Winding a Toroidal Ferrite Core

[Bob Scott]

Hi guys.

I have been delving into the wonderful world of inductors and transformers lately, with the intention of getting the ability to design my own high-power switching power supply. I have chosen the operating frequency of 30KHz. That is above the audio range yet still slow enough for efficient switching with power MOSFET. The ferrite core is capable of 1KVA at 30KHz:

Amidon type FT240-77. 2.4" O.D., 1.4" I.D, 0.5" thick. Mu of 2,000. Core cross sectional area Ae=1.57cm^2. Effective core length Le=14.4cm. Al=2,740 millihenries at 1,000 turns.
Maximum allowable flux density at 30KHz is 1,600 Gauss.

I need to wind this core with 14 gauge enameled copper wire on the primary winding. I am assuming the ferrite core is as fragile as a coffee mug handle. The core is a standard toroid with a "squarish" cross section.

How do I bend and form the big solid copper wire around this squarish 1/2" X 1/2" ferrite cross section without breaking the core?? I have to do it neatly, without wasting winding space. Can I use stranded wire?

Thanks in advance for any suggestions. Part of my apprehension is not knowing how strong the core is. I don't have any experience breaking cores. I suppose the only real way to get familiar with the strength of ferrites is to break a few. That is, apply enough forming pressure to the copper to get a feel for how much force is too much. I'd use 12 gauge copper, but I don't think it would be possible to get it to form-fit around the tiny core, except maybe for the secondary winding.

This reminds me of a talk I had with a really good mechanical tech, remembering how he got so good at what he does. "I cringe when I think of how many classic old Chevy parts I damaged when I was a kid".

[ltx71cm]

I've never tried to wind something as thick as 14 gauge wire but the cores are, in my experience, fairly sturdy. Someone correct me if I'm wrong but I don't think breaking the core is a risk but chipping off pieces along the edge is a very real possibility. My word of caution is to not pull too tight. I used to pull so tight that the edge of the toroid would penetrate the enamel and cause serious problems.

Do you have an old power or USB cable laying around with a lump in it? If you'll never need the wire cut out the ferrite chunk and go about destroying it to gauge relative strength for future reference. When I was younger (don't ask why because I couldn't tell you) I used to try to split toroids with pliers and never had much luck.

[jwax]

How many turns do you need on it Bob? Never wound one myself, but a layer of sturdy paper/plastic on the core first seems prudent.

[MrAl]

Hello there,


The first thing to do is wrap some turns of tape around the core. That protects the wire making a softer interface between the core and the winding and also makes some wiggle room for when the wire heats up and expands. A tape often used for these cores is mylar tape, but im sure there are other types too. If you can you can use tape between layers too, and that's more important if the layers have a larger voltage across them (depends on the application). Of course the tape used must be able to stand the expected temperature range.

Second, dont use #14 gauge wire if you dont have to for some reason. The best way is to go down in size so that you can use two strands, wound bifilar. You just have to use two wires who's area is the same together as one strand of #14.
Incidentally, it's more efficient too as well as a little easier to wind. The two wires, even with the same area as the #14, have less AC resistance than the #14.

[dyarker]

Mr Al already said what I was going to. So, spent a couple minutes reading Wikipedia pages on American Wire Gauge and Skin Effect. I think 3 18AWG windings in parallel will carry the same current at 30KHz as 14AWG does at 60Hz, maybe a little better. And a lot easier to handle!

Cheers,

[MrAl]

Hi again,


Yes three 18 gauge wires would work too, but then again three 19 gauge might work too,
or just two 17 gauge wires.

A common rule is that if you divide the area by half then you go up by three wire sizes. Another rule is if you divide the area by 3 then you go up by five wire sizes. More accurately however we have these formulas:

AWG2=AWG1+2.989164528727223
AWG3=AWG1+4.737713686518475

Thus if AWG1 is 14, then AWG2 (using two wires) would come out to #17, so two #17 wires would work also. For three wires AWG3 would come out to a #19 gauge wire, with a little less total area than the #14 gauge wire, but not too much less.

I would think two #17 gauge wires should work pretty well without being too hard to wind.

[dyarker]

True.

