Building Triodes With Blinker Fluid

The triode is one of the simplest kinds of vacuum tubes. Inside its evacuated glass envelope, the triode really is just a few bits of wire and metal. Triodes are able to amplify signals simply by heating a cathode, and modulating the flow of electrons to the anode with a control grid. Triodes, and their semiconductor cousin the transistor, are the basis of everything we do with electricity.

Because triodes are so fantastically simple, they’re the parts most commonly crafted by the homebrew tube artisans of today. You don’t need a glass blowing lathe to make the most basic vacuum tube, though: [Marcel] built one from the light bulb used in a car’s tail light.

The light bulb in your car’s tail light has two filaments inside: one for the normal tail light, and a second one that comes on when you brake. By burning out the dimmer filament, [Marcel] created the simplest vacuum tube device possible. In his first experiment, he turned this broken light bulb into a diode by using the disconnected filament as the anode, and the burning filament as the cathode. [Marcel] attached a 1M resistor and measured 30mV across it. It was a diode, with 30μA flowing.

The triode is just a diode with a grid, but [Marcel] couldn’t open up the light bulb to install a piece of metal. Instead, he wrapped the bulb in aluminum foil. After many attempts, [Marcel] eventually got some amplification out of his light bulb triode.

The performance is terrible – this light bulb triode actually has an “amplification” of -108dB, making it a complete waste of energy and time. It does demonstrate the concept though, even though the grid isn’t between the anode and cathode, and this light bulb is probably filled with argon. It does work in the most perverse sense of the word,  and makes for a very interesting build.

67 thoughts on “Building Triodes With Blinker Fluid

      1. Any pressure will create xrays at high enough voltage … The higher the vacuum the more it will produce … This is safe as long as you don’t put it under your pillow every night for 10 years

        1. It you tried to put more than a few kv across this bulb it would ionize and melt the internal structures of the bulb before you were able to get any measurable x-rays out

          1. Yes – that is why colour CRT television sets may never have a EHT voltage above about about 28kV (can’t remember exact voltage), no matter how bright they are. Since energy / frequency of radiation produced is related to tube voltage, any higher anode voltage produces radiation harmful to humans. (The energy of particles emitted is even measured in the units of “electron-volts”. Energy is directly proportional to the frequency of the particle/wave)

          2. The energy of the X-rays is measured on eV. If a field potential of 20kV across a vacuum accelerates a stream of electrons from a rest state to cause them to impinge upon a target at 20kV (relative to the source of electrons), then the energy of the electrons at the point of impacting the target is 20keV. The current only increases the electron beam current (the number of electrons).

    1. It’d still be awful, far as I know he hasn’t taken the glass out from between the cathode and the anode. But the principle is great! Everyone knows how this works, but there’s something about having it in your hands, having made it yourself.

      If I had a vacuum pump and a jam jar I might try something like this myself. Wonder if a repurposed aquarium pump or something would do?

        1. Nonsense. My clothes dryer has a heating element that glows orange all the time, and it’s operating at 760 torr, and hasn’t burned out after many hours of service. Many vacuum tubes have cooler filaments than this. Some, like vacuum-fluorescent display tubes, don’t even glow hot enough to be visible.

          I don’t know how good a vacuum it takes to make a respectable vacuum tube, but burning out the filament is not your main worry. The main problem with air in a vacuum tube is that if the plate voltage is high enough to ionize the air, it will always conduct. There won’t be much change in conduction due to grid voltage, so the performance will suck. Maybe not -108 dB suck, but suck just the same.

          But NO, an aquarium pump won’t do it. They just aren’t designed for a lot of differential pressure. I HAVE seen hand-operated vacuum pumps made for refrigeration service that are essentially bicycle pumps with a better piston seal and the check valve operating in the opposite direction. Also look at brake bleeder pumps, which are also hand-operated vacuum pumps.

          Hack away!

          1. A nichrome heater element in a dryer is totally different than the tungsten filament in a tube. Trust me, I have seen what happens to tungsten when it is hot in a vacuum system that is not under high enough of a vacuum. You get all sorts of nice yellow and white oxides.

            Those hand pumped things you see for refrigeration are for pumping compressor oil back in to the system while it is under pressure, they are not for drawing a vacuum. The vacuum pumps used for refrigeration are rotary oil pumps similar to what is used to back diffusion or turbo pumps, just smaller. But it is still orders of magnitudes higher than you will be able to get.

            Vacuum tubes need a hard vacuum, traditionally they were pumped down with a diffusion pump and then a getter is fired to get rid of the rest of the reactive gases. It can be done at home, I have the equipment to do it and so do a lot of others, so it is not out of the reach of hackers, but you will not do it with a hand vacuum pump.

          2. 1) who said anything about using tungsten as the heater?
            2) do you KNOW how hard a vacuum tubes need to work? References?
            3) the refrigeration pumps I was talking about (that look like bicycle pumps) most certainly DO pump vacuum – that’s what they were made for and what I used them for when working on refrigerators. The hand pumps I mentioned, used for brake bleeding, most certainly do suck. I don’t know what hand pumps you’re talking about.

          3. Tungsten is almost exclusively used for thermionic emitters.

            At least 10^-6, Page 524, https://archive.org/details/MaterialsAndTechniquesForElectronTubes Not only do you need a high vacuum to preserve the filament you mist have a long mean free path so a large percentage of the electrons get to the anode without colliding with remaining gas molecules.

            I have never seen one of these hand pump, all the HVAC techs I have seen use rotary mechanical pumps. Have a pic of one? Either way the ultimate vacuum will be several orders of magnitude too high.

          4. You win on the pump – it was about 20 years ago I was doing refrigeration, when CFC recovery was just getting big. It looks like nobody is using hand pumps for evacuating refrigeration systems any more, and in any case you’re right – you’re not going to get a vacuum better than about 1 torr with hand pumps.

