Unconventional electrolysis

  • magicsound

    Nice setup, way more efforts than it deserved given that the first runs reproducing my experimental conditions will be likely brief.

    Seeing the power.jpg photo you posted I have to point out that so far I have used the coil on the negative (cathode) side, but I haven't checked yet if it makes any practical difference to the process.

  • magicsound

    Nice setup, way more efforts than it deserved given that the first runs reproducing my experimental conditions will be likely brief.

    Seeing the power.jpg photo you posted I have to point out that so far I have used the coil on the negative (cathode) side, but I haven't checked yet if it makes any practical difference to the process.

    Brief or not, the basic setup should be durable. I used a short lead with connector on the cell, so it can be easily removed for cleaning or adjustment. I expect to do many runs, guided by your prior work and ongoing suggestions. For example, I set the spacing at 1 mm (bottom) and 0.5 mm (top), and will immerse the electrodes about half way. I selected the jar used because it has flat sides for better imaging, but the depth is more than the electrode length so more fluid will be needed.

    I just finished calibrating the current shunt for the scope. The bottom trace will show 4.83 amperes per volt. The noise bursts visible on that trace seem to be coming from the power supply, perhaps caused by the variable-speed fan control.

  • If these experiments look promising, you may want to make it more reliable and reproducible by adding an actuator to allow you to control movement of one of the electrodes relative to the other one. That way you could decide when to start it by using the actuator to move them together. The actuator could also pull them apart to clear shorts. You could also run an AC current to the actuator and experiment with different frequencies of operation.

    Many different kinds of actuators could be used, including solenoids, piezo stacks and voicecoil actuators. The type that seems best is a piezo bimorph bender similar to this:

    Piezo Ceramic Bimorph 40x10x0.5mm 2 KHz


    $19.80 /2 Pcs Set

    Dimensions: 40 x 10 x 0.5 mm

    Resonant frequency fr: 2 KHz±5%

    Maximum Deflection δ: 2mm (min)

    Maximum Input Voltage: 100 Vpp

    You would just attach one of the electrodes to then end of the bender with the neutral (nonactuated) position at a point near the nominal position of your fixed-position tests. A positive voltage to the bender would short out the electrodes and a negative voltage would clear out shorts. Fixed frequency oscillations could be driven by a simple center-tapped transformer. With a little electronics, or a piezo driver like this, you could drive custom waveforms from a signal generator or processor.


  • Robert Horst  
    Yes, lots of possibilities, including Parkhomov's "woodpecker" device, a carbon rod through a solenoid coil for self-actuation.

    The video stream is about to start (if Utube cooperates). The jar is half-filled with deionized water. Some HCl will be added after power-up.

    Looks like ManyCam no longer has integral streaming support. I'll look for a fix beforethe next test. Sorry...

  • Thank you Magic. This is what LF is mostly about. All the gossip is only to keep the audience interested until the real show starts.

    Maybe someone other than the man running that show in this case (who is busy enough as is), could provide some commentary for the blind? Tell us what we are seeing as it progresses.

  • I started with 1 ml of dilute HCl added to 175 ml of water in the cell. There's a lot of white foam forming. The acid I used is a cleaning product and may have some other stuff (soap?) in it. The current is slowly increasing but still only ~250 mA.

  • Lab computer shut down as I was adjusting the scope settings. Stand by....

    After reboot, the same failure happened when I launched the scope's desktop application.

    More work is needed to debug this complex setup. With four USB instruments, the video camera and GPIB for the scope all running through Google Remote Desktop, it may be too much for the CPU to handle. Or it could be power supply failure.

    Sorry for the tease, but that's why it's called testing. I'll let the cell run for a bit longer, and will make a fresh start tomorrow.

  • magicsound

    From what I managed to watch through the live stream recording it does appear that current increased slowly following HCl addition. In the latest runs what I typically do is adding HCl in small amounts (drops from a plastic straw), wait a couple minutes, then add more again if no significant reaction is observed. After a few times this is performed, current jumps to several A and the process begins. In the very latest ones where I used less water than usual the reaction started immediately after the first addition. I use plain 10% HCl from the grocery store with no perfumes or detergents added.

    One main difference from my tests is that you're using a significantly larger volume of water, which might need proportionally more HCl.

