Understanding the Royer Oscillator

[ZeroResistance]

I'm trying to understand the royer oscillator from http://www.rmcybernetics.com/projects/DIY_Devices/diy-induction-heater.htm



I can see that the fets alternatively come on and there is a half sine wave on their drains.
What i want to understand is since it is a center tapped trasformer in parallel with the tank capacitor.

Will current in both halves of the center tapped transformer flow in the same  direction or in the opposite directions?

[bktemp]

The circuit consists of two parts:
One is the LC resonant circuit formed by L1 and C1. The power oscillating between inductor and capacitor is much higher than the current drawn from the power supply. Therefore the dominating current waveform is a sinewave.
The second part is the centre tapped inductor and both mosfets. Its only purpose is to replenish the energy lost in the LC tank circuit. Assuming a large L2, the current flowing here is almost a square wave, alternating the current between both halves.
If you add both currents (from the LC tank and from the mosfets) together you have the current in both halves of L1.

Simulating the circuit and looking at all currents simultaneously is probably the best way for understanding this circuit.

[ZeroResistance]

Thanks for you answer bktemp,
will the current at any instantaneous momemt be flowing opposite to each other in the 2 halves of the CT transformer?

[Ian.M]

At any instant: in the same direction.
LTSPICE sim attached

[bktemp]

At any instant: in the same direction.

Being pedantic, no: Because of the DC offset caused by the current flowing through the mosfets, the current passes 0A and therefore changes the direction at a slightly different time in one half than the other one.
But the DC current is small compared to the sinewave current in the LC tank, therefore it is negligible, and you can they the current is nearly identical in both halves except for very high losses in the LC tank.

[Ian.M]

Justifiable pedantry acknowledged. 

[ZeroResistance]

At any instant: in the same direction.
LTSPICE sim attached

THis is amazing, you are a champ! did you cook that up in under a minute?!!

[Ian.M]

More like 20 minutes.  I had to guess some reasonable values for the inductors in the CT transformer then fiddle around a bit with various loads etc. to check it was working properly.

I only bothered because I wanted a good Royer sim for a VFD HT and filament driver I'm working on.  The bipolar Royar circuits are too fussy to wind the transformer for as you need a CT base drive winding and I hadn't previously considered using MOSFETs

[timb]

Check out Jim Williams' app notes on CCFL backlighting (there's several of them). They are great resources for Royer Converters/Oscillilators. Another app note from Jim about "Ultra Low Noise DC/DC Converters" has some good circuits and discussions as well.

[Ian.M]

Thanks for the suggestion.

The requirements for a VFD are a little inconvenient.  Its the AC filament supply that's the problem - if you use DC you get a brightness gradient across the display, and high crest factor waveforms are also undesirable.  I'm currently using a 1W DC-DC converter to boost from 5V to +30V (25V stacked on top of the 5V rail), and DC for the filament.  The intention is to replace it with the Royer converter.  However the need to maintain a reasonably constant filament voltage means I've got to keep the load on the HT rail approximately constant hence the shunt Zener.  The alternative would be a far more complex circuit to maintain a constant amplitude on the LC tank.

[ZeroResistance]

I only bothered because I wanted a good Royer sim for a VFD HT and filament driver I'm working on.  The bipolar Royar circuits are too fussy to wind the transformer for as you need a CT base drive winding and I hadn't previously considered using MOSFETs

The circuit at RMCybernetics seems to be pretty basic there are other versions on the web that have adequately sized gate protection zeners for the Fets, and some others too with a NON CT based transformer. Be aware that the voltage on the drains would be almost 4 times the supply voltage so you may have to select a mosfet with adequate Vds rating.

[Ian.M]

Basic is good. If it gets too complex, I'd be better off with a push-pull SMPS controller!  Vds and Vgs limits are not a major issue with only a 5V supply. I've even figured out a method of regulating the amplitude by varying the bias which stabilises the output without significantly altering the filament voltage.