My planned detector topology & Voltage regulation for HV.

Linear and non linear

Re: My planned detector topology & Voltage regulation for HV

Postby chrismb » Sat Sep 24, 2011 5:20 pm

So, with the board I have now made up, as per other thread, attached to another (on uncopper-ed perf board) with the HV bits on, here is the tube socket and preamp stage, as envisioned above:

P9240806s.gif


I decided I needed a bit more space than the 110x90 type size of the smaller project boxes, so went for 120mm x 120mm, and when I looked at what RS had there was a h=80mm box, so I decided I might as well use the height by installing AA battery boxes. So these are currently loaded with 5 x eneloop 2Ah, which should supposedly give 1000hr duty before a recharge (and 100hr once similarly installed in the second box with the HV supply in, where I'll use the same battery scheme).

...I am planning to make a second one of these to the same 'plan', so any comments for improvements/extra features welcome.
chrismb
 
Posts: 620
Joined: Thu Aug 05, 2010 6:32 pm

Re: My planned detector topology & Voltage regulation for HV

Postby Doug Coulter » Sat Sep 24, 2011 5:40 pm

Make sure your CCFL or any AC carrying wires from it are fully shielded. A box inside the box is not too much (and might not be enough if the wires coming out of it have current pulses on them). I'd add an analog output along with your thresholded one, so you can know if you've got drift or other issues. I usually do a lot more testing before I assume some TTL output is good to go than you did. Lots of things can go wrong...a scope output at analog level is nice to have to know about that instead of blindly trusting some sudden vast increase in counts as real neutrons that was really noise or oscillation or drift that puts the intrinsic noise above threshold.
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: My planned detector topology & Voltage regulation for HV

Postby chrismb » Sat Sep 24, 2011 6:02 pm

That's what has taken the 'weeks' of work - testing, testing and more testing, so it is right first time I build it.... I started with the HV PSU from the crate which gave me the stable supply I needed to get to grips with the tube's signals, but once I had established these basics I have moved on. The ccfl supply has already displaced this and has been the 'in-the-loop' supply for the last few weeks of test.
chrismb
 
Posts: 620
Joined: Thu Aug 05, 2010 6:32 pm

Re: My planned detector topology & Voltage regulation for HV

Postby Doug Coulter » Sat Sep 24, 2011 7:27 pm

Good show. Now all you need is a couple weeks of runtime with a known stable neutron source.... ;) At differing temperatures. Nah, being only half as hard on you as fusor.net would be :lol:
For example, see my next post, which is going to be on the standard counter available thread. Looks as if my sensor (the hornyak) has something like "phosphor fatigue" which is why the silver still rules the roost. Just did a great run, got the silver really hot, but look at the neutron counter output graph and log...something not quite right there?
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: My planned detector topology & Voltage regulation for HV

Postby chrismb » Sun Sep 25, 2011 5:38 pm

I have one stable source - terrestrial background (notwithstanding the occasional cosmic 'surge')! The issue for testing a measuring device is to have two (or more) sources of known difference, so you know your not measuring something random that just happens to look like the signal you expect!!

In regards the point about noise/oscillations, it's actually the reverse situation and I'm more confident over the ttl discriminator output I've designed into the circuit than the analogue. This is because the discriminator stage has a slow response, for various reasons such as an anti-bounce cap I have included in the circuit which prevents the response doing anything quicker than a 10us time scale. This is why if you look at my scope plots there is a 'signal delay' between the tube pulse and the output.

The output doesn't appear to respond to anything shorter than around 10us (combination of that cap and just the slower signal response of the LM311). But also the signal coupling cap from the tube means it doesn't respond to much longer than 200us. So there is not only magnitude discrimination, but also a temporal one. This means that the circuit appears never to really 'feed back' on itself - it just doesn't look quick enough to get feedback. (I did tests where I deliberately caused feedback oscillations by reducing the coupling to ground so that the preamp stage could begin to float wrt tube and be driven by any prevailing interference in the circuit, and, overall, the circuit still damped itself and settled back down again after a sub-second period of oscillations.)

