ferrite cores in push pull topology

[Warpspeed]

Yes you are quite right Flapjack.
I was confused between mT and Gauss.

A design figure of 140mT peak, will swing 280mT when driven in both directions.
That may be fine at 20 Khz but the increased core loss at 50Khz would make it run fairly hot.

 

[risingstar]

1T=10^4 Gauss.

 

[c_mitra]

YThat may be fine at 20 Khz but the increased core loss at 50Khz would make it run fairly hot.

Energy loss per cycle is approx const; hence power loss will be proportional to the oper frequency. Copper losses will be determined by skin effect (else will be constant) and the power output.

 

[Warpspeed]

Its not that simple.
Hysteresis loss will be proportional to frequency, but eddy current loss goes up approximately frequency squared. Also larger volume cores tend to run much higher centre core temperatures than really small mini cores.
Its the temperature in the middle you need to worry about, not the surface temperature.

Manufacturers data will recommend flux swings at varying frequencies for different core sizes and geometries and materials. The most recent core materials are excellent, some of the traditional core materials that have been around for 50 years, not quite so good. If its from Asia, the core material could be absolutely anything, whatever it has printed on it.

You also need to consider the effects of staircase saturation and flux doubling. The flux swing may not always be well centred within the BH loop.
Even using current mode operation, or active current symmetry correction techniques to avoid core saturation in one direction, its prudent to be very cautious whenever bi directional drive is used with a circuit that does not allow a full core reset each cycle.

Anyhow try it and see.
If you start regularly popping mosfets, something will need to be done.

 

[c_mitra]

Its not that simple.
Hysteresis loss will be proportional to frequency, but eddy current loss goes up approximately frequency squared. Also larger volume cores tend to run much higher centre core temperatures than really small mini cores..

True that the eddy current losses are proportional to frequency squared but I presumed that for most ferrites they will be negligible, even at MHz ranges. I may be wrong, of course.

Also true that core cooling is next to impossible but a stacked pair of toroids will be better in this matter than one single (jumbo) one. Only slightly worse than a single one (of the components)..

Perhaps you have not noticed, Chinese ferrites have started getting better and better in recent time. And they are getting costlier at the same rate (good old days are over; for cheap ferrites). But they still have nothing printed on the top...

 

[Warpspeed]

Ah yes, Dragon brand ferrite....
And Panda brand ferrite......
Ah so !

Maybe its super high technology material, or maybe its just baked black mud.
I can tell you for sure, the Chinese powdered iron toroids come in many different colours, and are nothing like US made Micro Metals cores of similar colour.

If its for some mundane 20Khz switcher it probably does not matter much.
But if its some 500+ Khz switching wonder....
I wish you luck with those cheap e-bay unmarked parts.

The best Chinese engineering and technology is very good indeed, world class. the worst is truly awful.
Stuff from Taiwan is usually excellent.