Home | Software | Rules and About | Atom feed
Admin | Edit
published: Tuesday 28 February 2017
modified: Thursday 20 July 2017
author: Hales
markup: textile

Diodes part 3: a working diode steering circuit

I've made something interesting with my homemade diodes.

This first video is of a diode steering circuit being used to choose which segments of a 7-segment display to turn on. The diodes are the white blobs on the silver-looking board, the black dots are permanent marker to remind me where not to dab diodes:

Video download links: h264+vorbis mkv or vp9+opus webm

Apologies for sounding very monotonous in this video, it was the second or third take due to exposure issues. If you want to hear excitement then head to the video at the very bottom of this page, shot sometime around midnight a week or so ago.

Here is another video showing four diodes being tested individually, in case you suspect evil trickery in the first video:

Video download links: h264+vorbis mkv or vp9+opus webm

What's going on here

I've made what I call "wet diodes" using aluminium foil, bicarb soda, water and a mostly unreactive metal (in this case: solder).

They're slow. They dry out. But they work, and you can make them at home.

The second video I have above, of the diode matrix, is effectively a lookup table or an encoder . Encoders, with the assistance of transistors/decoders, can perform any calculation that does not require memory.

Similarly you can make logic work to a certain degree using things like diode resistor logic .

Why is this exciting

Two common routes to making and using diodes:

1. Industrial mass-manufacture fab, solder them to your board
2. One-off home fab using whiskers/point contact, then adjust as necessary

Buying mass-manufactured diodes is cheap, but soldering is slow. If you want to use a lot of diodes then you tend to redesign to use ICs instead.

One-off homemade diodes tend to be unreliable or fiddly. I've seen some pretty nice setups, but these tend to require a lot of parts and assembly. Not something that's easier to do than buying a diode and soldering it in.

Being able to print reliable diodes directly onto PCBs would be interesting. You don't need a pick and place or lots of soldering. And it's something both a factory and a home kitchen/bedroom can achieve.

Why this isn't exciting

(1) They only work when moist

I tried one particular clear sealant and it didn't work at all. Presumably it didn't like sticking to the wet diodes.

I've been given a few ideas to try, including PVA glue mixtures and freezing the diodes before spraying them. I'd love to hear about other ideas too.

(2) Only one half of the diode gets printed in the designs I show in this post. The aluminium side of them still has to be laid down manually.

This is what I want to be able to do (from a previous post):

Now that I've worked out how to make these diodes reliably I'm going to start working on this double-paste again.

Tinned PCBs are easy to make, and it would be a dream if you could just dab two compounds on it make diodes.

(3) They're slow/capacitive.

This is partially related to their scale (commercial diodes have tiny contact areas) but it's also due to their method of operation (more on this later). Making them smaller will definitely help.

(4) Poor performance compared to off-the-shelf diodes

Forwards: they drop around 5V at a few mA. My previous post has curve traces of this chemistry:

Backwards: you can see in the videos above that they leak. LEDs are pretty unforgiving in this area, as their light output (and the way our eyes work) is non-linear, so the actual numbers are better than it looks.

(5) Long term operation

Unknown. In previous tests I laid them directly on copper PCB and the copper dissolved in, ruining them.

I'm currently setting up some air-tight (and hence water-tight) sealed diodes to better test how these behave in the long term.

(5) They're not transistors.

Need I say more? You still need external components most of the time to get things done.

My understanding of how these diodes work

This is my best guess as to how these operate. Please share your thoughts if you think you know what's going on.

These diodes are very similar to electrolytic capacitors. Electrolytic capacitors charge up in one direction (eventually blocking current) and 'fail' in the other (conducting current).

Traditionally we describe simple P-N semiconductor junctions in terms of mobile electrons and holes. In these aluminium diodes it seems both electrons and ions are mobile.

When operating the diode in the normal (low-resistance) direction: the ions move to a side where they don't do any harm (or protect the aluminium from harm). The aluminium just near the junction stays mostly pure and conductive, or becomes so if it was previously turned into aluminium oxide.

When operating in the reverse (blocking) direction: the ions move to the opposite side as before and the aluminium touching the junction converts to aluminium oxide. This oxide is more resistive than the original aluminium.

