L1-C5-reflected load impedance is the resonant circuit that determines the self-resonant frequency of the converter. The load impedance reflected via the transformer is usually capacitive in nature.
What does the ZVS do in the circuit shown below?
ZVS means Zero Voltage Switching. It means that the switch turns ON when the voltage across the terminals is zero.
ZVS can be used only at turn-on. At turn-off, the voltage across the switch is very close to zero since the switch is a short. So ZVS means really "zero voltage turn-on".
Turn-off can be current-based, since a variable current is always available to measure and use for timing when the switch is closed - but not when it's open. The abbreviation is ZCS - zero current switching.
Does it produce a sine wave or it just makes the generated square wave much smooth[er]?
How close to sine wave it gets depends on the Q factor of the LC circuit. A high-Q LC circuit, i.e. one having very low losses, would produce a clean sine wave. High-Q is not possible though since there's a load in series with the LC circuit. It acts like a parallel RC circuit in series with L and C. So the voltage waveform will be that of a low-pass filtered square wave.
What is the purpose of using capacitors on the IGBTs?
It is to limit the dV/dt across the IGBT. IGBTs, and most other switching devices, have limits on the dV/dt rates they can survive.
Since dV/dt is limited, the EMI is also slightly alleviated. But nothing is free: while the capacitors limit dV/dt, there's current flowing through them - in fact, the more they limit dV/dt, the higher the current will be. So they are critical elements in the switching current loop and need to be laid out and selected carefully.
how does this circuit ensure that the current and voltage gets zero before switching?
The diagram shown is a block diagram. It is not a complete circuit! There are many parts missing - notably there's no feedback of any kind, no overload protection, etc. It illustrates a concept at a very high level.
For ZVS, the controller needs to be measuring the voltage across the switches so that it can determine when the voltage is zero.
The controller also needs a measurement of the current flowing through the load L1-C5-T1.
So, at the very least, the controller will need the following feedback signals:
- DC Bus (+) voltage relative to DC Bus (-),
- voltage at point A relative to DC Bus (-),
- voltage at point B relative to DC Bus (-),
- current flowing through the load circuit.
The controller may also use a feedback voltage from the high voltage side of the transformer, or it may measure this voltage on the primary side of the transformer as it's reflected back, or even may use an approximate transformer model along with the voltages and currents in the list above to derive an estimate of the transformer output voltage. It depends on how the circuit is applied, and what are the requirements on the stability of the output voltage.
The feedback sensing can be arranged in a different way, but overall the controller will need this information. If it doesn't directly measure it, it'll have to derive it from other measurements.
For example, instead of measuring the voltage at point A and point B relative to (-), there can be a measurement at point A, and a measurement of total voltage across the load, i.e. point A relative to point B. Such alternatives can be chosen based on how they affect the performance and price of the feedback circuit. They are all options that lead to the same end result.
By the way, the "paper" you link to is something any engineer working with SMPS design could write in an afternoon while drinking heavily to dull the blaring horns of the ethical watchdog in their mind.
The "paper" is useless in every sense of the word, other than as an example of how not to write papers.
Even if we go past the "pay to publish" anything-goes attitude demonstrated by that publisher, the multiple times they retracted hundreds of papers/articles because they were plagiarized etc., there is still a way to write a laboratory/practical design paper that can be useful to people and yet would not fit into the scope of what leading journals publish. Alas, people who can write that usually just publish it as a blog/report on their own page or the page of their employer, or on Hackaday, etc.
Yes, some stuff on Hackaday is nothing fancy, but at least there are mostly reproducible results to be seen there.
If you want to learn more about the subject of the switching power supply design, you should look for good books on the topic, written by people with engineering experience. I'm sure you can figure out how to find PDFs of some of those books online.
There are also excellent course lectures/notes available online on that subject.