diyAudio (
-   Power Supplies (
-   -   Ferrite core for Boost PFC inductor. (

treez 23rd September 2012 08:21 AM

Ferrite core for Boost PFC inductor.

i need to make a Boost PFC inductor. (Its for a CCM PFC stage which will be follwed by a 330W, split output, 2 transistor forward converter for Class B amp supply.)

Inductor spec:
L = 331uH
I(RMS) = 5A2
I(PK) = 9A7
Switching frequency = 60KHz.

Do you know of any good families of ferrite cores that i can use to wind this?

A the moment i am intending to use the ETD 54 core with 2mm gap as follows....

I intend to use 38 turns to get 331uH , however, i am worried about the openness of the core, and how it may interfere with other circuitry via its fields, and also the 2mm air gap in the centre post may overheat the windings near to it?

rikkitikkitavi 23rd September 2012 08:44 AM

core loss can be calculated by flux swing, and is mateeial dependant. as frequency is quote.low i would look into material sutable for 100+ kHz operation.

Perhaps you could.look into the better iron alloy materials like Sendust , a distributed gap gives less fringing losses around as these are.hard to calculate.

check out.powermagnetics (pace) webpage, they.are UK bbased and sell to EU for privateers ;)

Terry Given 24th September 2012 04:09 AM


try a PQ core. unless you use KoolMu, you will be horrified at the core loss of iron powder. even KoolMu is bad, although if its far in CCM it might be acceptable.

treez 24th September 2012 05:22 AM


Thanks but PQ cores, made by Ferroxcube and Epcos, do not come with gaps, so they will saturate at the current levels that i am using.

I would need to add a gap of 1.2mm to the PQ5050 core and this would mean the fringing field being significant and the usual reluctance method wont work to calculate the new inductance of the turns on the gapped core.....3D magnetic simulation is reguired and it costs $1000's.

i cannot understand why gapped cores of big ferrite cores are not available?

rikkitikkitavi 24th September 2012 05:30 AM

i think you are up for some experiment then, testing the core to see the heating effect of the.losses.

as for larger pq cores with an huge air gap- why would they manufacture these? one major pro for the pq core is the better shielding (but con is worse cooling) and the shielding effect is reduced by the gappin'

perhaps you can shims the core and put a flux band. still need to test for losses though. dont waste money on a simulation, spend it on testing instead. then you will know.

Terry Given 25th September 2012 02:15 AM

Treez, for a center-leg gap, you'll have to either gap the core yourself, or send it to someone to get gapped.

For an ETD54, if you pick a ferrite with a decent high-temperature saturation flux density like say 3C96, and choose Bmax = 300mT you will need a 2.4mm gap and 38 turns.

If you use a PQ5050, you will need a 1.8mm gap and 33 turns.

all calculated :)

I have transcribed the entire ferroxcube databook into Mathcad, along with lots of data on core materials etc. and spent a long time working out how to design all this stuff numerically - including variation of Bsat with temp etc. all fun and games. Its handy to be able to create ones own Hanna curves - I invariably find the ferroxcube databook hasn't got a Hanna curve for the core I'm using. Grrrr

darkfenriz 26th September 2012 08:29 AM

Another vote for toroid Sendust (aka kool-mu or super-mss). Less field and way more forgiving on saturation.
Like this one maybe:,33_8,0.pdf

Terry Given 26th September 2012 02:37 PM

"less field" isnt necessarily true. whats the permeability? hint: if u_r is, say, 99, then 1% of the stored field is in the surrounding air. low-permeability iron powder cores spew flux everywhere.

rikkitikkitavi 26th September 2012 04:29 PM

I dont think that leakage flux is avoidable, the air .must get in there either localized (air gap) or .distributed.

question is which is worst?
A centre gap , covered with winding and perhaps a. flux band OR a powder core , covered with winding around the radius with a flux band ?

I dont know , I am just an amateur :) I guess that is what experience is for :)

Terry Given 27th September 2012 04:30 AM

1 Attachment(s)

Originally Posted by rikkitikkitavi (
I dont think that leakage flux is avoidable, the air .must get in there either localized (air gap) or .distributed.

question is which is worst?
A centre gap , covered with winding and perhaps a. flux band OR a powder core , covered with winding around the radius with a flux band ?

