I know that in Colorado, where I grew up, radon is a problem in basements. That might have little bearing on where you live. But the fact that you haven't seen the signature seems a pretty good indication that it would be something else.
MFMP: Automated experiment with Ni-LiAlH
-
-
Since I had little idea of what I was looking at, I thought it was older code (EDIT: eventually, I found it was almost 5 weeks old code).
The spectrum above (as well as those in the attached document) isn't averaged, although in the previous deleted comment I posted one containing average-subtracted data. I recalled that this was done last time but I didn't give much thought as of why. Of course one would actually want to subtract the background.
So, that's what I've done in the attached file below, using the first spectrum in the series as the background. The result is a bit noisy and probably not too useful, however.
-
-
Yes, unfortunately the signal at the spectrometer barely rises above the average. There are odd small peaks sometimes arising in particular in the low energy part but it's hard to say whether they're anomalous at this point.
Below is attached a final multipage graph of the entire experiment.
-
-
Yes indeed. Thank you can for your analysis, and to BobHiggins , for his many hours of work on design and experimentation.
-
Well done Bob! Your design is really showing its promise now, payback for all the hard work.
I am not as subject to Fukishima fallout and radon as Alan in Santa Cruz
I'm about 1 km from the Santa Cruz seafront, and under my topsoil is about 10 meters of sea floor mud and then sandstone. I have never detected a Radon signature in any of my long background spectra. But a few miles inland, up in the hills near the fault lines it is a real problem. One experimenter up there found high concentrations in his furnace filter.
-
I went through previous GM data. Hard to say if there's anything at all. The signal does seem to always start low at the beginning of the experiment (see attached images).
EDIT: I also tried looking for calibrations, but for some reason the Geiger counter and the spectrometer were disabled in most cases, except for the pre-calibration for the first experiment ("Round2"), performed around mid-March:
Since this seems to show an increase of about the same order of that seen in other experiments, I guess it has to be due to a variation of the background signal or some effect related to the temperature, as it appears to start low here too.
-
The only possible temperature-related effect I can think of is natural convection caused by the presence of a hot object increasing the air-flow around the whole experiment. This could in theory (big perhaps here) present more naturally radioactive particles to the counters. In the same way that a simple particulate filter - as used in aircon systems, say - picks up more particulates when the system is working than when it is idle.
-
...unless somehow heating the alumina tube in its current configuration actually causes gamma counts to slightly increase over the background, regardless of the load. It would be a novel effect!
-
Note that the temperature rise at the GM counter is relatively low and circulation is poor there due to the presence of the heat shielding. Still, there could be convection changes as the heated exterior of the K-26 bricks cause convection. Don't forget that there is the possibility of correlation in time, not just temperature. Generally the GM detector and the gamma spectrometer are started before the experiment, but usually not more than a couple of hours before the experiment, so there could be some correlation to time from instrument turn-on. I looked for correlation to local solar angle. For the graphs that can shows, the key solar times in Cloudcroft are:
On 3/10-12 the sun transits at 19:13 UTC in Cloudcroft, rises 13:18 UTC, sets 01:08 UTC
On 3/31-4/03 the sun transits at 19:07 UTC in Cloudcroft, rises 12:49 UTC, sets 01:24 UTC
On 4/21-23 the sun transits at 19:02 UTC in Cloudcroft, rises 12:25 UTC, sets 01:37 UTC
On 5/4-6 the sun transits at 19:00 UTC in Cloudcroft, rises 12:11 UTC, sets 01:48 UTC
Visually, I am not seeing a correlation. I just need to do more meteorology with the GM detector and the scintillator as Eric Walker has been asking.
-
In the Google folder for the experiment, in the Data_GammaSpectrometer folder, I added an additional folder having 20 hours of gamma spectra with the cold reactor tube still in place.
-
Spectra calibrated and all in one single document in the attached file below.
They don't seem significantly different than the ones I previously posted: there are still sometimes small narrow peaks on the low energy range visible particularly when a linear scale for the counts is selected (which is not the default view with the Spectrum Techniques MCA Software).
-
-
-
Yes, but to be rigorous, it may not be as simple as providing a running average of the average spectrum. What you really want to do is to adjust the smoothed curve in some way so as to minimize a weighted residual from the whole data set. Another problem is that imperfections in the calibration can cause small wiggles in the smoothed spectrum. For example, below about 600keV in your smoothed result, the integral is well below 0. If all of the spectra are null for other lines, then the residual must be zero average. This is where median combination instead of averaging may also help.
-
Just curious, but I'm wondering whether noise cancellation techniques would even be needed if a candidate signal were several sigma above the noise floor. I suspect you would be able to simply eyeball it.
-
I was only interested seeing more clearly what part of the spectrum exactly varies with time, but I can't seem to get useful results.
I think an issue here is that it's not a decreasing signal, it varies up-down over time in a way that doesn't seem completely random.
Here are total counts for every spectrum. I've also added the uncalibrated channel for the 1461 keV peak to see if there is any correlation with its drift over time:
-
Of course, if the signal were large and were due to a line source, it would be visible even in the raw data. However, almost any distributed signal and any other small signals will require photometric background subtraction to be visible. Big signals, multi-sigma over background are clear indications - the "Signal" from Alan Goldwater's Glowstick experiment had that kind of intensity. This experiment does not. However, we are looking for any metric upon which to optimize the fuel processing, formulation, and the protocol.
I like your analysis. The total count is less prone to calibration error than the count from the 40K line. The 40K count will depend on the bandwidth you use for measuring the 40K line, and the accuracy of the energy calibration (an error in energy calibration shifts the 40K peak left or right and will affect what you count from it). The variation shown in the 40K count is about +/- 1.5% and the variation in the total counts is less, at about +/- 0.7%. It would be interesting to compare the variation in total count in the GM detector to the variation in total count in the scintillator for approximately equal integration periods, but this will require taking some meteorology data that included the time domain counts.
-
To clarify, I'm not showing the 40K count, but rather the uncalibrated channel of the 40K peak. Maybe it's confusing that in this graph I'm showing it alongside the total count value.
In other words, ideally the 40K peak would always be at channel 1461, but during this run it drifted smoothly between channels 1395-1450.
EDIT: you can also easily see it in the raw data for example by comparing PostExp_1.spu with PostExp_10.spu in the Spectrum Techniques MCA Software.
Wouldn't it be odd if the signal variation better visible with the Geiger counter was correlated with the channel drift of the spectrometer?
