Push-Pull converter with a gapped transformer ??

[Joseph Hynes]

Here's my short question:

If I use a ferrite transformer with a 0.25mm gap in a push-pull forward converter, what side effects could I expect? Is it safe?


the long story:

Ive taken apart 10 dead Chinese SMPS's which used forward trafos with 2 identical primaries (1reset) and want to reuse the magnetics. Problem is, the cores are assembled so sloppy, that glue and varnish got in-between the core halves, creating a gap. I've sacrificed one trafo, by soaking it in paint stripper overnight, to disassemble it. This revealed the unintentional gap. A 30 turn winding had 600uH before disassembly. After cleaning the ferrite so the halves mate perfectly, 30 Turns measure 4.5mH. The paint stripper damages the wire insulation and the bobbin also, so I don't want to do this to the remaining trafos. They are well made, with shields, margins etc and the turns suit me perfectrly. I'd like tu use them as-they-are.

So a push-pull with a .25mm gap will store some energy during one voltage excursion, hoping to release it to the secondary with opposite polarity in the next excursion, right? But the push-pull already generates a secondary voltage of opposite polarity in the next excursion anyways, so the two processes may be synergistic! Will this stored energy oppose the normal (forward) energy transfer in the next cycle, or help it? To whoever can figure this out, I will concede the rights to patent this new topology: the "flyward" or the "push-back"!

PS. I think maybe many commercial forward trafos may have unintentional glue-gaps in them, since they are assembled by automated machinery, or unhappy slave-workers, and the cores are not closed perfectly. And we live with this...

Joseph

[poobah]

If there is enough primary inductance remaining for your design, gaps and all, you should be fine...

You will be using this in a push-pull rather than forward mode... right?

[Joseph Hynes]

Thank You Poobah,

yes, Push-Pull.
but I thought push-pull is really just like forward (full-forward) as opposed to the one-switch fwd, which would be half-forward.

I see your point, though.

The half-fwd would not send a negative half-cycle to the secondary, and would not rectify the neg. on the secondary so the energy stored in the gap would not have a way to escape..
is that what you mean?

[Joseph Hynes]

Yes, the L(p) is ok for me, 600uH I can use.

There is , expectedly a wide veriation (part-to-part) in L(p) due to the inconsistent gapping provided by the glue layer.

The reason I was concerned, is because I've seen that 0.25mm is a legitimate gap size used in many flyback designs, so there will be for sure flyback action going on here.

BTW, I forgot, the gap is in the outer legs too, so that makes 0.5mm

Joseph

[poobah]

Right Joe,

You have to reset the core in a FORWARD converter. My concern was that if you have all that energy stored in the gap; that you not use any "R" in snubbing/rectifying the reset winding. Efficiency (1/heat)being the reason.

In a true PP it should be fine... I have the the forgotten other ramifications of big gap... hopefully because they were minor! -

What control strategy, voltage mode? current mode? do you plan on?

[poobah]

Yeah,

You will be storing energy, like a flyback, but it wont be coming out the secondary... the secondary has the wrong polarity (the diode points wrong way). That energy comes back at ya in the PRI.

[Joseph Hynes]

Thanks again poobah

Darn!, Iwas sort of hoping that flyback energy could be recovered at the secondary.
What if I full-wave rectify the sec, can't that get the stored energy to come out on the secondary?

(I know , it's asking too much... doing something sub-optimal, and expecting a over-optimal result. )

I use the UC3846 contoller.

And I don't have Aux winding, so I was thinking of a simpler way to supply Vcc to controller:

(OK somebody is going to slap me for this)

use triac lamp dimmer circuit --pot replaced by fixed resistor-- to give 35 V rectified and filtered (the 3846 likes 35V, and the triac only starts reliably above 30V out).

Joseph

[poobah]

Relax Joe,

The energy stored in the gap is returned to the input rails, via the diodes of the OTHER two transistors as soon the CONDUCTING pair shut off.

You may want to parallel the FETS wih some "good" diodes depending on your power level. Look at Inernational Rectifier "FETKY".

Also what is your dc input voltage?

[Eva]

If your design has center tapped secondary windings feeding a buck stage (LC filter), then the transformer will be shorted during off-time so the stored energy in the core and the gap won't discharge until the next cycle begins. This means that no special clamps or diodes are required in the primary side.

I've actually tried to gap push-pull transformers and the result is just increased current ripple as seen by the switches (and the control circuit in case it was current mode). The current at the beginning of the cycles gets smaller, and at the end gets bigger than without gap. I've actually used that feature quite successfully to improve noise inmunity with peak current mode control.

The drawback is stronger stray magnetic fields. However, you may try to clean the cores with sandpaper placed over a perfectly clear surface or even with a file. I've done this sometimes to convert gapped cores into ungapped when I hadn't the proper core at hand, and the resulting cores doesn't yield as high inductance as properly factory machined ones, but they get quite close to specs.

[Joseph Hynes]

The energy stored in the gap is returned to the input rails, via the diodes of the OTHER two transistors as soon the CONDUCTING pair shut off.

What "other" 2 transistors? I've only got 2. This is push-pull, right?

240V input, 75W output.
The mosfets don't really get hot, I can keep my hand on them. I could maybe run this without heat sink, since the controller is limiting current pulse-by-pulse anyway.