Two transistor forward converter with bootstrap high side drive

[Orson Cart]

Hello Treez, most engineers would say that an air gap is not required in a fwd or full bridge or push pull Tx.

reducing the gap to zero maximises the Lpri, which minimizes the Imag, usually a good thing, [except in resonant converters]

If you are running a core up to Bmax (not usually a good idea) then you may need a gap to prevent saturation and the peaky increase in Imag at full power and high temps, this also often happens as Vin goes down and ON time goes up to compensate, the reset time (& voltage) is suddenly too small, and the Tx walks up to Bmax with peaky Imag currents.

Better to design a Tx for a FWD so that none of the above happens, therefore you need control that limits max ON time and a decent amount of Voff to reset the Tx in the OFF time available, also a small amount of slope comp is needed near 50% else you will get 1:2:1 or 1:3:1 oscillations that ramp up the peak current well above design max.

 

[Warpspeed]

A much better way to do this would be a very fast acting current limit, rather than gapping the core.

Its certainly possible to sometimes see transient conditions where the control loop overshoots, and you can have momentary wide duty cycle at the same time as having maximum rated supply voltage.

It may hit the saturation wall, and let the smoke out.

Or perhaps it would be simpler to just go up one size in ferrite core, and be able to sleep more soundly with a clear conscience.

If you are in business for yourself, or working for an employer, they will not be too pleased if you save them ten cents, but cost them thousands in failed products and lost customers.

 

[treez]

#29 post shows a gap is always needed

Hello Treez, most engineers would say that an air gap is not required in a fwd or full bridge or push pull Tx.

...many ferrite cores don't look like they've a gap but they have...its an "integrated gap"......the company will of course tell outiders that ther is no gap....but they are keeping the secret to themselves...I worked for one of the biggest TV manufacturers and we took apart a 40w offline flyback transformer...to see that there was no gap!!.....then we realised it must obviously be an integrated gap......we never had datasheets for these cores...it was top secret.

 

[Warpspeed]

Michael Faraday wrote all the equations for magnetic induction about 200 years ago.

None of this is top secret, its all well known time proven engineering.

 

[mtwieg]

#29 post shows a gap is always needed

No it doesn't.

If you are running a core up to Bmax (not usually a good idea) then you may need a gap to prevent saturation and the peaky increase in Imag at full power and high temps, this also often happens as Vin goes down and ON time goes up to compensate, the reset time (& voltage) is suddenly too small, and the Tx walks up to Bmax with peaky Imag currents.

To clarify, adding a gap does not decrease flux swing ΔB. The only valid reason I've ever seen to gap a core in a forward transformer is to reduce remnant field Br. Thus your peak B will be lower with the same ΔB, which can result in more saturation margin. But a gap won't prevent flux walking.

 

[treez]

No it doesn't.

so you would tolerate a widely varying (with temperature) primary inductance in a forward/bridge converter?

 

[mtwieg]

Yes, so long as the core never saturates. Between a transformer where Imag is %20 of Ip with 1% precision, and another transformer with Imag as 5% of Ip with 3% precision, I will prefer the latter.

 

[treez]

Without a gap, and you will have precision on Imag(peak) of far worse than 3%

 

[mtwieg]

I mean 3% relative to Ip, not to Imag. So 5% +/-3% covers a fourfold change in Imag, which is what your graph shows. But the variation relative to Ip is all that matters.

 

[treez]

so you would accept a fourfold change in Imag?..ie, imag peak could be 100mA, or 400mA?

 

[FvM]

so you would accept a fourfold change in Imag?..ie, imag peak could be 100mA, or 400mA?

If total Iprimary is several A? Why not? Still waiting for a quatitative example.

To add an estimation myself, forward converter Imag will be below 10% of total Ip, more likely below 5%.

 

[Warpspeed]

so you would accept a fourfold change in Imag?..ie, imag peak could be 100mA, or 400mA?

Yes, as its only going to be a tiny fraction of the normal primary operating current in a forward converter, or a normal transformer application.

