Estimating the Number of Turns of an Inductor

[BFoelsch]

I had to take a year of Electrical Machinery in college, including
labs with big transformers and motors and stuff. I learned a lot from
it.


Is that because they conduct short circuits too well?

Primarily. In olden times transformers were designed for the best possible
"regulation." After power systems got stiff enough, however, the concept was
changed to allow transformers to limit fault current, and the old measure of
"regulation" was replaced with the modern day "impedance."

Extra credit: What is the definition of impedance, in the transformer sense?

Answer: The percentage of rated primary voltage which will produce rated
current through a short-circuited secondary.

 

[Jim Adney]

Suppose that I have an inductor that's covered with epoxy or similar
that prevents me from seeing or finding out how many turns of wire are
on the core. The core is open, so that it's uncovered and most of the
magnetic field is outside outside of the inductor. Obviously it's a
bobbin type core.

Why not just wind some turns as you suggest and then drive one coil
with a known signal while measuring the signal on the other coil. I
think that since the signal would be proportional to N this would have
less error in it than going by the inductance.

As a check, you could swap primary and secondary.

-

 

[Tom Bruhns]

Yup, you're right John. Tony's idea seemed plausible, but as I think
about it, I realize that the effect is the same magnitude. Faraday
was close enough to being right for pretty much all engineering work
so far. Should still work to _measure_ the leakage inductances;
in fact one could do it just with ratios of currents and voltages and
never actually worry about the inductances, though you would want to
be sure the inductive reactance was large compared with series
resistance and that the operating frequency was low enough to ignore
parasitic capacitances.

Apply a constant primary voltage (adjust to hold constant as needed).
Measure primary current = is with secondary shorted, = io with
secondary open. Measure primary voltage = vp and secondary voltage =
vs, with essentially no secondary loading. Then N =
vs*is/(vp*(is-io)), if I didn't make any mistakes, and ignoring
winding resistances and parasitic capacitances.

Cheers,
Tom

 

[Phil Allison]

"John Larkin"
"Phil Allison"
Not a bit. But the outer coil has - surprise! - a bigger diameter than

the inner, so more of the return flux is flowing *inside* the sense
coil, in the direction that reduces the induced voltage. Given a
typical drum/bobbin type inductor, I'd guess that the resulting error

might be in the 50% sort of turf;


** Q. Is a "drum inductor" one wound a bobbin *made* of ferrite ??

These are known to me as "ferrite bobbin inductors" or "bobbin core
inductors".

Up to saturation - and an drum core will usually vaporize before it
saturates - the voltage ratio, whatever it is, will be independent of
drive level; the coupling is linear.

** You have missed the point of primary voltage drop.

voltage.

With a closed, high-permeability core, voltage ratios can track turns
ratios to a part per million, as in a precision AC ratio box.

** Inductors are most often wound on just such cores - pot cores, RM
cores and transformer cores of all kinds - then a small gap is added if
needed.

A torroid is ideal for close coupling. That's not the case with a system dominated by air
gap, because the flux is scattered all over in space.

** Large air gap is a thing you failed to previously mention.

Transformar manufacturers routinely use DVM-looking gadgets that
indicate turns ratio, and can easily and accurately resolve whole or
half turns. But only when leakage inductance is low, as for a closed,
high-mu core with tightly-coupled windings.

** I just gave you an example where no close coupling is needed and leakage
inductance is irrelevant.



............ Phil

 

[John Larkin]

** Q. Is a "drum inductor" one wound a bobbin *made* of ferrite ??

These are known to me as "ferrite bobbin inductors" or "bobbin core
inductors".

Drum, bobbin, spool cores are pretty much synonymous. The
cross-section is the letter "I", with the wire wound on the center
post. The flux returns from the top to the bottom through air. These
are common and cheap, used as switcher filters and such.

Like these:

http://www.vishay.com/docs/34015/ihb.pdf

** You have missed the point of primary voltage drop.

OK. Please explain it again.

** Large air gap is a thing you failed to previously mention.

The thread started with Watson's statement:

====

Suppose that I have an inductor that's covered with epoxy or similar
that prevents me from seeing or finding out how many turns of wire are
on the core. The core is open, so that it's uncovered and most of the
magnetic field is outside outside of the inductor. Obviously it's a
bobbin type core.

====

so I figured the problem was the air path. If not, the issue is
trivial.


John

 

[Phil Allison]

"John Larkin"
"Phil Allison"

Drum, bobbin, spool cores are pretty much synonymous.

** That is not the point.

OK. Please explain it again.

** Go read my posts again, it was clearly explained.

The thread started with Watson's statement:

====

Suppose that I have an inductor that's covered with epoxy or similar
that prevents me from seeing or finding out how many turns of wire are
on the core. The core is open, so that it's uncovered and most of the
magnetic field is outside outside of the inductor. Obviously it's a
bobbin type core.

====

so I figured the problem was the air path. If not, the issue is
trivial.

