With the pure gas (the old stuff had been collecting impurities due to outgassing I expect - even with a base pressure of 3e-8 mbar, there's always some), the results got better.
So, here's the absolutely raw data for that:
- Raw data aq plot
On the left plot, red is the main high voltage, green is current, and blue is pressure (logged). So you can see where I turned off the main HV, then opened the valve to the forepump and got it clean (more or lss, it'll go lower but this is4 orders magnitude below where we run normally). You can then see where I turned on the HV again, and started letting in gas till it "lit off". Note the difference in neutron output. I turned it off at 350 seconds or so, and got the silver out of the neutron oven so as to watch it decay - that's why there's a tapering off of the geiger counts at the end with nearly no neutrons (just cosmic ray background there).
So, we obviously do better with purer fuel. We get around 5 9's net at first, obviously it gets worse sitting in the tank with outgassing going on. We bought 6'9s good gas, but lost a 9 in the transfer to the smaller tank (despite heoric bake out and so on).
Now here's another set of plots, these analyze Q, in units of neutrons counted per second per watt input. (not any practical units, but good for comparing). I had to edit the log files and take out some serious outliers to make these about the same scale.
We sometimes get a fairly big burst of neutrons at onsets - before the current monitor can respond, due to a limit in this data aq system. That makes for Q readings on this scale of thousands. Nope, it's not that good, better to edit out the points and not tell lies. Also, since this all autoscales, so I can see some detail in the parts we know are correct. Guess which is the stale gas in this:
- If you can't tell, something's wrong
Note with the stale stuff, Q goes down at higher currents much more than with fresh. I'm pretty sure we can figure a reason for that. Like most systems, most of my contamination is water...