If I were to take the analogue output directly and try and count that, I would also pick up any old feedback and interference that doesn't come through the discriminator stage. You can see that on some of the scope plots where EM noise can be seen as a bi-directional pulse, whereas tube responses are only -ve going. The output doesn't respond to those bi-directional signals very much. (It can happen if, for example, there is a 10% 'duty cycle' noise for 100us, or 50% for 20us, but noise and feedback is typically sub unit us, and signals longer than 10us don't radiate well within the box to cause feedback.)

I have also done microphoning tests (going around tapping everything/HV cables, etc..), and sparking-emission tests (a spark gap firing close to the kit, routed via a long unshieled wire). The temp is going to be pretty stable and is right in the middle of optimum whre the electronics are designed to work, so I don't see any reason for temperature tests. The only other concern I have, which is true for any electronics (but more so for a hand made board without production-standard cleaning processes) is humidity, especially as it only uses a mA or two so there is no 'self-heating' at all. Any residual ionics from the soldering could produce non-trivial changes in resistance, but, again, the impact is mainly going to be a change to the gain of the amp stage, so I would expect to see the background rate drop off markedly if it affected the gain significantly, or otherwise I would get continuous counting if, somehow, the noise floor reached the threshold, which is extremely unlikely as I have set it quite high (so high that I lose ~20% of the 'legitimate' counts from the tube).

I rather suspect that I have done quite a lot more testing than most would ever do with their bought-in 2nd hand kit. The difference is that I now know intimately the behaviour of this setup so I can recognise something as 'anomalous' or 'normal' very easily. I think this is usual for anyone who's created their own circuits - being able to spot anomalous behaviour is one of the key benefits of home-brew electronics over buying something one supposes is 'ready' made for a job.
chrismb
 
Posts: 620
Joined: Thu Aug 05, 2010 6:32 pm

Re: My planned detector topology & Voltage regulation for HV

Postby Doug Coulter » Sun Sep 25, 2011 6:45 pm

You can of course clean your soldering up with various solvents - I do when high impedances are involved. But a little work with "normal good practice" built stuff, then finding out how much the noise of a fusor can affect that made me not-mad at how much crap I got on fusor.net about my measurements at first. They were right! I proved it to myself that they were, though I didn't like them forcing people into buying BTI's (not cheap, and die young), and them mentioning that some cargo-cult gear and 3He tubes would be trusted, but that the same guy saying that mentioned he'd cornered the market, but none were for sale at any price. That really made me mad (and is part of why this board exists).

The sparking test is the one, though you must consider DC drift up to and including any comparator as moving the threshold can let in noise, or remove real counts. Usually that's minor compared to a noise burst from an arc someplace else in the gear suddenly counting your counter up 10k counts - and you'll never know if you weren't watching it all the time (or do like std counter does, report every second so any outliers are obvious). The instant you go digital you lose all kinds of information. For example, the analog waveform, if monitored, can often show you things about what's real and what's not. It all depends on the setup of course. A "spark" vs a 50kv 1 amp arc...not quite the same thing, but at least on the same page.

I was just down running my fusor for something -- testing a new detector, actually. It's totally shielded, tight as a drum (literally airtight as well) -- all coax to the scope and digital counter, all things floated off ground except for one point -- separate power supplies, the works. Yet, a puny 20kv supply in an ion source with the ion source going in and out of "light on" was enough to make 5v positive pulses show on the scope out of something that doesn't even have one single positive power supply in it vs ground! And yes, they looked "different enough from the real thing" to be obvious on the scope, but not to the counter, which happily counted on overshoot in the "expected" direction. So, silver still rules, nothing else is even close. This other stuff is fantastic for tuning things, taken for what it's worth - don't know how I'd have got along without it. But in any case of disagreement, or any new claim of huge improvement, if I don't see silver counts, I remain a complete and thorough going skeptic, for good reasons -- I've been fooled myself too many times, and with stuff built "perfect" that some way, some how, I had a ground loop, an accidental antenna that could get enough millivolts on it to fake the millivolt sensor signal, you name it, it's happened.

