This is unprocessed data from either hugnet or google (I don't remember - it's been a while since February) plotted 1:1 in Excel. Zoom in to original size to see the time span (and the magnitude of the issue). I remember that it all looked totally innocent and not a bit discontinuous in hugnet's visualizer. Even with this dataset, the discontinuity completely disappears as soon as you apply a running average over as few as 10 data points.
MFMP Provides Update About Me356
-
-
Quote
It could well be that radiation and excess heat are not seen at the same time. We have heard before that the radiations seem to occur in some start-up transition
This is the problem with weakly predictive hypotheses such as you have here. They can be made to fit almost any unusual or noisy data. Thus given an anomaly, you can claim support for LENR. My point is that there are more plausible "Murphy's Law" error mechanisms for any such anomalies that are one-off and therefore cannot be properly instrumented. Of course "noise" is also weakly predictive.
The weak predictivity of such hypotheses is why most scientists give little weight to one-off unexpected observations in experiments that cannot be replicated.
-
Quote from THHuxley: “No normal scientist would spend much time trying to analyse this artifact. They would repeat the experiment and either measure it to death (if it exists) or reckon it was one-off artifact if it cannot be repeated.
I can't…
Indeed the MFMP tried to replicate, and while, during that attempt, they didn't see the same artifact, there were other types of "signals" detected, one of which was correlated with pressure/temperature changes, and repeated once or twice within the same experiment
-
Quote
Indeed the MFMP tried to replicate, and while, during that attempt, they didn't see the same artifact, there were other types of "signals" detected, one of which was correlated with pressure/temperature changes, and repeated once or twice within the same experiment
From which we can conclude that most likely the original "signal" was one-off noise-related (mechanism as proposed above or other unknown).
We would then need to do a separate analysis for the other "signals". Again, if repeatable they can be traced to a definite cause: whether that is an LENR cell or something else. If not repeatable they are most likely one-off noise etc.
The correlation with pressure/temperature changes is interesting, but since there are many artifact mechanisms that so correlate it does not by itself get us very far.
All of these lacunae are expected in any experiment with sensitive sensors and high power inputs with large temperature changes. They cannot therefore be correlated to make something bigger unless the specific anomalies are coherent. I have not heard evidence of that?
Remember, if there is a definite effect it will be reproducible, and we are in a whole new ball park where it can be chased down and properly identified.
-
Thanks, Bob, MFMP, and others who roll up their sleeves, and then freely put themselves and their experiments' results out there--warts and all.
Wonderful, really, to witness the internet mastication of LENR experimental raw meat. Perimeter grazer-geezers, like me, get a brainfood snack, while hoping that humanity will get a feast. Yum.
Cheers, Penswrite
-
We would then need to do a separate analysis for the other "signals". Again, if repeatable they can be traced to a definite cause: whether that is an LENR cell or something else. If not repeatable they are most likely one-off noise etc.
We cannot be absolutely certain that the signal was not noise. However, the spectrometer is a completely different, separate autonomous system from the temperature data gathering system. Noise in one does not correlate with noise in the other. Those of us who have been close to the experiment would have pointed out a noise pathway had we found one. At the moment, our conclusion, after hours searching for a noise pathway, is that the signal cannot be explained as noise or by environmental radioisotope contamination. This leaves the reactor as the source to be the most likely explanation.As I said earlier, Alan has no means of "close" replication at the moment. Who knows what small variable had to be just right for the signal to present itself. What will help with replication is to automate the experiments - I.E. run them from a script. Then if you want to replicate, you just execute the same script again. I have tried to replicate a long manual experiment myself and it is virtually impossible to not deviate somewhere along the way - if not frequently along the way.
-
Quote
Noise in one does not correlate with noise in the other.
It does. You may not be able to explain why but it remains fact that your TC data went berserk around the time of "the signal". It is strange that you would dismiss an obvious correlation like this one but - only a few posts above - seriously consider an outrageously far fetched one using the words "It could well be that radiation and excess heat are not seen at the same time." That is a prime example of confirmation bias. -
it remains fact that your TC data went berserk around the time of "the signal
Jami, this is only your opinion, not supported by facts or analysis we have seen yet. A more detailed analysis (not just a picture) might help support your contention. For example, if you were to download the publicly-available data and show that the standard deviation increased by (say) an order of magnitude coincident with the spectral signal, it would be a useful clue to the event, and a valuable contribution to Live Open Science.
I say this as an invitation, not a criticism. Much of our data analysis was done by volunteers from this very community, for which we are grateful.
-
The signal could mark the beginning of the excess heat production stage. The first thing that occurs is a change of state in the LENR reaction that is marked by the signal, then the excess heat stage begins. From what I have read here, the data does not contradict this scenario.
