I have found this weird DC-DC converter topology in datasheet of MC34063 from ST.
The primary charging is single ended, so at first it seems like flyback, but it has four diodes on the output instead of only two. Is this some forward or flyback hybrid? What is it called? Or is there a mistake in the datasheet?
It is a flyback.
Don't let the number of diodes confuse you here - that bridge rectifier is just necessary because the circuit strives to do full-wave rectification.
A flyback converter's transformer secondary is in principle isolated from the primary, so the output differential voltage can be referenced to any voltage.
This has some characteristics of a forward converter, (or it uses "forward" topology). See this wikipedia article or this guide.
Specifically, the current in the secondary of the transformer may occur at the same time as current in the primary.
However, it also has a characteristic of a fly-back.
Specifically, because the primary side of the circuit has no method to "reset" the flux induced by the DC component of the pulse train, (for example a diode and "reset" coil), the transformer must store (some) energy during part of the cycle, and release it to the secondary during another part of the cycle.
So, perhaps it is a hybrid between flyback and forward topologies? I originally thought it was a simple forward converter, but discussion with Jano952, gives me doubts.
Because the primary circuit goes open-circuit when Q1 is off - there is no snubbing or core-reset circuitry - I'm in the flyback camp.
IMHO - The reason for four diodes is that there are ((two)) outputs and it should improve efficiency.
In a standard, one-output flyback circuit, there is no transformer secondary current when primary current is moving because the output rectifier is reverse biased. That is why schematics show opposing transformer polarity dots. The "transformer" in a typical flyback circuit is not actually a transformer in the traditional sense; it is a pair of coupled inductors.
Anyway, in this circuit, the center-tapped secondary can create bipolar outputs with only two diodes. When Q1 is on and there is primary current, there also is secondary current because the other two diodes create a path for that direction of secondary current. that is one possible reason there are no transformer dots. Of course, the other is that the designer left them out. Still ... Cute.
Fly-Forward is a valid topology. This can run with either a gapped core (preferred), or solid ring toroid. Lack of any primary side snubber or clamp or any quasi-resonant capacitor, is bit scary though. Leakage inductance energy can be kept small, only if the primary current stays small (~ < 100mA ) and the core depth of magnetization is low ( < 30%). This would imply the core is over-sized for the power being transferred. The transistor output of the MC34063 can clamp a small amount of avalanche energy (and it has internal C-E capacitance), but personally I think that's not a highly reliable thing to depend upon. This circuit might have high EMI emissions. The core reset is handled by the secondary diode clamping (fly-back mode). If you want to do this without any primary snubber (or clamping or resonance), stick to really low power ( < 1W ) and wind the transformer as twisted pair pri-sec windings. The high inter-winding capacitance will partly absorb the low residual leakage inductance energy at switch-off, but unless you pick an optimal oscillator frequency (might be ~100khz) for the MC34063, you'll likely get terrible ringing. I can see this circuit not doing so well if the output load is too high or too low. Even though it has the short-circuit resistor, such a circuit might not survive a secondary short-circuit. The MC34063 allows for high duty cycle (~ 85% ) which for forward converters using reset ratio of 1:1, the core will saturate. Its possible this design actually relies upon the current limit getting activated, and uses that to enforce the duty cycle stays roughly below 50%. Without any details about core, windings, operating frequency, input voltage, or current limit resistor; we are left in the dark for deeper analysis. For sure however, if the power transferred is very very low (and thing runs ~100Khz or higher), this circuit can work. At the very least, a snubber (or bi-directional TVS) on the primary would be much wiser, than nothing at all.
