"Kelvin's Thunderstorm"
Lord Kelvin's water-drop
electrostatic generator

Bill Beaty, 1995

NOTE: avoid using wood to support metal parts! See "debugging" notes at the end. See FURTHER INFO at end too.
It is possible to build a very simple high-voltage generator which has no moving parts and is powered by the energy of falling water. By dribbling water through some old soup-cans, several thousand volts magically appear. The "magic" lies in the fact that water (as well as everything else!) is made of vast quantities of positive and negative electric charge in perfect balance. It's not too hard to cause an imbalance. Water normally has zero net electrical charge because it contains equal and opposite charges. "Kelvin's Thunderstorm" is a gravity-powered charge un-canceller.

                         +  + +
                        +|||||| +
                         ||||||   Water
                         ||||||   Dripper
    Negative charge is   ||||||
    'induced' at tip      \  /
    of dripper           - || -
                          - _ -

                            _           + + + + + + +
                           -o-        + -------------- +
                            -        + |       A      | +
                                     + |  Positively  | +
                            _        + | Electrified  | +
                           -o-       + |    Object    | +
                            -        + |              | +
          Negatively                 + |              | +
          electrified                 + -------------- +
          water droplets    _           + + + + + + +

                     -             -
                  - |               | -
                    |               |     Collector-can
                  - |--__----____---| -   becomes Negative
                    |               |
                  - |               | -
                    |               |
                  - |_______________| -
                      -    -     -



Even though water has no overall electric charge, it is full of movable electric charges (called ions). Half of the water's charges are positive and half are negative. It is not hard to separate these charges; simply hold an electrified object near the water. The electrified object will attract the "unlike" charges to the water's surface. It will also repel the "alike" charges away, deeper into the water.

In the above diagram, the positive object attracts the water's negative ions and repels the positive ions. This draws an excess of negative ions into the tip of the water dripper, while repelling an equal amount of positive ions off to the other end of the dripper. When the water drop detaches from the tip of the dripper, an overall negative electric charge is still trapped in the droplet. The falling water droplet carries away negative charge, leaving the dripper slightly positive. If we catch the falling droplets in a container, the container will become negatively charged.

In the above diagram, negative water droplets will be continuously created forever as long as the water flows. However, this process does not exhaust the imbalanced charge on the positive object. Sounds like perpetual motion, eh? Actually no. The electrical energy is being created by the work that gravity does in pulling the negative droplet away from the grounded dripper, and away against the attraction of the positive object. The electrical attraction-force from the positively-charged object keeps the tip of the dripper charged negatively, but the positively charged object does not supply energy. YOU supply the energy, since you LIFT the water to a height to fill the dripper. It's like the generator in a hydroelectric dam, but without the turbine or the spinning coils or magnets. The water itself becomes the moving parts of an electric generator.

(Note: the charge polarities can easily be reversed. If the above "object" is made negative, the droplets would come out positive.)


If somehow we can make a positively-charged object, then we can create negative-charged water. But where can we get a positive object? If there was some way to CHANGE the negative charge on the water into a positive charge, then we could use the water to charge up it's own "positive object". We would then have a a self-sustaining generator.

There's a simple way to do this: build TWO water-drop devices like the one in figure one! See the trick? The device in figure one uses a positive object to create negative water. It uses "plus" to create "minus." If we build a second device, we could use the "minus" from the first one to create "plus." Then we could hook the two devices in a circle. The first one would create an imbalance of negative charge, which could be fed to the second one which would create an imbalance of positive charge, which would be fed back to the first one again. It might sound crazy, but it really works.

We'll build two of the drippers in Fig. 1, set them side by side, then collect the electrified water droplets from one side and use them to electrify the "charged object" on the other side, and vice versa. We'll cross-connect the lower and upper parts with wires. One side will have a positive "object" and will make negative droplets, while the other side will have a negative "object" and will make positive droplets. We'll also connect the drippers together so they remain neutral. Then we will have a self-sustaining electrical reaction.


