Core material for Resonant inductor

[skn96]

Hi All
I have a problem selecting the right core material for 25uH resonant inductor for my power supply project. This is a Phase-Shifted Full-Bridge dc-dc converter 2.1 kW , 400 V input (from PFC) ,170V output and 200kHz switching frequency seen by output inductor i.e. 100Khz seen by primary . Since the resonant inductor sees peak current up to 11 Amps and the current is AC waveform the common sense choice would be to use ferrite material such as R material (magnetic inc.) or 3C95 (from Ferroxcube). However, I got confused when I saw some reference recommend using MPP material or Kool mu or powder iron -2 material. These material have very high core loss characteristic. Some other reference recommended very low loss high Q low permeability powder iron from Micrometal inc. used in RF application. The MPP or RF cores have better core loss characteristic but still not comparable to ferrite. I understand that ferrite saturate faster (saturate for B>.3T ) but I can simply limit the B to 0.2T by using PQ2625 and 3C94 material and gaping the core to provide AL 100nh/(T)^2 .
II want to know if I’m missing something here or not? Is there anything that prevent me form using the ferrite in this application?

 

[Warpspeed]

A couple of basic truths that may help.

Flux swing in the core is entirely a voltage phenomenon, current has absolutely no effect and can be ignored for pure ac applications (such as transformers).
Only where there is a residual dc need you be concerned about current.

Gapping the core has no effect on the required ac flux swing either.

So all you need to derive core loss, is the calculated flux swing due to turns and voltage, and the frequency.

If you also need to precisely control the inductance, that can best be done with an air gap, but that will have no effect on flux swing or core loss.

Its a lot easier to pick a ferrite core for loss and overall dimensions, then just gap it.
With a toroid, what you buy is what you are forced to use, with no easy way to change the inductance if things don't quite work out as anticipated.

 

[RichardHead]

I've used both a Micrometals T130 Yellow/grey (RF catalogue not power catalogue) and a gapped ferrite in the past for resonant chokes in PSFB converters. I prefer the gapped ferrite like Warpspeed suggests. It gives you more freedom to fine tune the inductance, if required.
Just remember that at 200kHz you will be loss limited and not saturation limited and to get the loss to a reasonable value you will probably have to limit your flux swing to way below 200mT peak. Check the graph "Power loss vs freq" for different flux densities and keep below about 1kW/m^3 and it shouldn't run too hot. Obviously use a low loss material. There's an exponential (Steinmetz) relationship between power loss and flux density at a given freq. Don't worry too much about using a custom gapped ferrite that has a gap way bigger than is available as a pre-ground gap. This indicates that you are using to core effectively.
NB. If you decide to use a Micrometals core be wary of cheap Chinese counterfeit versions.

 

[Warpspeed]

+1 on the Chinese toroidal cores.

They come in various colours, and I am sure there is somewhere a Chinese guy with a paint brush and pots of different paint that can produce the whole range of different materials, any colour you want.

 

[RichardHead]

Sorry, delta B below 200mT. Not peak.
Yeh, the Chinese invasion hasn't left any manufacturer untouched.

 

[skn96]

I've used both a Micrometals T130 Yellow/grey (RF catalogue not power catalogue) and a gapped ferrite in the past for resonant chokes in PSFB converters. I prefer the gapped ferrite like Warpspeed suggests. It gives you more freedom to fine tune the inductance, if required.
Just remember that at 200kHz you will be loss limited and not saturation limited and to get the loss to a reasonable value you will probably have to limit your flux swing to way below 200mT peak. Check the graph "Power loss vs freq" for different flux densities and keep below about 1kW/m^3 and it shouldn't run too hot. Obviously use a low loss material. There's an exponential (Steinmetz) relationship between power loss and flux density at a given freq. Don't worry too much about using a custom gapped ferrite that has a gap way bigger than is available as a pre-ground gap. This indicates that you are using to core effectively.
NB. If you decide to use a Micrometals core be wary of cheap Chinese counterfeit versions.

Great advice. Thanks. The switching frequency is 100Khz. The output inductor sees 200Khz but the resonant inductor because it is on the primary and before output bridge , sees 100kHz. Based on the loss and saturation curves at 100kHz the saturation and loss both are limiting factors. I'm trying to keep delta B below 300mT. Based on my calculation the core won't get high with this much flux.

