don't use Facebook so I can't see that page. :(
(or any other 'social networking'; I may have a Google+ account, comes compulsory of using gmail and/or an Android phone, but I don't know for sure and I've never tried using it).

Re: Criticize my neutron detection setup

Posted: Tue Oct 24, 2017 6:44 am
by Richard Hull
Likewise here. I have never used any of the common forms of social media. I fear it is a mixed bag of "here, look at me and what I am doing now" and also some valuable info., I am sure. I have chosen not to jump into the mixing bowl. I don't need to be bouyed by 5,432 likes or have 20,604 friend me's.

Fusor net is about the extent of my on line presence. My life is truly lived off-line, doing things. But enough philosophy.

Richard Hull

Re: Criticize my neutron detection setup

Posted: Sat Nov 11, 2017 3:11 am
by Rex Allers

This reply was stimulated by your post about TIER for one of these tubes.
Exercise TIER calculation for SNM-11 tube

Richard replied that he would expect to see more counts from the test set-up you have.

Earlier I made a reply to your BNC connector post.
Making a BNC adapter for SNM-11 boron neutron counters

In that, I mentioned the excellent paper by Bob Higgins about using these Corona Neutron Detector tubes. It is the best reference I have seen on the subject. You said you had read it in great detail, but then you said some things about voltages that make me wonder if you really follow how these tubes are intended to be used. You said,
"However the SNM-11 works best at 700V. In fact I've had one test where the tube worked correctly at 670V. I didn't try lower voltages.
In fact even other corona tubes such as the SNM-32 which are also said to work at 1500-1700V; my experience is that they are unusable at voltages above 1275V because they start oscillating at high frequency (and I've had independent confirmation from somewhere else that it's indeed their normal and expected behavior based on a sizeable sample of them). But they work well at 700V. "
The ~100 Mohm resistor in series with these tubes is critical. I was trying to point that out in the BNC thread.

Here's a table with specs for a good number of these Russian tubes.
Translated specs.pdf
(21.85 KiB) Downloaded 172 times
This was google-translated from Russian. The places where it says "crown" should be "corona". It has listings for both the SNM-11 and SNM-32. The SNM-11 is a Boron tube while the SNM-32 is 3He but both are expected to be run in corona mode. The letters SNM are westernized from Russian, I've seen CHM, CNM, etc. all seem to be equivalent.

These corona tubes are intended to have an internal light corona discharge all the time they are operating. The normal configuration has a 100 Mohm resistor (or maybe a bit more) in series with the anode (center conductor) of the tube. The supply voltage goes through this resistor, not directly to the tube.

The pdf table above has 3 key columns of values: Nominal operating voltage, Starting voltage, and discharge current max. The nominal voltage is the HV that is applied across the series connected tube and 100 Mohm resistor. Per the table, for the SNM-11 or 32 this should be in the 1500 to 3000 V range.

As this voltage is first applied, the tube appears as an open connection, drawing no current, so the tube will see the whole voltage. This starts the corona discharge in the tube, causing it to draw current. The current through the 100 Mohm resistor causes voltage drop across it making the tube see a lower voltage. The system finds a stable operating point with 600 to 700 V across the tube and the corona discharge drawing about 10 to 20 uA current. In the table, the Starting Voltage column gives the voltage across the tube in corona mode and the discharge current max column is the highest current that the tube should be allowed to draw in its idle mode.

The 100 Mohm resistor in series with the HV supply effectively makes a constant current supply for the tube in the 10-20 uA range. In his paper Bob did some calculations. He assumed the corona voltage would settle to the midpoint of its start voltage range or 650 V. He picked a supply voltage of 1800 V, so the voltage across the 100 M resistor is 1800-650 = 1150V, so the current through the resistor (and tube) calculates to 11.5 uA which seems OK and well below the max current. This is a ballpark calculation. The tube will find its own operating point but it should be close to this.

The 100 M in series also helps to quench the tube after a neutron pulse. The pulse draws more current making the drop across the big resistor greater and lowering the voltage across the tube a bit.

I think you said you bought your solution from someone else. The circuit feeding the tube needs to be roughly constant current, either via the 100 Mohm or some other method. If I was you I would try to figure out the circuit and ensure you have a very large resistor (~ 100 Mohm) from the HV supply. The specified operating voltage of 1500 to 3000 V is into this series resistor, so the voltage across the tube should settle to around 600-700V. If you want to measure this directly bear in mind that the effective impedance of the tube is around 650/12u or about 55 Mohm, so your meter should have an input impedance of 500 Mohm or more to not shift the reading much from what just the tube sees.

