Pulses out of SNM 18-1 (at last)

This is bound to get mixed up with things in Electronics, check both. Physics-specific stuff here, mostly.

Pulses out of SNM 18-1 (at last)

Postby chrismb » Thu Aug 25, 2011 3:41 pm

I bought a couple of 18-1 tubes some time back. This was earlier last year - before these 'corona tubes' got much of an outing in the forums, and I wasted a considerable amount of time trying to get the *** things working - and failed!

All this talk recently encouraged me to look closer. Well, with a new oscilloscope I bought in the intervening time period, it picks out the pulses! Cool!

snm18-1_pulse.gif


Both 18-1 tubes work, giving around 4 to 10 pulses per minute background (little groups of bursts distort the minute-by-minute statistics, it seems). I also have two SNM-17 tubes, I tested one that seems to give only one pulse every several minutes, but bear in mind these have no moderation around them at all, so we're looking at direct cosmics, here.

I ran the tube with 100Mohm serial (seems to work fine - is there any benefit of the higher serial resistance?) and 100k to ground (this gives bigger pulses but doesn't actually improve the S/N that much). Whatever resistors I choose, the pulse seems to be around the 10pC worth of charge. I guess in previous attempts, my circuits were simply suppressing so little charge and I didn't notice the pulses just didn't become 'visible', and got swallowed up in noise.

Now I turn to detector circuits... oh boy!!.. I went around this loop for ages - dozens of 'man hours' before already to no success. Similarly today and yesterday I've spent around 12hrs experimenting with damned circuits. Yes, Doug, I tried yours early on, at least with the transistors I have available. Maybe that is critical. But I found out something else, which makes me hesitate on your circuit. In one test, I manually adjusted the bias on a sensitive signal transistor to give a response. The circuit was set up to be 220kOhm input impedance. I did get signals, but it was significantly missing many out, it was only picking up the very biggest. So even with that high impedance, the signals still got pulled down.

I tried further and further, but the more I tried to push up the impedance of the detector, the further I seemed to get from detecting those pesky 10 pC's!

I don't know if the output on these tubes is a bit different to the ones you are running, but I am also sure I'm not skilled to make your circuit 'sing'. So I am planning on going to buy some JFET opamps and see if upping the input impedance gives me a more suitable circuit. I've spent too long now with the parts I have to try carrying on with them, so I'll try some new components, see if they give me the 'answer'. The cheaper JFET opamps can be found with 10^12 input impedance and 3MHz response, so I'll go for that level of spec first.

Here's all I have to show for a day's worth of 'effort'...
snm18-1_detector_attempts.gif
chrismb
 
Posts: 620
Joined: Thu Aug 05, 2010 6:32 pm

Re: Pulses out of SNM 18-1 (at last)

Postby Joe Jarski » Thu Aug 25, 2011 4:16 pm

Chris, good to to see you finally got 'em working! I've been struggling with my B10 tubes, though I haven't spent much time on it since I first hooked everything up. Waiting for the random N to enter the detector when you aren't sure that it works is kind of tedious, but it looks like you're getting a decent number of CPM there.
User avatar
Joe Jarski
 
Posts: 231
Joined: Thu Sep 16, 2010 8:37 pm
Location: SouthEast Michigan

Re: Pulses out of SNM 18-1 (at last)

Postby chrismb » Thu Aug 25, 2011 4:40 pm

Hi Joe, yes my SNM-17s are B10, I think they are the slightly longer versions than the ebay ones, but the same diameter. If they are the same, but for being a little shorter, then seeing now what mine do, I'd say you'll only get one count every 10 mins or so. I just put the scope on 'single' trigger and left it for half an hour, because nothing showed up while I was sitting there watching it!! As interesting as watching eggs - anticipating they might hatch!...
chrismb
 
Posts: 620
Joined: Thu Aug 05, 2010 6:32 pm

Re: Pulses out of SNM 18-1 (at last)

Postby chrismb » Thu Aug 25, 2011 6:01 pm

Just looking back on the posts for the other Russian tubes, gee.. the pulses from mine are (at least, as the scope sees it) muuuch longer. The trace, above, is set at 100us/div, whereas on the other posts the scope says 5us. Am I misreading this?
chrismb
 
Posts: 620
Joined: Thu Aug 05, 2010 6:32 pm

Re: Pulses out of SNM 18-1 (at last)