I tend to suggest "do-able". 18AWG magnet wire is available at Mouser and Jameco. Odd gauges would take ... umm ... more intense shopping.

Not competing. Trying to add something. Perhaps being brief conveys bad attitude.

Cheers,

[MrAl]

Hi again,


Oh well i agree that three #18 gauge wires would provide a little more area than the single #14 gauge.
In fact, it would be about 19 percent more cross sectional area, and about 89 percent more surface area.

[Bob Scott]

How many turns do you need on it Bob? Never wound one myself, but a layer of sturdy paper/plastic on the core first seems prudent.

I played around with ‘Faraday’s Law.

B = (V * 10^8) / 4.44 * Ae * N * f)

where this is the CGS system. B is in Gauss, Ae is cross-sectional core area. I wondered quite a while where this “4.44” came from. I think that the equation was originally simply written as :

B = (V * 10^8) / Ae * N * w

Where “w” is actually a lower case omega meaning radians per second, but this message board seems to have lost the ability to post symbols.

Then substitute “2 * pi * f” for “w” so we can use cycles per second instead of radians/sec. Then since this equation was mostly used for designing 60 Hz power transformers in the olden days, they added a 1.414 (the square root of 2) multiplier to the numerator for finding Bmax at peak sinusoidal current.

Bmax = (1.414 * V * 10^8) / (2 * pi * Ae * N * f)

The 1.414 / (2 * pi) equals 1 / 4.44. That's how I figure they came up with this "4.44". For square waves, my book says to use 4.0 instead of 4.44. I don’t know how they calculated the “4” Anybody??

Anyway, using that formula to calculate the number of tuns with B at the max recommended B of 1600 @ 30 KHz, and adding 10% for power line surges, I need 4 primary windings of 32 turns each. Voltage is rectified line DC of 170VDC with 120VAC in.

Depending on the driving curcuitry I choose, I can:

1, Install a filter cap in series with the winding to 0V and the other end to +/-85V with a couple of FETS (half an H bridge), all windings in parallel.

2, Drive the primary with an H bridge, 2 series sets of 2 windings in parallel.

3. Drive all 4 windings in series, with a tap in the middle to 170V, 1 N channel FET to 0V at each end.

All windings must fit in the core's ID of 1.4" Half for primary and half for secondary, the secondary windings get a 1" dia space in the middle. The primaries must fit in the space between, 0.2 inch thick band around the secondaries, with an average diameter of 1.2". Winding area for primaries is then 0.785 in^2. That's enough room for 180 turns of #14. I only need 128 turns. I think maybe perfect bends might not be needed.

[Joseph]

Hi Bob, 128 turns sounds unusually high in number. I usually set my primary turn figure by experiment and base the secondary on that, so i can't easily check your math with the formula. You gave a nice possible history of the formula, however.

[MrAl]

Hi,


The difference between the 4 and 4.44 is probably due to the average values. It's been a long time since i looked at the derivation of these formulas though.

[Dean Huster]

I tend to suggest "do-able". 18AWG magnet wire is available at Mouser and Jameco.

Magnet wire is the key here. It's softer copper, bends more easily and doesn't spring back like harder copper. Also, stranded wire would be easier to wind, but finding it with thin or enameled insulation is the problem.

[Bygar]

Wire sizing for transformers!
What type of service will it be used for?
Amps per circular mil, your choice.
AWG, 10 gauge wire =10,000 circular mil.
As you move higher in gauge E.G. to 13 gauge = 5,000 circular mil.
For every 3 change in gauge divide circular mil by 2 or multiply by 2.
Depends on if getting larger or smaller.
Bygar

[Dimbulb]

The square edges could mean it was intended for coating.
Always put a radius, dremels work. once or twice around
with plummers teflon tape is super.

Depending on the end use. For simplicity the larger gauge
and gently flatten with small rubber mallet as you go.
gaps may not be good

Test wrap 10 turns with what you intuitively feel
will work then stop and cut it. measure the inductance, divide by ten
to get the actual uH per turn then see if it will measures out,
then purchase what you need, pull out the length plus some extra.
better too much than not enough.

I believe for 77 material you may get at least 2000 mu
initially but probably more and don't forget about grid dip
oscillator, sounds like fun!