            Thanks for the reference book on vacuum tubes! Lots of useful information in that one.

            As for the pressures required, when you’re looking at technical books like this, you’re seeing accepted GOOD practices for making vacuum tubes. We’re not talking about high quality here – we’re talking about hacking. I’m aware that any gas at all in a vacuum tube degrades its performance – a blue-glowing tube is a bad tube, but then again, sometimes with a blue glow, the tube still seems to function. We’re talking about making improvements on a hack that in its current state yields a gain of -108 dB.

          5. Fridge compressors make very good vacuum pumps. They are accumulative.
            Gas in the vacuum of a valve causes noise for noble gasses and phonic emission for non-noble gasses (like neon lights) and also rapid erosion of the cathode / elements for non-noble gasses.

            Valves include a “getter” which is loosely something like burning light bulb element to consume remaining gasses but the getter in a valve is purpose made.

            A lot of what has been mentioned here related to a *quality* valve that is expected to last a long time. You can get away with a lot of it if you are prepared for things not lasting long and sometimes breaking.

            The break light valve mentioned here would have worked much better if the anode was better placed or much better still if it were internal. The bottom of the bulb has a place where the two elements can be split.

            A gate has to be in the path between the cathode and the anode with no alternate shorter paths.

            Some basics –
            The cathode is heated so that the extra energy provides free electrons that can be attracted to the anode.

            The only thing that makes a cathode a *cathode* is that it is heated. If you heat the anode a valve would work backwards with much lower gain.

            The relationship for electrostatic attraction is roughly Fa = V / s^2 or the force of the attraction is the voltage divided by the square of the distance.

            The gain (excluding the effects of secondary emission) is then the square of the distance from the gate to anode divided by the square of the distance from cathode to gate.

            The center voltage for the gate is just a proportion of is linear distance from cathode to anode. At this voltage it has no effect – at lower voltage it reduces current flow by repelling free electrons back towards the cathode. At higher voltages it attracts the free electrons so that they are then closer to the anode for the anode to attract them further.

            For those old enough to correct me – please do so – politely.

    1. No, no, no.. you got it all wrong, he’s using a BRAKE light.. so, ergo he needs to use some BRAKE fluid. Maybe the high temperature racing type fluid… since if he runs a lot of power on that triode, it’s gonna get hot.

    2. instead of using a 5w/21W dual filament position/brake light bulb, you could use a H4 bulb and burn out one of the filaments (or use one where one of the filaments already is toast.)

        1. But it has a gas fill – not good for tubes. The smaller – non-halogen bulbs can work. But they can become dark silver-black from the evaporated and condensed tungsten. One time I changed a 12V car lamp. The glass was completely metallized and dark. I burnt my fingers as the bulb was still glowing, but no light could escape, so I thought, the filament was broken.

    1. No, he can’t. I was wondering why the filament had to be burned out, then realized that this is because one side of both filaments are connected together.

      HOWEVER, I don’t think they’re connected together inside the glass, so if you carefully remove the brass base, you should be able to have all four wires available, and then this just might work. Probably not very well due to the argon, and due to the filament not being in the path between cathode and plate, but probably better than -108 dB.

        1. I agree, this is quite workable if more attention was given to the physical locations and the location of the anode. Someone suggested a H4 bulb and these have a reflector that would be greatly useful if it is separately wired.

  1. He should have tested it also with the heater turned off. Just to verify if the results obtained are really from the vacuum tube effect and not capacitive coupling between the elements.

    1. That is very true, I should have written about it because actualIy did test it. When you apply voltage to the filament you can see the anode current slowly increasing so there is no doubt about it. What actually surprised me is that it reached a maximum only after 5-10min of heating up which is far more than I expected. I will add this to the project description. Thank you!

  2. Didn’t Lee DeForest discover / invent the triode doing something like this? He wrapped a wire around the outside of a thermionic diode, and it influenced the electron flow enough to amplify.

    This has the electrodes in a different order, but same sort of thing. Why did he burn out one of the coils? A heated coil makes a better electron emitter. It’s the therm in thermionic.

    1. You only need one heated element. Burning out one of the filaments disconnects the two filaments, since one side of each is connected together inside the base of the bulb.

  3. Somehow replying to a comment won’t work for me, but as “Dodo” pointed out it did just test the setup to see if there is unwanted capacitive coupling responsible for the effect and wrote about it on the project page. Thanks to him for pointing this out!

    1. how to: burn away the upper filament which has a shield behind it, in a car this is normally ‘high’ headlight. i was lucky to make it burn, so the half of the wire/thread, still is presented. i’ve used a 30 vac 5 amp ps to burn it fast.

  4. Recently i have experiment with tungsten thermionic emission, one of biggest mistake i have make is the tungsten require higher temperature to have emission. I don’t have fancy vacuum setup with diffusion pump or turbo pump, i only use with mechanical pump i think as long as 0.05Torr or below will work i guess.. So i think you get so low emission maybe due to the tungsten is not hot enough. The temperature is really sensitive from my experiment when 4.2V i get zero emission when increase to 4.5V i get about 1mA and go up 5.2V i able to archive 9mA.

    I make my tube to test it how it can work as amplifier,
    https://youtu.be/XXmLhUA6zDk

    after i make more proper bias it sound less distortion and more powerful
    https://youtu.be/4K_HI2kFO48

    The noise it from vacuum pump and also the motor leaking current i forget to earth it so the noise go into the tube and amplifier into speaker.. I not sure the car tail lamp is vacuum or gas filled . i having same ideal i haven’t try.. so i think it worth to try again.. To add Variac to your filament and increase between 12V-15V and see get better emission.

Leave a Reply to CRJEEACancel reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.