# Help winding my own inductor?

Discussion in 'Electronic Basics' started by DaveC, Dec 19, 2003.

1. ### DaveCGuest

20 millihenry, 1 amp, 25 KHz. Pot core, I presume. Don't know enough about
these to know if I need a core with a gap or not. Willing to learn...

I have a 20 mH inductor (off the shelf) but is getting hot with the current
I'm passing through it.

Where do I start? References and advice welcome.

Thanks,

2. ### Robert BaerGuest

If your 20mH inductor is getting hot, then (obviously) you are forcing
more current thru it than it is rated for...and you are very likely
saturating it, meaning the actual inductance may be 1uH or less.
First, with a given core configuration (with or without a gap), you
determine the effective pernmeability in amps per turns squared; then
calculate the number of turns needed for the inductance you would like
to have.
The ampere turns gives you the flux density, and from the effective BH
curve, determine if it is saturated.
Increase the core area and/or the gap and re-calculate; repeat as
needed.
Then use wire of a size as follows: cross-sectional in circular mils =
milliamps of current rating for the inductor.
That wire size, and number of turns can be used to calculate if it
will fit in the available space.
Tis not symple.....

3. ### John PopelishGuest

A practical inductor that you describe will need a gap to store the
energy of the inductor without magnetic saturation (either pot or EE
core) or a distributed gap (powdered metal toroid) and also the wire
window to carry the wire size that will handle the current for those
turns. The easy to wind solution (filling a bobbin) would be a pot
core or some other variant of an EE core. 20 millihenries at 1 amp
would need a substantial core size.

The Fair-rite catalog has a design guide for their core types that is
pretty applicable to any brand.
http://www.fair-rite.com/fr_catalog-14thed_rev3.pdf

Getting your hands on a selection of cores and bobbins about the right
size is the hard part. I have been collecting them for 20 years (most
recently shopping on ebay), just to have the parts to make prototypes.

4. ### GenomeGuest

That's a mother big inductor. Are you sure you meant 20mH

Your starting sum is

Aw.Ae = L.Irms.Ipk/Bpk.J.Kcu.Kw

L is the value of your inductor in Humphreys
Irms is the rms current you want it to carry in amps
Ipk is the peak current you expect it to suffer from in amps
Bpk is the maximum flux you want the core material to hack in Teslas
J is the operating current density of your wire. 4E6 amps/meter^2 for a
nominal 30C temp rise.
Kcu is the utilisation factor of your wire.... 0.7 for round copper.
Kw is how well it fills the available winding area. 1 for a single winding.

Aw is the winding area
Ae is the effective area of the core.

Aw.Ae = 20mH x 1 x 1/0.3 x 4E6 x 0.7 x 1
Aw.Ae = 23.8E-9

Square root it

Aw = Ae = 154mm^2

Ooooh maybe it's not so big.

Find a pot core that sort of does that. And they won't...... Bugger

Number of turns

N = LIpk/BpkAe
N = 20mH x 1/0.3 x 154E-6
N = 433

Just make sure that when you wind it you lock the door and take the phone
off the hook. Begin 1.... 2..... 3.......30.... 80...... 333. 'How many
sugars do you want in your tea?' Arghhhhhh, bastard.......
1.....2......3.....28

Size of wire.... err

154E-6/433 = 0.35565mm^2

Square rooted.... 0.596mm diameter, enammelellmeleded heavy insulation...
AWG 24 or 25 if you am an American.

Al value is L/N^2 or 106nH per root turd. Now that's just got to be an
almost standard set. N27 or 3C80 or some such.

DNA

5. ### DaveCGuest

Use maybe 78 material from this catalog? It is recommended for applications
<200 KHz, but it says it's for power supply applications (meaning 50 - 60
Hz?). Other materials seem appropriate, also, but don't come in pot cores.

What is a good mail-order supplier for such cores and bobbins?

Thanks,

6. ### DaveCGuest

Yes, I'm sure. I measured it with reliable L meter, at 25 KHz.
Don't understand what "And they won't" means... Can't use a pot core for this
application?
Hmm... the one I've got now is about that wire size. Is it overheating
because it's core is saturating? Maybe I can just put it in another pot core?
Or is the wire size the *only* thing that determines whether or not it gets
hot?

(Yes, 'Merican.)
Where would I find these "standard" pot core sets? A mail order supplier,
maybe? Suggestions?

Thanks,

7. ### John WoodgateGuest

I hope that anyone who thought Genome was an idiot is now firmly
disabused. Although he still couldn't resist misspelling 'turn'.

8. ### DaveCGuest

Being a non-pro, I was wondering about that term Thought maybe I'd slept
through some important topic in electronics lecture...

9. ### Mark FergersonGuest

Nah, our Genome merely has a mild case of whatever James
Joyce suffered from. I say a "mild" case because one can
usually figure out what Genome's on about in a single

Mark L. Fergerson

10. ### Tim WescottGuest

Oh, when you're halfway through winding a large toroid you start thinking of
them in those terms.