How to make your own

To make these diodes you will need:

To make the test circuit you will need:

(1) Prep the diode paste

Pour a few cm of water into a glass. Add enough bicarb soda that you get little mounds developing at the bottom, enough that we can scrape it out later.

After a few minutes of settling you will have something that looks like this:

Now leave this glass sitting uncovered for a few hours. I've found that using the mixture immediately produces poor diodes. Leaving it until the next day is better.

I'm not exactly sure what's happening, but I presume the liquid is interacting with the air (eg altering the mixture's pH, changing what's dissolved). If you have issues getting your mixture to work then try boiled water or distilled water (and again leave it sit after pouring the soda bicarb in).

(2) Make the substrate

Cut a small piece of electrical tape and sit it on your worksurface sticky-side-up:

Stick aluminium foil to half of it and trim off all the excess:

Carefully place your coin onto the tape, as close to the foil as possible, but not touching it:

If your coin is relatively non-reactive then it should be fine. Don't use pure copper coins or the like, as you will damage them (and ruin your diode).

Ensure that your coin and your aluminium foil are not touching using a multimeter.

(3) Lay down the bicarb blob diodes

Your glass of water and bicarb should look like this:

You need to be able to tilt your glass and access the bicarb solids at the bottom, like this:

If you have too much water: gently pour it out until it is at a manageable level. Make sure to leave some behind, as it's useful to rejuvenate the dried diodes later.

I use the tip of a small flathead screwdriver to collect the bicarb solids and lay them on my boards. As mentioned previously, tilting the glass to expose the solids allows you to scrape them out:

Gently dab a tiny amount of the wet bicarb solids between the coin and the foil. You want this dollop to be as small as possible, eg less than 3mm in size, as shown in this first photo. Larger dollops, like the one shown in the second photo, do not work as well with our test circuit (but would probably be fine at higher currents).

(4) Test your diodes

A simple test circuit, as shown in the second video at the top of this page, is basically just a (1K) resistor, LED and a (approx 12VDC) power supply:

Ignore the fact I was using two LEDs in the above photo. I was still thinking I was using an AC source.

Try your diodes in each direction and see how they behave.

If they work in both directions:

If they do not work in either direction:

(5) Re-wet your diodes when they dry out

Generally these diodes only work for less than half an hour (depending on your local atmosphere). I use a cheap plastic pipette off eBay to re-wet them. Gentle dabbing with bamboo skewers, toothpicks or damp towels can also work.

Use the water left in your bicarb and water glass. This is already in the right condition for use. Distilled also works.

More advanced: making arrays and matrices on PCB boards

You will need:

Prepare your bicarb mixture as per the previous instructions.

How to make aluminium sticky-tape

I use thin double-sided tape (not the thick foam stuff) and aluminium foil:

I have tried some off-the-shelf aluminium sticky-tape, but it always seemed to short to whatever it was stuck to. If you want to use this sort of tape then laminate it with a layer of electrical tape underneath.

Diodes on a PCB strip

I'm using a small piece of singled-sided copper clad board, cut from a bigger piece. I get mine from eBay and cut it with kitchen scissors.

Here you can also see that I've put it on top of a piece of wood. We are going to hacksaw grooves into the board, and the hacksaw physically won't allow you to cut thin objects on your bench/table without raising them up first.

If you have extra double-sided tape and screws then use them to temporarily hold everything down. This makes things easier to cut. If not, hold things by hand and be careful.

Now gently cut the top copper layer of your board away to make several separate copper islands. Make sure they are at least 1cm wide or they will be difficult to dab and contain diodes on.

We need to cover the surface of our copper with solder. I clean my surface first with steel wool and then only tin half of it, which is (as you will see later) enough, but feel free to gently coat the whole surface.

Make sure to cover right to the edges and right into the corners. We don't want our diodes going green because they were in contact with some copper.

Clean the surface of your copper and solder. We need to ensure there is no water-based flux left on our board.

I scour my board with steel wool and then rub it clean with methylated spirits (alcohol). I would not recommend rubbing your board down with steel wool if you use leaded solder. Lead dust is not fun. Just stick to the metho.

Now cut and place a piece of aluminium tape across the centre of your board like so:


There are variations on how you can do this. I like leaving some of the copper exposed because it looks fancier and gives you somewhere to put your probes that is further away from the delicate diode mounds.