** I see your point re bobbin cores now.




............ Phil

 

[Tony Williams]

Yup, you're right John. Tony's idea seemed plausible, but as I
think about it, I realize that the effect is the same magnitude.
Faraday was close enough to being right for pretty much all
engineering work so far.

I used the method a few years ago to find out
what was the source of problems on 500A/400Hz
CT's that were out of spec. It allowed a core
watts-loss problem to be distinguished from a
turns-count problem.

The CT's had a 1000 turn secondary and had a
100-wire connector pair that could be passed
through the CT and plugged together to form a
100 turn primary. So the primary was just a
bunch of ptfe covered flexibles that passed
through the hole. Fairly poor leakage-L.

 

[Tom Bruhns]

Belay that. That's not right. Maybe I can blame it on some red wine
or something.

Cheers,
Tom

 

[Tom Bruhns]

Subtitle: flies in the ointment.

Perhaps it's not possible to determine the turns ratio in the way I've
been suggesting, unless the coupling is very close to unity (in which
case there's no need to measure leakage inductances).

Consider the following coils/scenarios. Make three bobbins, all the
same dimensions. Make two of them from some nice dielectric like
Teflon or polystyrene. Make the third from a low-loss ferrite. Put
identical windings on all three. I'll assume here that the
permeability of the ferrite and the shape of the bobbin and winding
result in a coil with a self-inductance exactly four times that of the
other two coils. Neglect resistance and parasitic capacitance.

Now position the two "air-core" coils so the coefficient of coupling
is 0.5. Excite one with 1V. Measure the voltage induced in the
second without loading it: I believe it will be 0.5V.

Replace the second coil with the one wound on the ferrite bobbin, and
adjust for the same k=0.5. Now what voltage do you measure in that
coil? I believe it will be 1.0V, even though the turns ratio is the
same.

Sound right this time?

Cheers,
Tom

As has been pointed out in other postings to the thread, the
coefficient of coupling is important. Whatever flux from the primary
(driven winding) does not couple to the secondary will not induce
voltage in the secondary, and the measured turns ratio will be low as
a result. However, by measuring the inductance of the primary when
the secondary is open and again when it is shorted, and doing the same
with the secondary, you can find the leakage inductances and therefore
the coefficient of coupling, fairly accurately. (The second
measurement is really a check for consistency.) No need for xrays.
You could further improve the accuracy, I suppose, by including a
resistance value for each winding; ideally it would be the AC
resistance at the operating frequency. It will probably make for
easier calculations if you load the secondary very lightly for the
measurement.

....

 

[Terry Given]

Thanks for the interesting info. I would expect the core to be more of
a bobbin. But when it's covered, it's not always certain.


If I don't know the number of turns to begin with, do you expect me to
UNwind the coil to find the number of turns?

As I said, the coil is usually covered or potted in epoxy.

I design lots of magnetic devices. One of the first things I do with a
prototype transformer is dismantle it - I look at insulation, wire type,
winding pitch etc, AND I specifically count the number of turns (my
Leatherman has gutted hundreds of transformers I also measure coupling,
saturation, DC resistance, capacitance etc

If I had 2 of those bobbin core coils, I would just unwind the damn thing
and count the number of turns.

Ultimately, pay attention to John Larkin. A mate once had a 3-phase inductor
made (for a 600kW Butterworth filter that was hopeless - fringing flux
and leakage totally ruined the current distribution, making it glow red hot.
OTOH it made a great heater

Cheers
Terry

 

[Terry Given]

OK, let's suggest something different.

1. Measure the diameter of the wire in the existing coil.

2. Measure the DC resistance of the existing coil, and calculate the length
of the wire needed to generate that resistance.

3. Figure out how many turns will use up that length of wire.

You did say ESTIMATE, did you not?

??!!??

Nice

Cheers
Terry

 

[Watson A.Name - \Watt Sun, the Dark Remover\]

One way might be to in-case the coil in epoxy resin and saw it in half

and simply count the windings.

Well, yeah, if you don't mind destroying the coil.

I worked for a guy who paid quite a bit of money to buy a competitor's
filter so he could reverse engineer it and get his product to do the
same thing. So he took it to a friends of his, who happened to be a
dentist. He and the dentist put it on the X-Ray machine and took a
picture of it, and found that the potted filter was just a few pieces of
coax cut off at the right wavelength to eliminate the transmitter's
spurious outputs.

I could X-ray this, too; I don't want to destroy it. And it would be
extremely difficult to count the number of turns by your method if the
wire was 32 gauge or finer.

 

[Rich Grise]

There are many formulas for calculating inductance. All of them admit
to being approximations - but that's all you need. For example:

"For a coil of rectangular cross-section, of thickness t inches, length
l inches and mean diameter (average of inside and outside) d inches,
Hazletine's formula is L = 0.8d^2N^2 /(12d + 36l + 40t) uH"

Now if your entire coil, including the ferrite, is potted in epoxy, it
is a different situation. But I don't see that in any of your posts.

