Unconventional electrolysis

Robert Horst

It is certainly way more painful than I expected, but I wanted to see to the end how viable the process was after installing the webcam on a fixed mounting point and if what I had already saved could be somehow salvaged. Regardless of the outcome, I think at least the test (and the previous one) was successful in showing what it takes and what problems could arise in the process.

By design the system is intended to allow an as high as possible current during transients (short circuits at the electrode assembly) within the limitations of the cabling and the power supply. During such transients the total resistance of the coil+electrodes can be as low as roughly 0.3 Ohm. A 0.1 Ohm resistor would affect this appreciably - I'd need a smaller coil to compensate. I haven't determined with certainty yet if the coil is strictly required (i.e. if it can be replaced with a resistor), but from a short test made during the 2018-12-14 session, placing a ferromagnetic core within it appeared to have a positive effect.

So far I've been testing with improvised (scavenged, even) equipment trying to lower costs as much as possible, but in part also in an attempt to prove (to myself at least) that there should be no need to invest a lot of money to observe appreciable results. It would seem odd within this context to spend almost nothing on the actual experiment, but hundreds on the measuring equipment.

I'm wondering if the bandwidth of a computer sound card (96/192 kHz) would be sufficient to properly work as an ultra-low cost oscilloscope here, once it is determined that the coil can be replaced with something else. The data could be stored as an audio wave file and be processed later on with software tools.

https://www.build-electronic-c…/sound-card-oscilloscope/

https://makezine.com/projects/sound-card-oscilloscope/

Robert Horst

I made a quick test and the microphone/line input of the integrated sound card of my PC indeed seems to detect only pulses and not a small constant voltage. So it's a no go.

A Hall sensor looks like it would be more tolerant to temperature and current spikes and be significantly less expensive than a precision current shunt so that seems attractive. I would still need an oscilloscope to read it and determine where I'd need to use it exactly, though.

Perhaps for the purpose of measuring the overall efficiency of the setup using the current clamp I already have (which has true RMS AC readings) on the AC side of the power supply would be the simplest choice. The switching PSU I'm using has active PFC. It's been measured to have 77% efficiency and a P.F. of 99% at about 25% DC load (~123W): http://www.jonnyguru.com/modul…Reviews&op=Story2&reid=18

I just made a test and when turned on and with no load, the PSU shows a current usage of 0.10-0.11 A (at 238V).

magicsound

Thanks to you for still committing to reproduce the effect and measure it with better equipment, regardless of the possible outcome. I will try checking out over the next days if it can be reproduced with similar materials of different dimensions, in addition to the current set of rather worn electrodes (couldn't manage to last time, but I had one of the components holding them together fail; excessive residue accumulation in the aqueous solution probably didn't help either).

Today I attempted checking out if by starting from a cleaner state (cleaned/scraped electrodes and new aqueous solution) I could reproduce again the resonant (?) noise even though the electrodes are quite worn, and I can confirm that I did.

I used a slightly different configuration this time. Instead of using zip ties I used clips normally employed for holding small objects like paper sheets, banknotes, etc. Furthermore I tried covering the backside of the electrodes with mica sheets similarly to what I originally planned to do (ideally slightly thicker mica sheets composed of one single piece would have been used). I used 5x 0.09mm mica sheets as spacers, although since the electrodes are quite worn, the effective starting gap will be larger than this. But it will also change dynamically when operating.

The hissing noise seems to be more related to water being close to boiling temperature rather than being directly due to the short-circuits causing plasma formation. Those can occur relatively silently. Today, perhaps due to the mica sheets causing electrolysis to occur more inside the gap rather than along the entire immersed electrode surface, I had more intense reactions, at times almost impressive, under the aqueous solution (initially distilled water + a few drops of 10% HCl slowly and progressively added a few minutes after the experiment started).

