Unconventional electrolysis

  • Recently I've been doing some tests (calling them experiments might be too much at this stage) that could be defined as "unconventional electrolysis"; nothing really rigorous or in any way involved, but I thought that perhaps I could open a dedicated thread instead of scattering comments about them around.

    Basically, I have two small electrodes composed of the same material (mildly magnetic stainless steel of unknown composition; likely not austenitic) placed on top of each other and separated by a very small gap. In the gap a saturated aqueous electrolyte solution (of sodium carbonate in this case) is provided through an opening on the top electrode. The electrodes are therefore not immersed in water, but a thin water layer exists at their interface. A relatively large current (a few A at 3.50~4.75 V depending on conditions) is passed through the electrodes during active conditions.

    My idea (however misplaced) was initially that ordinary electrolysis would occur as if the electrodes were immersed in water, with the difference of the anode oxidizing and the cathode getting reduced only at their interface, but it quickly became clear that something slightly different occurs instead. Here is a schematic representation.

    Material from the anode appears to form a relatively thick non-stoichiometric oxide deposition layer on the cathode. The anode does progressively get eroded at the same time. From close-up observation at a low magnification factor this layer on the cathode appears like it could be porous, or anyhow very far from being smooth or shiny. The higher the current applied (again depending on conditions which are not necessarily controllable in my case), the coarser it becomes, and with this, tiny reflective speckles also start appearing.

    I tried giving some thought on what could be occurring. I think the following processes might be taking place roughly at the same time:

    • Electrolysis causes Hydrogen and Oxygen ions to be formed at both electrodes to some degree (obviously)
    • Hydrogen-Oxygen recombination possibly occurs within the electrolyte to some extent due to the close proximity of both electrodes
    • The higher than normal temperature involved causes a certain amount of water to vigorously evaporate
      • This also provides a sort of "cushion" which to some extent prevents the electrodes from shorting out (at least in the beginning stages)
    • Electrodeposition from the anode also occurs, but metal ions might be getting partially oxidized in the process
    • As electrolysis and electrodeposition continue, hydrogen atoms might get incorporated or adsorbed within the mixed metal-oxide/carbonate layers formed, which can have a brittle quality
      • I previously briefly reported in another thread of crackling noise occurring from the deposition layer at the cathode after a period of operation, which might be supporting this

    I haven't performed any objective measurement yet, but potentially an environment containing layers of mixed oxides of Fe, Cr and Na (in addition to traces of C from the alloy) would be similar in some ways to the catalysts used in many industrial processes. Within this week I should have a Geiger counter which although it's not a very sensitive device, might allow to check out if anomalous emission, as improbable as it could be, is associated with the processes observed in these non-standard tests.

    For now I am left with a few questions:

    1. What could be actually occurring from a chemical standpoint besides or in alternative to what I listed?
    2. Besides Randell Mills (who does not consider his work to be LENR) and the examples given below, are there other ones in LENR research or related with LENR where dissociated water gets catalytically recombined on purpose over suitable catalysts?


  • Since I'm mostly copy-pasting text from my personal notes, the text below will be in a sort of blog format.

    Within this week I should have a Geiger counter which although it's not a very sensitive device, might allow to check out if anomalous emission, as improbable as it could be, is associated with the processes observed in these non-standard tests.

    Summary: not unexpectedly, following some quick tests no anomalous emission was observed. Still, there was something to learn in other areas.



    It took a good while for the postal service to deliver it, but my Japanese NetIO Geiger counter finally arrived earlier today. It seems to work fine, but the bright display might not make the 2x1.5V AA battery box last very long. It has a surplus Russian SBM-20 Geiger tube as expected. I noticed a couple small indentations on it but they do not seem to be affecting its functionality.

    This morning I tried it for some background measurements at my place; it initially varied from 25 to 45 CPM. Then took it for a stroll around my floor and sometimes it measured 50-60 CPM, peaking up to 65 CPM in some spot especially close to some walls. It seems in general it varies from 28-30 to 50-52 CPM; it's a bit higher and more variable than what other people often report with their own Geiger counters in their own experiments.

    I tried putting it close to a few bunches of ripe bananas and I didn't seem to be getting any real response from them. It could be that their signal is lower than my local background.

    Earlier I briefly inadvertently touched one of the exposed Geiger tube terminals with my hand and this caused counts to immediately jump off quite a bit. Clearly they are not meant to be touched.