_   __________________  \  Water Supply
  \ \                   \ \
    \ \                   \ \
      \ \                   \ \ Drippers (metal,plastic,glass)
       ||                    ||   
       ||                    ||
       ||                    ||  Connect the water supply to a
       ||                    ||  metal faucet using a wire, or 
       ||                    ||  to the screw on an electric 
       ||  Wire not shown,   ||  outlet or wall switch.
           see next diagram
    |    |                 |    | Bottomless metal coffee cans, 
    |    |                 |    | or wire rings, or bundt pans,
    |    |                 |    | or metal disks with large holes 
    |    |                 |    | (supported by insulating rods.) 
    |    |                 |    | Called "Inducers."  These
                                  act as the "charged object."

    |    |                 |    |  Metal cans on insulators
    |    |                 |    |  (styrofoam? insulating rods?)  
    |    |                 |    |  The "Collectors."
    |    |                 |    |
    |____|                 |____|
     |  |                   |  |   The "inducers" and "collectors"
     |  |                   |  |   should be separated from each
     |__|                   |__|   other by several inches


See Fig. 3 below. Wires connect the two sides together. The negative droplets touch the lower Collector-can of the first side. The collector can is electrically connected to the upper negative Inducer of the second side. The negative Inducer will cause the second side to make some positive droplets. The positive droplets of the second side will touch the second lower collector can, and this will charge the upper Inducer can of the first side positively. (This makes the first side produce negative droplets.) The grounded drippers must be connected to each other and to ground. See Fig. 3 below to see how the wires connect things together.

Highly recommended: ELECTROSTATICS by A. D. Moore (lots of projects), also others


But where does the first charge come from? In fact, if you build such a device, it will usually create voltage all by itself, spontaneously, without being pre-charged. During dry conditions everything near the generator ends up with a tiny electric charge just from being handled. If one of the upper cans is slightly negative, it will cause the water to have imbalanced positive, which will start up the other side of the generator, which will make the charge on the negative side become larger, etc., over and over.

It's like balancing a penny on edge: it's hard to start out with a perfect balance, and usually it falls one way or the other. Same with this generator. If there's a tiny electrical imbalance at the start, the generator will amplify it over and over, and the voltage will "fall over" to either one polarity or the other. A high voltage will magically appear from nowhere. (But nobody knows which side will start out positive and which will be negative.)

Here's another viewpoint: each section is an electronic inverter! If we place a negative voltage on the inducer-ring, a positive voltage will then appear on the water droplets, and upon the lower can. By cross-connecting two such "inverters," we form a flip-flop circuit. When first turned on, it randomly "decides" to store either a logic-high or a logic-low. Ah, but what would happen if we instead built THREE of these segments? And connected them in a loop? Why, then it's unstable, and must form an electronic ring-oscillator. And that's exactly what happens: "Euerle's dynamo" is a Kelvin Thunderstorm with three segments hooked in a loop, and producing some (very slow) three-phase AC output. See Self-excited ACHV Generation Using Water Dropelts AJP 1973


The metal parts of the generator must be supported with insulating materials. A large vertical sheet of acrylic plastic works well. So does styrofoam plastic. Don't use wood for the supports, it's too conductive. Fasten the collectors and inducers to the plastic sheet with screws or silicone caulk, or make holes in the sheet and tie them to the sheet with string or wire. Some people have used plastic rods or plastic strips to support things. Other people use plastic water pipes. The plastic must be clean and dry. The inducers and collector cans must be spaced away from each other by several inches horizontally and vertically. The lower collectors must be kept away from the table surface.

Bare wires are used to cross-connect the four cans. These two diagonal wires must be far from any other conductive object, and the wires must not touch together. Use bare wire, this will let you create sparks between the wires, or to later flash a NE-2 neon bulb.

Connect the ends of the diagonal wires directly to the metal of the cans. If you use plastic-covered wire, strip off the plastic coating from an inch of each end of the wire. You can use tape to hold the wire against the metal, as long as the wire touches the bare metal directly (not, for example against the painted part of a coffee can.) Alligator clipleads (bought from Radio Shack stores) work well for this. Or poke a hole in the metal near the edge of the can, stick the wire through the hole, and bend it and tape it so it doesn't fall out.