Postby Doug Coulter » Thu Aug 25, 2011 6:44 pm

Chris -- yes, the transistors are important to that circuit, and if you substitute other types, you'll have to readjust the bias resistor (at least). The input pnp is a very low noise, very high gain type (but still cheap at most suppliers -- ). Obviously, if the circuit is biased so a transistor is all the way off or on - no output. It did take some fiddling to get the bias so the output DC resting value would stay in range over a couple of volts of power supply range -- that's why the big emitter resistor in the NPN, to cut down the DC loop gain some so it would work over the life of a 9v battery. The output collector has to be somewhere in the middle of the supply volts for the thing to work right. IIRC it was 4-5 volts here when tuned right (you have to check that there's a couple volts CE on the output transistor, because it could have a collector voltage, but the emitter volts are the same in saturation). A rising supply voltage turns on the pnp more and that turns on the NPN more and if it has zero CE voltage, well, you're not going to get much out of it....If you substitute a crummier pnp for the input (390x) then you'll have to fiddle with the bias feedback resistor to get things in range. The one specified is a "hot mamma" and has very high current gain (300-500) under the conditions. The bias is set up for lowest total noise with the input impedance. If you've got the circuit biased up right, touching the input with a finger should produce a full output wave (3v or so pk pk) at mains frequency or the local radio station.

As a side note, only a pretty experienced analog/rf designer should try substitutions in circuits I provide. I tend to take full advantage of the parts specced, and the "wires and parts that don't need to be there because I'm clever and used this to cancel that, etc". This circuit for example way outperforms the "not quite as simple charge sensitive amp" they went on and on about over at fusor.net -- with about 1/10 the parts. That's not accidental - 40 or so years practice with low level high speed analog design here.

You're likely seeing longer pulses because at the higher R, with the same C (the tube itself plus wiring) -- longer time constant. Takes a long time for the current through 100 megs and almost no volts to recharge the cap.

Not sure what type scope you have there, but mine (GW instek) when set up with "normal" triggering simply saves the last trace till the next trigger. For this low-rate stuff, it's a real advantage to have that, you get to see what happened. A regular analog scope isn't so great for this. I've also been able to use really slow sweep speeds, by putting it in "peak detect" mode. That keeps the sampling rate real high so it can show very skinny pulses even with a 10 second or longer sweep time -- or the cool "roll" mode. That alone is pretty good for testing your physics and tuning, as you can really see the rates as you tune things easily as they roll on past.

Putting a pretty hot rock up to the tube (or other rad source) shouldn't make any more full scale pulses, but you might see some more "grass" on the baseline. In fact, you should if things are working right. In use, you set your threshold to cut that, but pass the "real" pulses.

I have a long b10 tube too and it works fine here. It says it's a CHM-11. I actually like the B10's better when ultimate sensitivity isn't key, because they use a better gas fill to make a counter work. With 3he, things are a lot trickier -- they have a tendency to become a relaxation oscillator, like a neon bulb would -- and this gives a bunch of spurious pulses if things aren't all just right. This might in part explain a longer pulse time.

There's always some moderator around -- the air, the bench etc...

MMBT5087.pdf
2n5087 data sheet
(100.03 KiB) Downloaded 745 times


As you can see, that's not a garden variety switching transistor. These were quite popular for ultra low noise phono preamps back in the day, nothing beat them for medium input impedance work. You can trade off voltage vs current noise with collector current (bias). Your basic switching transistor need not apply here, not enough DC gain, too much noise, lousy low level performance.

Looks like your tube is not only a long one, but fatter. Mine are all 3/4" (about 19mm).
Posting as just me, not as the forum owner. Everything I say is "in my opinion" and YMMV -- which should go for everyone without saying.
User avatar
Doug Coulter
 
Posts: 3515
Joined: Wed Jul 14, 2010 7:05 pm
Location: Floyd county, VA, USA

Re: Pulses out of SNM 18-1 (at last)

Postby chrismb » Thu Aug 25, 2011 7:02 pm

Yes, the SNM18 size is about twice the width (and twice the working length of the 'shorties'). It's got the same volume as a 22"/1", in fact. Arguably a better length dimension that 22", for being able to keep the 'counting' area closer to our 'small' devices.
chrismb
 