    Preliminary tests and considerations

    I need take a long term CPM reading because they appear to vary noticeably even when apparently doing nothing special; I might attempt this later with a webcam. As a lower and upper limit I got 20-60 CPM in the room where I did in the afternoon some electrolytic tests which didn't seem to indicate that I was seeing elevated radiation emission. If anything, reported CPM might have even been lower than the normal background I was previously getting.

    Evening thoughts

    I just added USB power taking advantage of the extra cable provided, so batteries are not an issue right now. Data logging is, unfortunately. I planned to add a serial cable connection to my PC, but it needs 5V and the serial port from my PC outputs -11V. It might be cheaper to get from Amazon an Arduino clone starter kit which should handle 5V serial transmission directly than purchasing the officially made USB cable or making one with a signal conditioning circuit.

    Longer term background measurements

    In the evening I took a video of a 90 minutes background at my place. Converting this into usable data will require some work.


    Video data analysis

    The NetIO Geiger counter uses a pixel-based text display, so a simplified OCR program for segmented displays which I used in another thread for experiments by Alan Goldwater won't work. Google's Tesseract OCR engine will be used instead, but obtaining good results with it is not as straightforward as it does not work well with pixelized fonts. I placed a cheap HD webcam in front of the GC and left it recording at a 0.5 fps framerate for about two hours (in retrospect an even slower rate might have been desirable).

    ffmpeg processing sequence

    Every frame of the recorded is individually saved as image file using ffmpeg. Several image operations are performed in sequence in order to improve character recognizability (see ffmpeg filters documentation). It's important to scale up the images significantly as Tesseract appears to work best with large images:

    1. Convert to grayscale
    2. 180 degrees video rotation
    3. Crop desired value
    4. Apply gaussian filter
    5. Apply erosion filter
    6. Adjust contrast with curves
    7. Correct perspective distorsion
    8. Scale image x0.25
    9. Scale image x6
    10. Add 5 pixel white border
    1. ffmpeg -i video.flv -vf "colorchannelmixer=.3:.4:.3:0:.3:.4:.3:0:.3:.4:.3,rotate=PI,crop=160:52:385:140,gblur='sigma=2',erosion,curves=all='0.00/0.00 0.67/0.00 0.71/1.00 1.00/1.00',perspective=0:0:W-7:0:1:H:W-1:H-1,scale=w=0.25*iw:h=0.25*ih,scale=w=6*iw:h=6*ih,pad='iw+10:ih+10:5:5:0xFFFFFFFF'" -f image2 ./output/%05d.jpg

    From this:

    This is obtained:

    OCR with tesseract

    A simple (perhaps too simple) bash script executes in sequence OCR on the saved processed images and appends the result from stdout into a single file. Some flags are enabled in order to make tesseract work better with digits; errors are suppressed.

    1. #!/bin/bash
    2. BASEPATH=~/Videos/convtest/output/
    3. OUTPUTFILE=~/Videos/convtest/gc.txt
    4. rm $OUTPUTFILE
    5. for FN in `ls -1 $BASEPATH`; do
    6. tesseract --psm 7 $BASEPATH$FN stdout nobatch digits quiet -c tessedit_char_whitelist=0123456789 | head -n 1 >> $OUTPUTFILE
    7. echo $FN
    8. done

    Manual processing in Libreoffice Calc

    Excess whitespace from the saved text data is trimmed, then the file gets imported into a spreadsheet to create a suitable chart. Extensive manual tinkering is unfortunately required due to OCR failing in numerous instances, so overall the process isn't sustainable on the long run, at least with the current settings. The resulting graph is an almost 2-hours background measurement.

  • I tried taking another prolonged background reading, which unfortunately I had to abort after 40 minutes due to inadvertently bumping the webcam, disrupting OCR. If I am to assume that this is a normal background reading and that the Geiger tube is working correctly, I guess that unless I will get a very large signal I won't be able to tell whether I'm actually producing higher than background emissions.

    The attached photo shows today's recording arrangement. What I've put around the Geiger counter in theory not have any positive effect on the readings.

  • I had other business to attend to and setting up OCR is a somewhat involved process. The data up to that point was fine. I set it up again later on and it showed slightly lower values than earlier in the morning, but still higher than yesterday evening.

