flybacks & transformer leakage

quite so poo bear, but is this guy making a consumer product? if so what the hell is he doeing asking for advice here?

At that power level almost certainly.

It all comes out in the wash. It's a nice simple circuit with few components.

I'm currently working on a half bridge forward converter for ~ 3kW.

So many more bits !

Well, actually you normally go flyback, single transistor forward, half bridge forward, full bridge forward as the power goes up.

Flyback's ok up to a couple of hundred watts.

Graham

Another good reason to choose flyback is the easy implementation of multiple secondaries. All other things being equal, a good way to further reduce leakage inductance is to use a core with a longer window and/or some way to reduce the need for clearance/creepage margins. The overall goal is to reduce the amount of volume of insulation between windings. Triple insulated wire, provided that it is the very thin type, is one way to achieve both of the above. Leakage inductance that is 1% of magnetizing inductance is achievable using the above techniques. Paul Mathews

Not unsurprising.

Have you looked at a 200V TVS ? E.g. P6KE200. Gives an assured clamping voltage.

Too complex for 25W !

Graham

The leakage inductance of a flyback transformer is typically about the same as that of any other topology's similarly sized transformer. It is its own deliberately lowered magnetizing inductance (to which its leakage inductance is proportionally larger) that makes leakage inductance seem greater than it really is.

A flyback transformer is best thought of as an ideal inductor in parallel with an ideal transformer (winding resistances and leakage inductances are added outside this ideal core). The ideal transformer part behaves like any ideal transformer, where (for two windings) the windings only conduct simultaneous currents of equal and opposite- ampere turns.

The ideal inductor part may conduct some ac current, or, if large, may only conduct substantial dc. As a thought experiment, imagine an ideal flyback transformer consisting of infinite inductor (initialized to one amp) in parallel with an ideal 1:1 transformer, all driven at 50 percent duty cycle. The inductor can pass no ac and the transformer can pass no net dc. Think about it.

Regards -- analog

a) thay aren't actually cheap when you're talking about 1000's.

b) you can optimise the design for your application - show me any consumer product that uses an Astec off-the-shelf power supply for example.

Graham

At 5% you should start asking yourself whats wrong.

Flybackers have lower parts count in total and less of the costly parts.

The flyback design primary side can be easily protected from shorts at load because the short circuit protection doesn't have to act as quickly.

This doesn't have to be true. The snubbing and resetting circuits can return a goodly portion of the leakage inductance energy back onto the input capacitors.

You don't have to use a gate drive transformer at all if the smarts are on the primary side and the feedback is either brought over by optoisolators or you just sense the extra winding's voltage and lose some regulation.

I'm building a 25W off line flyback SMPS, and wondering what are typical leakage inductances for the gapped ferrite E core transformer. I'm winding the secondary sandwiched between a split primary and getting about 2.0% leakage. Without interleaved windings then leakage inductance gets too embarrassing to mention.

Another question: Is flyback my best choice.

Can I ask why flyback is common for < 50W power levels ? Sure, flyback converters eliminate the largish output inductors and use a single switch. However its simplicity seems superficial because the designer is challenged with high peak currents, EMI/ripple filtering, and magnetics design.

The forward converter reduces these high peak currents but requires output inductors. Energy stored in the transformer leakage inductance still gets wasted.

Ok, so leakage inductance energy can be utilized with the two switch forward converter, and additionally halving the peak voltages across the switches. But this topology adds one gate drive transformer to the system.

With the need for a gate drive transformer, then seems almost no more effort to go for half bridge topology and gain the benefits of even better transformer core utilization. Comparing the half bridge to the flyback converter in the 25W .. 100W range;

Flyback disadvantages:

• high peak currents and ripple

• large ferrite core with air gap

• high switch peak voltages

Full bridge disadvantages:

• needs a gate drive transformer

• needs output inductor

• an extra FET switch

• slightly more complex control circuitry

When does one topology over the other get more economical/easier to build ?

The half bridge seems simpler. Adam

Thanks for the reassurance. When I was doing calculations for the RDC snubber in order to clamp peak MOSFET Vds to 700V I get 1.2W dissipation in the resistor ! Is this typical for 25W offline supply ?

How about the "two switch" discontinuous flyback ?

With the triple insulated you can make things a lot more compact and avoid stuff like having run tape up the inside of the bobbin to keep the creepage high enough.

Best regards, Spehro Pefhany

Hello Cbarn,

For series production you can design them a lot cheaper than the modules cost. I do that all the time.

Even for a onesie I occasionally whip one up, just because I don't have to wait for the Digikey package to arrive. It also cuts down on the amount of pet hair that flies around after our dogs cussed out the delivery driver from behind the door ;-)

Regards, Joerg

Well yes , series production can be cheaper but this guy is making 1. Also bear in mind that Digikey and their like have a 100% + mark up. I'd admire anyone who can whip 1 up in an afternoon, even more so if they can get it to work as well. Another aspect for series production is reliablity, 1 error that comes to light after you have 200+ units in the field can cost you dearly. Then there are the regulations to comply with, if you make your own you have to cost all that in as well.

I had to start somewhere too !

Graham

I guess I'm just stuck with what I've got, at %2. If I could abandon the use of safety margins and internal foil shielding it would go a long way to lowering leakage inductance. The smaller the transformer the worse things become because clearance margins must remain fixed at 5mm.

This is part of a one off hobby project, and also partly doing it to learn something. It would be nice to be able to whip one up in an afternoon, assuming I had PCB and all the generic at hand. Can you actually buy part assembled DIY offline flyback converter kits ?

Adam

Good advice although processing very thin wire may be problematic.

I'm using TEW on my design btw. Solves all those margin problems.

Graham

Use triple insulated wire..

You may be able to get a small sample enough for your needs.

Too exotic I reckon for such a thing.

Graham

It is also possible to reduce leakage using a pair of "flux cancelling" windings, each with Nc turns. Wind Nc turns first on bobbin. Then wind transformer. Then wind Nc turns (so one winding is "inside" the other "outside"). Connect the windings in parallel. because the inner and outer windings have different flux linkages due to leakage, current will flow thereby creating a flux which counters the imbalance.

this trick gets used in HV transformers, where the secondary is a triangular profile to minimise capacitance (and keep HV the hell away) and a split primary is generally not used.

I normally do what Paul suggested, and consider 1% leakage to be acceptable (except in a planar transformer, where its crap). reducing leakage helps quite a bit with snubber/clamp losses.

Cheers Terry

Hello Cbarn,

True. But after a couple decades or so you build up a "repertoire" and then most such designs become deja-vu. It's mostly a matter of having the right toroid cores and stuff in the lab supplies. I always make sure there is a pound or two of #77 in stock.

True as well. Then again, claims by module mfgs that theirs are "compliant with xyz" sometimes turn out to be, well, a bit of a stretch. At the end of the day the whole system must pass anyways. In the medical world even the isolation requirements have to be dealt with separately because it is often next to impossible to meet EMI plus low leakage without an iso transformer.

Regards, Joerg

Since this is a hobby project, and the transformer is not getting wound professionally then I'm predicting it will be difficult sourcing small quantities of triple insulated wire. I've got nearly 10 kg of various sizes of the standard enamel winding wire at my disposal.

Adam