Processing math: 100%

ZVS Circuit

ZVS stands for Zero Voltage Switching. It is a technique used in power electronic circuits for minimizing switching losses by switching the power transistors only when the voltage across them is zero. In the hobbyist community, a particular power oscillator circuit is associated with the term ZVS, namely the oscillator circuit in the figure below.

MISSING IMAGE

It is an LC oscillator and the switching of the MOSFETs indeed occurs when there is close to no voltage across them (this can easily verified by setting up e.g. an LTspice simulation). A practical realization of this circuit can be seen in the figure below.

MISSING IMAGE

I built it with IRFP260N power MOSFETs, MUR460 diodes and the used 470Ω resistors are rated for 10W. The Zener diodes are standard 12V, 1W ones and the 10kΩ resistors are rated for 0.25W. I made the choke, i.e. the inductor at the top in the above schematic, from an unknown ferrite or iron powder core. The capacitor of the LC tank circuit is made from four MKPH capacitors, each of them having a capacitance of 0.68μF and rated for 800V at 50kHz (I salvaged them from an induction stove top). The four capacitors are wired up in pairs of two in series and these pairs are wired in parallel, so the total capacitance is still 0.68μF, but the stress is shard among all of them.

The coil of the tank circuit is made from 1.5mm2 home installation wire. Its construction is illustrated in the figure below, i.e. there are actually two coils overlaid for good magnetic coupling. The point where the two overlaid coils are connected forms the center tap of the complete coil. In the practical setup, each of the two overlaid coils consists itself of six separate strings of 1.5mm2 home installation wire. To make this coil, I first took twelve wires in parallel, twisted them together and wound them to a coil. Then, I selected six of these twelve wires to form one end of the coil. The other ends of these selected six wires form the center tap and were then connected to the remaining six wires, the ends of which form the other end of the coil. Several zip ties assure that the coil holds its shape.

MISSING IMAGE

So the coil has effectively 16 turns, a diameter of approximately 60mm and a length of approximately 85mm. According to the online inductance calculator https://wetec.vrok.de/rechner/cspule.htm, this should give us an inductance of approximately 8.1μH. Together with the 0.68μF tank capacitor, the resonance frequency follows as fres=1/(2πLC)68kHz.

The circuit is built such that the tank circuit can easily be replaced. Another tank circuit can be seen in the two figures below. It has a capacitance of 120nF and the coils inductance, according to the online calculator, is 3.4μH, resulting in fres=249.2kHz.

MISSING IMAGE
MISSING IMAGE

Test Results

With both tank circuits, a supply voltage of 12V suffices for reliable oscillation. With my old analog Tektronix 2215A oscilloscope, I measured an oscillation frequency of 71.4kHz with the first tank circuit, and 250.0kHz with the second one. The oscillation frequency varies only very slightly with the supply voltage.

In the image below, the voltage across the tank circuit (the first tank circuit, i.e. the one with the lower resonance frequency) for a supply voltage of 20V resp. 37V can be seen. The current drawn from the power supply is 0.77A resp. 1.45A, corresponding to a power dissipation of 15.4W resp. 53.7W. Neither the MOSFETs nor the capacitors heat up significantly, so it seem like most of the power is dissipated in the coil, which indeed heats up a lot. The voltage across the tank circuit seems to have a relatively clean sinusoid shape, with slight distortions at the zero crossings. The amplitude is approximately 63V, resp. 115V, from which the amplitude of the current flowing in the tank circuit follows as approximately 18A resp. 33A.

MISSING IMAGE
Vertical setting: 20 V/Div | Horizontal setting: 2 us/Div.
MISSING IMAGE
Vertical setting: 50 V/Div | Horizontal setting: 2 us/Div.

With the second tank circuit, the voltage waveform is more distorted at the zero crossings. The MOSFETs also heat up much more, but the coil is again the part which heats up the most.

When putting a metal object into the coil, the current drawn from the power supply rises and the metal object heats up due to eddy currents induced in it by the alternating magnetic field of the coil. In the figure below, you can see a 4.1mm diameter nail heated up until glowing brightly. Since the power supply used in these tests can only deliver 3.2A, I was limited to small metal objects. This limit is already reached with the 4.1mm nail put only half way into the coil. The MOSFETs and the capacitors still did not heat up significantly. The amplitude of the voltage across the tank circuit and the oscillation frequency were hardly affected by the nail in the coil.

MISSING IMAGE

With the second tank circuit, I was not able to heat the nail up until glowing. The coil got too hot too quickly.