I dont know , I am just an amateur :) I guess that is what experience is for :)

sorry, that came across wrong - prose fail. you're quite right. FWIW I figured you weren't an amateur.....

there really isnt a "best" - it depends entirely on what you're trying to achieve and why, and what you can actually get. every approach is bad if done wrong, and if done right they are all pretty good.

about the only hard and fast rule I have is this: never, ever use bobbin cores. they have the air gap on the outside..... the shielded ones arent too bad (but not as good as your suggestions), but unshielded is a nightmare.

and make sure you take skin and more importantly proximity effect into account.

Ahem. I got a bit carried away, and generated the attached graph for DIYAudio. Its a plot of AC-to-DC resistance ratio Fr, versus phi, which is the ratio of equivalent wire thickness (usually given the greek symbol Phi) to skin depth, for varying numbers of effective layers (usually called "p").

If you interleave your windings (eg 1/2P,S,1/2P) there is a line of symmetry through the center of the secondary - the windings "look" the same above and below this line. the number of effective layers in each winding is now HALF of the actual no. of layers.

eg we have a 4-layer primary with a 1-layer secondary on top. p_pri = 4, p_sec = 1

but if we interleave it then we get p_pri = 2, p_sec = 0.5

Phi is a horrible function of bare and insulated wire diameter, bobbin winding width, core center-leg length and turns-per-layer. I did some re-arranging, and came up with a really simple approximation thats good for wire with thin insulation. it takes into account all the packing factors (averaged across entire families of core/bobbin sets) and only needs wire diameter and switching frequency. MKS units, so m & Hz. And I inverted the approximation, so you can pick some Fr, find Phi then calculate wire diameter.

to see the advantage of interleaving, use the above p-values, with Phi = 2.

Fr_pri = 18 so Rac_pri = 18*Rdc_pri
Fr_sec = 1.9 so Rac_sec = 1.9*Rdc_sec

Fr_pri = 5.1 so Rac_pri = 5.1*Rdc_pri
Fr_sec = 1.1 so Rac_sec = 1.1*Rdc_sec

the primary Ac copper loss went down threefold!

the advantage of a gap, center-leg or core separation, is that the gap and hence fringing field is in a known & controlled place. so you can put a belly band on it to shield it, and not point it at something sensitive. center leg gaps are the best, but you do need to be careful of excess losses in the winding.

belly bands almost always help. in say an etd core, most of the winding is hanging out in free space - good for cooling and leaking flux. wind a planar xfmr on say an EI22, and half the damn winding is hanging out. Pot cores solve that problem, and replace it with thermal and "where do the wires come out" problems. PQ's are a compromise, giving quite a lot of core coverage but with a decent chunk exposed for cooling and leadouts. and its pretty easy to cover the big holes with a fat belly band.

One approach to avoid fringing flux related copper loss but that makes the winding look ugly and wastes a chunk of the winding area is to make a "lump" in the bobbin around the center, to lift the winding away from the gap. if the lump holds the winding a couple of gap lengths further out, and extends a couple of gap lengths either side of the gap, it can make a noticeable difference. litz is good too, if you can get it.

the distributed airgap cores have less problem with excess copper loss as the amount of fringing flux in any one place is a lot less - because its leaking a bit everywhere. And if you have a large AC component core loss will be a problem - and the less lossy a material is, the lower its permeability (generalising). so as you try to reduce core loss, you end up spraying flux everywhere (and coupling to any other cores doing likewise). I found that out the hard way, trying to reduce core loss by using -2 material.

the koolmu type cores are pretty good for losses (ferrites still win, with room to spare) but wow. $$$ even in volume. like 10x the cost of a -52 core the same size. funny thing is, thats probably much less of a problem for DIY stuff (assuming you can actually get them), and being able to buy a bag of 500 -52 cores for like 20c each is probably equally unappealing. I havent looked for a few years, and based on this thread its pretty clear there's a bunch of materials to look at. thanks :)

When there is a big DC component (so the AC core losses are small) iron powder is great. one sneaky gotcha is that the permeability drops off with frequency - and when it gets to one, its air cored. but thats at the MHz or so mark, eg I use -52 material over say -26 for that reason (it plays hell with many an EMI filter).

toroids are a pain to wind, especially with big fat wires. sore fingers.

to be honest the method I learned was "only choose cores you can actually get". by picking a core, then having the xfmr manufacturer or core vendor say things like MOQ 50,000 pieces, 26 weeks delivery. repeatedly.

All times are GMT. The time now is 04:30 PM.

Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2018 DragonByte Technologies Ltd.
Resources saved on this page: MySQL 18.75%
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2018 DragonByte Technologies Ltd.
Copyright ©1999-2018 diyAudio