_   _______________________ \
  \ \                       \ \
    \ \                       \ \
      \ \                       \ \
       ||                        ||
       ||                        ||
       ||                        ||
       ||                        ||

    |  o |                     | o  |
    |    |                     |    |
    |  o |+    +         -   - | o  |
  + |    |----\            /---|    |  -
    |  o |  +   \ +    - /  -  | o  |    (no connection
                + \    /                  between the 
       o            C/           o        crossed wires!)
  - |    | -       / \       + |    | +
  - |  o | -   - /   + \     + | o  | +
    |    |_____/         \_____|    |
  - |----|   -             +   |----| +
    |____|                     |____|
     |  |                       |  |
     |  |                       |  |
     |__|                       |__|


The upper rings (or cans) must be positioned near the place where the water droplets break from the rest of the water. If the droplets break away right at the tip of the nozzle, then put the nozzles within the upper cans. If a solid stream of water comes from the nozzle and breaks up further down, then move the nozzels up, so that the water-break spot is inside the upper inducer cans.

For the Drippers, you can use glass or plastic "pipettes". You want to have a very small hole, so that the dripper makes lots of tiny droplets. If you cannot get a pipette, try poking holes in a plastic container. See below.

Adjust the water flow so it drips VERY fast. The faster, the better. Even better, use lots of drippers instead of just one.


Once you have the water dripping, you can expect high voltage to immediately appear. After the device runs for a minute, touch one of the coffee cans gently with a finger and listen for the tiny snap of a "static" spark. If you don't hear a spark, the machine is running weakly or not al all. See the "debugging"" section near the end of this article.

If you can't hear any spark, you can try detecting sparks electronically with an AM radio. Place a radio a foot away from your generator, tune it between AM stations (or tune it to a very weak station), and turn the volume up. Run your Kelvin machine. Touch one of the cans with your finger and listen to the radio. You should hear a "snap" noise as your finger touches the metal. Touch one upper can, then the other, then the first one again. You should hear a "snap" each time. Even when sparks are too small to hear or see, a radio will sometimes still detect them.


Once your machine is able to produce sparks, you can also make it flash a small neon lightbulb. Normal flashlight bulbs won't work, you need a small neon "pilot light" bulb instead. Obtain an "NE-2" or other similar neon light, the kind which looks like a short glass tube with two parallel wires inside and two bare wires sticking out of the glass. Hold the bulb by one wire, look at the tube, then use the other wire to touch one of the cans of your Kelvin device. You should see a dim orange flash inside the bulb. (It might help to turn off the lights and work in a dimly lit room.) Hold one bulb wire, then use the other wire to touch the positive can, then the negative can, then the positive, and you should see a tiny orange flash each time.


Don't connect the NE-2 bulb directly across the two generator wires. It will short out the generator and prevent high voltage buildup. You can make the generator automatically flash the neon bulb by making a "spark gap". First twist one wire from the neon bulb around one of the generator's diagonal wires, then bend the bulb wires so other short wire from the neon bulb is very close to the other generator wire ( but not touching). Small sparks will occasionally jump across the small gap and flash the neon. The smaller the gap, the faster (and dimmer) the flashing. Try a 1/16 inch gap (1 mm) at first. If it works, increase the distance to get slow, bright flashes. Also, eliminate the sharp wire tip of the bulb by bending tip over to form a little ring. The sparks then jump to the edge of the little ring. This sometimes lets the voltage rise higher before a spark jumps, which eliminates any air-leakage and lets the bulb flash more brightly.


To "see" the high voltage surrounding the cans, tape some strips of tissue paper to the cans. Put tape only at the top of each strip of tissue so the strip hangs down against the side of the can. When the can charges up, the strips should lift slightly outwards. The higher the voltage, the farther the strips move. When a spark jumps, the strips jerk because the repulsion force suddenly becomes less.


The energy that builds up between the cans comes from the falling of the water. As the stored energy grows, the water has to do more and more work every time it adds a bit more imbalanced charge to the cans. The electrified droplets feel a repulsion force as they fall towards the alike-charged lower cans. As the voltage increases, the droplets will fall more and more slowly. The sound of the splashing water will change. The droplets may even start bending their paths, even occasionally falling upwards!