    As a side note, in the early tests I performed in this series I would typically let electrolysis in deionized water proceed for a while (hours), leaving the cell semi-unattended. This would eventually cause the formation of a relatively thick deposition layer which shortened the effective gap width between both electrodes. Then, once stable conditions were reached, just adding very slight amounts of HCl would immediately cause current to ramp up substantially, together with discharges and noises, and the solution to blacken from the dissolved metal particles.

    While I later found that all that waiting was unnecessary, it could still be helpful knowing that a shorter gap can make the process easier to start, but also less manageable (more likely to cause unrecoverable short circuits which need manual intervention to be cleared), which is why I eventually settled for a larger gap width on average than I initially thought would be ideal.


    The experiments shown so far essentially use pulsed DC. Ultrasounds could be a very good idea to try getting rid of persistent shorts, and probably even help achieving more stable/reliable operation, but I originally intended to keep these tests as simple as possible and preferably self-regulated if possible.

    As they are right now, they can be potentially reproduced just using one or several 12V batteries (or a readily available, sufficiently large 12V DC power supply) and a few dozen meters of wire. It's hard to imagine they could get any simpler.

  • To make sure yet again that I could easily reproduce the resonant noise that magicsound was interested testing with more appropriate instrumentation, earlier I did another test. Following some suggestions I wanted to try setting up a narrower gap at the bottom and a larger gap at the top while keeping the rest very close to previous experiments. It's possible that a condition similar to a Jacob's ladder could arise, although this wasn't my original plan. There's also the chance that a larger gap at the top could improve overall reliability since it tends to get filled up with conductive material that does not get easily removed with usage.

    EDIT: for those who haven't followed these tests all along, below is an older photo showing more in detail how the clip is typically installed (mica insulator sheets are used to prevent the clip from shorting the electrodes right away. It's not optimal but it works for quick tests):


    • Spacers
      • Bottom spacers piled to obtain a 0.4mm gap
      • Top spacers folded to obtain roughly a 1.4mm gap
    • Electrodes
      • Cleaned with a plastic brush under warm water
      • Will try to install only a clip at the top
        • The bottom spacers will be held by residual pressure. For this to work the top clip has to be located just below the top spacers
    • Coil
      • Will still be the one with the perhaps-slightly-improved core composed of mild steel bolts and tools
    • Instrumentation
      • To make the experiment quicker to perform, only audio and Geiger count recordings will occur.

    Experimental notes

    Photos taken during the experiment


    Command used to convert videos (semi-proprietary mpg interlaced video from a 12-years old Panasonic SDR-H20 video camera) into a better format:

    1. ffmpeg -i MOV001.MOD -deinterlace -c:a copy -aspect 16:9 unconv-yyyymmdd-n-comment.mp4
    • 001
      • The experiment just after adding some HCl drops. The perceived frequency of the hissing noise seemed higher than usual.
    • 002
      • A restart attempt after stopping the experiment due to excessive current load. The water at this point has become dark. It overloaded again.
    • 003
      • A failed restarting attempt due to the usual issue. The noise started at a low rate in the beginning.


    • With the inverse gap spacing arrangement I could still reproduce the same effect quickly, so again at least this does not seem to be an issue or due to just luck at least in my case.
    • Reliability however is indeed still an issue. I think excessive accumulation of debris on areas of low current density is occurring near the clip that was placed to hold the electrodes together. This eventually causes an unrecoverable short circuit.
      • A different method for keeping the electrodes together will be needed.
    • I didn't notice by eye significant discharges, but they had to be occurring.
      • Upon watching the videos I made, I noticed some visible discharges there. Could it be that the camera CCD sensor has a better sensitivity to them?
    • The black conductive material deposited on electrode parts and components after some time and with the help of low amounts heat becomes brown.
      • The same occurs with the electrodes themselves. It appears as if they quickly superficially rust after a period of operation. This might be due to HCl/acidic residues promoting oxidation upon exposure to air.

    EDIT: some excerpts from the audio (recorded from the headphones output of the AM radio tuned to 1600 kHz, placed near to the jar and coil) close to where the pulse/discharge rate seemed the fastest.

    112 samples @ 192 kHz = 0.00058333s = 1714 Hz

    Going along what is often seen in oscilloscope readings with automotive ignition coils, a high voltage spike should correspond to the initial brief signal peak. The average voltage along the duration of a single discharge event should be significantly lower.

    In the denoised spectrogram at least 7 harmonics could be seen at times:

    EDIT2: no significant change in Geiger counts (counter located roughly 60 cm above the cell) during the experiment. Red dashed lines denote start-end. As usual I had the window open during the experiment.

