LION-AG Experiment

Quote from can

However you have to be careful not to be looking too much into coincidental patterns that can arise for example with different rolling average sample sizes. These spikes get also easily averaged out with larger rolling average windows.

Related: I suppose an interesting event would stand out above the noise floor and be pretty easy to identify by eye; if not, the putative event is probably too subtle to really persuade anyone of anything.

Another challenge: establishing a causal connection back to the system under test.

Regarding post $81, continuing...

The diapads will adhere to the magnet because the substrate that the micro diamonds are affixed to is nickel, a magnetic material. The diamonds will be the highest conductive projections from the smooth surface of the magnet and will generate the most electrolysis.

Quote from axil

The LENR reaction is a weak force based reaction, a particle decay reaction that has nothing to do with fusion. The meson is the sub atomic particle usually generated by the LENR reaction, not neutrons. Sometimes fusion does occur as a secondary reaction produced by mesons/muons. Tritium is usually not a product of the LENR reaction.

Axil, since you have a well established theory of how LENR works, can't you build a working device for others to replicate? I don't understand why you let people waste time trying to replicate LION when you already have it figured out. What was the highest COP you got on any of your devices?

Quote from PhysicsForDummies

Axil, since you have a well established theory of how LENR works, can't you build a working device for others to replicate? I don't understand why you let people waste time trying to replicate LION when you already have it figured out. What was the highest COP you got on any of your devices?

I am not well suited to be an experimentalist. I know well that most experimentalist produce bad theory. This includes Rossi especially.

Regarding: "What was the highest COP you got on any of your devices"

LENR does not usually produce heat. Many of the experiments that people perform generate the LENR reaction, but not heat. People fail to looks for sub atomic particle production. This production is the true test for LENR.

As an example, MFMP produced the "signal". This gamma ray bust was a sure sign that LENR was active in that experiment.

I have taken a look at some of Alan G's data. In particular, here are radiation data from the 4 different counters that Alan's group uses, shown during during one of the ~32 minute cycles that have been going on for the past day or two. Using Alan G's labeling the traces are: grey - GM; yellow - gamma 500-520; orange - Li6l neutrons; blue - He3 neutrons

And here are the concurrent temperature and input current traces

Because there are many repetitions of exactly the same input waveform it is possible to average different the radiation responses together. This drastically reduces the standard deviation of the time series of the radiation counts. Below is an average of radiation responses to 20 identical input waveforms. The vertical scale is the same as in the top figure

The averaging has decreased the noise by a factor of 1/root(20). I see no differential radiation signal associated with any part of the input.

It is possible to add more sweeps to the average. The data are available. Conceptually, no matter how small a radiation signal is that you want to spot, you can do so by adding in more sweeps. As data sweeps are averaged in with the others the noise will be go down and down until, eventually, even the tiniest coherent signal should become obvious.

[Data for averaging taken from Exp1-2-TC.xlsx between timepoints Apr 10, 5:45:55 and Apr 11, 3:39:40AM.]

Quote from Eric Walker

Related: I suppose an interesting event would stand out above the noise floor and be pretty easy to identify by eye; if not, the putative event is probably too subtle to really persuade anyone of anything.

I would agree.

Quote from magicsound

I'm surprised the temperature cycles are not consistent. The PID is simply looping a program cycle: [...]

Now I see. I assumed that the cycles were in total 15 minutes long and since I was having other problems with the graphing system I was trying to set up I didn't investigate further.

EDIT: I removed another portion of the comment as I was not sure that I was obtaining proper results.

PhysicsForDummies

Quote from PhysicsForDummies

Axil, since you have a well established theory of how LENR works, can't you build a working device for others to replicate? I don't understand why you let people waste time trying to replicate LION when you already have it figured out. What was the highest COP you got on any of your devices?

LOL - Cancel SETI - higher lifeforms are amongst us! ...cruising our fora, sweet-talking our kids 😯

magicsound

I tried setting the major grid spacing interval to 32 minutes on a graph showing power (Watts, in blue). If you notice, it quickly goes out of phase with the power/temperature cycles.

After a couple tries it appears that the interval should be 33 minutes:

The red lines are the "pstate" and it's 1 when Power is >120W and 0 when power is <120W. This initially appeared to work better, but I eventually noticed that the periods of "low" pstate defined this way are on average about 4-5% longer than the "high" pstate. So this wouldn't have worked properly either.

Alan Smith

Averaging your data will help your analysis immensely. It will definitely solve your radiation analyis problems.

You can make life easier for your data guy if you add some sort of sync signal to your system i.e., a pulse coincident with the beginning of some interesting input. If this is not available, however, I am finding with Alan G's data that event-triggered averaging is pretty straightforward if you use sudden changes in the current input as the trigger.

can

I noticed a bit of jitter in the starting times of the cycles and also a handful of occasions when here was an delay that went beyond jitter. I dealt with this by aligning things by hand in a spreadsheet but obviously the effort involved in this this technique doesn't scale up well.

Does the data analysis package you are using have the capability of triggered selection of data segments? If it does then I suggest event triggered averaging based on your "pstate" variable going from low to high. That would get around all issued of jitter etc.

Bruce__H

I'm not a data analyst, just interested checking out what's going on with more advanced tools than spreadsheets. With what I've got pretty much anything is possible as the data and the graphs are prepared with programming code using mainly http://pandas.pydata.org/ and https://matplotlib.org/ .

What I wanted to accomplish was grouping 'delta' counts into 1/2 period chunks (16.5 minutes) so that it could be possible to differentiate between "high" and "low" power states simply by using the odd or the even chunks.

Subsequently, by summing cumulatively the so-formed chunks it becomes possible to check out if the rate of increase in the total count is constant or varies throughout the experiment, and if it's different between the two power states. This also avoid averaging the data, which could eliminate small but potentially real signals (in my opinion).

So far the only interesting changes appear to be with the He3 neutron counts, but there doesn't seem to be a clear correlation with the state of the experiment, so I'm guessing that it can be part of the long term background variation. It's also possible that a signal is buried in the jitter in start/end times like you mentioned, but since nothing seems to clearly arise, if there's anything it must be very small.

With the usual caveats:

Data from the beginning

From 2018-04-12 through the end

Quote from Alan Smith

Hi. AFAIK The Netto geiger timebase (1 sec) actually provides the time signal for the whole system. Data is collected every 10 secs and for the Netto is it's own rolling average over 1 minute periods - it shows you CPM at any given time. I am just investigating major system upgrades, so this may change. I will pass your comments on. Thank you.

To be clear, a rolling average is magnificent if you are interested in slow changes in average behaviour. But it is like a low-pass filter and smudges out the time characteristics of signals. If you have a suspicion that faster events (i.e., bursts) or events that are locked to particular time-features of your stimulus input are important, then a rolling average is your enemy and triggered averaging across different data segments is what you want. It is an extremely useful technique and can be set up on an oscilloscope or in DAQ software for use when an experiment is ongoing. The power of real-time analysis during an experiment is that you can wait for the signal you are looking for to emerge as the random noise is averaged away. You then know when you can stop the experiment because you know when you have collected enough data.

can

Our analyses are complimentary.

The numbers on your charts are too small for me to read. What is the time stamp associated with the biggest difference in the He3 signal in your top plot? I will have a look at the corresponding stretch f data using my analysis technique.

Bruce__H

Try opening the images in new tabs. LENR-Forum is making them small. You can also click the "view source image" button to achieve the same.