The only time you might really have to fret about permeability change with temperature might be in a circuit where critical tuning or resonance is involved.

 

[Velkarn]

If you have a gapped transformer in a forward converter and remove the gap (while keeping everything else the same), then the core will not saturate.

maybe that is not true for all the cases, because residual flux value will rise if you remove the gap.

citations from ON semi appnote TND378-D.pdf (improved 2 sw forward):
- For resetting properly the core, a minimal magnetizing current is
needed to reverse the voltage across the winding.
(Enough energy must be stored so to charge the capacitance)
– Rule of thumb: Magnetizing current = 10% primary peak current

i suppose its about parasitic capacitance

 

[FvM]

As previously mentioned, an air gap may ease flux reset. But this point is beyond the discussion started by treez.

 

[Velkarn]

IR appnote AN-978 shows how to use bootstrap driver ir2110 in 2 sw forward. i tryed this shematic - it works even with no load conditions. but i have to change shem to GD transformer because of ICs turn on tresholds.

 

[Orson Cart]

To clarify, adding a gap does not decrease flux swing ΔB. The only valid reason I've ever seen to gap a core in a forward transformer is to reduce remnant field Br. Thus your peak B will be lower with the same ΔB, which can result in more saturation margin. But a gap won't prevent flux walking.

Adding a gap increases the current level at which the core saturates, most sources driving a transformer are not pure voltage sources able to provide infinite current, so at the higher currents there is a volt drop in the components supplying the Tx which causes less volt-seconds to be applied which often keeps the flux in bounds, not the greatest hidden negative feedback action, but it is the reason adding a gap solves some saturation problems in some instances...

 

[treez]

Thanks Velkam, I see your app note..but it needs two extra transistors but the method I bring forward here just needs one....(ie my post #1)

AN978 (page 24 shows IR2110 doing 2 transistor forward fet drive)
http://www.irf.com/technical-info/appnotes/an-978.pdf

 

[treez]

oK Thanks for these discussions, back to the original question, and then this answer..

Its only needed where the upper mosfet duty duty cycle goes to extremes, and there may be insufficient flyback energy from the transformer during the off period to keep the bootstrap working reliably.

...sorry but surely its not just that case, it is all cases, after all if a bootstrap high side drive is used with a two transistor forward converter, then the bootstrap capacitor cannot refresh itself because the freewheeling diode will not let current flow as such to recharge the bootstrap capacitor. Do you agree?. This is why I am saying that if a bootstrap high side drive is wanted, then the circuit I show in #1, using a full bridge driver, is the only way to do it.....(ok there is the way that Velkam pointed out aswell, but that uses more components...see it in #57 link).

 

[mtwieg]

...sorry but surely its not just that case, it is all cases, after all if a bootstrap high side drive is used with a two transistor forward converter, then the bootstrap capacitor cannot refresh itself because the freewheeling diode will not let current flow as such to recharge the bootstrap capacitor. Do you agree?

No, normally the magnetizing current will force both freewheeling diodes to conduct, which brings the source of the high side FET one diode drop below GND. This allows the bootstrap cap to charge. So long as the magnetizing energy is enough to force the clamping, then the boostrap caps will be fine.

This is why I am saying that if a bootstrap high side drive is wanted, then the circuit I show in #1, using a full bridge driver, is the only way to do it.....(ok there is the way that Velkam pointed out aswell, but that uses more components...see it in #57 link).

Your circuit and Velkams are basically identical, except yours makes use of the complementary outputs of the controller. The first transistor implements this in the absence of the complementary output. If you really have complementary outputs then your circuit is fine though, but not necessary if the magnetizing energy of the transformer is always above some level.

 

[treez]

No, normally the magnetizing current will force both freewheeling diodes to conduct, which brings the source of the high side FET one diode drop below GND. This allows the bootstrap cap to charge. So long as the magnetizing energy is enough to force the clamping, then the boostrap caps will be fine.

thanks, but you do agree that the bootstrap cap does not recharge via the lower freewheeling diode, rather, it recharges via the magnetising current going through the primary and then on thru the hi-side freewheeling diode?