Two transistor forward converter with bootstrap high side drive

[treez]

thanks but did you see the graph referred to in #5...I think that changes things.
A ferrite core will need a gap.
BecAuse else AL value is too variant.
Magnetising current needs to be well known, for sizing of low side sense resistor (I am talking 'forward 'here)

the graph explains all, unless people see the graph then all is nothing. FvM pointed this MDT thing out to me in the first place.

 

[D.A.(Tony)Stewart]

Forward and flyback can use the same core materials but agreed, flyback is power limited in the gap, while forward conv's generally don't need gaps until very high power levels >>1kW to raise % transfer impedance and lower short circuit currents.

Although winding complexity is high, smaller chokes are needed and high coupling of turns ratio permits PC PSU to track only primary output with excellent secondary tracking with good cross & load regulation, using more than 4 outputs.

TDK , the best ferrite company in the world, now owns Epcos.

 

[treez]

thanks, of course I still say the graph from #5 ref Is the key to why all forward converter transformer ferrites need gaps....albeit very very small gaps.

you have to download it but when you see the graph you will agree with the statement that all forward ferrites need gaps.

 

[FvM]

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

I don't see the claim substantiatiated in your posts. If a forward converter is operated in current mode, it's the secondary rather than the transformer current that dominates the primary current. Energy storage is performed in the secondary inductor, not the transformer.

all forward ferrites need gaps.

The statement is wrong, respectively.

 

[treez]

thanks, but did you see the graph which explains what I mean...ie in #5

from the ideal point, yes forwards don't need gaps.

you have to know the magnetising current peak, so you can downsize the low side sense resistor accordingly.
I know in a forward, the gap is not required for energy storage like in a flyback.....but you tend to need to put gaps in forward cores partly because of the graph in #5, and partly because otherwise you end up with too big core, ie you saturate on the magnetising current unless you make the core too big.

 

[Warpspeed]

If its saturating from excessive ac magnetizing current, you need to add more turns not an airgap.

If you revise Faraday's law of induction for calculating ac flux density, there is nothing there about air gaps or permeability.
The only actual core parameter that matters for calculating ac flux density is cross sectional area.
All the other factors such as voltage, frequency, and number of turns are factors external to the core itself.

 

[treez]

I agree with you but again we are arguing at cross purposes, I still stand by everything ive said.......I have done my leg in so cant sit at the computer for too long, when I can, I will paste the graph from #5, that will make everything clear.

I am arguing that today is Tuesday, and everyone else is arguing that tomorrow is Wednesday.......were both right.

 

[FvM]

There's no graph in post #5 as far as I see, just a download link for the well-known ferrite tool. If you want to show a graph, you'll insert the image to your post...

You are still talking about magnetizing current being relevant for the primary current measurement which is simply wrong in a typical forward converter. The dominant primary current is the transformed load current.

There may be other reasons to use an air gap in a forward converter, depending on the used method to "reset" the flux. But that's beyond the reasons mentioned in your posts.

 

[treez]

In a forward, if you don't gap it, you end up saturating on magnetising current unless you use a big core.....we want small cores..

I will try and fetch the graph.

Ok...so if you download the "MDT" from epcos as per #5 post..
-go to the "core calculations tab"
-then go to the "ue vs T" tab.

then put in "0" for the gap (called "s") and press "calculate"
then put in 0.1mm for the gap and calculate again.

...then you will see why you need a gap in a forward converter...its because it makes your "ue" value constant over temperature...which also means your AL value is constant over temperature.

ive attached a printout of the chart for your perusal..this is for the ETD39 ferrite of epcos TDK

 

[FvM]

The graph shows that the inductance of a transformer will be temperature dependent. It's also current dependent (non-linear). But you didn't manage to clarify why this should be a problem.

To avoid core saturation, two points have to observed:
- the transformer must be correctly designed for the maximal Vt area (input voltage multiply maximum on time)
- the circuit must allow transformer flux reset

 

[treez]

The graph shows that the inductance of a transformer will be temperature dependent

..As you know, it certainly will be if its not got a gap.

We must know the pretty exact inductance of the primary in the forward...because then we can evaluate the magnetising current waveform, and size our low side current sense resistor accordingly. As you would agree, we simply cant use an ungapped core as its inductance would vary far too much with temperature, as the graph shows.......I am not saying a big gap is needed....even a tiny tiny gap levels out the graph of Ue vs temperature. -but we do need that gap

 

[mtwieg]

..As you know, it certainly will be if its not got a gap.

We must know the pretty exact inductance of the primary in the forward...because then we can evaluate the magnetising current waveform, and size our low side current sense resistor accordingly.

You still haven't explained why this is true. I mentioned that some engineers rely on magnetizing current for slope compensation, but most do not, in which case the magnetizing current does not need to be precisely controlled.

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. The magnetizing current will decrease. Maybe it will decrease from 10% of Ip to 5%, or to 1%, or somewhere in between. But the variation isn't really relevant, because it's still a decrease, and that's still good.

 

[treez]

the magnetizing current does not need to be precisely controlled.

but I am sure you will change your mind when you think of the sizing of the sense resistor. The magnetizing current flows in it (of course you may be using a CST, but same difference)

By the way did you see the graph in #29 post?

 

[mtwieg]

but I am sure you will change your mind when you think of the sizing of the sense resistor. The magnetizing current flows in it (of course you may be using a CST, but same difference)

I don't see why this should matter. Current mode control will still work fine, even with a small unknown ramp current added in.

 

[treez]

you may go above the sense reference voltage though.
On max load/min vin, the sum of the current sense voltage from the referred sec currnt, and that due to the magnetising current must be just under the current sense reference voltage.

 

[Warpspeed]

The current sense resistor is there to sense full load current, of which the magnetizing current should hopefully be only a very small proportion.

In other words magnetizing current becomes pretty irrelevant to sizing a current sense resistor in a forward converter.

 

[treez]

if using a very big core, then I agree with you, if using a smaller core, then magnetising current ends up being significant and you've gotta know it.

 

[Warpspeed]

If you wish to deliberately design it with zero margin to run right up into core saturation, the risk is entirely yours.

But one thing is for sure, fitting an air gap is only going to decrease the inductance and increase the magnetizing current even more.

 

[FvM]

if using a very big core, then I agree with you, if using a smaller core, then magnetising current ends up being significant and you've gotta know it.

How "significant" actually? Would you mind to give a quantitative example?

Furthermore, even if the magnetizing current has to be considered in the calculation, it will be less significant if you omit the air gap, isn't it?

 

[treez]

thanks, but the graph in #29 shows that we need an air gap, even if only tiny one.