Here's Bob's simple diagram of the detector.
Corona block.png
He is coupling the neutron detection pulses from the anode through a DC-blocking 27 pF capacitor to the amplifier and comparator. He put the battery-operated detection and HV supply circuitry very close to the corona tube. Bob feeds the resulting pulses through a coax to some kind of counter.

If I understand your configuration, you may have the same general circuits but you are planning to run a coax from the tube to the circuits to carry both the high voltage and signal. It would look like this.
with cable.png
This may work OK but it puts the capacitance of the cable from the anode to ground. This may make it harder to get a clean pulse and might make the whole set-up more likely to oscillate. The connections to the cable here are high impedance so more susceptible to noise. I think you'll do better if you can put the front end circuit close to the tube and send the processed pulses from a low impedance driver through the cable as Bob did.

Bob Higgins' paper covers all this in more detail than I have provided, but your mention that 700V works better than >1200V makes me think you might not be seeing the distinction between the supply voltage and the corona voltage across the tube.

Re: Criticize my neutron detection setup

Posted: Sat Nov 11, 2017 2:17 pm
by Silviu Tamasdan
I understand perfectly well what you're saying, and I had much of the same concerns from the beginning. I plan on investigating exactly how the inner connections are made in the MCA I use - it is built like a tank though and I think it will be difficult to open. Nonetheless I'll find a way to do it non-destructively. I think that the voltage I'm setting in the software is the voltage across the tube only, not tube+resistor, and that would be in line with the tube specs and your post.

Re: Criticize my neutron detection setup

Posted: Sun Nov 12, 2017 12:47 am
by Richard Hull
I will say that I wonder at the max count rate of such a setup. Rex did a great job of explaining the purpose and need for the 100 meg resistor and its action in establishing the corona.

The reason for the wonder at the max rate is that this tube may have the same problem of early GM tubes prior to quench agents. The earliest of GM system demanded a 100meg resistor or better to achieve "quench", (end the conduction once the gas had broken down). Just as Rex noted, once the gas fully conducts, (creates a pulse), to end the pulse the increased voltage drop across the resistor kills the corona and the tube shuts down. Basically, the power supply voltage to the tube buckles once it conducts. This action creates an effective, fixed, recovery time before the next count based on the net RC time constant of the system. 100meg with any capacitance will slow the recovery time consant, limiting the upper level rate of counting.

With GM tubes, an alcohol quench was used in the earliest internally quenched tubes to stop the conduction, this allowed the ballast resistor in series with the tube to be reduced to under 10 meg with a nearly 100 fold increase in count rate. Unfortunately, the alcohol was chemically bound up and thus, used up, limiting the life of such early quenched tubes to about a billion counts. Modern tubes, since the 50's have a Halogen quench agent which molecularly, uncouples after quench and can live forever. Today only about a 3 meg resistor is needed in series with the best GM tubes.

Even if a quench gas is used in the Russian tubes here. The rate will forever be limited by the RC time constant. This should not matter in fusor neutron counting as the rates are low in all but the finest of fusors that go over the mega neutron/sec class.

Richard Hull

Re: Criticize my neutron detection setup

Posted: Sun Nov 12, 2017 10:45 am
by Silviu Tamasdan
I think I read somewhere that a halogen is used in the corona boron-lined tubes. Don't ask me for a quote becaise I don't remember where that came from; I have read so much in so many places about them. OTOH I don't think a quenching agent is used in the helium corona tubes.

Re: Criticize my neutron detection setup

Posted: Sun Nov 12, 2017 6:25 pm
by Richard Hull
Like I said, even with quench gas the time constant will limit the rate with the 100 meg resistor as it is used as a ballast much like the 1920's GM tubes.
Chemical quench in the GM tubes was used to get rid of the early 50-200 meg quench resistor that limited the count rate.

Richard Hull

Re: Criticize my neutron detection setup

Posted: Wed Nov 22, 2017 6:56 am
by Finn Hammer

One thing about your data puzzels me, because they are a couple orders of magnitude away from my own data. I seem to have an output in the 1-3 volt range, whereas your setup appears to output in the 10 - 30mV range.
So you trigger in the low tens of millivolt, whereas I would be comfortable to set a trigger level around 1.5V.
Operating in a noisy environment, a high signal level is an asset, so I would like to ask you this:
Do you have any explanation of the low output from your tube? This low output could be caused by the resistor and capacitor network that is connected directly to the end of the tube, it looks like it is covered by just alluminum tape, any chance you want to have a peek under it?

Cheers, Finn Hammer