Question about isolated DC-DC converter design

Hello KH,

The two main switching topologies that you mentioned, flyback and forward, as well as almost all of the others, involve duty-cycle control and filtering (averaging) to achieve regulation of the output voltage.

An overly-simplified design using only a bridge-rectified transformer has a number of short-comings that make it impractical at best, and damaging at worst, to implement. The first is that there is no voltage regulation possible. When Q2 is on, Vin is applied across the primary winding and the output voltage is equal to Vin/N. If Vin is always fixed, then N can be chosen to deliver exactly Vin/N to the output while Q2 is on. But if Vin varies, Vout will also vary as Vin/N no matter how much the duty cycle of Q2 changes. When Q2 is off, there is no defined voltage across the primary winding, so Vout (during Q2 off time) is also undefined. In an ideal situation, the winding voltage is zero, so Vout would also be zero. The requirement for regulated Vout is that Vin is fixed and duty cycle = 100%. The case for an overly-simplified flyback is even more unrealistic. 100% duty cycle is impossible because energy is only delivered to the output during Q2 off-time. This energy can be regulated with duty cycle but the output voltage is not constant. Vout goes to zero during the Q2 on-time. Without regulation, excess Vout may damage the load, and without current limiting, excess peak currents may damage the switching components.

The main point is that switched-mode regulation relies on state-space averaging to accomplish output voltage regulation, and the averaging is done by output filtering. In a forward-mode topology, an L-C filter is used where the output inductor L limits the amount of current that can be transferred through the transformer during the Q2 on time. In a flyback topology, an output capacitance is sufficient, because the transformer primary inductance limits the peak current during the on time. Now, in either case, the Q2 duty cycle can be modulated by a controller to be able to regulate the average output voltage to the load even as input voltage and output load levels vary. And the “size” of the output filter determines the amount of residual ripple voltage superimposed on the regulated average output voltage.

There are special cases where simplified switching and rectification of the transformer can be used, such as in certain point-of load distribution schemes, but these are indeed special cases. In these cases, all of the regulation and filtering has already been accomplished by another part of the overall power conversion architecture, and high-frequency transformer switching (at ~100% duty cycle and without filtering) is done locally to allow a pre-regulated pre-filtered high voltage to be distributed at low loss, and transformed to low voltage at the load, for small size.

I hope this answers your question.

Regards,
Ulrich

Hello KH, both primary and secondary must both have an average voltage of 0V.

Transformers work from a Volt x Seconds applied basis.
T1: Primary will apply Vpri for Tsec (VxS), building up flux in the core.
T2: The transformer core must reset itself, so after T1 the primary voltage will reverse and bring the flux back to where it started.

If the transformer has a DC component then eventually the core will saturate.

This picture illustrates the voltage action on a transformer.
Looking at the primary voltage, V+ is applied for +VxS and then the flux forces the exact same reverse voltage -VxS
The net result on the primary is 0V average.
There is deadtime after the core resets until the next cycle.

The secondary will look exactly the same, except the voltages are scaled by the turns ratio.
Vpri/Vsec = Npri/Nsec where N is the turns ratio.

5VWINDING264.TIF