Mazilli ZVS inverter
A certain person, Vladimiro Mazilli, has developed a inverter which got a high popularity thanks to simplicity, parts accessibility and impressive stretchable flame-like arc. It's well known as ZVS driver. It's wrong, because ZVS stands for Zero Voltage Switching, which means switching, when voltage is equal to zero, giving no switching losses. In pracitce, there's no any inverters that switch perfectly at zero voltage. The voltage can be close to zero, but not equal. ZVS is a one of switching topology, not a name of this inverter. The correct name is Mazilli inverter!
It's worth to keep in mind few facts:
- inverter works in resonance, which means that the voltage on primary coil is equal to 2Π times input voltage (for example, with 30V input, we heave around 90V peak), and it must be considered when selecting parts
- transistor works in push-pull mode - they conduct current through primary coil in opposite directions
- resonance capacitor and primary coil forms a LC circuit and their values decide at what frequency inverter will work
- Zener diodes cuts down transistor gates, so they're not overvoltaged
What do we need to construct inverter?
- Resonance capacitor
-
It must be supposed to work in pulse application, and have correct rupture voltage (at least, supply voltage multiplied by 3.14 plus dozens volts for safety) and be a high quality made. The best capacitors I've had were 0.68µF 450VAC and Miflex X2 1µF 275VAC - below I posted a photo of them. Practically they didn't heat! The first one is a film type, second one is polipropylene metalised, which has few names, i.e. MKP (because it's a X2 class capacitor, it can work at voltage higher by dozen voltages than described 275V). Another the same type capacitors should work. The bigger capacitance is, lower frequency is and vice versa. That's what a equation for resonance frequency of LC circuit says
L - primary coil inductance in henries
C - resonance capacitor capacitance in farads
- Transistors
MOSFET and IGBT type transistors are fine for this converter, especially the first one due of lower rise, fall time. In choosing we must be mindful about few criterias. Firstly, drain to soruce voltage must be at least three times larger than supply voltage, if we don't want to overvoltage transistors by primary's amplitudes. As I've mentioned, they're capable to pump humongous current at incredible frequency, but they have weaknesses and overvoltaging is one of them. Second, they must have low internal resistance - as I wrote before, tens of ampers are passing through converter, so, lower resistance means less heating. According to scheme, R(ds)ON (a resistance in drain-source track at opened transistor) must be lower than 150 mΩ. You can try with higher value, but you must be aware of heat dissipation issue and eventually burn transistors. IRFP250 has, according to datasheet, quite nice parameters - 85 mΩ resistance, 180W dissipated heat (although it's not important because the resistance is low and converter works very efficient and heat losses are little) and maximum working amperage is 33A (two times more than we need). The only one thing hurts it's a drain-source voltage, which is only 200V. It means that we can supply with maximum voltage 50-60V. This transistors costs something aroung 1 euro in my town, but a lot of electronic shops sells them for much higher price (average 2 euro, even I've seen 4-5 euro). c4r0's converter worked very well with IRF640 transistors (200V, 18A and 180mΩ). You might try to buy a transistor package on internet auction, like ebay. Once time I bought package of 200 IRF840 transistors, but due of complications with finishing transaction, I've received 100 pieces without paying, even for shipping ;-D
- Diodes and resistors
Pair of fast diodes for voltage like as transistors and reverse time in tens of nanoseconds (although c4r0 used normal 1N4007 diodes and they worked well). 1UF4007 is easily accessible, cheap and fast. Zener diodes from 12V to 18V (I suggest to try 15V or 18V - higher gate voltage might increase overall switching perfomance in my opinion) and be at least 0.5W wattage (0.25W burned for me). Resistors don't need to be high power, 0.5W is enough. Remember, Zeners and resistors are in converter's part, which switches transistor gates, not pumps the whole current. Their suggested resistance is 10kohm
- Inductor
Iron powdered toroidal cores - it's what you're searching for. Each computer power supply unit has at least one in yellow and white colors. The amount of coils must be enought to make output current a sinusoidal. Too low amount means low autoinductance and current will be looking more a square. Too large inductance will give sinusoidal, but lower also amplitudes and core saturation might occur, where signal is ruined and core heats a lot. It's best to use osciloscope and add coils, until current will become a beautiful sinusoid. If you don't have this device, wind something around 20-30 coils of emmanelled wire with diameter minimum 1mm.
According to scheme, I've done the first version of converter, which photos are below:
Unfortunately, this version didn't work. This time I had a little knowledge about transistor converters, so I've combined, combined and comined, until I've decided to make again from start. Also, capacitors, which said no and for goodbye thrown a smoke made to this decision. Third version worked and it turned out why the previous one didn't. I've connected fast diodes and primary in incorrect way. Diode's anode have to be soldered to gate, katode to second transistor's drain. A cross-over configuration, as you see. I didn't know it makes different and I've soldered diodes separately, each for one transistor. Correct primary is a one coil winded in one way with center tape (bend it out, so you can solder wire there). In that way current coming from supply flows in different directions. My error was caused by two coils winded in different ways, so current was flowing in same directions. A little things, but they decided if converter works. Photos:
The whole setup was looking like that:
In top we have main toroidal transformer, 100W and output voltage 24V AC. I plugged to his output a rectify bridge and smoothing capacitors. The total capacitance was 2500µF. In that way I obtained 35V DC, which powered ZVS converter. Converter makes DC into sine wave with tens of kiloherces frequency. Changing magnetic field created by primary of high voltage transformer induces high voltage in secondary. HV outputs were connected to gold-looking thing and to bamboo stick with electrode at end. Also I did put a probe, so I knew if there's a high voltage, or not. Being prepared, I powered on circuit and saw this:
Arc acts like as flame, it's white and has little orange color on top. It's stretching, ignities when range between electrodes is ~4mm., and can be stretched to 3cm. Unfortunately, resonance capacitor blown out. It thrown out a weird white goo that looked alike to spider web. Even so, I was happy of my first success and became more hungrier for better arcs. Next I rebuilded converter:
Later I've done some modifications, but I can't recall them. Probably I was fiddling with power supply. Thanks to that, I could achivie more thicker and longer arcs. It could be stretched to 6cm. Below are photos of arc:
Inverter, thanks to advantages, can be used for various applications. It can be used to make induction heater, where instead high voltage transformer heating coil is used. However, it's not a high power class heater. Well, it pumps a lot of power, but the low frequency (heaters works best if frequency is at hundred kiloherces range) makes us hard to expect impressive effects like melting threads. CompWiz on 4hv.org forum described own successes and problems with ZVS based induction heater. Next, Vasil has made Tesla Coil with this converter. Results were really impressive. Addidionally, he checked how it would work with light bulb and did it with nice results.