The natural background is for all intents and purposes zero -- cosmics are many thousands of times higher than that (guessing, it might actually be in the millions or billions or trillions). I have two neutron sources I made with very hot staticmaster Po sources I laid Be foil over -- should be making 5-10 neuts/second by theory. Nothing can see them (other than a BTI and 48 hours getting about one more bubble with than without), maybe I could prove it with a 3He and some fancy statistics and long term counts. You definitely cannot tell by ear or by hand counting the counts. I was messing with you - I know you don't have a neutron source that's EMI free hot enough to be seen on one of these, almost no one does, they don't grow on trees. In fact, your RF driven thing might be the first practical one to exist, once it works. A mere few hundred watts of RF is nothing compared to what you can get in an arc during the peak -- there it's not impossible to break into the 10e12w region, and I've done it.

You really only need one source to test, preferably of known strength. You only need two to set scale factors.

You''ll just have to keep watching the analog, not that you can't count the thresholded version too. I feel pretty sure you'll see the things I have, or I wouldn't bother cautioning you so much -- no point getting up false hopes only to have them dashed -- I didn't enjoy it, I suspect no one else would either. But you now have this fine scope, and I hope you'll get computer screen capture going - it's really worth a bit of effort to be able to easily document your work, even if only for your own use that way. You save net effort -- it's a good investment to add one BNC or whatever to be able to do that, as you can use the human mark I eyeballs and signal processor to detect all sorts of errors that are otherwise hard to anticipate and correct before they occur -- life just can't be done perfectly feed-forward.

I've had a bit of trouble here with the Russian 3He tubes just deciding to count say 20-100 hz for seconds or minutes on end, no obvious reason for it, then settle down and work right for the rest of the day. The waveform coming out is only a little different from "the real signal" too - you really have to look close at the scope to tell (other than that you know there's no neutrons being made, so why is this thing going nuts). I'm going back to the boron ones with better behavior (damn, my spell checker doesn't like the UK spelling of that word), as I really don't need the sensitivity here and already do have a really serious 3he tube that's pretty reliable. But -- it's still not as rock-solid as the older boron tubes. Even that Hornyak, "deaf as a post" is plenty good for any real neutron production, and very hard to fool by comparison - at least here.

Hey, I'm just trying to save you trouble and disappointment here -- it feels pretty dumb to make a claim that turns out to be wrong later, that's all. These are super for tuning in real time, nothing beats a good neutron detector of some sort for that. It just better not be your only way to get it done, that's all. "Trust, but verify".
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: My planned detector topology & Voltage regulation for HV

Postby chrismb » Sun Jan 15, 2012 4:20 pm

Doug Coulter wrote:As for remote-ing the preamp from the tube -- just say no. Don't do it. ... Richard and I have both had problems using a preamp box the screws directly onto a connector on a 3He tube top - it's that picky ...


I was getting some pick up on it recently, so thought I would do some more tests and make a few changes/add some ferrites. Not sure this shows that much, but, hey, what more can I do to check it isn't likely to count false!?...

http://www.youtube.com/watch?v=hX6QUOuSPN8



...[Skip] The first 1:30 just confirming the normal background count (~5cpm)....
Last edited by Doug Coulter on Sun Jan 15, 2012 4:53 pm, edited 1 time in total.
Reason: Put video inline (Chris, you edit this so you can see what that looks like - you'll see how I did the tags). Good show man!
chrismb
 
Posts: 620
Joined: Thu Aug 05, 2010 6:32 pm

Previous

Return to Analog

Who is online

Users browsing this forum: No registered users and 4 guests

cron