[ §1 · tank ]

The DC blocking capacitor: SSTC or DRSSTC by value alone

Its value, and nothing else, decides whether you have an SSTC or a DRSSTC. People find this out by accident, and the way they find out is that the transistors explode.

CommonGnd4 August 202610 min read

There is a capacitor between the bridge and the primary on an SSTC too. Its value, and nothing else, decides whether you have built an SSTC or a DRSSTC. People find this out by accident, and the way they find out is that the transistors explode.

What it is and why

A bridge does not switch perfectly symmetrically, so a DC component appears at its output, and on a half bridge an imbalance in the divider adds one as well. What that DC meets in a Tesla coil primary is not a magnetic circuit. The winding is air cored, so there is no core to unbalance and none to saturate. What it meets is the winding's own resistance, and against a few turns of thick wire a DC component is very nearly a short circuit. Two published coils do put ferrite in the primary's magnetic path, and what coupling actually buys keeps those two on a line of their own for exactly that reason; every coil in the ordinary band is air cored.

The core story gets told about this part anyway, and the builder who tells it says where it came from. Gao writes that the DC blocking cap "comes from switch-mode power supply designs, where saturating a transformer can destroy the transistors" (SSTC 2). An SMPS transformer has a core. A coil's primary has not, and the two failures Gao names for a coil are the ones worth carrying: a DC bias current in the primary, and a device held on long enough to discharge the bus capacitors through itself and the winding.

So there is a capacitor in series with the primary, to keep DC out of it. On a full bridge, in series. On a half bridge, in series, or as two capacitors across the bus rails with the primary running from the bridge output to their midpoint, which is the arrangement Gao describes and the one his own half bridge uses.

What you decide

There is one decision and it is the whole article, which is why this page is short.

Whether it resonates with the primary

A capacitor in series with a coil is a resonant circuit. If its resonance comes anywhere near the working frequency then you have unintentionally built a DRSSTC, and a DRSSTC without overcurrent protection destroys itself.

That gets written as two rules, a reactance small enough to ignore and a resonance far below the working frequency. They are one rule written twice:

X_C / X_L = 1 / ((2*pi*f)^2 * L * C) = (f_res / f)^2

A margin of three in frequency is a ninth of the primary's reactance cancelled, a margin of ten is a hundredth. Measure whichever end you can and the other follows.

What the record here holds is two builds by one person. Gao's figures, off his own pages:

Ramped SSTC 3   primary 2.12 uH at 452 kHz    X_L 6.02 ohm    X_C 1.035 ohm
SSTC 2          primary 7.412 uH at 252 kHz   X_L 11.74 ohm   X_C 0.134 ohm

The inductances, the frequencies, both X_L and the 1.035 Ω are his; the 0.134 Ω is ours, from his 4.7 µF at his 252 kHz. Put as the ratio the line above asks for, R-SSTC 3 is cancelling 17 per cent of its primary's reactance and SSTC 2 1.1 per cent, which are frequency margins of 2.4 and 9.3 [derived]. Both coils work. That is the whole of the published evidence this site can put behind the criterion.

The other end of the scale is on the same page. Gao says that bringing the total tank capacitance on R-SSTC 3 to 58 nF puts the primary in resonance and turns the coil into a DRSSTC, and that a 68 nF capacitor in series does it. So on that machine one slot holds about 340 nF for an SSTC and about 58 nF for a DRSSTC, a factor of six in one part [derived].

Two thresholds this page used to print have been withdrawn: an absolute reactance in ohms, and a fivefold frequency margin attributed to an author who was never named. Neither had a source, and neither survives the record above, where a published SSTC runs happily at 2.4. What is established here is the identity, those two points, and that the DRSSTC condition is a margin of one. A number in between is not established here, and anyone who wants one has to say whose it is.

How to read a half bridge's pair. R-SSTC 3 carries two 680 nF films, and Gao's own sum treats them as a 340 nF equivalent, the two in series; his route to resonance agrees with that reading, since 68 nF in series with 340 nF is 56.7 nF [derived] against the 58 nF he quotes. Seen from the primary with a stiff bus the pair is in parallel instead, 1.36 µF, and then X_C is 0.259 Ω, the margin 4.8 and the cancellation 4.3 per cent [derived]. His page does not say which reading its bus deserves and the criterion passes on both, so nothing above turns on it. Your own coil may be less lucky: the two readings are a factor of four in capacitance and a factor of two in margin, so settle it before you do the sum.

What sort of part

Very low ESR, which means the film parts that go into a tank bank rather than whatever is in the drawer. This capacitor carries the full primary current. Gao's own choices are a 4.7 µF MKP polypropylene on SSTC 2 and a pair of 680 nF films on R-SSTC 3, and the range he gives for the part is 1 to 6.8 µF, film.

The numbers

The criterion: X_C / X_L = (f_res / f)^2. One sum, either spelling.
Two published SSTCs: 17 per cent of the primary's reactance cancelled at a margin of 2.4 on R-SSTC 3, 1.1 per cent at 9.3 on SSTC 2 [derived from Gao's figures].
The DRSSTC condition: a margin of one. On R-SSTC 3 that is 58 nF against the 340 nF fitted, both his figures.
Typical values: 1 to 6.8 µF, film, which is Gao's range and not a rating. The value that is right for your coil comes off the sum.
What is not established here: any threshold between the margin of one that makes a DRSSTC and the margin of 2.4 that is running in public.

What will get you

One sum before the first power up, and then a decision.

Take the resonance of your primary inductance with the blocking capacitor, and divide your working frequency by it. Square that ratio: it is the fraction of the primary's reactance the capacitor is cancelling, and the current is what the primary alone would pass divided by one minus that fraction.
If the ratio is near one you are building a DRSSTC whether you meant to or not, and the part that keeps it alive is current protection. If it is comfortably clear of one, take the largest film capacitor the budget allows and run the sum again on the value you actually fitted.

And a counterweight, from the same builder. Gao's position is that plenty of coils have been built with no blocking capacitor at all, that it is not necessary on small ones, and that it is cheap enough to fit anyway because it can save the day. Read the part as insurance rather than law. What is not optional is knowing where its value puts you on the margin, because the values that turn your coil into a DRSSTC are the small ones, and a small film capacitor is exactly what a drawer is likely to yield.

What goes wrong

Transistors explode on a coil that was designed as an SSTC. The blocking capacitor resonates with the primary somewhere near the working frequency. Do the sum, and if it comes out near one, change the value or fit current protection.
The current is higher than the primary's reactance alone allows. Same cause, caught earlier. The capacitor cancels part of that reactance, and the published case is 17 per cent cancelled for 21 per cent more current [derived].
The bridge dies on a coil that has no blocking capacitor at all. Two routes, and Gao names both: a DC bias current in the primary, or a device held on long enough to discharge the bus capacitors through itself and the winding. Neither of them needs a core, and the coil has not got one.
The capacitor gets hot. ESR. It is carrying the primary current, so it wants a film part, not whatever was in the drawer.

Where next

The bridge and what kills it, which is what this sits between.
An SSTC primary is a different part, the other half of the pair this capacitor is in series with.
Sizing a tank capacitor by RMS current, the same component chosen the other way.
Overcurrent is not a setting, which you need the moment the margin comes out near one.

Attribution: every figure on this page that is not marked as ours comes off Gao Guangyan's two build pages, SSTC 2 and Ramped SSTC 3.

The SSTC department diagram draws the tank across the middle of the bridge with no capacitor in it, which is the whole difference between that department and the next. On the QCW page you can take the capacitor out yourself and watch what happens to the arc.

DC blockingSSTCresonance