-
Quote
for which we are grateful
Nice words, Alan, but I'd prefer it if you guys learned from this stuff and do this kind of basic analysis before you call "the end of the carbon age" next time. I already ran a basic PCA for Matthieu (because he too couldn't be bothered to do it himself BEFORE he claimed success) to prove that his infallible flow meter reported nonsense and that Celani's congratulations were unfounded - with the result that the next experiment was again designed without any sanity check on data at all. You do this yourself now. Four or five lines of matlab coding won't kill you. You know what you'll find. -
Nice words, Alan, but I'd prefer it if you guys learned from this stuff and do this kind of basic analysis before you call "the end of the carbon age" next time.
Not my words Jami ! If you want to object, talk to Bob Greenyer. I just run the experiments and publish the data.
I've calculated the standard deviation of GS5.2 T_ambient, for each 4-hour interval of the 48-hour test period 2Feb & 3Feb. The result shows some increase due to the induced AC noise I mentioned, but is always less than 1 √°C.
I used the 30 second average data file for this first pass. The raw data file would have about 173k rows and would need some patience to work with in Excel.
-
I'm going to take the discussion here as a proxy for the old thread on this site that it seems TC did not look at.
The question is whether these interesting one-off (at the time) results come from so-called "inner" Bremsstrahlung (IB) generated by the cell or Something Else.
Sorry THH alias TC. We extensively discussed this spectrum months ago. Your answer below is a bad joke. The same signal was seen later again in an overlay with one or two other signals. You are simply late...
The data we do have shows a remarkably smooth curve. I'd therefore hypothesise that this might be a software response to an electrical noise signal on the scintillator photo-detector lines. The software generating the spectrum identifies isolated peaks and counts the area underneath them as indicative of the number of photons received coincidentally, and generates a count at an energy proportional to the peak area. All such counts are summed to generate the spectrum.
Let's call this spectrum the TC noise curve.
It is my opinion that the Signal was caused by the power glitch. I do not expect this to be tested, however. Interrupting the power supply is likely dangerous to the equipment, so testing this idea may be rather expensive, whether this was the cause or not.
@P: The signal reoccurred many times and what you see is an integral of different spectras not a glitch.
I see absolutely no need to discuss this again, because some (low/high) bands were not measured and neutrons ignored...
Let's wait for more data! -
Quote
I just run the experiments and publish the data.
Ok, so if your job is to run the experiment and publish the data and Bob G's s job is to shout "hurray" and Bob H's job is to quote Celani anecdotes from old Rossi (yes, Rossi) demos - then who does all the work to confirm "the absence of contradictory evidence" you claimed a few posts ago? This sounds like your definition of "live open science" is that you guys simply claim stuff and as long as nobody volunteers to prove you wrong, you assume you're right. That should be the other way round. Your approach to noise detection is just plain wrong. The minimum time resolution for that data file is one second (I assume you either couldn't log sub-second interval timestamps or you decided they weren't important and cut them off at the full second to save a few KB of disk space). But whatever the reason - THAT is where you start. Not some arbitrary multiple you pull out of thin air based on whatever made you think 30 seconds was a good idea. In this particular case, you needn't bother with more than 8 seconds (and use matlab or python or fortran for God's sake - doing this in Excel will take forever).Another thing. Yes, I know you're only running the experiments and publish the data - but can you honestly tell me that you ever looked at the raw data (assuming you did) and took a step back and thought "Yes, that is perfectly ok. I see nothing wrong with this at the time in question. All as it should be."?? Honestly?
-
Quote
We cannot be absolutely certain that the signal was not noise. However, the spectrometer is a completely different, separate autonomous system from the temperature data gathering system. Noise in one does not correlate with noise in the other. Those of us who have been close to the experiment would have pointed out a noise pathway had we found one. At the moment, our conclusion, after hours searching for a noise pathway, is that the signal cannot be explained as noise or by environmental radioisotope contamination. This leaves the reactor as the source to be the most likely explanation.As I said earlier, Alan has no means of "close" replication at the moment. Who knows what small variable had to be just right for the signal to present itself. What will help with replication is to automate the experiments - I.E. run them from a script. Then if you want to replicate, you just execute the same script again. I have tried to replicate a long manual experiment myself and it is virtually impossible to not deviate somewhere along the way - if not frequently along the way.
I agree it is a very tough job to track these things down when they are intermittent. Normally, while this process is incomplete, researchers would not assume the intermittent signal to be a signal of new physics. The key here is that intermittent incoherent signals are most likely to be bugs. That does not preclude the possibility they are not bugs - and if so, eventually, it will be possible to show this. My experience (perhaps different from yours) is that however hard you look unexpected error mechanisms remain so the correct approach is to view any surprising and incoherent behaviour as some not understood bug, rather than some new physics, even given the apparent certainty that all such errors have been ruled out (this turns out illusory).
