Current ringing in SMPS

[wrenchone]

With a coupled choke, you want to be careful that both sides aren't too well coupled, as there will inevitably be some difference between the voltages (reflected and real) driving each winding of the choke. If the choke windings are too well coupled, minor differences due to differential voltage drops can result in large amounts of circulating current in the choke. Usinfg bifilar windings on a coupled choke results in coupling that is too good. A reasonable compromise in the case ot a toroid would be to put windings on opposite sides of the toroid. This will introduce enough leakage between windings to pad out small differences in driving voltage between the two windings. Leakage can be your friend...

 

[Eva]

If the choke windings are too well coupled, minor differences due to differential voltage drops can result in large amounts of circulating current in the choke.

This would happen when driving the inductor from very low imopedance voltage sources, but since there is one diode in series with each end of the inductor, the effect that you have described will just never happen. If one driving voltage is higher than another, then the current will come from the higher voltage source until the output capacitors charge accordingly and both diodes start to conduct. The system will always tend to balance itself.

What has to be reduced in that kind of inductors is capacitance, and starting both windings from the same side certainly reduces the amount of switching transients coupled to the output, but at the expense of increasing input capacitance.

In these circumstances, the only thing that can help is operating the circuit at low frequencies thus allowing for slower switching transients without yielding too much switching losses (at 40Khz the switches may be turned on much slower than at 100Khz, thus producing less EMI while still yielding lower total losses and keeping fast turn-off).

 

[carvinguy]

I tried a few different winding arrangements with the ouput inductor. I first tried winding it with both outputs bifilar, in opposite directions (closely coupled turn for turn in opposite directions.) This has decreased the current ringing dramatically in the primary! I still have a very high frequency leading edge spike, but I think this is to be accepted and unavoidable (correct me if I'm wrong) I also tried another output inductor winding experiment where I wound it similar to a common mode input filter inductor- each winding occupying separate sides of the toroid. This also performed well, but I had more high frequency noise, although the lower frequency 100kHz ripple was lower than the other winding.

I also am playing with adding a small inductor in series with the primary to help snuff out more of the current spike.

I'm wondering if I can now go back to standard MUR860 output diodes again since I was trying to solve the problem with super fast doides. Something else to try...

Overall this system is working quite well and most of the noise on my amp is gone and efficiency is acceptable. I am considering dropping the frequency as Eva suggests, but I find the troubleshooting challenging, and a terrific learning experience that I can use for class-D projects I am considering.

Thanks for all your help so far!

 

[Eva]

Be careful with these diodes, being super ultrafast doesn't mean that they have a soft recovery characteristic, in fact some models suffer from very sharp reverse current spikes. On the other hand, MUR860 is slower but softer, and the 200V version would be even better if this voltage rating is enough for your circuit.