(Editor's Note: To see a linked list of all entries in this series, click here.)
To see a video on this topic by the author, click here.
Have you considered using a single-ended primary inductor converter (SEPIC) topology for a bias supply? If you don't need isolation, it just might make sense. The SEPIC has several features that make it more attractive than a non-isolated flyback. MOSFET and output rectifier ringing are controlled to reduce electromagnetic interference (EMI) and voltage stress. In many cases, this lets you use lower voltage parts, which may cost less and be more efficient. Also, a multiple-output SEPIC improves cross regulation between outputs, which may eliminate the need for linear regulators.
Figure 1 shows a SEPIC converter which, like a flyback, has a minimal parts count. In fact, this circuit would be a flyback, if C1 were removed. This capacitor provides voltage clamping of the semiconductors to which it connects. When the MOSFET is turned on, the reverse voltage on D1 is clamped by the capacitor through the MOSFET. When the power switch is turned off, the drain voltage rises until D1 conducts. During the off time, the MOSFET drain voltage is clamped by C1 through D1 and C2.
A SEPIC converter with multiple outputs puts a constraint on the winding ratios. One of the secondary windings needs to have a 1:1 turns ratio to the primary, and C1 must connect to it. In the example circuit shown in Figure 1, the 12-V winding has a 1:1 ratio, but it could have used the 5-V winding instead.
Figure 1. Shown is a multi-output SEPIC converter.
The circuit in Figure 1 has been built and tested. It was operated both as a SEPIC with C1 in place, and as a flyback with C1 removed.
Figure 2 shows the MOSFET voltage stresses in both operating modes. In the flyback mode, the MOSFET drain went to nearly 40 V, while in the SEPIC mode the drain voltage was only 25 V. So the flyback design would have to use a 40- or 60-V MOSFET, while the SEPIC design could use a MOSFET rated for only 30 V. In addition, the high-frequency (greater than 5 MHz) ringing would be problematic for EMI filtering.