Again, if indeed it was a forward converter (which sounds likely at this point) -- then, the switch and primary are arranged much as Arthur's schematic above. But that's a flyback supply: note that, when the switch turns off, the primary voltage (the non-dotted side of the winding) shoots up (flyback), and the secondary voltage (non-dotted) shoots up in sync, until it's clamped by the output diode.
In a forward converter, the output filter is a choke-input type, which means there's no clamp on the inductor voltage. The secondary is phased oppositely, so that the diode conducts while the switch is on, charging the inductor. (The inductor maintains its current during off-time, with a second clamp diode from GND.)
That means the transformer's flyback is unconstrained, which will destroy the switch.
A note about the transformer itself: in the flyback supply, the transformer is made with a low inductance (it's better thought of as several coupled inductors; inductors store energy, transformers merely transform it). In the forward converter, the transformer is made with a high inductance, so as not to store much energy, and this means the flyback pulse tends to be weak, only delivering a few watts perhaps. (The excess voltage alone is enough to destroy the transistor though.)
There are three typical ways to handle the transformer flyback:
One is a clamp diode into a capacitor. This acts like a boost converter on the primary side, creating a high voltage supply (say 600V). The supply is just a few watts, and isn't useful for anything else, so we burn it off in a load resistor. This is called an RCD clamp snubber: it uses a resistor, capacitor and diode, and the capacitor is relatively large, so that it acts to clamp the voltage (the change in capacitor voltage is small during a cycle). (This would be D1-C4-R8 in the above schematic, except with a larger value for C4.)
Another is to add an auxiliary winding on the transformer, which we can connect to one supply rail (GND or +V) on one side, and with a diode to the other rail. Now the flyback is clamped at exactly +V, just as in the above schematic, but instead of delivering useful output power or burning it in a resistor, it is "stirred" or recycled back into the supply, costing minimal losses. Great, huh?
The last is to use two switches and a single winding. This seems over-the-top, but semiconductors as a group are one of the cheapest parts of a power supply design, so we can afford to use more of them if it offers other savings. In this case, we can switch the top and bottom ends of the primary winding, so that the whole winding itself is reversed in polarity during the off cycle -- it serves as its own reset winding, greatly simplifying the transformer design.
This is called "2-switch forward", and also works for flyback with the bonus that the transistors can never be over-volted by a suddenly open-circuit load, and that the transformer's leakage inductance is safely controlled (and indeed recycled), whereas the energy stored in leakage is usually just burned off in a one-switch flyback (which is the intended function of D1-C4-R8 in the above schematic). The downside is, the flyback voltage obviously can't exceed the supply voltage, so some adjustments need to be made relative to the traditional flyback design (a somewhat lower nominal duty cycle).
Tim
Home
Help
Search
Login
Register