Full 7-segment steering matrix

I'm using a piece of 10cm by 7cm single-sided copper clad board from eBay. This size is serendipitous -- we want 10 input numbers and 7 output segments anyway. I mark my piece off in seven 1cm intervals lengthways:

In order to cut the top surface into separate tracks I use a hacksaw. Unfortunately this tool does not work if I try to cut the board on my bench, so I have to lift it up a bit on a scrap piece of wood:

If you have extra double-sided tape and screws then use them to temporarily hold everything down. This makes things easier to cut.

Now, as best you can, cut away the top copper layer along the lines you have drawn:

Use long strokes of your hacksaw and try to apply pressure evenly. I find the blade bows a bit and you end up cutting the edges faster than the centre.

It's worth checking that your copper strips are properly separated using a multimeter. Sometimes tiny pieces of copper remain.

Next we need to cover the whole board with a thin layer of solder. The surface is easier to solder to if you first clean it up with steel wool:

If there is one thing I hate in the lab it's steel wool. The little fibres get everywhere, including on the floor (I just had to pull another OW! bloody piece out of my foot). All of my hand tools end up very slightly magnetised too, so the stuff sticks to them like Christmas. Sweep this detritus into a bin as soon as you can -- large paintbrushes are useful.

Soldering this large area takes a bit of time with most equipment, so make sure to setup a comfortable environment. I still have my board stuck down from earlier and I have a computer fan running to move the smoke out of my face. The 3rd law of soldering is that smoke always takes a B-line directly to your face, regardless of where you are standing, so even an old small 80mm computer fan really helps.

Make sure to cover every corner. We don't want our diodes going green because they were in contact with some copper:

Clean the surface of your board. We need to ensure there is no water-based flux left on our board as we don't want this to interact with our diodes later.

I scour my board with steel wool and then rub it clean with methylated spirits (alcohol). I would not recommend rubbing your board down with steel wool if you use leaded solder. Lead dust is not fun. Just stick to the metho.

Now we need to mark out spacings for our aluminium strips. Mark out ten 1cm intervals in pen on the board, perpendicular to how we cut the board into slots:

Using pencil here would likely be fine too (it's probably not conductive enough to matter) but I err with caution.

Next cut your aluminium tape to a width a few cm larger than your board:

This allows you have 'aluminium wings' off the sides of your board, as shown in my video. I integrate the wooden block into the design, which is very convenient.

Cut your aluminium tape into ~5mm strips:

Lay them on your board, covering half of each pen-drawn region, as shown in this next photo. You also need to solder jumper wires to your PCB tracks:

If possible: make your life easier and write what each PCB column and aluminium row represents on your board:

Only certain intersections between the PCB columns and the aluminium rows must have diodes installed. If you are too lazy to work this out yourself then steal my table:

I recommend marking out where you don't want to put diodes down using a permanent marker. This makes things much easier:

Always make sure your substrate is not shorted

A few times I've only found out about this after putting the diodes on. I've been bitten by (amongst other things) my own poor cutting and dodgy aluminium tape.

Grab a multi-meter, put it on buzz or ohms mode and randomly test things.

End notes

When I was able to get my first homemade diode matrix working I was slightly excited. It was near a midnight somewhere and I couldn't believe my eyes:

Video download links: h264+vorbis mkv or vp9+opus webm

The problems I had with the first few digits were entirely due to them drying out. Some quick dabs of water fixed them immediately after this video.

You should go and make yourself some diodes. Grab a glass, prep your mixture and leave a coin on your desk. Share any thoughts, issues or results.

Commenting below requires no sign-up. Otherwise, feel free to shoot 'whales' an email at halestrom.net.

If you are curious for more background then I have written previous escapades into home-diode fabrication land:

  1. Homemade copper semiconductors
  2. Homemade aluminium semiconductors

This page has not come from no-where. I've slowly been making samples over the last few months whilst dodging between pieces of other work:

A blogger's posts are always erratic :)

vitor - Tuesday 7 March 2017

That's nice! I always wonder what would happen in an apocalyptic scenario where electronics manufacturing knowledge would be gone since very few people know the processes to build the components. Maybe technology would go back to relay logic.
That's why I think your work is amazing.
By the way, what's the voltage drop in that diode?

Dan - website - Tuesday 7 March 2017

Hi we are talking about this on hackaday.com at the moment. I have listed a few ideas for you to consider.

