Tesla coil calculators
11 working calculators for coil design. Each one tells you where its numbers stop being reliable.
Everything here was derived while building, and most of it exists in no other tool. Where a calculator overlaps something published, its page says so.
One page here is no longer a calculator. Tesla coil primary and secondary geometry stopped computing, because JavaTC does the whole geometry at once and a second set of numbers next to it only adds noise. The page still covers two things JavaTC does not: where Wheeler's forms stop being right, and what a tapped primary does to a published set of figures.
Every calculator opens on the same reference coil, and so does every interactive figure in the articles. The two exceptions are soft switching, which is for an SSTC, and the snubber's stray inductance, which has not been measured on that coil. We use it because its numbers are known all the way down: the primary counted turn by turn, the tank measured rather than assumed, the frequencies taken off the coil. Other people's builds publish what their builders chose to publish, and the figures a calculator needs are usually among the ones left out. Two pages differ by a few per cent, as the articles do (a 3.6 against a 3.9 Ω burden, 160 against 170 A), and the bank is ten 120 nF capacitors, 12.0 nF, against the measured 12.2. The phase lead starts its ramp at 490 kHz, where the articles set the lead, rather than at the upper pole's 519.1. Put your own numbers in.
What to have to hand
Every field on these pages is a number you can measure, read off a datasheet table, or take from JavaTC or the calculator before it. With the seven things below, nine of the eleven work.
- The coil's dimensions, for JavaTC: the secondary's diameter, winding height, turns and wire, the topload's two diameters and height, the primary's inner and outer diameter, shape, turns and tap, and the distance to the walls and floor. JavaTC turns them into the secondary's frequency,
L_es,C_es, the coupling and the primary's inductance. - The tank: which capacitors, and how many in series and in parallel.
- The switches: the part, how many share each position, full or half bridge. Most common parts are in the lists, and picking one fills in its datasheet figures.
- The peak primary current and the bus voltage you are aiming for.
- The burst: its length in microseconds and how many a second.
- The current transformer: turns, core (from the list), burden, the lead inductor, and how long its leads are and whether they are twisted.
- The gate drive: the gate voltages, the gate resistor and the driver's supply.
The other two need one extra thing each. The UD2 oscillator needs its timing capacitor. The snubber needs the distance from the bus capacitor to the bridge, how far apart the wires run and how thick they are. The coupled poles page also asks how long your arcs are.
A scope is never required, but it improves some of these: the running frequency off the burden, the whole delay chain with one probe on the burden and one on the switch, a driver's dead time with a probe on each gate of a leg, and the ring at switch-off. Capture the ring once as built and once with a film capacitor of a nanofarad or two clipped across, and you have the bridge's stray inductance. An LCR meter gives a current transformer's magnetising inductance.
You only type a figure once. The peak current, bus, burst, running frequency, burden, switch, and the CT's turns and core (which also set the feedback page's ratio) are kept in this browser once you enter them on one page, and filled in on the others. Those pages say so and offer a way back to the reference coil. Nothing is sent anywhere. A link under each result carries it to the calculator that uses it next: the tank to the poles, the poles to the phase lead, the phase lead to the CT pages, the feedback page and the snubber.
For the whole coil geometry at once, primary and secondary together, use JavaTC. It is Barton B. Anderson's, with Paul Nicholson, Phil Slawinski and Matt Lewis named on it, and it has been maintained for years. Nothing here replaces it. These eleven answer questions it does not cover.
JavaTC's coupling figure is an estimate. It asks for the ground radius, the wall radius and the ceiling, so it does account for the room, but nearby metal and walls throw it off anyway and it cannot know what is standing beside your coil. Two builds on the record show how large the error gets. BrOdin wound a secondary to JavaTC's 0.185 and got massive racing sparks at 2000 A, and brought the coupling down until they stopped. On another coil Uspring read the builder's file at 0.134 and suggested dropping the secondary; the builder lowered it and the topload by 2 cm and reported 0.172, a quarter more coupling out of two centimetres. Take the printed number to start with, then measure the coil you built.
To measure it, sweep impedance across the tank with the secondary standing where it will stand and no power anywhere. You get three readings: two dips, which are the two poles, and a peak between them, which is the secondary's own resonance. With the secondary's frequency read off the peak, the two dips give both the coupling and the primary's frequency, measured on the coil as it is rather than computed from its dimensions. The inversion, the algebra and the bench note are on the two poles, and what you can do with the answer is on coupling.