Need help on flyback converter.

[blueskull]

Hi guys,

I've be debugging my power supply project for a while, and tonight (maybe morning now, my local time is 3:14 am ) I finally got it worked (only the flyback stage, for more info pls check out my usb mini power supply topic).

The efficiency data is pretty nice, I've got a 91.5% yield. However, the drain waveform is very, very disturbing, pls see the attached files.

The problem is NOT on oscilloscope grounding or MOSFET gate waveform, I doubled checked them, and they are just fine.

Generally you expect the voltage goes down when MOSFET turns on, then the voltage goes up to flyback output voltage/turn ratio (2 in this case), then when all energy in the inductor was depleted (I operate this thing in DCM), the voltage should go back to supply voltage.

I can understand there will be some sort of ringing overshoot, but the waveforms shows huge ringing undershoot, which I can not explain it using my knowledge.

The switching frequency is 800 kHz (precise value is 798.18 kHz), and the primary inductance without bias current is somewhere around 2.7 uH measured with smart tweezers st-5s at 10 khz, 0.9 vrms setting.

I can observe NO significant voltage fluctuation on both MOSFET source and supply voltage (I measured directly in the transformer's solder joint).

Any suggestions or analysis are super appreciated.

[T3sl4co1l]

That's normal.

If you're very picky, you can add an R+C across the inductor (or output diode, or MOSFET) to dampen the ringing, but you will reduce efficiency as well.

Instead of a clamp RCD snubber, a dV/dt snubber is often better: with proper values, it serves all three purposes (absorbing turn-off energy, reducing dissipation in the transistor; slowing the initial overshoot and dampening the subsequent shallow ringdown; and dampening the ringdown to +V).

Schematic of a recent example; values are for a quite high current current level, hence the small resistors and large capacitors.



Tim