4.4 Power Transformer

The main purpose of a transformer is to translate AC voltage/current from one level to another. The second is to provide galvanic isolation between different reference grounds, especially if power is drawn directly from the grid. In that case, the reference potential is earth (PE connection), therefore, all the lines coming out of the system (considering it as a black box) must be galvanically decoupled from that potential, allowing no possibility for direct current flow against the PE connection (earth, metal housing etc.) One problem solved; many others generated.

Every transformer is a source of Common mode noise. This is a challenging topic when working with high-frequency power converters. One priority is to prevent high-frequency noise being coupled across the transformer. A larger transformer coupling capacitance will allow more high-frequency noise to cross. But decreasing the coupling capacitance will increase leakage inductance, which directly influences the converter operation and can cause significant performance degradation.

For the DAB topology, the main energy storage element is an inductor in series with the transformer primary winding. Because of the huge current swing of 2x I_PEAK(MAX), this inductor cannot be a simple storage inductor but instead, must be a shim (or resonance) inductor.

This shim inductor can be bulky due to special construction requirements. With the shim inductor being in a series with the transformer primary winding, the leakage inductance of the transformer (which behaves like a shim inductor) is used to relax the spec for the shim inductor. The total inductance used for energy storage is given by Equation 4-1.

The proportion of L_SHIM versus L_LEAKAGE is not set in stone, but experience has shown that a ratio of L_SHIM to L_LEAKAGE of around 2:1 works quite well. This ratio can lead to lower transformer parasitic capacitance which can help EMI performance. This ratio may vary with frequency and converter power level or currents.

The turns ratio was fixed based on arithmetic mean value of $\raisebox{1ex}{$2x{V}_{BUS}$}\!\left/ \!\raisebox{-1ex}{$(V_{OUT\left(MIN\right)}+V_{OUT\left(MAX\right)})$}\right.$.

The idea is to have the optimum (unity) working point in the range of the largest working area.