Two transistor forward converter with bootstrap high side drive

[treez]

Hello,
The attached is a two transistor forward converter which , (unusually), has a bootstrap high side FET drive. The bootstrap high side drive is enabled by using a full bridge controller ic.
Why don’t all two transistor forward converters use full bridge drivers like this?..After all its so much easier than a gate drive transformer.

 

[Warpspeed]

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.

With a forward converter there is no air gap in the core, and you try to minimise the stored magnetizing energy in the transformer core. The flyback energy under some operating conditions can be a bit weak, and the small pulldown transistor ensures the bootstap still works even if the lower flyback diode does not go into conduction sufficiently hard.

This same idea is sometimes used with a bootstap used with a buck regulator that must work right down to zero load. The duty cycle can fall to almost nothing under no load, but a very narrow pulse from a small pulldown transistor keeps the bootstrap working.

Its not a new idea, and is not often seen, but it is a sneaky way to fix some of the difficulties that bootstrap systems can have working under extreme limits of duty cycle in some topologies.

 

[treez]

Thanks,

With a forward converter there is no air gap in the core

ok, I know what you actually mean, (and that you are being 'efficient' with your wording) since as you know, there is , in reality , always going to be an air gap in any smps transformer core, simply because the AL value of a totally non-gapped core is too widely varying. So I take it that you actually mean that there will be a gap in any forward transformer smps core, but that it will be small.

There is only one specific two-transistor-forward converter control chip in the world (that has an on-chip driver for both fets)...the HyperTFS by power integrations...
https://ac-dc.power.com/products/hiper-family/hipertfs-2/


..as you can see, the chip also incorporates a flyback smps controller and its the flyback that gives the high side supply.....the hyperTFS literature says that its chip is useful to avoid using pulse transformer drive for the top fet, but as we can see, the method I present in the top post does it without pulse transformers, and looks easier than the hyperTFS controller.

 

[Warpspeed]

Thanks,

ok, I know what you actually mean, (and that you are being 'efficient' with your wording) since as you know, there is , in reality , always going to be an air gap in any smps transformer core, simply because the AL value of a totally non-gapped core is too widely varying. .

If you wind your forward converter on a high permeability ferrite toroidal core (a good choice) where is the air gap ?

The transformer design for flyback and forward converters are completely different.
In forward converters the inductance (Al value) is really of little direct importance, it mostly only determines the magnetizing or no load current.

Flyback converters always have too much inductance once you allow minimum turns count to set the desired flux swing.
Hence the air gap then has to be adjusted to get the required inductance, which is a very critical design feature.

 

[treez]

The transformer design for flyback and forward converters are completely different.

..I agree

In forward converters the inductance (Al value) is really of little direct importance, it mostly only determines the magnetizing or no load current.

I agree with bits of this, though again would say that the magnetising inductance value does end up being a significant parameter, but I believe you agree with this anyway.

The point is, and again, I believe we agree on this, is that if you look at the epcos ferrite datasheets, you can see how the AL value of a ferrite core set that is ungapped, is very widely variant over temperature compared to an ungapped core set.

- - - Updated - - -

As such, if you download the Ferrite Magnetic Design Tool (MDT) software form the epcos website, then pick a ferrite core, then pick a zero air gap, then look at effective permeability versus temperature.....then do the same again, but this time with an air gap of any size, even tiny...then you will see that the effective permeability is contant over temperatue change , which is what we want.

without the air gap, the variation in effective permeability with temperature is shocking...even in a forward converter I don't want that, in fact, in a forward converter, I want to know exactly what is my magnetising inductance because it impacts on the sizing of the low side sense resistor......

EPCOS ferrite magnitc design tool...
https://en.tdk.eu/tdk-en/180490/design-support/design-tools/ferrites/ferrite-magnetic-design-tool

......I would like to see the same graph for the toroid which you speak of, I suspect it will have the same problem....of course, many toroids are made with "integrated gaps" so they do have a gap even though it doesn't look like it.

 

[Warpspeed]

Why do you feel that the primary inductance of a forward converter is so very critical ?

What effect would doubling or halving the inductance have on output power, voltage regulation or efficiency ?

 

[treez]

the magnetising inductance creates a ramp of current in the primary, as you know, the primary sense resistor therefore needs to be downsized in ohms in order to accommodate this.
Also, the ramp of the magnetising inductance is effectively slope compensation, so we need to exactly quantify it for use in our feedback loop equations as per the basso book (switch mode power supplies) page 225-235. If you look at the feedback loop equations there you see that sloPE compensation needs quantifying, for use in the respective equation.

If I design a forward converter, I definetely want to know exactly what is the magnetising inductance.

 

[Warpspeed]

I am afraid you have lost me.

Sure, it causes the ramp up on the current waveform, but that is always going to hopefully be made very small.
Unless you are deliberately trying to make the transformer very inefficient.

Slope compensation is only required for topologies that use current mode control and have duty cycles that exceed fifty percent in one direction, to prevent sub harmonic instability.
Even simple low power single ended forward converters rarely if ever use more than fifty percent conduction in the forward direction.

And of course with full bridge, or half bridge diagonal, or push pull forward converters over fifty percent in each direction is impossible anyway, so slope compensation is just not applicable.

