Pond Machine III uses standard 4-ft fluorescent tubes, but runs them in "plasma finger" mode. Each tube is driven by a two-watt miniature Tesla coil device. These are the Five-dollar Tesla Coil available from several electronic surplus companies.

• Tesla Coils from All Electronics
• 3/16" copper adhesive foil (50 yd spools from stained glass suppliers)
• 1" copper adhesive foil ( Vetco electronic surplus, Bellevue WA)
• 4ft fluorescent tubes, Philips F40T12/ALTO

If we connect the high-voltage terminal of this Tesla coil to one end of a fluorescent tube, then apply the required 5volt DC power supply, the tube lights up. No Ground connection is needed, since the high frequency AC creates a virtual return path to ground through the space around the tube. This particular type of Tesla coil gives proportional output voltage, so if the DC power supply is varied between 0.9V and 5V, we can smoothly control the brightness of the tube. At the low end of the power supply voltage, a "plasma finger" effect is seen, where the plasma first ignites at the tube electrode, then it grows longer as the Tesla Coil output is increased. However, the brightness of the "finger" is very low, and it extends the full length of the tube long before the DC supply is cranked up to the full 5 volts.

The plasma-finger brightness can be easily increased by bringing a grounded metal plate near the tube, but this also shortens its length, and the DC power supply voltage must be increased to preserve the same length but at higher brightness. Brightness is also limited by the 2-watt Tesla coil output. To tailor the plasma-finger length to the drive wattage, I replaced the metal ground plate with an adhesive foil strip 3/16-in in width, adhering this strip along the length of the tube and connecting the far end of the strip to a Ground connection. The plasma-finger length is not totally proportional to DC power supply voltage, but the nonlinearity can be removed by using a long triangular ground foil: starting at 1/4" wide at the HV end, and tapering to 1/8" wide at the far end of the tube. If you use tubes of a different type or different length, you'll probably need a foil strip of a different width. (And for applications where all sides of the tube must be visible, replace the adhesive foil strip with hair-fine magnet wire wrapped in a loose spiral around the glass tube.)

My first prototypes used 40-watt four-foot fluorescent tubes. I soon found that the common and inexpensive tubes of the 34-watt or EW "environwatt" type DON'T WORK!. They contain a different gas mixture than the older 40-watt tubes, and they require a much higher voltage than these tiny Tesla Coils can provide. The older 40-watt tubes remain available, but they cost twice or more than the "EW" types.

The original "III" version shown at Seattle COCA gallery used a direct connection between the Tesla Coil output and one pin of the fluorescent tube filament. I found that this was a bad idea, since driving the plasma at a high voltage with an unheated tungsten filament causes a metal-sputtering effect which slowly destroys the internal electrode and sprays unsightly dark metal onto the fluorescent coating at one end of the tube. (Lightning engineers call this drive method "sledge-hammering", and rapid blackening of the tube ends are the well known result.)

In version "III.i" I switched over to Nikola-Tesla electrodless drive method to avoid the black contamination. The early history of fluorescent tube lighting is unknown to most people: gas-discharge tubes were popular playthings during the era of Victorian science, but then near the turn of the century the American inventor Nikola Tesla harnessed them to light his laboratory in NYC, later showing them off as part of the Westinghouse exhibit in the 1893 World's Fair in Chicago. Tesla's flourescent lamps used no hot filaments as today's do, but instead were either lit wirelessly using the intense e-field from a nearby radio transmitter (Tesla Coil,) or they used a metal electrode wrapped around the outside of the glass at one end of the tube. This method is known as "electrodeless discharge" or "RF light source." There is no filament to fail and no problems with black metal-dust contamination, so tubes powered in this way could theoretically last forever.

The foil electrode has a minimum size. The value of high voltage from this particular Tesla coil determines the minimum value of series capacitance between the plasma and the metal sheet outside the glass, and that determines the area of the foil. I found experimentally that the minimum area of adhesive foil needed for an outside contact was one square inch. I used a couple of square inches of 1" wide copper foil adhered to the tube (locating it a couple of inches from the end, so that the plasma cloud doesn't touch the metal internal parts and cause blackening.)

The microcomputers of course did not directly supply the 300mA 5-volt drive signals for the tesla coils. I used power FETs, one per tesla coil, soldering the pins directly to each small PCB. I used type IRL530 FETs from Digi-key (with a 2.0v turnon threshold,) open-drain circuit, then added 1.8 ohms in the source leg to give a limited gain for more linear response. To allow 100Hz PWM drive input I added a 4.7uF tant cap in parallel with the FET Gate terminal and a 68K in series between the computer and the gate. To allow adjustments to compensate for the different FET's gate threshold voltage I added a trimmer network consisting of a 33K from Gate to ground, and a 100K trimmer pot from Gate to +5V DC supply, with a 56K in series with each 100K trimpot. This allows a smooth change between 0 and 5VDC, low current signal, to control the drive current of each Tesla coil between 0 and 300mA.