High power Boost PFC inductor?

[treez]

Hello,
We need to produce a 180uH Boost PFC inductor for a 3.7kW PFC. (Fsw = 45kHz).
Minimum VAC input is 216VAC.
Inductor RMS Current = 25 Amps max.
Inductor peak current = 39 Amps
Vout of PFC is 400V.

None of the Epcos range of ferrite cores with bobbins can manage this. We have been forced to use stacked Toroids by Magnetics Incorporated. We find that we will need to glue together four toroids of part number 77615A7. We will then need to wind 26 turns of litz wire round this four_torroid stack. The litz wire will comprise 9 strands of 22SWG enamelled copper wire (0.71mm diameter). The total winding resistance is 41 milliohms, leading to a winding loss of 25 Watts at minimun VAC(in), max P(out).


Toroid datasheet (part number = 77615A7)
https://www.mag-inc.com/home/Advanced-Search-Results?pn=77615


This torroid is of “kool mu” material. Its permeability changes with the instantaneous current flowing through it. Thus over one mains half cycle, the inductance of the boost inductor will change very significantly. I tried to arrange it so that the inductance of 180uH is what we get at the inductor peak current value. The torroid datasheet has a graph of AL value vs Ampere.Turns. I used the AL value that corresponds to an Ampere turns of [39 Amps * 26 Turns].

Do you know of any more suitable ferrite core for this, that doesn’t change its inductance with the instantaneous current level.?

Also there is no bobbin or mounting assembly for this DIY torroid stack. I think we will have to just “glue” it to the PCB with a load of high temperature silicone. Do you agree?

 

[Warpspeed]

Not saying its the ideal solution for this project, but...

Have you tried looking at an air cored inductor ?
Its going to be big, and will probably radiate like mad, but it can never saturate, and will have zero core losses.
A "Brookes" type of coil provides max inductance for minimum copper.

Its worth running some numbers on a trial air cored design, you might be surprised.

 

[treez]

thanks, but sorry it will be too big, we are looking for that rare toroid that doesn't vary its inductance with current.

 

[Warpspeed]

we are looking for that rare toroid that doesn't vary its inductance with current.

Rare indeed.

 

[FvM]

My first general suggestion is, ask a specialized manufacture for customized power magnetics rather than designing it yourself. It's obvious that you have serious difficulties to identify an appropriate solution.

If my hand calculation is right, you have about 15 % reduction of initial permeability at peak current, sounds moderate for a PFC inductor. Don't know about expected core losses?

Stacking four toroids gives a winding geometry far from optimal shape and respective extra copper losses. I also doubt that a toroid core is first choice for large inductors when applying objective criteria.

 

[Dan Mills]

I wouldn't be doing this with a toroid, for all that the Magnetics Inc stuff is awesome (It wants a big gap, and that makes for copper losses if you are not careful, the distributed gap on a toroid is fixed and you really want to be able to define this).

Talk to a transformer company, tell them the application (PFC), current (average and peak) operating frequency, and required inductance and target form factor and see what they come up with (They can get cores that you would have a lot of difficulty sourcing).

There are also ease of production issues when you go to production that a transformer winder will understand but you may not, it is very possible to do something for a one off that will cause a BOM cost nightmare when you order a few thousand of the things.

Regards, Dan.

 

[mtwieg]

Coincidentally, I once built a 20kW PFC and tried using stacked powdered iron toroids from Amidon (their material 26, which is apparently quite different from the material you posted). The power dissipation was far higher than expected, and looking at the loss curves and comparing to a good ferrite explained why. Easily a factor of five or more worse for the powdered iron. I think it's the same with your Kool Mu material. At B=0.1T, f=100kHz, the kool mu Pv=800kW/m^3. For N96 ferrite its Pv=55kW/m^3. Again, not sure it they both define flux swing the same way, but even using the 200mT curve the N96 still is only Pv=310kW/m^3, still substantially lower than the Kool Mu.

A gapped ferrite is going to be your best bet. You're not going to find appropriate bobbins on digikey for this power level, so you need to talk directly with your manufacturers.

If you absolutely want to use a toroid, then go for a tapewound ferrite toroid instead of powdered iron.

If you use average current mode control for current shaping, then some droop in inductance is not a problem.

 

[Easy peasy]

A large PM core with gap should get you pretty close N27 or N87 for 300mT Bpk

 

[treez]

thanks. ill look into these again, though I have now done the calculation for two stacked "Kool mu" cores, and this means 39 turns gives 183uH at 39 Amps.
There is worst case 9 Amps of delta_I , which gives rise to 58.5mT of delta_B, I believe....though I believe it will be even less delta_B than this because permeability increases as current reduces, thus the swing in delta_B will not be so far, so in fact, worst case delta_B of less than 58.5mT.