Progression of water color during the test (the first photo was before actually starting it. I added some more distilled water after that photo):

Videos

Here sometimes there were flashes occurring underwater, but the jar was silent for the most part.

https://streamable.com/zbwy3

(Direct link)

This video is a bit blurry unfortunately. This is after the electrode started hissing again. It can be seen through the dark water (which was initially clear) that the electrodes look incandescent and/or emit relatively strong light intermittently in several spots, but in particular on the most bottom part. Toward the end of the video the reaction suddenly intensifies and I had to turn off the PSU to avoid (or mitigate, at least) spilling.

https://streamable.com/864ec

(Direct link)

More of the same on a later moment of the test. From the light reflected from the Al dish, it seemed it was even more intense on the side opposite of that under view.

https://streamable.com/y2kdf

(Direct link)

Sorry for the harsh crop. Here the cell was hissing relatively loudly, drawing at times large amounts of current. Eventually it short-circuited completely causing the PSU to trip and the current clamp to complain that current was >40A. I couldn't restore normal operation after this and the experiment was subsequently terminated.

https://streamable.com/5x9pu

(Direct link)

Alan Smith

Thanks, although bright lights might be all that was produced in the latest test. Using the same method of measuring water evaporated (or that left the cell), but more accurately than in the previous ones, and calculating power from I and V, it doesn't look like there have been hints of excess energy. Update 2018-12-22: It looks like I previously made a silly mistake in the calculation which halved the amount of water leaving the jar. Hopefully correct calculations:

- Jar weight before the test began = 136.35g

- Water added from refills during the test with a syringe = 4.03g

- Jar weight after the test finished = 131.81g

- Jar weight before the test + water refilled = [136.35g + 4.03g] = 140.38g

- Jar weight difference = [140.38g - 131.81g] = 8.57g

- If this was only due to evaporation = [8.57g * 2.257 kJ/g] = 19.34 kJ

No elevated signal at the Geiger counter either, but that might possibly require different methods/arrangements than what I'm using at the moment. The dashed lines indicate when I started and stopped the experiment. It seems odd that the signal would decrease coincidentally during these events, but it might possibly be due to external factors like opening the room window during the test. Furthermore It usually decreases at about that time regardless of what I'm doing.

Perhaps the most useful information from today, which magicsound could take advantage of as well in his attempt in January, is that covering the backside of the electrodes with mica sheets (or any other chemical and heat resistant covering) appears to make electrolysis and thus the reaction stronger inside the gap, or at least that's my interpretation of what happened.

I'm wondering if changing the water solution with new one also decreased the apparent excess heat previously reported, or if it is just a matter of error margins/inaccuracies in the calculations. Possibly both.

In the spoiler tag are notes I took while I ran the experiment. That's what happens in a typical run like this.

By the way, it looks like a kind of plasma fireball (?) got produced during one of the videos.

magicsound

What kind of paint, more in detail? I have the idea that perhaps ceramic-based high-temperature coatings could be preferable over ordinary silicone-based spray paints, but they might end up being excessively expensive and specialized for this kind of experiment. I don't have any practical usage experience with either, though.

Alan Smith

After reading around it does seem like J-B Weld has most of the requirements for this kind of application (even acidic environment resistance), in addition of being readily available. I wonder how it would fare against local high temperature transients, though. According to the company's FAQs, heating above 600 °F (315 °C) removes it, and it can withstand 10 minutes at this temperature. The average electrode temperature will be much lower than this, on the other hand.

I was partial to the thick mica sheets idea since in addition of being good insulators and having excellent chemical- and heat- resistance they can be quickly disassembled or replaced, without extensive downtimes from curing times or further deteriorating the already messy conditions of these tests (for example the black aqueous solution formed when the experiment is running tarnishes everything). From how the electrode area below the spacers - which are not held by a huge clamping force - appears to have resisted against erosion so far, a perfect seal might not be that required either.

I don't have them yet, but I intended getting some of those thicker/larger universal sheets for replacing waveguide covers in microwave ovens, to be cut into the desired shape.