My suggestion wrt correlations with temperature is that circuit characteristics (for example poor joints) would be expected to correlate with temperature of the apparatus. I was not suggesting that noise in two different systems would correlate.
More specifically, sometimes you can get hints of what a bug is from things that correlate. What are the anomalous results that you consider to be worth putting much effort into understanding? Is it just the original "signal" or is it other things as well.
Now to the details. I'll reply to Wyttenbach and Jami together.
Quote from WyttenbachSorry THH. We extensively discussed this spectrum months ago. Your answer below is a bad joke. The same signal was seen later again in an overlay with one or two other signals. You are simply late...
Quote from JamiYour approach to noise detection is just plain wrong. The minimum time resolution for that data file is one second (I assume you either couldn't log sub-second interval timestamps or you decided they weren't important and cut them off at the full second to save a few KB of disk space). But whatever the reason - THAT is where you start. Not some arbitrary multiple you pull out of thin air based on whatever made you think 30 seconds was a good idea. In this particular case, you needn't bother with more than 8 seconds (and use matlab or python or fortran for God's sake - doing this in Excel will take forever).
Another thing. Yes, I know you're only running the experiments and publish the data - but can you honestly tell me that you ever looked at the raw data (assuming you did) and took a step back and thought "Yes, that is perfectly ok. I see nothing wrong with this at the time in question. All as it should be."?? Honestly?
@Jami - I have sympathy with your view that detailed scrutiny of the raw data should be the first thing anyone does in this case. However, I did this, prompted by you. I even wrote a script to do some checking of "raw data noise". There was a lot of TC (possibly USB earth?) noise, especially high when the "signal" occurred but as high at some other times too. The correlation between "signal" and "noise" was not persuasive, although I can't say on that basis I could rule it out.
Quote from BobHowever, the spectrometer is a completely different, separate autonomous system from the temperature data gathering system.
Which knocks on the head any direct connection - but there still could be indirect connections.
Coming back to Wyttenbach's comment. I think he is arguing that a one-off supply glitch could not result in this signal because the same pattern (above normal background smooth spectrum with characteristic LF bias) occurs on other segments as well as the 24 hour segment where it was most obvious. That is helpful - and does point to something different from the one-off supply glitch being the culprit - but it does not invalidate my comment.
The noise on the photodetector lines creates the "normal" background - in conjunction with the spectrometer software and any residual real scintillation. These "signals" of varying height above background, in all cases close to background, look to me like an artifact caused by varying noise level. Whether that noise is electrical, or even optical, remains to be determined. We do not know the exact spike processing used by the spectrometer (at least if it is a separate bit of kit I believe this is the case, and if the software referenced at one point on MFMP site digging out the spike processing and precisely characterising it is difficult). So we cannot tell what changes will lead to modulation of this noise floor. You can imagine both DC and AC signals doing this, interacting with existing noise. It is complex when you get down this low and have "spike capture" algorithms with their own "noise suppression" to contend with.
A fruitful line of investigation would be to set up the spectrometer with the same kit as when the signal was measured (but without the magic powder hypothesised to show new physics behaviour) and look very hard for even tiny changes in its background output level comparable to those that Wyttenbach notes happened during the active experiment. We are talking here about small raw count rates (of order 1 per second - I forget...?). The best way to follow this might be simply to look at total number of counts (the sum of bin n multiplied by n over all bins, I think) over say 5 minute periods (we are summing all bins so get much better temporal resolution on the total) and see what modulates this background. Whether this is contaminant beta emitters, ambient temperature, IR going into the spectrometer, or some electrical noise source. The problem is that the spectrometer software will have some built-in calibration stuff and also perhaps DC level stabilisation stuff that complicates things, and what would be most helpful investigation would be the raw noise data, but perhaps the calibration can be bypassed and the level-shifting stuff is relatively harmless.
Anyway if Wyttenbach is correct - that the signal at very low levels was present on multiple segments of the captured data - it opens up this line of investigation.
If MFMP is correct - such control investigation will have null or inconsequential results.
If I am correct - MFMP's "active experiment" investigation will have null or inconsequential results.
If neither I nor MFMP are correct (quite possible) then both investigations stay null or inconsequential and the original artifact remains a mystery.
MFMP's motivation for the more negative line of investigation (which in the good old days, when starting with calorimetry and getting loads of false positives, they did pretty well) should be that no-one will take them seriously unless they follow normal procedure and do it. And having investigated equipment characteristics in more loving detail they have better understanding of what is going on.
LENR enthusiasts point out that skeptics are often biassed and hence do not look creatively and enthusiastically for LENR characteristic anomalies. That is possible.