If the device is run for very long, the lower cans fill up. How to get the water out of the cans without discharging them? Here's my addition to the classic Kelvin Waterdropper: use the "faraday ice pail" effect, where a conductive hollow object always has no charge-imbalance on its inside. To do this, connect an exit tube inside each lower can as below, so the water DRIPS out, as in Fig. 4 below. If water falls in a solid stream, the cans will discharge and the generator will stop working. So, drill lots of tiny drip-holes in a flat plate?

Or even better, make your generator look like a VandeGraaff Machine. Make a metal sphere using two 14" Ikea Stainless Steel salad bowls. Carve holes in the top and bottom to pass a "shower" coming down from the inducer-rings. Suspend a plastic bucket inside, using heavy nylon fishline. The center of the poly bucket is perf by many tiny #60 drill holes. Run a wire between the inside of the metal sphere and the water inside the bucket. When the charged droplets from a "shower head" pour through this metal sphere, they're intercepted by the plastic bucket (and wadded plastic screening to prevent splashing,) and all their electric charge is grabbed by the metal sphere. The bucket then streams the water out in a narrow shower of small uncharged droplets, which pass through the hole in the bottom. Water never touches the metal sphere. Problem: if the sphere gets charged up to huge voltage, the top stream of droplets will repel and fly outwards. The upper hole may need to be larger than the lower. Cover the holes' sharp edges with slitted hose filled with RTV silicone.


          ||                 ||
          ||                 ||
          ||                 ||
          ||                 ||   For best results, no sharp 
          ||                 ||   edges or burrs anywhere. 
          ||     WATER       ||   Or, cover sharp edges  
          ||                 ||   with thick bead of
          ||     __   __     ||   RTV silicone caulk, or 
          ||    |  | |  |    ||   use heavy Tygon hose,
          ||    |  | |  |    ||   slitted lengthwise
          ||    |   U   |    ||
          ||    |   O   |    ||
           ======       ======
                       Uncharged droplets
                    O  exit from bottom


                        o  o     
                         o  o   
                          o      holes cut above, below
                __----- o  o  -----__   
              _/         o  o        \_  
            _/             o           \_ 
           /      |      o  o     |      \   
          |       |--___----__---_|       |    HOLLOW METAL SPHERE
         |         |             |         |  ( 14" SS bowls fm/Ikea, 
         |   plast. \_         _/          |    Blanda Blank(tm)  )
         |   bucket   \_______/            |
          |             o   o             |
           \_           o o            _/  
             \__        o   o       __/    plastic bucket hung by
                --____  o o   ____--       heavy fishing line, with
                        o   o              many tiny holes drilled
   Water never          o o                in center bottom
   touches sphere       o   o   
                        o o      
                        o   o   

Or, even simpler, install a cone-shaped piece of metal window screen inside a bottomless can, so the water droplets touch the screen and continue through. Make sure the screen is centered vertically within the can, so that the point of the cone doesn't extend past the lower lip of the can. Don't let the water drip from the edge of the can, otherwise it will carry charge away with each drop.


With a little catcher-tray and a fountain pump, you can make the system recirculate. Or, you can stack all four parts of one Kelvin device in a single row, for an in-line waterdropper generator. See my article on "Inline Kelvin Thunderstorm Device" found on my site at


Note that the Inline version is more tricky to make work. Build the above device first before attempting the one below.


             \ \
               \ \
                 \ \
     Grounded     ||
      Dripper     ||
                |   |
        Neg     |   |
        can     | o |
                |   |

       Pos      |   |
       can      |...|
      w/screen  |   |     Connecting wires not shown, see 
                |   |     ikelv.html article for more info

                \    /    Connect pos to pos, neg to neg
    Grounded     \  /
    Funnel        ||
                |   |     Note that this is a more advanced
        Pos     |   |     project, and is more difficult to
        can     | o |     debug than the side-by-side version.
                |   |

       Neg      |   |
       can      |...|
      w/screen  |   |
                |   |
Fig. 5  l IN-LINE VERSION (wires not shown)

The water supply need not be a "dripper", it can be a high velocity spray, as long as the water jet divides into droplets, not a contiguous stream. And multiple jets can be used, sort of like a shower head. The faster the flow and the larger the number of separate streams, the higher the total output current. (Higher current gives faster recharge rate after a spark, and it lets the generator self-start more reliably when humidity is high.)