Robert Horst

That looks interesting for a solution which on a first look at least on Windows wouldn't require any coding to be used. However, from what I read (since I also considered eventually getting an Arduino in some form) most of these low cost data logging systems don't have sampling rates higher than a few kSamples/s for a single sensor at their maximum resolution and the one linked seems to offer 1 kSamples at most. Wouldn't this be insufficient for more than basic oscilloscope needs?

Most of the company's efforts software-side appear to be on the Windows tools, while the Linux ones are limited to a simple simple command line utility to read/write values from the device, that hasn't received any update in years: https://github.com/baycom/iCP12 . To make this useful some sort of script or program would have to be written (not a huge problem) but at this point this wouldn't be much different than an Arduino or similar boards, unless I'm missing something here.

Since availability outside its official store also seems kind of limited, wouldn't an Arduino Mega2560 R3 clone (newer, faster hardware compared to earlier and slightly cheaper Uno models) provide more flexibility and support and be quicker to replace in case of failure? It can be found for prices as low as 14 euros where I am. If I haven't one yet at this point it's because I tend to not impulse-buy materials and equipment just for the purpose of doing these experiments.

For the record and those who might care, yesterday I tried a couple more tests after adding to the same 25ml solution the equivalent of about 0.1M KOH (a small 0.22g KOH flake). In the first attempt I couldn't manage to get the electrodes working as intended, but I used them exactly as I left them last time (no cleaning performed). I haven't taken many photos of this quick test.

In the second attempt after removing excess residues from them just with paper towel, the reaction (mainly electrolysis) would proceed on at a rather high current, but with no plasma-forming reaction or resonant noise produced except for a few flashes and fizzes at the beginning; the AM radio was also for the most part silent too. Upon cleaning the electrodes after the experiment with a brush and warm water, the black iron layer that usually is rather difficult to remove would now get off easily. I have to conclude that at least to observe a rather "active" jar, KOH addition is not particularly helpful or must be counterbalanced by an appropriate amount of HCl.

It's likely that the pH wasn't low enough and that I should have added more 10% HCl. From a quick calculation it looks like I should have added about 1.5g of it, which is way more than I added in the form of small drops. I'm aware that pH meters exist, but the black aqueous solution formed would probably ruin them quickly. pH test strips might be preferable here and unless anybody has a better suggestion I might eventually get some in the coming weeks.

No coincidental sharp drop or rise in Geiger counts was noted as I performed the experiment, but they still did drop somewhat. I think this is part of a regular signal that regularly occurs at around that time of the day or in part associated with opening the windows of the room/house, but I'm not entirely sure. No comparable drop occurred in the day before yesterday. Red dashed lines here denote the start/finish of various test sessions. The one on the left part of the graph was associated with the flashy experiment of which I posted videos earlier.

A large problem I had this time despite initial issues with uncleaned electrodes was the coil heating up significantly, not cooling down at a sufficiently fast rate and forcing me to stop the test. This is mainly due to the intermittent action/small short circuits not occurring as intended causing prolonged current surges. Furthermore the more it heated up, the higher its resistance should have been. This time the measured apparent heat (I'm aware that the error margins are high for various reasons) seemed lower than in the past few tests, but on the other hand there is the chance that this could be signaling that current measurements with my current probe are more accurate with less distorted DC waveforms.

Other semi-random observations:

- I noticed that spraying water on the electrodes while they were operating caused a load surge which in turn caused the inductor to attract nearby ferromagnetic materials. This could be seen in the above graph with power spikes associated with water refills (the time for them is not perfectly accurate)

- I'm wondering if with the Chlorine in HCl there might also be some relation with the halogen cycle in light bulbs.

Alan Smith

In that context I meant the brown coil located outside the cell, as visible in the screenshot below from a previously posted video, depicting how the setup is typically arranged:

Following some of the prolonged current surges, portions of the coil were too hot to touch and the coil itself would start smelling like "almost burning" plastic. According to some sources:

Quote

The Temperature Coefficient of Copper (near room temperature) is +0.393 percent per degree C. This means if the temperature increases 1°C, the resistance will increase 0.393%.