11. ### Tim WescottGuest

Amidon (http://www.amidoncorp.com) has been the standard place to buy this
stuff for amatuer radio operators, and they have a reasonable selection of
cores. They seem to list the fair-rite numbers in their ferrite cores (they
also have iron powder cores).

Most of the posters are on track as far as I can tell, here's my summary and
two bits:

- Design for the inductance you need _at_ your maximum current. The
effective inductance of your coil will go down as the core saturates, take
that into account.

- If you put a constant voltage on your core you can watch the current and
see if it's saturating: a constant inductance will be indicated by a
constant slope as the current increases. As the core starts to saturate the
current will curve upward slightly. If the current _really_ curves upward
before you've reached your design current you've got problems.

- If you're making a core for a switching power supply you probably don't
have to worry about gapping it -- gapping the core reduces the inductance
variation with variations in current by limiting the overall reluctance of
the core. If you need controlled inductance for some reason, run away.

- The winding will exhibit loss due to current x voltage drop. The core
itself will also show a loss due to hysteresis. The core loss is
proportional to the coil voltage x frequency (actually the core flux x
frequency). You need to size both winding and core, but if you use a
material that's recommended for your frequency and application you'll
probably get the right size core.

- One final note: if you do this again for much higher frequencies
(100-500kHz), make homeade litz wire to reduce skin effect: calculate the
wire cross section you need, and make it up with three to five strands of
smaller wire, loosely wound. This will bring the effective AC resistance of
your wire down closer to the DC resistance.

12. ### John WoodgateGuest

I read in sci.electronics.design that Tim Wescott
Yes. I bought a demister antenna kit from Maplin that required a small
ferrite toroid to be wound with 1.6 mm wire. In two words - im possible.

13. ### John PopelishGuest

They mean switching power supplies. Just what it sounds like you are
doing.
I wish I knew. If you can't find any supplier, I may be able to
scrounge 1 or two core sets and bobbins from my personal stash to get
you started. This is the easiest component to make yourself that can
compete with a commercial product in quality, but it is very hard to
find the parts in low quantity. And that is a real shame.

14. ### John PopelishGuest

The first turn probably wasn't so bad.

15. ### TweetldeeGuest

Yeahhh I was laughing my ass off while reading his answer. That was a
riot!!! I'll bet he's a good one to have at a party...
--
Tweetldee
Tweetldee at att dot net (Just subsitute the appropriate characters in the

Never take a laxative and a sleeping pill at the same time!!

16. ### TweetldeeGuest

True enough, but the rest of those turds were just plain nasty!!
--
Tweetldee
Tweetldee at att dot net (Just subsitute the appropriate characters in the

Never take a laxative and a sleeping pill at the same time!!

17. ### GenomeGuest

Well sort of. Humphreys are more cuddly.

Snippy
Pot cores have an alarmingly extreme mismatch between Aw and Ae. It's not
their fault, it's just the way they are and it prob don't matter. Rooting
things give a guess as to the bigness. Chances are a pot core will have a
bigger Ae versus its Aw but if its Aw times its Ae does the biz then I
wouldn't argue with it.

Snippy
I seriously hate bastards who *define* their *meaning* with arseholelicks.
That would be Asterixs.

Oh well,.......

Yooz losses, like heating things cum from the core and the copper. Now,
being Mr thick brick..... Oh bolicks its got something to do with exorcising
the BH loop and prximity layer such like. Simple.

And stop giving me Chritmas cards, I'll have to retaliate.
Ahhhhhhhhhhhhhh. Manys the time I've fantasized about a bit of the anal with
Hillery Clinton's. Licky licky.
You phone up your local TDK rep and listen to an Indian rolling about
laughing when you don't want to buy a million off.
Not a problem.

Do not rely on others to define your existance. Go work it out for yourself.

They will hate you for it and you will be seriously fucked up.

Nerrrr Nerrrrr nerrrrr
DNA

18. ### BrianGuest

Using a program we have developed over the years (which will do coils and core
selections, plus many other things as well). I plugged in your numbers to come
up with a core size. Take a look at it at http://www.fncwired.com/CoreExample

As you can see, we could have used a little bit more information. If you would
like us to change something, let us know and we will enter in the new data (and
display it for you). This program will also give you all the formulas used.

The core used is "Magnetics" molypermalloy powder core (MPP core).

19. ### DaveCGuest

Sorry. How 's THIS instead? =-o
Thanks (I think...)

Not impossible
What you need is a shuttle.
If the hole is real small you can use a needle as a shuttle.
If it is larger then you can take a wire coat hanger and cut a section off
and with a riffle file, file notches in a section of it and wrap your wire
around the notches in the wire (Or a metal or wood rod) that will keep your
wire neat and nice while you pass it around the core.
I have wound many of them

As an alternet use a pot core if applicable.