GIGANTIC VERSION Kelvin's waterdropper

I've always wanted to build a gigantic version like the one below, with hollow metal toroids. (Use halves of VandeGraff spheres, the halves with the holes). Or maybe use metal 55-gal drums. But those drums may have sharp edges, and we can't attain millions of volts if the edges aren't bulbous. A foil-covered truck innertube should support about a million volts before air-corona leakage stops the voltage from rising any higher. Or do as Tesla-coilers do, and make a skeletal torus from bent, coiled pipe.


                                            Four tori
        \\                           \\     (shown cross-
          \\                           \\    sectional)
            \\                           \\
             ||                           ||
             ||                           ||   shower heads
             ||                           ||   water spray
             ||                           ||
     ___           ___             ___           ___
    /   \         /   \           /   \         /   \
   |     |       |     |         |     |       |     |
    \___/         \___/           \___/         \___/ 

     ___           ___             ___           ___
    /   \         /   \           /   \         /   \
   |     |\     /|     |         |     |\     /|     |
    \___/   \_/   \___/           \___/   \_/   \___/

    Conical screens in lower torii touch droplets and 
    release, discharging them.  Entire screens must be 
    deep within the "hole" of each donut so the torus can 
    shield the departing water droplets from the electrical 
    fields on the outside.  If you can see the screens
    sticking out, your device will barely work.
(or foil-covered inner tubes, or a spiral of pipe.)

        _____             _____    
       /     \           /     \   
      |       |         |       |  
     |         |\/\_/\/|         |   LOWER TORUS WITH
     |         |       |         |   DROPLET-TOUCHING METAL
      |       |         |       |    SCREEN ACROSS THE
       \_____/           \_____/     DONUT-HOLE

I formed cone-shape screens, then punched in successive circles to form a concentric "ripple" pattern in the screen. This stops water from running off the cone-tip as a long stream (which would discharge the whole thing.) Also keeps the sharp cone tip hidden within the metal cleft: electrically shielded. DON'T let water run along the torus and drip from the bottom.

High-velocity shower heads and cross-connecting conductors made from large-diameter pipes will complete the scene above: it's a "VandeGraaff Generator" version of Kelvin's Thunderstorm apparatus!

NEWS: I suspended a bundt-pan by fishlines, then sprayed water through the center, so that the water did not touch the metal. I used a garden hose with a "water breaker" (a sort of shower head attachment.) I charged up the bundt pan with a 10KV power supply, and then measured the electric current between a collector pot and ground. It was 2.5 microamps! This doesn't sound like much, but it's at lease as much as some VandeGraaff generators can create.

I found that if I disconnected the power supply from the bundt pan, the current did not vanish. The charge on the bundt pan stayed the same as I watched for about 30sec. And this was in high humidity conditions! Fishing line makes a VERY GOOD insulator. The system kept working until I touched the bundt pan with my finger, then the electric current coming from the collector can fell to zero.


The above generators can be used to run a motor, if the motor is my Pop Bottle Electrostatic Motor at:


I find that these small Kelvin Waterdrop Generators are a little too feeble to keep the motor going continuously. Instead they make it slowly pulse. They build up a charge imbalance, then the motor starts turning and rotates a few times. This exhausts the charge imbalance, the motor stops, then it builds up again and repeats. This happens a couple of times per minute. A bigger waterdrop generator is needed if you want to run the bottle-motor continuously.


I put multiple drippers on my waterdrop generator and this improved things considerably. The rainstorm should be like an actual rainstorm, not just a single stream. The best generator uses lots and lots of tiny drops falling in parallel, with the drops being created as fast as possible. If we use a cluster of dripper nozzles, the inner ones probably act as electrical shields for the outer ones. This is bad, since this will prevent the inner ones "seeing" the charged inducer cans, and they won't make any electrified water. Therefore a CIRCLE of nozzles is probably best. Or if you want to really get fancy, use a cluster of tiny squirt-holes, but run lots of thin wires across the "hole" of your metal donut, positioning the wires below the squirters, so they don't get hit by droplets. Then every single squirter sees the charged wires close below, yet its stream of opposite-charged droplets just passes through without touching anything. You've made a water-proton gun (or water electron gun, for the neg. side.)