Posts: 620
Joined: Thu Aug 05, 2010 6:32 pm

Re: Pulses out of SNM 18-1 (at last)

Postby Joe Jarski » Thu Aug 25, 2011 7:13 pm

Ugh, keep it up and you guys are gonna to convince me to get one of those new fancy little 'scopes with the color display!
User avatar
Joe Jarski
 
Posts: 231
Joined: Thu Sep 16, 2010 8:37 pm
Location: SouthEast Michigan

Re: Pulses out of SNM 18-1 (at last)

Postby Doug Coulter » Thu Aug 25, 2011 7:48 pm

Let me recommend a GW Instek then. I have two and love them both. The "low end" one is two channel, 25 mhz, GDS-1022. The "high end" one is 4 ch, 2.5 ghz (sample rate) GDS-2204. I believe the local distributer is in Mass -- I can go look him up, he's a good guy. Marlin P Jones also has the low end one for about $400, but I think the disty is cheaper. Both will save screens on SD or USB sticks (depends on the model which), print to a USB printer (haven't tried that) and best of all - talk to a PC over usb, and you can do *anything* from the PC, including fast screen grabs and setting up all the scope parameters. There's free (of course) linux software for all that (which works right) as well as the very crappy/crashy windows app that comes with them.... The linux software also lets you suck out the trace data as ascii numbers so you can munch on it further or use something like Gnuplot to make pretty-plots on the data.

It's like finally having a storage scope that works right, by default. In normal trigger mode, they hold the last sweep till the next. Both do FFT's on the signals and other math if you want. They really work as well as the better Tektronix scopes -- trigger is especially nice, lots of options and noise filtering if you want. And unlike an analog scope, you can set it so you see the stuff leading up to the trigger, neat. Although I did have to close my eyes and hold my nose to pay for the high end one, I'd not turn it loose for anything. Even the probes are high quality. The cheaper one is now built into my tinker bench kind of permanently and I've not particularly noticed a problem with speed for what I do there. But it sure is nice to have that Ghz one too. I got a 100:1 HV probe from Marlin for these too - it's nice to have one that's hard to blow when working around HV...and 100 meg input impedance.

The only time I fire up my older teks now is when I need real good analog XY plots (like from an MCA that has that type output). The digitals don't do that quite as nicely and don't have a Z input to make pretty graphs.
Posting as just me, not as the forum owner. Everything I say is "in my opinion" and YMMV -- which should go for everyone without saying.
User avatar
Doug Coulter
 
Posts: 3515
Joined: Wed Jul 14, 2010 7:05 pm
Location: Floyd county, VA, USA

Re: Pulses out of SNM 18-1 (at last)

Postby Joe Jarski » Thu Aug 25, 2011 10:25 pm

Thanks Doug. I'll have to think about which way to go on the scope, although the low end might get me back in business quicker. Right now I'm unable to pick out any pulses that I would consider "real" - probably similar to what Chris has been experiencing. And that was using 2 different scopes, one was was a storage scope, both with and without the preamp. I can get the trigger set so that it fires at about the right rate for background, but nothing identifiable in the trace. The corona noise, about 2mV worth and ~5mV of ripple from the HV supply are easy and the onset of the corona noise is right at the 700V in the datasheet. So, everything is "right" for it to be working, I just can't pick out the signal with them being so far and few.

Edit:
I don't want to get off topic, so maybe this deserves another thread... looking at the GDS-1022, it has a 250MS/s sampling rate and the pulses on your scope captures are ~5uS, so assuming that I followed the hook-up in the tube datasheet then there's a pretty good chance that I still wouldn't capture the pulses, or at least only a small percentage, right? I don't know what Chris' sampling rate is, but he may have been *lucky* that he used a 100M resistor to stretch out the pulse?

I guess I need to be looking at the higher end scopes for this sort of thing...