  • I noticed from the MFMP Facebook page that this comment was posted on the post associated with magicsound's experiment, which might explain the lack of updates yesterday on this matter:

    Martin Fleischmann Memorial Project wrote:

    After a short run, the motherboard of the lab computer failed and so Alan is swapping in a spare.

    Hopefully (or not?) this was not due to the electrodes sending out emissions harmful to the equipment after adding HCl. For what it's worth, so far I've had odd instances where the multimeter would occasionally show garbage values in earlier experiments with narrow-gap electrolysis experiments where I used KOH, but not on those with HCl. In the experiments with KOH, material deposition did not occur to any appreciable extent and discharges did not occur at all.

    On the other hand sometimes upon adding HCl on the newer discharge experiments, especially after letting the cell run for a while with standard electrolysis at a low rate with distilled water, I would sometimes feel light-headed for a short while. I don't think this was due to the emission of harmful chemical compounds (far more would be emitted under high current conditions and water getting dissociated and evaporating) but it's well possible this perceived effect could have been some form of autosuggestion.

  • I noticed from the MFMP Facebook page that this comment was posted on the post associated with magicsound's experiment, which might explain the lack of updates yesterday on this matter:

    I traced it to power supply failure. Yesterday I upgraded both lab computers with new 400-watt supplies and did some full-load testing with good results. I also added a current shunt for the PA1000 power monitor. This morning I will calibrate that input and also move the camera closer to the cell. Then time permitting, a short run with fresh electrolyte this afternoon.

  • I'm beginning a test run with fresh electrolyte. The streaming video with chat sidebar is available at


    21:10 1 ml HCl added. New log file started.

    21:30 1 ml HCl added.

    21:40 1 ml HCl added. Bubbles starting to appear.

    21:50 1 ml HCl added.

    22:00 1 ml HCl added.

    22:10 1 ml HCl added. Sharp increase in current seen.

    22:34 1 ml HCl added. Yellow color starting to appear in the electrolyte.

    22:44 1 ml HCl added. Current is up to 3.4 A. Black particles falling off the electrodes.

    22:55 1 ml HCl added. Current is 6.3 A. Vigorous bubbling and the cell is getting noticeably hot.

    23:10 The electrolyte seems to be boiling around the electrodes. Steam escaping and sizzling sound heard. Current is 4.7 A.

    23:18 Current down to 1.6 A. 1 ml HCl added, no change in current. Bubbling has mostly stopped.

    23:35 Current increasing to ~4 A accompanied by intense bubbling and some spark events. Voltage transient 21 volts seen at electrodes.

    23:55 Current up to 10 A. Sparking has stopped and electrolyte is boiling at the electrodes.

    00:40 About 1/3 of the electrolyte has boiled off. Current has dropped to 3.5 A and the cell is quiet.

  • magicsound

    Seems to be progressing slowly, possibly due to the large amount of water (is it even more than last time?) compared to what I typically use.

    In several instances I found reactions to more likely occur with the water close to boiling or boiling, so this might be a factor too in your case.

    Do you have an AM radio to monitor the reaction as well? It could be useful/interesting to add subjective observations of the changes noted over time with one.

    EDIT: perhaps under your conditions a narrower electrode gap will work better and faster. The arcing comes from electrodeposited material short-circuiting both electrodes. Since the electrolyte in your case is indeed getting dark, it could be a matter of allowing such material less chance/time to get transported out of the gap.

    If this is a change that you can do in real time by temporarily getting the electrodes out of the solution and insert thinner spacer material, perhaps you will be able to reproduce the sound and arcing during this testing session.

    EDIT2: I'm noticing that current has jumped back to higher values and that it is now irregular, so it could be close to short-circuiting intermittently if it isn't already.

  • It seems like you would need a sustained short in order to dump energy into the inductor during the current pulse. Then when the high current clears the short, the inductor delivers the high voltage to cause the sparking.

    If your initial electrode spacing is slightly too large, there would never be enough plating to short out the electrode. There might also be problems if the electrodes are not flat and parallel enough. If a tiny bump shorts out first, a small current would clear that short too quickly and not allow enough energy to be dumped into the inductor.

    Also Magicsound's inductor may have much higher L than CANs because the bolt he is using appears to be a much better core. The higher the inductance, the longer it takes for current ti build up, meaning that the short has to stay longer in order to charge up the inductor.

    This is why I suggested an actuator to force the electrodes together to start the reaction.