Gapping a ferrite core for Full Bridge SMPS transformer?

[Warpspeed]

What it shows is that current mode control with soft start would be a far safer way to prevent disaster at start up.

 

[treez]

Thanks, the simulation of post#20 is in current mode.
Even if you have a soft start, it will be ‘dis-armed’ in cases of restart (ie after an OFF-ON incident)

 

[FvM]

The start-up behavior of the circuit seems to be rather chaotic and far from deserving the name "soft-start". It can be surely improved, but as long as current mode prevents large saturation currents, partial core saturation for a few 10 ms or even seconds doesn't hurt.

 

[treez]

Thanks, indeed the ltspice simulation of post#20 doesn't have a soft start. That is deliberate ,because I wanted it to be like for a non-soft-restart.

but as long as current mode prevents large saturation currents, partial core saturation for a few 10 ms or even seconds doesn't hurt.

Though what Bmax for a non-gapped core would you say Is allowable?.....the 0.397mT that the sim of post#20 gets to is high, and 10ms is all it needs to start to go into "flyaway" saturation. -Because the sim of post#20 is based on a non-gapped design.

 

[FvM]

what Bmax for a non-gapped core would you say Is allowable?

No specific limit, a matter of tolerable current and core heating. As thermal time constant is in a several minutes order of magnitude, transient saturation is no thermal problem.

 

[treez]

If you look at the Figure 3 graph on page 92 of the attached, you can see that 3C94 ferrite is deeply saturated at 397mT.
I just cannot see it ever being acceptable for the full bridge smps core to saturate to that extent, as especially with an ungapped core, it will suffer more “brick wall” saturation and fly off putting a short circuit across the primary…BOOM!

https://www.ferroxcube.com/FerroxcubeCorporateReception/datasheet/FXC_HB2013.pdf

 

[mtwieg]

No, if anything it means you need more turns on the transformer. Simply adding a gap won't change flux density, given the same applied volt-time product.

 

[treez]

OK thanks, just increased the turns while keeping np/ns the same, and the peak flux does indeed go down.

 

[FvM]

Current mode controller current limiting is pulse-by-pulse with little delay. On the other hand, power ferrite saturation characteristic isn't very steep. Why shouldn't the current mode circuit not be able to switch it off safely?

Gapping the core doesn't increase the saturation flux respectively accepted Vdt integral of a transformer, in so far it won't avoid saturation caused by asymmetrical duty cycles as in your simulation.

 

[treez]

The bottom of the following article states why a gap is good even for a forward type smps transformer....
**broken link removed**

Finally, even discontinuous forward designs will benefit from an air gap, as the residual flux value will be nearer zero, allowing a larger working flux density range.

 

[FvM]

You'll notice that the consideration doesn't apply to the full bridge forward converter originally discussed in this thread.

 

[asdf44]

Current mode controller current limiting is pulse-by-pulse with little delay. On the other hand, power ferrite saturation characteristic isn't very steep. Why shouldn't the current mode circuit not be able to switch it off safely?

Gapping the core doesn't increase the saturation flux respectively accepted Vdt integral of a transformer, in so far it won't avoid saturation caused by asymmetrical duty cycles as in your simulation.

Yes this logic seems rock solid to me. From a heating perspective the primary winding can handle X current. If current mode control keeps primary current below X it won't overheat. It doesn't matter whether that current is from a real load or saturation (and in-fact, if saturated there will be no corresponding secondary current, so overall transformer heating is less).