 

[Velkarn]

i wouldn`t use cool mu core for pfc boost inductor. large dB swing leads to high looses in iron core.
here is for example pfc boost inductor 400uh (PREMO) with copper foil winding


4xE65x32x27 cores 5-10mm gap

 

[treez]

A large PM core with gap should get you pretty close N27 or N87 for 300mT Bpk

thanks, I tried the PM87, and I got up to 6mm gap, and it still was above 300mT at 39 Amps peak current when it was 183uH....so I stopped, and started looking elsewhere.

Thanks Velkarn, I like the foil windings, but for the first few prototypes, I am limited to Enamelled copper wire.

I wouldn't be doing this with a toroid

I think it's the same with your Kool Mu material

i wouldn`t use cool mu core for pfc boost inductor

The PFC inductor current is 39 Amps peak, with at maximum, 9 amps of current ripple pk-to-pk, at 45khz. I calculated that this leads to a delta B of just 39mT.
(considering it’s two stacked pieces of the 77615A7 kool mu core with 39 turns round them).
The graph on this page (below) shows that the delta_B of 39mT gives rise to a power density of 17mW/cm^3 = 17000W/m^3

**broken link removed**

The volume of the 77615 core is 51800e-9 m^3, therefore the power loss in one of the stacked cores is 881mW. This doesn’t sound too much core loss.

 

[schmitt trigger]

First of all, Kool-mu is a "hybrid" magnetic material, with far lower losses than powdered iron, and gentler saturation characteristics than ferrite (but higher losses).
And...I have to agree that stacked toroids may not be the optimal shape for this power level.

Having said this, I would have to ask you something: how far advanced is your design?
I'm asking this, because at that power level, an interleaved boost topology is recommended. Not only will you split your magnetic components into more manageable sizes, but the losses in the switching components will be easier to tame.

 

[treez]

thanks, we are interleaving two 3.7kw boost PFC's.

 

[treez]

Hello,
The following is a guide on how to calculate the number of turns needed to give a certain inductance with stacked kool mu toroid cores from mag-inc.com.
This is for a Boost PFC inductor. (Fsw = 45KHz)
I would like to get this published on the mag-inc website.

It also shows how to calculate the core loss and winding loss.

The calculation shows that in spite of a maximum of 9 Amps of pk-to-pk inductor ripple current at 45khz, the core loss is just 1.3W per core. –And this is a very overstated value since that peak-to-peak current only happens at the mains peak time.

This shows the huge benefit of these cores, though unfortunately, they are difficult to mount on the PCB. The low permeability of kool mu means that stacked cores are needed, to get the reluctance down.

 

[FvM]

If you prefer Kool Mµ material over ferrite for some reason, there's also an option to use larger toroids or E cores. So still not obvious why stacked toroids should be used.

 

[treez]

I cant find a ferrite E core big enough for the 39 Amps peak current, and 180uH.

- - - Updated - - -

To be honest, since the torroids are so flat, its easier to stack them and make use of room vertically.

 

[Warpspeed]

If you prefer Kool Mµ material over ferrite for some reason, there's also an option to use larger toroids or E cores. So still not obvious why stacked toroids should be used.

Kool Mu cores do offer a huge advantage in multiples higher flux density than ferrite, and with low loss. They are not made in large sizes unfortunately.

Stacking multiple cores is totally impractical for production.
Machine winding is impossible, every part will need to be individually hand wound by an expert.
Your production manager is just going to laugh when you present him this.

 

[treez]

the stacked core is like on page 17 of the following...
**broken link removed**

Also, we took apart a 3kw charger, and it had stacked dual torroid core for its two interleaved boost pfc stages. This was by a well known, reputable manufacturer.

There is no off-the-shelf core with bobbin for this power level (3.7kw PFC stage).....we would have to have 4 lower power boost pfc's in parallel if we were to use a easily mountable core.

If two torroids are glued together, then why cant they be wound by a torroid winding machine like a single core torroid?...after all, when two torroids are glued together, its just one part.
Also, Toroids offer compact size and self-shielding.

- - - Updated - - -

I see your point though, maybe we should use 4 smaller pfc stages instead of two that have non-readily mountable inductors.

 

[Warpspeed]

If two torroids are glued together, then why cant they be wound by a torroid winding machine like a single core torroid?...after all, when two torroids are glued together, its just one part.

Fine, if you can find a machine that is built to hold (and mechanically clear) THREE stacked small toroids.

It becomes far easier in the larger sizes, but Kool Mu don't come in large sizes.

The proportions of most of the smaller toroids are such that stacking them produces a shape like a pipe. And it becomes rather difficult for the round shuttle hoop on the winding machine to pass through the centre of this "pipe", and still leave enough clearance remaining at the ends as the copper winding builds up.

It may be possible (say) if you only need to wind these core with a single layer. But if you can do that, it may be better to use fewer toroids and more turns which would suit the proportions of the machine better.

- - - Updated - - -

How about winding three of these cores individually for 60uH and connecting then electrically in series ?
Its inefficient with regard to total wire length (and copper loss) but it would be an entirely practical solution that would take up hardly any extra board space.

 

[Mr.Cool]

check out Micrometals, part # T300-52