by Doug Coulter » Sat Mar 05, 2011 4:52 pm
There isn't really a solution to the time vs running timestamp problem, even NSA had me build "something" to help with that for them, but getting them locked in super close when there's more than one timing source, say better than a uS, just isn't going to happen without "heroic" work on that problem, which I can describe but it takes a lot of words. I use a watch xtal in my uP data logger, but that only has to be decent for 5-10 minutes at a time. There are some nice timing quality standards out there surplus, we've got one from NBS (before they renamed themselves NIST) that is very very good. I think Charles Wenzel's outfit makes some things that are very very good, but I'm not sure if he's into absolute rate-accuracy, but more like stability and low sideband noise. It won't matter if you can compare with NTP once in awhile anyway.
I'd say injecting the NTP time once in awhile should do it (with some back tracking in software later). It kind of depends on what you're looking for, you may not need sub millisecond accuracy if you think it's related to solar activity, but you'd sure like it better than 10 minutes!
I think those tubes are really stable. If they have that "grille" structure on the side like mine do, they will let alphas in there - the grille protects a part of the tube that is a thin Be window for that.
They are completely uncritical about supply voltage and gas pressure, and are run a very much sub micro-watt powers and do not drift or self-heat with counting unless something else is very broken. Having built a few tubes from scratch, a rather large factor of gas pressure variation doesn't have a lot of effect either. The physics in this case is in your favor all the way -- they either avalanche or not, and there's enough energy put in there by a beta or gamma to make them trigger quite reliably under a very wide range of conditions.
You will have the usual statistical false alarms -- John Strong's book and Haliday's book (see listings elsewhere on the board) both have some very good explanations of how much you have to count to get a number you can trust to some percentage, and it tends to be crazy big to get to single digit percents, much less fractions of percent.
The main issue you may have with those is their response time. After a tube counts, it's dead for a little while, then puts out smaller sized pulses for another little while after that. Whether you count those will make a big difference when the count rate is high. Typical times are in the 100's of microseconds, which means at 10k cpm, you'll be hitting that "all the time" -- check with a scope and a good HV coupling cap right at the tube signal. If counting 1/4 or 1/2 height pulses (or not) is going to affect your accuracy (and it will) -- you'll have to go to a much less hot test source to avoid that happening many times a second. Remember, radioactive decay is (at this point) the very most random thing known to man -- you can have two near simultaneous pulses even at very low average count rates, and that's going to be most likely the most important error source using these.
I think Haliday said it best. In U decay, alpha's present themselves at the coulomb barrier of the nucleus at about 10e22 times a second, yet U has a half life of what? In the billions of years, depending on the isotope. You can see how such a process would be pretty darn random, making brownian motion look very regular by comparison.
Chris, I very much doubt your theory on that one. These run pretty high pressure, plenty temp conductive as a pirani gage would show (doesn't even read much less than atmospheric). The powers involved for a pulse are very tiny if you're doing this right -- megohms in series with half a kv and for a few microseconds at a time (during which the tube voltage is near zero, so very little power is involved - it's a switch) -- these tend to run in the region of 1/10 atmospheric pressure to A: be near the bottom of the Paschen curve and B: have enough gas in there so radiation is more surely going to interact at all with it. Strong (with 1930's tech) did notice a difference in tubes made with rubber stoppers, due to outgassing, with temperature. That's not going to be a problem in a tube already sealed up (with no rubber) for a few decades. That's pretty far in the past. I would bet any money your background variation is simply a real thing -- I see it here and it does not matter whether I'm using a geiger tube, or a scintillator, or for that matter, a neutron tube -- they all count "up" during heavy cosmic ray periods which are themselves pretty doggone random, but also tend to have a time of day component if you stack counts over 24 hour runs. Also depends on what the sun is doing to the earth's magnetic field just then.
This big background variation IS the big issue in counting silver activation, of course, since you're not all that far "out of the noise" there at a mere 1k cpm or so.
We see on a particular sensitive geiger a range of 6 cpm to about 120 -- quite normally, and when it's counting hot, so are all the other detectors. It's just the way that is.
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.