I'm pointing out that LENR enthusiasts are often biassed and do not look creatively and enthusiastically for artifacts. One sign of this, in some cases, is a continual change in experimental conditions and instrumentation in an attempt to "optimise" or "detect" the effect which also of course makes tracking old artifacts more difficult and potentially introduces new artifacts. Adding new instrumentation is a great idea - if it allows previous claimed anomalies to be investigated in more detail. If it just adds new artifacts to the equation it is actually unhelpful.
-
Thoughtful post THHuxley.
MFMP has put a great deal of effort into analysis of this signal - and I have personally put in a lot of time and analysis. I do not have access to Alan's particular spectrometer, but I have a similar model. The spectrometer is a Spectrum Techniques UCS-30, a commercial instrument. This metal boxed spectrometer connects to the scintillator via coaxial cables. The scintillator itself is housed in a continuous light and RF tight Faraday cage and has mu-metal internal shielding around the scintillator's PMT. Inside the spectrometer instrument, there is a small analog front end section and then the MCA processing, integration, and storage are all digital. Add to that, the scintillator is surrounded by 500 pounds of lead. All of this provides a great deal of interference protection. Alan has created strong electrical interference in testing and it did not show in the spectrum - nor would we expect it to. Electrical power failure would not have affected the spectrum without causing the internal processor to reboot. There is still the open question of whether strong line interference could have affected the analog front end of the spectrometer - that immunity test is harder to conduct (a pun).
Neither Alan or I have seen any other instance of this commercial instrument failing to operate as designed. If we find any evidence we will report it. In the mean time, the most rational conclusion is that this instrument worked as it was designed and recorded properly the photon spectrum that was present.
We have examined scenarios that could cause that photon spectrum and have been unable to find a spurious contaminant that could be responsible. Temperature shift could not be responsible as this has been measured and compensated in the photometric analysis.
Intermittent radiations have been reported before from LENR experiments (Focardi, a well respected and experienced nuclear scientist reported measurement of low energy gamma). This becomes another documented report of a radiation measured. In the future when/if other similar phenomena are reported, perhaps the median conditions that cause it (or caused an error) will become known from the accumulated reports. If we were to fail to report what we observed, we deny the scientific community the body of evidence needed to move forward. If it is never observed again, it will fall into obscurity as an unexplained anomaly and will have no effect on the field.
At minimum, the report may cause other researchers to expand the instrumentation on their experiments. As I said, we are already seeing this happen.
-
All as it should be.
Aah, you're here for An Argument. Sorry, that's next door. This is Being Hit With Data lessons...
Lest my comment be misunderstood, it's intended to lighten our discussion a bit:
https://www.youtube.com/watch?v=wdoGVgj1MtY -
The noise on the photodetector lines creates the "normal" background - in conjunction with the spectrometer software and any residual real scintillation.
@THH: This comment is more constructive. In the Holmlid thread secondary radiation caused my muons was discussed. There might be even more particles, we don't know yet, to be there.
I went through hundreds of mfp spectra and the picture was sometimes complex, with different signal overlays/time constants. The most interesting spectra were obtained during the pressure decrease (H sweating phase) where the creation of inverse Rydberg matter was strongly supported. I asked them to focus on this effect and to add other measurement equipment for X-UV/soft X and higher gamma energies.
It's not about COP, its about the physics happening on the surface! -
I went through hundreds of mfp spectra and the picture was sometimes complex, with different signal overlays/time constants. The most interesting spectra were obtained during the pressure decrease (H sweating phase) where the creation of inverse Rydberg matter was strongly supported. I asked them to focus on this effect and to add other measurement equipment for X-UV/soft X and higher gamma energies.
Yes, the pressure change CPS thing was neat. More work should be done on that effect.
However, I disagree that the "creation of inverse Rydberg matter was strongly supported". There was an unexplained effect, and no test whatsoever for special states of matter that could support or refute a special state of matter hypothesis.
-
If it is never observed again, it will fall into obscurity as an unexplained anomaly and will have no effect on the field.
Does MFMP publish the 'null-results' from replication attempts somewhere?
I think at the time when Bob trumpeted that 'The end of the carbon age is nigh", he also said that SKINR is prepared to run a replication of the GS 'signal' experiment.
Where can we find the results of this replication. On any MFMP web-page? -
Does MFMP publish the 'null-results' from replication attempts somewhere?
I think at the time when Bob trumpeted that 'The end of the carbon age is nigh", he also said that SKINR is prepared to run a replication of the GS 'signal' experiment.
Where can we find the results of this replication. On any MFMP web-page?
The MFMP data is available on the web - though you may have to ask to find the links. Like most, MFMP only reports on the data that seems notable - but at least all of the data is available for anyone to examine.You will have to email Bob Greenyer to ask if there was any data shared from the SKINR "replication". From what I saw of it, it was not a "replication" at all, but an experiment of their own convenience using some of MFMP's experiments to select their protocol. Perhaps some of that work will be presented at ICCF-20. As I recall, SKINR did not even deploy radiation detection.