I drilled a circle of eight tiny holes in a plastic bowl (using #64 drill bit), and this gave good results. A crude version of multiple-dripper: use a soup can, and punch a bunch of holes in the bottom by using a tiny nail. I've also seen a shower-head thing called a "water breaker" in the garden supply section of hardware stores. If a circle of tape was stuck to the center of one of these, it would plug up the middle holes and form a ring of about 100 tiny water jets.

SALT WATER? Heh, "Kelvin's Plastic-storm!"

The liquid has to be conductive. But will adding salt improve things? Will it stop working if we use distilled water, or "DI" deionized water? These questions are easily explored if we note that the water in the water-stream an in the hoses, is a resistor. How large can this resistor become before it screws up the generator's operation? First, use basic electronics concepts: see the generator as a voltage-source with a large impedance in series. If the electrical resistance of the water in the hose is much higher that this natural "series impedance," then the water acts insulating. If it's much lower, then the water acts conductive, and the generator will work.

  • Typical open-circuit voltage: roughly 10,000V, for a 1cm spark
  • Typical short-circuit current: a fraction of a microampere. Let's guess 1/100, so 10 nA (nanoamps.)
  • Zseries V/I = 10^4 / 10^8 = 10^12 OHMS HOLY CRIPES! A mega-megohm. TeraOhms.
So it looks like distilled water will work. Deionized water will work. Probably alcohol, gasoline, most solvents will work. I wouldn't be surprised if even low-leakage HV transformer oil still works! It's hard to make extremely insulating liquids. The slightest contamination by ionized molecules will provide moveable charges during voltage-drops at HV. Solids make much better insulators, since any ions are immobile and locked into the solid's internal grid structure. With solids, all the leakage is through surface contamination.

Cross-check, actually how low is the current? Got a nano-amp meter? Well, the capacitance of the metal cans can be measured. Guess: 10pF? (Get a C meter and find the real value.) If the recharge is roughly linear, then we can use equation V/T = I/C. For a 3sec recharge to 10KV and 10pF can-capacitance, 10,000/3 = I / 10^-11, so current I = 0.33 microamps rather than my guess of 0.01 microamps. Maybe our giga-ohms are actually 33X smaller. Only 300 Gigohm rather than a TeraOhm. Still really, really insulating, at least in the usual sense.

So, to stop the Kelvin T-storm "Generator Effect," maybe we have to fill our tank and hoses with little plastic beads, polyethelene or teflon. And even then it still might start generating a voltage if the beads and hoses aren't kept really, really clean, and operated under low-humidity conditions. Heh, Science Fair trickery: try using a tank of tiny plastic beads, but beads which are all contaminated from handling: all of them with a molecule-thin layer of oily hygroscopic human-hand contamination. Will your Kelvin Plasticstorm Device still slowly charge up to 10KV? If not, use metal tubes rather than plastic, so the only "insulating fluid" is at the tips of the droppers where the bits of plastic (or oil droplets) are detaching.


Whenever a spark discharges the generator, it also discharges the inducer rings. As a result, the generator takes quite a while to "ramp up" to full voltage again. This is exponential growth, and it's quite slow at first. There are several possible ways to solve this problem (I haven't tried them, you be first!) One solution is to insert very large resistors in series with the wires to the inducer rings (large = thousands of Megohms). Then always discharge only the collectors, not the inducers. The resistors will keep the inducers from instantly discharging. If the inducers remain charged, then the generator will quickly recharge with a fast linear voltage curve rather than a slow exponential curve. High-value resistors are expensive, so perhaps try making your own resistors. Use strips of paper with fine lines of india ink (india ink is conductive carbon.)

Another possibility: rather than using resistors, instead insert high voltage diodes. For example, use several 7,000-volt microwave oven diodes in series, available from Allied Electronics etc. Orient the polarity of diodes to allow the Inducers to charge but not to discharge. Diodes in one conductor should point upwards, and in the other conductor should point downwards. This way the collectors will charge up the inducer rings, but when you discharge the collectors, the diodes will turn off and become nonconductors. The excess charge on the inducer rings will remain high. Also, if you use diodes, the generator polarity would always be predictable, since the generator would not function if the polarity became reversed.

