What is the purpose of a schottky/zener couple on an isolated flyback topology?

This isolated flyback topology has been generated by TI's webench. It's a classic schematics with a feedback going through an optoisolator.

There are, however, two diodes connecting the switching side of the primary coil with Vin which are puzzling me.

The input is 5-15V but the Zener chosen is 22V. The schottky is rated 60V. I really don't understand their purpose. It seems like they are just going to introduce noise in Vin with voltage spikes generated at the end of the primary coil.

Could someone more enlightened than me take the time to have a look at it?

They are there to avoid over voltages on M1.

These can easily happen due to “leakage” inductance in the transformer.

No transformer is ideal so, in addition to the excess voltage when the transformer is discharging into the secondary, any current due to leakage inductance must also be discharged somehow.

Due to the needed excess voltage for normal operation of the secondary, the zener becomes necessary.

Depending on how much leakage inductance there is, it is possible to simply add an RCD snubber to dissipate the energy. But a zener and a diode is much easier to calculate.

• Simulating with LTSpice it seems you can simply ignore the spikes by picking a slightly higher drain source breakdown voltage. Say pick 100V instead of calculated 60. Wouldn't that be more efficient? – Tony Feb 24 at 1:15
• @Tony (1) simulations are not reality, have you considered all possible worst case parameter variations? (2) all things being equal higher voltage MOSFETs are more expensive than lower voltage MOSFEts. – Edgar Brown Feb 24 at 1:18
• @Tony no. The leakage inductance will always find a way. You need some sort of snubber. – TimWescott Feb 24 at 1:34
• I've seen a circuit that used an active snubber -- they dumped the energy from the primary leakage into a capacitor that powered a buck converter that fed into some power supply or another (I can't remember, but I think it was Vin). It raised the efficiency by one or two percent, which was pretty impressive because they were already north of 90% (I can't remember by how much). – TimWescott Feb 24 at 1:36