 

[Easy peasy]

for your AC choke t 100kHz, you will need to keep the peak B below 80mT, you need to know the peak current to work this out, you will need litz wire to avoid wire losses, powdered iron type materials will all have higher losses than a good ferrite, make sure you keep the litz away from the gap

 

[Velkarn]

When we have been designing power supply based on FSFB topology, we made several versions of resonant inductor. The best (less looses) was on gapped ferrite core (russian ferrite worse than N87 or 3C94) with RM configuration by several twisted wires 0,1-0,2mm (we had no litz wire unfortunally). And I agree with Easy peasy about 80-100mT peak B.

 

[Easy peasy]

Flux swing in the core is entirely a voltage phenomenon, current has absolutely no effect and can be ignored for pure ac applications

unfortunately the above is not quite correct B = uH = uo.ue Ni / Le i.e. B is directly proportional to I, in a choke it is the current, in a Tx it is the magnetising current (not the load current).

 

[Warpspeed]

Yes its the magnetising current, not the load current.
And magnetising current is proportional to the applied ac voltage.

Its why we Faraday's law (commonly used to calculate flux swing) does not need to include current into the formula.

 

[Easy peasy]

that is a bit like saying the power in a resistor has nothing to do with current only the applied voltage, you are seeing one side of the coin only...

 

[Warpspeed]

Yes indeed, Power = E squared on R
No need to work out, or know the current through the resistor.

Not saying the current is not there, or its zero, or totally unimportant.
Just that if you know the resistance and the voltage, and you wish to calculate power you can do it directly.

Faraday understood all that when he wrote his famous formula linking voltage, frequency, number of turns, cross sectional area, to flux swing.
The alternating magnetising current is definitely there, its just not necessary to know what it is.

 

[Easy peasy]

if you know the current - you don't need to know the voltage either, this is the other side of the coin, if you are totally ignorant of the current then you cannot factor in things like the source impedance of the driving circuit... and associated volt drop...

 

[Warpspeed]

if you know the current - you don't need to know the voltage either, this is the other side of the coin, if you are totally ignorant of the current then you cannot factor in things like the source impedance of the driving circuit... and associated volt drop...

The discussion here is mainly about choosing a core material, and relevant flux swing in an ac application.
Source impedance has nothing to do with it.
And neither does ac current, or voltage drop, effect the choice of core material.

The only thing that matters as far as the core is concerned, is the ac voltage across the inductor, the number of turns, and the frequency.

 

[Easy peasy]

if the inductance is low, and the current drawn by a choke (or Tx) is large, then there will be a volt drop in the source, thus lowering the volts applied to the choke/Tx thus reducing Imag, it is incumbent on an engineer to know what currents are flowing, it affects the choice of wire (litz) and the total losses in the choke/Tx, very peaky high freq currents require a core with higher intrinsic resistance ad lower hysteresis losses.

 

[Warpspeed]

Yes, its the voltage across the inductor that matters when designing for core loss, which is what I said in the first place.

Total losses and winding technique are a very different things to selecting a suitable core material for core loss, which is what we are discussing here.

There are other important factors, such as a sufficiently high self resonant frequency, but that also, has nothing to do with core loss.

 

[Easy peasy]

Actually the peak B determines the hysteretic core loss, you need the peak I to determine this peak B, as for a distorted, e,g, 3rd harmonic centre peaking voltage the RMS of the volts will not give you the peak B (that equation only works for a pure sine), the average volt seconds are needed (which assumes a zero ohm source - which is rare) so if you know the peak current in the circuit - this can safely be used to determine peak B and hence core losses at that freq.

 

[Warpspeed]

Flux swing is calculated from average voltage, not RMS.

Put it this way...
If you wind an ac inductor with X turns and no air gap, and apply a certain voltage to it.
The current increases until the back EMF equals the applied voltage, and that determines the ac current through the winding.

If you now gap the core, (or remove the core altogether) the inductance is reduced, and current rises, but the flux swing required to generate the original back EMF remains the same, as the number of turns have not changed.

Its only turns, voltage, and frequency that matter in determining flux swing, and hence core loss.

 

[Easy peasy]

actually if you remove the core, the current will now be limited by the R of the winding (and the R of the source), so for the same applied volts the flux will be lower...

 

[Warpspeed]

That's probably true, it may even go *BANG* or burst into spectacular combustion.

But the poor thing will definitely try to pass enough current sufficient to generate enough flux to generate enough back EMF to cancel the applied voltage.

Its interesting though, if the turns stay the same, and the required back EMF stays the same, the flux must be the same too. It just takes a massive amount of current to do it when you remove the core.