Edit2:
I finally got some real pulses spaced anywhere from 5-10 minutes apart (with a moderator). Since my old storage scope doesn't want to show the first little bit of a trace after it triggers, I managed to pick up the trailing edge of the pulse using a 1us sweep which gives me a nice distinctive arc all the way across the screen even though I miss the peak. Now back to the preamp.
User avatar
Joe Jarski
 
Posts: 231
Joined: Thu Sep 16, 2010 8:37 pm
Location: SouthEast Michigan

Re: Pulses out of SNM 18-1 (at last)

Postby Doug Coulter » Fri Aug 26, 2011 8:23 am

You'd see it fine with either scope. Say the sample rate is 25 mhz. That gives you 125 samples during a 5uS pulse -- which if displayed direct, would make the pulse about half the screen wide with a lot of detail on the pulse shape. If you use the "peak detect" mode with a slow sweep rate, the sample rate stays high, but more than one sample is shoved into each horizontal pixel width, in a min-max way, so you can still see one pixel wide pulses, even with ridiculously slow sweep rates, like 10 seconds/screen. I like the roll mode for this a lot. You just see the last N seconds scrolling horizontally, and the pulses show up as 1 pixel wide vertical lines off the baseline, and are very easy to see going by. I've had no troubles with the far shorter pulses from our NaI detectors as well and even with phototube/plastic signals (which are as fast as it gets -- a few ns). Don't know the specs on Chris' scope, but he's probably going to be able to do the same.

The fact that these, by default, keep the last screen up while waiting for the next trigger is a real boon with these "rare" signals. (They're not so rare once you get some D and some HV into a tank almost no matter what the configuration there).

What I see is that the real neutron pulses are well out of the background noise -- several times that is the norm, but not all the same height. This is because the n+B10 reaction may or may not throw all the reaction products straight toward the center of the tube, or into the gas at all, but some might just go sideways and bury themselves back in the tube wall before ionizing much gas.

The effect happens to an extent in any of these, even ones where the tube gas is the detection medium (3He or BF3). In use, we just set a threshold so the noise doesn't trip things, but any pulse well out of noise will. Since everybody's noise will be a little different, everybody's threshold will be too, and the net counts per neutron flux will also be different - you'll miss greater or fewer low level pulses depending on the threshold. A standard neutron counter is going to be a lot tougher to get to than what we're doing with the geigers, and this problem shows up all over the neutron literature -- these things are at best somewhat relative compared to the activation/measure method which CAN be calibrated pretty closely, but is only useful once you have a lot of neutrons. So, the idea is we use each tool for what it does the best. For tuning a fusor, a relative calibration is fine as long as things don't drift during a session, and immediate feedback on what makes it better or worse is really valuable. Later, once you've got a "tuning" and things are stable you can then do absolute measurement with the activation, and then have a number you can use to have at least a sloppy absolute scale factor for your tube detector.

By doing it that way, you can eventually come up with a fairly good cross calibration between the two. It won't be perfect as the time history of neutron exposure on something very short half life like silver[1] will vary the ratio of total tube counts to resulting activity -- if your neutron source is "hot" at the beginning of a run, but not so hot near the end, you'll get lower readings off the activation than otherwise, for example. Indium is much better than silver for that, but a lot more "numb" when the total neutron count is marginal for the technique.

All these tubes show me a few hundred CPS on a good fusor run, the exact numbers depending on distance to the fusor, the tube type, moderator, etc, but they're all in that range, which is also just right for audio monitoring -- and you don't have to look at it while your eyes and hands are busy elsewhere tuning fusor parameters. I like that as an adjunct to any other counting and logging as it gives you information that's lost in that (things like bursting over short periods).

[1] Silver is quite weird. Not only can it be activated, there are multiple metastable states of nuclear excitation, with half lives from a second or so on up. Since the counter can't tell gammas from betas, it sees all of that, and the decay curve is very steep right at first from all the short life stuff. This makes the neutron flux during the very end of a run the most important to the final reading. The often reported 12 minute half life only shows up later and is closer to the noise (background count) -- at first the counting is dominated by all this short stuff. Indium has a lot simpler single decay time, but it's longer which means fewer CPM for the same level of activation, so it's more useful once you get your source good and hot, and less sensitive to the time-history of neutron exposure. We'll be looking into some others -- manganese might be interesting (and cheap and available in primary batteries, as MnO2). So far gold hasn't gotten hot enough here to be easy to use -- that long half life means to count the same percent of activation takes longer (during which time you also get more cosmic counts).
Posting as just me, not as the forum owner. Everything I say is "in my opinion" and YMMV -- which should go for everyone without saying.
User avatar
Doug Coulter
 
Posts: 3515
Joined: Wed Jul 14, 2010 7:05 pm
Location: Floyd county, VA, USA

Next

Return to Metrology

Who is online

Users browsing this forum: No registered users and 3 guests

cron