 

[treez]

Sure, it causes the ramp up on the current waveform, but that is always going to hopefully be made very small.

...not always the case, as we are always trying to use small cores, and end up needing bigger than wanted gap to avoid saturation on magnetising current peak.

Slope compensation is only required for topologies that use current mode control and have duty cycles that exceed fifty percent in one direction, to prevent sub harmonic instability

-precisely.
But the magnetising current ramp means that we are stuck with a slope ramp whether we like it or not

And of course with full bridge, or half bridge diagonal, or push pull forward converters over fifty percent in each direction is impossible anyway, so slope compensation is just not applicable.

..I am sure you know that it is possible to have a full bridge etc with more than 50% Duty cycle..i am talking about the "real" duty cycle, ie that seen in the output inductor....the upslope = the on time, the downslope is the off time

 

[Warpspeed]

If your core is saturating at the magnetising peak, you are not using enough turns.
What sort of flux densities are you designing to ?

Fitting a gap and reducing the inductance is exactly the wrong way to go with that particular problem.

I have NEVER seen any full bridge inverter have more than 50% each way.
Its mathematically impossible to split 100% into two halves that are each greater than 50%.
Please explain exactly how you can do that.

 

[treez]

I have NEVER seen any full bridge inverter have more than 50% each way.
Its mathematically impossible to split 100% into two halves that are each greater than 50%.
Please explain exactly how you can do that.

we are talking at cross purposes...we are both referring to entirely different things.

I am speaking of the "real" duty cycle of the converter, not the individual "duty cycle" of any of the individual fets

- - - Updated - - -

What sort of flux densities are you designing to ?

300mT peak

- - - Updated - - -

If your core is saturating at the magnetising peak, you are not using enough turns.

as you know, with a given core, the more turns, the lower is the saturation current

 

[Warpspeed]

So what is the difference between duty cycle and "real" duty cycle ?
Still in the dark about what you are trying to tell me.

300mT is pretty adventurous, especially if it gets a bit hot.
I am not that brave.
That is very likely your problem.

 

[treez]

the real duty cycle of a full bridge, is that seen in the output inductor current.

 

[Warpspeed]

That is certainly a novel way of looking at it.

One other thought about your high flux density and core saturation.

It might be a little bit less trouble to raise the switching frequency just enough to lift you out of trouble.

 

[D.A.(Tony)Stewart]

Energy is usually stored in precise air gaps for precision regulators.

Otherwise gapless ferrite stored the energy between the magnetic particles which give a smaller effective gap that has much wider tolerances for saturation and thermal effects.

Eddy current and skin effect losses can be improved with Litz wire raising the SRF, f operating and reducing Imag and slight expense on fill factor.

Many ferrite types to choose and many major Japanese quality sources.

 

[Warpspeed]

Forward converters do not use energy storage in the core, they are just normal voltage transformers and utilize the turns ratio to transform the voltage up or down.
They are never fitted with an air gap as this reduces the inductance which is not a desirable feature.

The associated dc choke (placed after the rectifier), DOES work on the energy storage principle and will have an air gap.

Flyback, boost, and buck regulators all use inductive energy storage, and must either have air gaps or use suitable low permeability core material.

Its a totally different design approach.

 

[treez]

On the idealised level I agree, but in reality, all forward converters will end up having a small gap, though I agree that this is not for the purpose of wanted energy storage.
If you download the MDT by epcos (magnetic design tool), you will see why forwards and bridges end up needing a gap...it is because without a gap, the effective permeability of the core varies too much with temperature.

As you know, I believe we agree that in fact, the actual level of the magnetising current in a forward converter does in fact need to be known and quantified, for the multiple purposes as described above.

 

[Warpspeed]

The core permeability only needs to be known in respect to assessing its suitability for the application.
Likewise the grade of ferrite for the intended switching frequency.

But why would you want to deliberately increase the magnetizing current by introducing an air gap, just to make them all the same, when doing so is detrimental to the better ones ?

To be honest, I did once in the past have to resort to fitting a very small gap in a forward converter (someone else designed) but it had nothing to do with magnetizing current.

The purpose was a panic band aid cure to solve the problem of flux doubling at startup, as this particular POS inverter never had a current limit fitted to it.
Inverters were returning under warranty in large numbers, and a proper complete redesign was just not possible. So a fine slip of paper got us out of that particular disaster, but it was by any standards a Mickey Mouse cure.

 

[treez]

the core "effective permeability" figure (please not I said "effective") is related to the AL value....if you see the graph which I have referred you will know what I mean, please see the graph first. Please see post #5

 

[mtwieg]

To answer the original question, a gate drive transformer is suitable for a two switch forward converter because both FETs see the same gate drive, and therefore there's no need for two independent gate driver, just a GDT with two secondaries. But when using a GDT one must be sure that the duty cycle always stays within strict limits so that the gate drive voltage does not sag during extremes.

And no, forward converters do not always have gapped transformers. Having a large variance in your core reluctance isn't a problem so long as it always stays above a certain value, and adding the gap will always decrease that value. The one exception to this would be if you are relying on magnetizing current to provide slope compensation, in which case you would want a more tightly controlled reluctance. Most people don't do this though, I think. Adding an otherwise unneeded gap will hurt efficiency.