I've been experimenting with Aluminum/borax or aluminum/baking soda diodes, here are the results
This graph shows a typical peak ratio at about 100v. Note this isn't always
true and I don't know why. Sometimes especially after the diodes have been working for a long time the ratio approaches 1.33 for anything between 100v and 140v in.
The above diodes in legos and/or duplos worked, but the volume of fluid
wasn't enough to prevent thermal runaway even with light loads, also it was
too hard to keep oil on the surface and too hard to work with.
These diodes use the same coils as above, but the coils are bent in a semicircle
and there is 150ml of electrolyte in each diode.
Using Al diac to try and protect a silicon diode.
Graph of a sinewave being put across a back to back diode (really two aluminum rods.)
this is what I think is happening Al+H2O+NaCO3+e(470v)=>Al(OH)3 production occours when cell starts to conduct, voltage stays at or below 470v (think 470v MOV)
this is the back to back diode (really two aluminum rods.) used in the test.
There are apparently two different modes of this Al diode in reverse bias, the
first (in this case) is the glowing phase which works up to about 370 volts,
then some kind of breakdown discharge starts all over the aluminum surface, its
a percussive random bright pinpoints affair with blue/violet sparkles
gradually turning to "sodium orange" over time. Note that the brightness of
smaller plates of possibly different Al alloys glow much better than the 6061
shown here. This percussive sparkly phase uses a lot of current (200ma)
whereas the glowing bit uses more like 10ma.
The leakage current in reverse conduction with at 50ohm load (lightbulb) is only 2ma at 186vpeaks.
Putting AC through the diode, causes the aluminum to electrolyze, and the
coating to 'improve' the reverse breakdown voltage went up to about 380volts
before the sparklies started after doing this.
I'm noticing here that the alloys of Aluminum that block most of the
current in the reverse direction do not work well as displays and those that
work well as displays are a little leakier, no idea why yet. (I mean no idea
why some aluminum alloys work better than others, I understand energy for the
electrophoresis has to come from somewhere...)
2024 Al with 2-6% copper glows brightly 6061 hardly at all.
a 8 oz mason jar with 36 in^2 Al (2" pipe 2.75" high) makes a really good diode.
To drop the no load power consumption of these diodes 3 sugar packets in one 8oz
diode will cut the no-load power (with 100vdc out) from 18 watts to 1-2 watts (some of this power was
because the aluminum plates are so big.) In addition at full power the efficiency
goes from somewhere around 34% to 44%. It turns out this is a red herring, anything that changes the capacitance of half the bridge relative to the other starts to make the bridge more of a voltage doubler.
304 stainless rusts too quickly
24 tsp Baking Soda/gal. seems about right, nope 32 tsp/gal is better.
Ti cathode seems to limit the current way too much and prevents proper diode operation?
Al-Ti doesn't work well Ta-Ti doesn't work well.
Al-Stainless seems to work the best.
Al-Pb rusts red too quickly.
Best ratio of Al-SS diode dc out vs ac in is at 85vac in. 1.33* vout, up to 100vac in (slightly worse?)
the peak voltage where the best ratio now seems to be around 90vac sometimes 100vac in.
Threaded aluminum rod doesn't work well because of the sharp points? This evolved into a single coil
An oil layer should be used on top to prevent evaporation and prevent the crystallization of the baking soda out of solution.
Al+H2O+NaCO3+e(470v)=>Al(OH)3 production occours when cell starts to conduct, voltage stays at or below 470v (think 470v MOV)
I'm getting 54-55% efficiency at high loads equivalent to 3-11kw of total diode stacks of from 51-72 diodes with 7200vac input
Best arrangement now seems to be a cylinder of 1/4" aluminum 2" diameter 2" submerged for anode (6061 is fine, pure might be better), stainless steel 1/8" wire coil for cathode.
This diac was used to try and protect a silicon diode. It worked a little bit, but eventually the silicon diode died.
Test setup using neon sign transformer.
Test setup using neon sign transformer, and two aluminum diacs.