A high bus with a low bridge, and where it breaks

The bridge is not fed from the bank. It is fed from the buck's output. So if the buck never takes its output above 440 V, the bridge never sees the bus at all, and that opens a trick worth knowing:

mains -> DOUBLER at 650 V -> large bank -> BUCK limited to 440 V -> bridge on 650 V devices

Why go to the trouble

Because the bank's energy is CV²/2, and whether the ramp holds its shape depends directly on it. One 406 J bang against 9900 µF:

• mains direct, 325 V: the bank holds 523 J and the bang eats 78 per cent. The ramp falls apart.
• a variac at about 300 V AC, giving 424 V: 890 J, 46 per cent. It sags visibly.
• a doubler, 650 V: 2091 J, 19 per cent. It holds.

The second argument, which is less known

A doubler pulls the bus practically to zero every cycle, so no residual voltage is left between bangs and the ramp starts from a clean sheet.

It also takes both half cycles, which allows a longer ramp and uses the full rating of the bridge.

And the duty cycle is there

With the bus sagging from 650 to 514 V, a 440 V ramp top needs 86 per cent duty, which is fine:

• a 400 V top: 78 per cent duty, and 1.62 times margin on a 650 V device;
• 440 V: 86 per cent, and 1.48 times;
• 560 V: 109 per cent, which is unreachable, at 1.16 times.

The alternative that was rejected, and why

Raise the AC with an autotransformer to about 300 V and get 424 V without a doubler. It has one real advantage: the gate DC-DC isolation is over by only 1.06 rather than by a lot.

Where the trick breaks, which has to be read

You want either a hardware clamp on the buck's output capacitor, or bridge devices with real margin at 1200 V. And do not forget the isolation on the buck switch's own gate supply, which sits on the full bus.

While we are on frequency

The other reason a QCW lives where it does. The frequency sets the character of the arc, not just the pole:

• 50 to 150 kHz: thick branching lightning.
• 200 to 500 kHz, which is QCW: a long straight sword.
• 1 to 4 MHz, which is HFSSTC: a thin straight warm flame.

The mechanism, from Uspring: the higher the frequency, the more often charge is delivered into the channel, the hotter the channel runs, and a hot channel is straighter and branches less.

The ceiling comes from the switches. The bridge commutates at the working frequency, and at megahertz the switching loss plus an IGBT's current tail kills the device. Hundreds of kilohertz is about the limit for IGBTs; above that it is SiC or a valve.

---

The QCW diagram draws the buck between the bus and the bridge for exactly this reason, and its ramp readout is scaled from published coils rather than measured here.