Parametric Design Guidelines for MW Oven Inverter

Today’s Microwave ovens are often equipped with inverterized MW generators; compared with old style appliances, the magnetron is no more driven by a bulky ferro-resonant power supply, but by an electronic inverter, lighter, more efficient and, moreover, able to modulate the MW power, at least to a certain extent. The electronic power supply is generally designed around a resonant inverter which drives an HV transformer, followed by a voltage multiplying rectifier, needed to provide the 4-6kV DC requested by the magnetron. Single ended parallel resonant inverter is often a good choice, at least for MW ovens fed by 115V line. Being such topology also used for low cost induction heating hobs, it could appear similar design process may be applied; unfortunately, the presence of the HV transformer and of the voltage multiplier changes a lot the way the converter operates, and a revised design procedure needs to be developed. In this paper, parametric design analysis for single ended converter aimed to MW ovens is presented, and results compared with Pspice® simulation and practical measurements on a commercially available magnetron inverter. Inverterized Magnetron Power Supply Figure 1 is a sketch of a typical magnetron used in MW oven appliances; also, a typical schematic of its power supply is shown. The single ended parallel resonant converter uses HV transformer leakage inductance and primary resonant capacitor to generate resonant waveforms across the primary of the HV transformer. To limit the isolation requirements of the HV trafo, its secondary side usually provides half the voltage needed to polarize the magnetron, and a diode/capacitor voltage doubler configuration is then needed to develop the 4-6kVdc requested by the RF tube. Converter Parametric Analysis While the behavior of the single ended converter is well known, the combination of such converter with the HV transformer and the voltage doubler poses some challenges when parametric design is attempted. Generally speaking there are 3 constraints which have to be respected: a) The power to be delivered to the load b) The maximum Vce/Ice allowed by the power switch c) The need to maintain ZVS turn-on Figure 1: Magnetron sketch and inverter tolology. In the case of MWO, a general model of the converter is shown in Fig. 2, and constitutes the basis for the following discussion. Figure 2: General model for the converter Parametric design needs the output stage (HV trafo and voltage doubler) being modeled so that they appear on the primary side as a R_L network. Then, the design of the converter can be done as for the induction heating application. But, the operation of the inverter is NOT symmetric. In fact, by Fig 2 and assuming the HV trafo is not inverting, said Ton and Toff the ON and OFF times of the switch Z1, during Ton the voltage applied at the primary is Vin, while capacitor C9 is re-charged; during Toff, voltage at the transformer primary is Vce(t)-Vin, while capacitor C11 is re-charged. The average voltages across C11 and C9 are not equal to each other. As a “very first” approximation, we can say that V(C11) and V(C9) stay in a ratio which is function of Vcepk and Vin, Ton and Toff. More precisely : V(C9)ave = V(C11)ave * Vin/(Vcepk-Vin) * Ton/Toff The approximation comes by the fact the voltage drop on the HV trafo secondary leakage inductance is neglected, and such voltage drop is different between Ton and Toff as the recharge current is different. If C9 and C11 have a value which stays in the same ratio as their peak voltage, then also their voltage ripple will be similar. We also need to consider that, to avoid adding an external inductor on the primary side, the resonating inductor is achieved by ad hoc increasing leakage inductance of the trafo (see Fig.3). Figure 3: HV transformer primary side model To complete the converter model, the “equivalent” load is needed. Because of the effect of the voltage doubler, the reflected load on the primary side of the HV trafo, should be Rload/4. Actually, because the voltages across C9 and C11 are not equal, “different” loads are reflected at the input of the voltage multiplier, and so different damping of the primary resonant circuit occur during Ton and Toff. A detailed analysis is very complex; in line of principle, the equivalent resistances at the input of the multiplier are different between each other and are between RL/4 and RL/8. It is now possible to calculate the total impedances at the transformer primary, which are in the form: Zon N j , s Lj ⋅ 1 K − ( ) ⋅ s 2 K ⋅ 1 K − ( ) ⋅ L j ( ) ⋅ s K ⋅ Lj ⋅ Ron N 2 ⋅ +      s L j ⋅ Ron N 2 + + :=