I just used it about 43 hours ago to get an understanding of a circuit idea I had ([1] being the secondary of [0], so that the abstracted switch (some bidirectional transistor approach of still-unclear nature; suffering so much ripple (+- half input voltage) requires this rippling to be low-loss) in the center can turn on with no voltage across it), which was to use loose coupling between the inductors of the phases of a modified 3L-FC buck [3] (running in divide-by-two mode as a switched capacitor converter; plans are 5 phases at 6.7 MHz for the half-bridges and 27 MHz for the resonant inductors), that has been modified by splitting the flying capacitor into two series capacitors and placing an auxiliary switch in series with an inductor between the center of the flying capacitor and the output of the 3L-FC (in a buck, that would go through an inductor and then to an output smoothing capacitor; I only aim for a 2:1 DC/DC transformer so output voltage will be a little below half the input voltage (difference is due to resistive losses in the half bridges that mostly cover up the ripple losses of the flying capacitor)).
This auxiliary switch was initially meant to block only half the output voltage and use the inductor as an LC resonance with the output capacitances of the half bridges to flip them around after they just calmly turned off.
The calm turn off is fading over input/output currents of the converter to other phase(s) of the converter; this is to virtually eliminate any (inherent) need for filter capacitors at the input and output of the converter by ensuring the phase currents, when added up, result in a time-invariant current for the converter as a whole (like how 3-phase grid supply theoretically (i.e., with active power factor correction) doesn't need the (mains-side) smoothing capacitors in a computer power supply, because a the (virtual) resistor the grid operator prefers to see has constant power draw from the grid).
The issue was that the auxiliary switch still suffered from hard-turn-on (i.e., no ZVS-on for it) which turns out to be somewhat of a problem when the goal is to take in e.g. up-to 60kV (+20% peaks/surges on top of 50kV nominal) from an MVDC line, and convert it down to something a "just" 1.6 MW induction motor with it's inverter can comfortably handle (e.g. situation in a Siemens Vectron MS, except that those only go up to 25kV AC): the first stage would take in 60kV, output 30kV, and that switch would have to block +- 15kV.
At the expense of the auxiliary switch seeing up-to full output voltage, sightly coupling the inductors across phases should allow the coupled fraction of the full current of the phase who's auxiliary switch is currently on to fully charge/discharge the much smaller output capacitance of the auxiliary switches in the other phases while the entire current in itself charges/discharges it's own [phase's] half bridges.
I'm sorry the switches in the falstad links aren't automated; the modified 3L-FC switches between which of the two controls is on, and briefly activates the center switch while both control inputs are off.
The weakly-coupled transformer has one activate the switch when the red graph gets closest to the 0V center line, and deactivate it when the yellow current in the same scope has completed the sine-half-wave/crosses the 0A center line the first time (again) after the switch was activated. The result should be that the green voltage across the simulated output capacitance of the half bridges on the right scope transitions quickly to the other polarity while the switch is active, and then stays there with only mild ripple until the switch is activated again. I believe in the 5-phase converter it would be activated not the first time afterwards the red goes back to (almost) zero, but the one second time, so the (with-switch-active) current half-waves are distributed round-robin.
This auxiliary switch was initially meant to block only half the output voltage and use the inductor as an LC resonance with the output capacitances of the half bridges to flip them around after they just calmly turned off.
The calm turn off is fading over input/output currents of the converter to other phase(s) of the converter; this is to virtually eliminate any (inherent) need for filter capacitors at the input and output of the converter by ensuring the phase currents, when added up, result in a time-invariant current for the converter as a whole (like how 3-phase grid supply theoretically (i.e., with active power factor correction) doesn't need the (mains-side) smoothing capacitors in a computer power supply, because a the (virtual) resistor the grid operator prefers to see has constant power draw from the grid).
The issue was that the auxiliary switch still suffered from hard-turn-on (i.e., no ZVS-on for it) which turns out to be somewhat of a problem when the goal is to take in e.g. up-to 60kV (+20% peaks/surges on top of 50kV nominal) from an MVDC line, and convert it down to something a "just" 1.6 MW induction motor with it's inverter can comfortably handle (e.g. situation in a Siemens Vectron MS, except that those only go up to 25kV AC): the first stage would take in 60kV, output 30kV, and that switch would have to block +- 15kV.
At the expense of the auxiliary switch seeing up-to full output voltage, sightly coupling the inductors across phases should allow the coupled fraction of the full current of the phase who's auxiliary switch is currently on to fully charge/discharge the much smaller output capacitance of the auxiliary switches in the other phases while the entire current in itself charges/discharges it's own [phase's] half bridges.
I'm sorry the switches in the falstad links aren't automated; the modified 3L-FC switches between which of the two controls is on, and briefly activates the center switch while both control inputs are off. The weakly-coupled transformer has one activate the switch when the red graph gets closest to the 0V center line, and deactivate it when the yellow current in the same scope has completed the sine-half-wave/crosses the 0A center line the first time (again) after the switch was activated. The result should be that the green voltage across the simulated output capacitance of the half bridges on the right scope transitions quickly to the other polarity while the switch is active, and then stays there with only mild ripple until the switch is activated again. I believe in the 5-phase converter it would be activated not the first time afterwards the red goes back to (almost) zero, but the one second time, so the (with-switch-active) current half-waves are distributed round-robin.
[0]: https://www.falstad.com/circuit/circuitjs.html?ctz=CQAgjCAMB...
[1]: https://www.falstad.com/circuit/circuitjs.html?ctz=CQAgDOB0Y...
[3]: https://www.ti.com/lit/pdf/slyt807