A third method: build a big generator, but don't connect the collectors to the inducers. Then build a second tiny water-drop generator, and use it to charge up the inducers of the big generator. Then always discharge only the collectors of the big generator, and leave the other metal parts alone.

If you want to use your generator to power a pop-bottle electrostatic motor, the above idea is the way to go. Build a separate generator to power the inducers of the larger generator, that way the motor cannot "short out" the inducer voltage and make everything stop.

Now that I'm thinking about it, here's another type of generator to build: a half-kelvin device with only two cans. How can it ever work? Simple: use the screen of an old TV set to charge the inducer! I bet that this would even work when the humidity is really high. A TV set normally cannot produce constant electrostatic energy output unless you keep turning it off and on. But it should be able to supply enough energy to power the inducer ring on half of a Kelvin's Thunderstorm device.

WEIRDNESS: antigravity?

If your generator really works well, you will see water droplets slow down and their paths curve upwards! No, this is not antigravity, this is just electrostatic repulsion. Alike charges repel.

WEIRDNESS: really really gigantic generators

In a private conversation someone told me that there were patents on a wind generator based upon the Kelvin Generator. Build two big parallel vertical metal screens the size of outdoor movie theater screens (or larger). The upwind screen has coarse mesh, the downwind screen has fine mesh to gather water droplets. Suspend them on insulators which are good for millions of volts. Charge the upwind screen with a power supply. Spray a fine mist of water into the screens upwind, and let the wind push the spray through the screens. The upwind screen will attract imbalanced charges into the sprayer tips, and the water droplets will have an imbalanced charge of opposite polarity. The wind takes the place of gravity in the classic Lord Kelvin device. Wind pushes charged water to the second, fine-mesh screen. Water droplets touch this screen and deliver their charge. The wind is slowed by repulsion of the water mist, the upwind screen uses no current, and the downwind screen puts out amperes at millions of volts of electrical potential (amps times megavolts equals megawatts). Simply step down the megavolts of DC, then convert it to AC. Ta-da, a wind generator with no moving parts! An artificial thunderstorm, harnessed as a commercial generator, powered by the wind.


If your project will not work, it may be because the humidity is high and your device is having trouble "deciding" which side should be positive and which side negative. See my hints about humidity, found at http://amasci.com/emotor/statelec.html. With Kevin generators it takes voltage to make voltage. If your device starts totally at zero, it may take a minute or two to build up to maximum. Therefore give it a "goose" by holding an electrified object briefly near one of the cans while the water is running (for example: a balloon, a 2liter pop bottle, or some styrofoam, each rubbed on hair to electrify them.) This gives the generator a "kick start."

The two drippers must be neutral, so they need to be connected together electrically. To make certain they're neutral, connect a grounded wire to their water supply. These drippers should drip FAST. Many droplets per second. If your drippers are very slow, then your machine will charge up very slowly or not at all. To produce a fast drip rate, usually you need a tiny hole at the end of your water tubes. Plastic or glass pipettes are the usual way to accomplish this. Also, your machine will work better if you have many drippers.

If you really cannot get your generator to work, here's a way to "cheat." Put a piece of aluminum foil on a TV screen, connect this foil to one of the inducer cans on your generator, and then turn on the TV. This will make your generator run, at least temporarily. Don't let any water get near the TV set!!!


Kelvin Generators usually can tolerate fairly high humidity. Watch out though. Materials with large internal surface area, such as wood, cloth, masonite, etc., usually absorb moisture from the air and become slightly conductive. Therefore, assume that these materials are the same as metal, and avoid using them as supports or framework when you build this device. Wood provides a leakage path and shorts out the high voltage. One experimenter even found that problems were caused by supporting their cans with insulating blocks glued to a wooden panel. The short lengths of plastic must not have been sufficiently insulating, and there must have been a leakage path across the plastic and through the wood. Switching to all-plastic supports solved their problem.

If acrylic sheets such as plexiglas(tm) or perspex(tm) are not available, large styrofoam blocks work well as supports. Avoid using solvent-based glues with styrofoam, it makes it dissolve. The plastic MUST BE CLEAN. Use new plastic if you can. If you wash the plastic, don't use much soap, since soap can form a conductive coating. Rinse it and scrub it with lots of clean water to remove all the soap. If you wash the plastic, then afterwards dry it thoroughly with an electric hair-dryer gun (but be careful, don't melt the plastic with too much heat!)

