As mentioned elsewhere, we like to test things in real life situations. That's why I built up a semi-robust version of the upcoming sdaq device. No test like the real thing, so I took it down to the lab and lashed it up on the fusor and took some data. I think I'm going to like this! While the fusor itself is a little in need of work, it surely worked well enough for this, and for me to instantly note I'd not put a log plot option on the counter plots - so by the second run i'd edited the code to default that way, and of course now I should really add a button to the software too. I really took this data to get a couple log files, so I'll have some real stuff to work with in developing my multidimensional plots for this data.
OK, cut to the chase. AD0 is wired to the HB output sense line (arb scale). 52 kv is what the supply was putting out at the peak on the plot.
AD1 is the current. Most of the first run was set to 20 ma current limit, so you can see that getting hit and the HV dropping.
Note, these don't compensate the drop in the 100k ballast resistor - but we could!
C0 is a Geo-210 pancake (slightly more sensitive than the standard counter one, due to being larger)
C1 is the Eljen Hornyak, a phototube, our preamp. This is the same one that went to Richards and got us what turned out to be a very easy to remember calibration factor ~~1k/cpm per 1meg neuts/second, roughly.
The two plots here reveal a lot. I turned on the HV first, but with no ion source, so the HV rises to max, 52kv (from the meter reading, the plot isn't scaled) until I turned on the secondary grid ion supply. You then see the HV dip a lot - There was too much gas in there for the 20 ma current limit. I then removed gas in steps - you can see some of them between 100 and 150 seconds. Notice hitting max HV also made both X rays and neutron counts go way up. The rest of the run was a dance with the gas, trying to keep the HV maxed at as close to the current limit as I could. The big downspike after 250 seconds in HV was a time I let in more - and that was too much, so I went to the other button and took some out. Notice here that with the HV "not all that low", the X rays and neutrons are just gone for that time. For (I think) different reasons, they are both nonlinear functions of voltage. The pancake by the way, is at the operators position, behind led in the tank, lead glass, and a block of lead(!). Some of that has to be scatter, some *could* be very hot gammas from fusion that would laugh at my shielding. At about 370 seconds I turned things off, and went to get the silver to check activation. IT took me way too long, as I'd left some junk in the path, but you can see a bit of background counted, then a jump up when I got there with the silver, followed by some decay. About this time, I went, OH, you can't back extrapolate the activation on a linear plot! You need log so you can take the straight line back to the turnoff time. So I fixed that for the subsequent running.
For the next run, I was a little more warmed up and had removed the obstacle to getting to the silver.
On this run you can see me doing the gas dance as well, starting around 200 seconds. As it warmed up, the gas pressure and current went up, so I was taking gas out - as you can see in the steps on the voltage waveform. At 300 seconds I was hitting current limit hard again, but feared that I'd lose it if I took more gas out, so instead I turned up the current to about 30 ma. Note this didn't do much for the neutron count, but then we're not yet compensating for the drop on the 100k ballast - we lost a KV there even when the supply is reporting reaching its max (here still set to 52kv). It looks like in this run I hit about 1603 cpm on the silver, extrapolating back along the now-straight (log plot) decay curve back to shutoff. Gnu plot tells you the cursor location in the plot units, making this a fairly easy measurement.
Look how much better this is for learning about your gear! I like this. Now I have some data to go write more sophisticated plotting software against, too. There isn't even a reason to be stuck with pure inputs. I could plot watts vs neutrons - a Q map. I could correct for the ballast resistor drop. I could make a plot what would make it obvious if there were two different nonlinear effects for X rays and neutrons.
This is the way to go with this stuff. I knew about some of this - heating effects, how it gets dodgeier to control gas over a run, and so on, but here we "learned" this in two runs and plots.