    There's probably some long term diurnal variation taking place, but I'd need a more stable way to get data from the Geiger counter to make sure without interruptions.

  • Some parts of Europe - particularly those with a relatively young geology, like those parts in the earthquake zone do have high and fluctuating background counts - possibly due to things like intermittent radon release. I think that google may have maps covering some of these 'hot zones'.

    In Essex UK the normal background on the Thames Valley clay belt is (with a NettIO and SBM-20) is 25-30 cpm. Remember always that CPM is a 'moving average' count, that smooths out brief highs and lows.

  • Alan Smith

    Generally speaking I do live in one such areas according to several very broad "Radon maps" I found online. I am located within 25 km from at least one dormant volcano after all.

    Quite an annoyance still that background CPM can range 20-80 at any given time of the day, although so far I've seen the highest values in the morning. To avoid false results I would either need to have a very large signal, a highly repeatable signal, or heavy shielding (probably not worth the effort and not clear if it would help significantly).

    To compute the CPM I use the total count value displayed, which allows to select shorter or longer periods like in the example below (over 24 seconds). I'm aware that this GC can also directly output pulses from one of the pins (I think its signal would have to be conditioned), but its UART serial interface apparently only gives the internal CPM value as reported by the display unless firmware modifications are performed.

  • We currently have 9 of these coupled to the Labjack T7's - my colleague Martin has managed to get them logging counts per second - but he is rather skilled at this. Previously he sorted out all the Arduino-based logging we used to use, if you want to know how that is done send me an email.

  • Alan Smith

    I found I can simply take the previously mentioned analog pulse output to the microphone input of an external sound card I have and it works fine, although it's kind of hacky. The pulses are positive and all have equal length and amplitude and they can be easily logged by analyzing the signal amplitude in real time, which I'm doing with a cobbled up together Python script.

    Although it should be kind of obvious, I just realized that since now I'm only retrieving the time of arrival of the pulses instead of measuring their increment over a moving fixed time interval, calculating the CPM is not so straightforward anymore. In other words, a sample-based plot is going to give slightly different results that I believe will exaggerate the differences between the low and high count periods.

  • It ended up being the other way around than I initially assumed. The sample-based CPM calculation appears to give somewhat more averaged values, at least under the current conditions where there are no severe changes in count rate going on. It would be interesting to check with an actual radioactive source.

    However, I'm not entirely sure if I did everything correctly for the fixed moving period version. The series was resampled to equally-spaced 0.1 seconds bins. The click count was interpolated to the nearest value, so some precision is lost, but no fractional value was created. Resampling the series to 1-second bins (thus less precise) smooths very slightly the plot, but it remains peakier than the naïvely plotted sample-based version.

  • I took a longer background measurement with my Geiger counter over the past 16 hours or so. It appears that long term average values can fluctuate up to at least 50% above their daily minimum. This is in the order of the increase reported in some LENR experiments. It looks like I'm getting the highest values in the morning as I previously thought I observed.

    EDIT: graph updated.

  • The geiger counter might possibly be measuring some sort of solar-induced variation; not sure what to make of this.

    EDIT: graph updated.

    EDIT2: updated again.





  • With an arrangement similar (but not identical) to the one briefly described in the opening post but designed to operate immersed in water I'm getting, when power is applied, continuous broadband noise (white noise) that my small transistor AM radio appears to pick up well especially at the lowest frequency band that I can set (530 kHz). No apparent response from the Geiger counter located about 40 cm away, but I haven't changed its location nor placed any material between it and the experiment.

    I tried to locate the noise source and it appears to be the narrow electrode gap, not the PSU (which does emit its own set of electrical noise). I have little idea if this is to be expected conventionally or not, but in an earlier test today (2018-10-26) using just wires as electrodes I could obtain a similar effect with the electrodes almost touching each other, but not actually shorting (electrolysis and gas production stop when this happens, together with the white noise).

    As the noise appears to stop as soon as power is removed, my guess is that it's either the switching-mode power supply I'm using (whether indirectly or not) or something related to the electrolytic processes occurring in the gap at very close distances. How to make sure?

    Interesting links

  • Could it be temperature sensitive? I always find my smoke alarm batteries die at 4-6am when it's coldest.

    Ambient temperature barely moved today, while the long-term average background radiation signal changed in the usual pattern; I don't think it is correlated with it to any significant extent.