Nylon fishing line makes a good insulating support, especially during high humidity conditions. Long, very thin supports such as fishing line have small surfaces and therefore give less surface-current leakage than short, fat supports or flat panels. Don't use twine or string as supports, of course, since these materials become too conductive when the humidity is high.

If humidity is very high, even plastic can become slightly conductive. This can be temporarily fixed by using an electric hair-dryer gun to dry the plastic surfaces. Bathe the plastic in hot air for several minutes, taking care not to heat it so much that it softens! Try testing your generator again, and it may begin to work.

The water droplets must not touch the inducers. The droplets should pass through the inducers. The droplets should break free from the water while they are still inside the inducers. If continuous streams of water (not droplets) shoot out of the drippers, then move the drippers upwards to make certain that the droplets break away from the continuous streams while the droplets are inside the inducer cans.

The water droplets MUST touch the collectors. To start, simply let the collectors fill up with water. Once you have succeeded in getting sparks from your machine, then later you can try the trick with the cones of window screen (see far below.)

To detect the tiniest charge imbalance, build the RIDICOLOUSLY SENSITIVE CHARGE DETECTOR shown elsewhere on my web pages. (/emotor/chargdet.html) This device will detect a few hundred volts of electrostatic potential at a great distance from the cans. It is extremely sensitive. The tiniest sparks won't begin until the cans reach about 1,000v potential, yet the sensor responds to about ten times less. A sparking Kelvin generator can make the Charge Detector flash even if it is many feet away.

Don't neglect the balloon trick. If your device doesn't self-start, then momentarily hold a charged balloon near one of the cans while the water is running. (Verify that the balloon is really electrified, see if it raises your arm-hair when rubbed. Sometimes humidity is so high that the balloon will not aquire an imbalanced charge by rubbing on hair.)

A simple way to detect static charging: place a portable AM radio near the device, tune it to a blank station, then touch one of the cans with a finger. If your device is just barely working, there will be an imperceptible spark. But this will make a loud click on the radio! If you wear an AM Walkman headphone radio, it will extend your senses so that you can hear the electromagnetic pulses given off by the tiniest spark. Become a "Borg" from Star Trek, with the ability to hear electromagnetic impulses via a biointerface to electronic circuitry (Walkman headphones). :) Try spending the day wearing AM radio headphones tuned to an empty spot on the dial, and you will encounter all sorts of interesting electromagnetic "sounds" in the environment. You'll hear the "crack" noises of distant lightning even when the thunderstorms are too far away to hear the thunder. Electric fences in countryside farms make a periodic clicking. The overhead wires from electric city buses make all sorts of musical tones.


LOOKING FOR BOOKS? Try searching amazon.com:

(try "science fair" too)


   'Inline' version, stacked sections

   General debugging notes for electrostatics

   Solving humidity problems for VDG machines

   VandeGraaff Machines (big sparks)

   Tesla Coils (bigger sparks)

   What if lightning was slow?


  keywords: Kelvin's Thunderstorm (google)

  Kelvin's Thunderstorm davidwilliamson

  Sparking Buckets (Bizarre stuff made in your kitchen)

  Solaris Kelvin Devices (several)

  Version of above article

  UMICH w/diagram

  John D'Mura's device

  UMD photo


  UNIMELB diagram

  Diagram w/Leyden Jars

  THE PHYSICS TEACHER (magazine), May 1988, pp304-306, The Ting-A-Ling
  Machine, by Cliff Bettis   (uses mixing bowls)

  THE PHYSICS TEACHER (magazine), February 1972, pp100-101, Electrostatic
  Lobby Display, by M. Fast   (uses coffee cans)

  Scientific American Magazine  June 1960, THE AMATEUR SCIENTIST, by C. L.
  Stong, page 175

  Alvin Marks' "Power Fence", an HV wind power generator with no moving 
  parts http://www.rexresearch.com/airwells/airwells.htm

  Inline version of "Kelvin's Thunderstorm" electrostatic generator:

Want books? Try searching amazon.com:

(try "science experiment project" too)

Created and maintained by Bill Beaty. Mail me at: .
View My Stats