Hales - (site author) - website - Tuesday 7 March 2017


I've just run another few tests at the few mA level using both a large and a small dollop diode. They both started at around 6-7 volts forward, but then proceeded to drop over a matter of seconds. Interestingly they behaved very similarly otherwise.

5.0V 3.5mA
4.3V 4mA
4.0V 4.5mA

Even more curious is that our current is going _up_ as the voltage drops. The missing information here is time of operation. It seems they conduct better the longer you use them forwards. This could actually be an indicator of something bad happening (as I discovered with the copper dissolving into some of my previous tests). I'm not sure yet.

If we did go through an apocalypse and were trying to recover, I think we'd discover a lot of new things. Much of what limits our creativity are our biases based on what already exists.

Thanks Vitor.


Thankyou for the heads up. Getting a couple of comment notifications by email twigged me that something was up. I'll be replying to people's comments on both sites.

Hiro Protagonist - Friday 2 August 2019

Have you tried sealing the wet diodes with copolymer sealant? It's quite thick & sticky so may be difficult to work with, but it does stick to wet surfaces.

I presume you've seen this already, but in case you haven't: http://sparkbangbuzz.com/zinc-oxide-diode/zinc-oxide-diode.htm

Hales - (site author) - Friday 2 August 2019

(*adds Snow Crash to reading list*. Any good?)

Secretly behind the scenes I've complete a thesis investigating more on this topic. I focussed mainly on ways I might be able to get transistors to work using a similar chemistry. I didn't reach my goal, but I went through dozens of experiments and discovered a few papers I was trusting turned out to be outright wrong. A whole chunk of my report is dedicated to proving a device is a "useful transistor", with headings like "Amplification is not enough: easily misleading gain interpretations" and "Impedance modulation is not enough".

Through this I've been mainly focusing on functionality, not lifetime. I have not tried any two-part polymers yet, but that seems like a good idea.

Thankyou for the Zinc oxide rectifier link. I've read some of sparkbangbuzz's other pages, but not this one. I wonder how hard it would be to grow the oxide coating consistently enough to permit you to use a powdered-graphite + binder (silicone) contact material...

I've almost finished some other big life commitments, perhaps expect some more articles soon. Maybe. I'm not reliable with my promises :P

Hiro Protagonist - Saturday 3 August 2019

Snow Crash is good, it's actually many years since I read it though.

Copolymer sealant can be found next to RTV silicone at the hardware store. It's a single part sealant, looks like RTV, but it's a bit thicker. Unlike RTV it will work on wet surfaces.

Sparkbangbuzz is a cool site - the negative resistance and flame triode stuff is interesting. Pity it's not getting any more posts.

I agree getting away from 'point contact' would be a Good Thing in a practical device.

I hope to have a play with some of this stuff myself when I've finished my current house build.

Chiefrunningfist - Thursday 29 August 2019

Very cool!

Tyler Streeter - website - Wednesday 16 October 2019

I wonder if adding a humectant like glycerine (or even replacing the water entirely with glycerine) would help prevent drying out...

Tyler Streeter - website - Wednesday 16 October 2019

I was just reading a bit about glycerine/glycerin/glycerol, and I thought I’d add some info here in case it’s useful to anyone. Glycerol is a humectant, which pulls water out of the air. So in any given environment, the glycerol/water concentration will approach some equilibrium point which depends on the surrounding relative humidity.

This pdf contains a useful chart (Table H on p.12) which plots this relationship. Also see the discussion of this relationship on p.9. For example, at 50% relative humidity, the solution will approach 80% glycerol, 20% water.


Again, this might provide a way to hold onto enough water to keep the electrolyte operational...

Hales - (site author) - Monday 21 October 2019

Thanks Tyler. I was originally looking at Polyethylene Oxide, suggested by a nano-scale group I had advice from, but it doesn't seem to be hydrophilic enough to be useful at a macro scale (at least in the powdered form I was using it).

Glycerine definitely fits my "low-tech" requirements.

Add your own comment:

Email (optional):
URL (optional):
Enter the word 'irrlicht' (antispam):
Leave this box blank (antispam):

Comment (plaintext only):

If you provide an email address: it will only be used for the site admin to contact you, it will not be made public.

If you provide a URL: your name will hyperlink to it.