    The holes in the data are due to interruptions resulted from software/data-logging tests.

  • Wyttenbach

    The background signal I'm getting apparently correlates with other space weather/solar activity data in different ways, not just the X-ray flux. Some fit more, some less, as long as the data is scaled in amplitude and shifted in time. For example below I added "Average flux of 210 keV protons measured by telescope 1 corrected for dead time but not electron contamination" from the file "g15_magpd_19mp3_16s_20181026_20181026.csv" available here:


  • Hopefully this is informative too. After trying to debug why I would get slightly less (about 2-3% overall) "clicks" than the what the Geiger counter displays, I found that the analog pulse output actually appears to not show as many pulses as reported, or at least this is from what I could see by counting them manually on the computer screen.

    This is how a pulse usually looks like:

    The pulses (which are all positive) usually have a width of 120-150 samples at 192 kHz. I'm simply detecting them by applying a negative bias to the audio signal and counting the zero crossings. Every pulse should add exactly two crossings.

    Perhaps the counter uses a different counting method?

    In the lengthy debugging session I improved the script to retrieve and log the pulses, which is now in a shareable state. Also posting here for backup purposes.

    Attached are a typical .csv log (renamed to .txt), console output showing timestamp, total counts and counts per tick (1 second) in a sort of semi-graphical manner. As the log uses a slightly different format, I'm not using this right now, I'm waiting until current measurements are finished to switch.

  • I noticed that the Geiger tube was mounted with the incorrect polarity:

    According to this website this causes the tube to operate inefficiently:



    The tube is clearly marked with a "+" on one of the terminals. But theoretically it should collect charge on either terminal. So what happens if you reverse it?

    [...] Overall appears to be less efficient and works under a much narrower range. Presumably GM tubes are optimized for positive voltage though so maybe it could be improved a lot if you intentionally tried to design a negative center GM tube. In any case, the biggest takeaway is that under standard 400 V operation the tube works. This is actually unfortunate as you might not notice you installed it backwards (mechanically possible).

    So I tried reversing it and indeed, I'm now getting a significantly higher CPM reading: currently 100+ CPM which is almost twice what I was previously getting. However, are SBM-20 tubes supposed to be this sensitive? I will post an updated graph soon but this will invalidate my long term background readings.

  • can

    Looking forward to the new data. What is confusing me is that you are getting elevated CPM during local nighttime and minimal detections just after local noon. To my naive understanding this is backwards

    To my surprise the same thing seems to seems to happen for the X-ray flux recorded by GOES15. That satellite is parked in a geosynchronous orbit at 135degrees west which is UCT-10. If I have figured things out right, then the flux is maximal at local night-time and minimal near local noon which seems backwards as well. I wonder if the detector on the satellite is directional.

  • Bruce__H

    I think some instrumentation might be directional. For example from the file named g15_epead_p1ew_8s_20181025_20181025.csv, from the public directory for the current month:

    float P1E_UNCOR_FLUX(record);

    P1E_UNCOR_FLUX:description = "Average flux of 2.5 MeV protons from the B detector that faces either East or West depending on the yaw flip of the satellite with no correction applied to remove electron contamination ";

    P1E_UNCOR_FLUX:long_label = "protons-1-B (2.5 MeV) flux";

    P1E_UNCOR_FLUX:short_label = "p1B fx";

    P1E_UNCOR_FLUX:plot_label = "p1B(2.5 MeV)";

    P1E_UNCOR_FLUX:lin_log = "log";

    P1E_UNCOR_FLUX:units = "p/(cm^2 s sr MeV)";

    P1E_UNCOR_FLUX:format = "E10.4";

    P1E_UNCOR_FLUX:nominal_min = "1.0000E-03";

    P1E_UNCOR_FLUX:nominal_max = "1.0000E+05";

    P1E_UNCOR_FLUX:missing_value = "-99999";

    However there is no such indication in the X-ray flux files.

    On the other hand, I don't know how the orientation of the satellite changes over time.

    Perhaps I could try looking if other data better correlates with the observed changes. I just tried plotting the "Average total magnetic field measured by magnetometer-1" from the file g15_magneto_512ms_20181028_20181028.csv (and previous days): without time shift it seems to better align with my data (however here I applied Y scaling+offset to make the plot clearer)

    EDIT: attached the Geiger clicks I logged so far, also for backup purposes.