What if you took the dimensions of the existing coil, and guesstimate
down for winding cross-section because of the epoxy (I'm envisioning
a bobbin-shaped winding potted so you have a torus), plug in the
measured inductance, and it should give you _something_ !

Cheers!
Rich

 

[Watson A.Name - \Watt Sun, the Dark Remover\]

====================

Agreed. That's about the best he can manage. But what is not known is the
coefficient of coupling between the two coils. They are not wound in the
same volume of space or anywhere near to it. One is entirely outside the
other.

If the outside coil has a coefficient of coupling of 0.5 with the inside
coil then it is equivalent to a coil with only half the number of turns.

The arithmetic is simple. But what the coeff of coupling might be is
anybody's guess without knowledge of ALL dimensions of BOTH coils. Ask your
dentist if you could borrow his X-ray machine for the day. Even then a
hefty treatise involving higher mathematics on how to calculate the
coefficient of coupling between two coils would be essential.

All one knows is that the turns error, possibly very large, must lie on the
low side of the true value.

Its just occurred to me that with access to a precision X-ray machine or
electron microscope it may be possible actually to count the number of
turns. Try NASA.

Well, thanks for the advice.. I think.. :-/

How many Henrys is the thing anyway?

The coil on which I had considered doing this was only a hundred or so
microhenrys, but I meant it to be applied generally to any coil or
toroid that could have a few turns wound on it.

I've thought about other ways of estimating the turns. If I can see the
wire, I can roughly determine the gauge, and then measure the resistance
and get a rough estimate of the length. I could estimate a mean
diameter of the winding and then figure the number of turns from the
length.

But then I was just trying to find an easier way.

 

[Watson A.Name - \Watt Sun, the Dark Remover\]

As has been pointed out in other postings to the thread, the
coefficient of coupling is important. Whatever flux from the primary
(driven winding) does not couple to the secondary will not induce
voltage in the secondary, and the measured turns ratio will be low as
a result. However, by measuring the inductance of the primary when
the secondary is open and again when it is shorted, and doing the same
with the secondary, you can find the leakage inductances and therefore
the coefficient of coupling, fairly accurately. (The second
measurement is really a check for consistency.) No need for xrays.
You could further improve the accuracy, I suppose, by including a
resistance value for each winding; ideally it would be the AC
resistance at the operating frequency. It will probably make for
easier calculations if you load the secondary very lightly for the
measurement.

But I'm still not seeing any need to know the number of turns, other
than for idle curosity. "I need to know because I want to"??

If I want to make a reasonable facsimile of the coil, I have to know a
bit about it, like what kind of ferrite material and how many turns it
has. I thought that getting the number of tuens would be alot of help
with this.

 

[Watson A.Name - \Watt Sun, the Dark Remover\]

"John Larkin"
"Phil Allison"
are in
exact


** It is *unhelpful* because it is so damn ambiguous.



** It makes less sense to scorn a perfectly practical test method.




** The overwind is to be around the existing coil, wound in parallel and on
top of it, touching it - is that hard to comprehend ?

A further ( rather obvious) condition is that the inductor coil current for
the test be low enough to not generate a significant voltage drop across the
coil's resistance - or you calculate that drop and take it into account.


** I just took a small mains toroidal ( 30VA) and with the primary
energised at 230 volts passed a one turn loop through the core and measured
0.102 volts rms across the ends. The loop could be made as open as you
liked or tight wrapped as you liked with NO change in the measured voltage.

The primary magnetising current was only 1.5 mA and the primary resistance
was 94 ohms - so a negligible primary drop of 140 mV.

So I make the primary turns to be 2255 ( +/- the AC voltmeter's 0.3 %
error, or about 7 turns)

Thanks for a vote of confidence... One thing about a toroid, it
concentrates most of the field inside the donut so any added windings
would be coupled more strongly than with a bobbin core.

 

[Phil Allison]

"Watson A.Name
"Phil Allison"

measured 0.102 volts rms across the ends. The loop could be made as open as
you liked or tight wrapped as you liked with NO change in the measured
voltage.
resistance was 94 ohms - so a negligible primary drop of 140 mV.
0.3 % error, or about 7 turns)


Thanks for a vote of confidence... One thing about a toroid, it
concentrates most of the field inside the donut so any added windings
would be coupled more strongly than with a bobbin core.

** Strange then that a one turn loop could be up to a metre in diameter
with no change in voltage.




............ Phil

 

[Watson A.Name - \Watt Sun, the Dark Remover\]

That method of measuring the leakage inductance
(by shorting windings) gives a hint towards a
possible experimental method.... Short the sec
with an ammeter and treat the thing as a CT.

After all, CT's have a current-ratio that is quite
close to the turns-ratio, even though the coupling
can be poor (as in a CT with a bar primary). This
is because the leakage inductance (and R-primary)
can be regarded as being in series with a constant
current stimulus source. The major source of error
is then the sideways current due to the shunt loss.

So perhaps do a short-circuit current-ratio test,
then measure the sideways shunt-current taken by
just the primary, at the same equivalent voltage.

So it first was voltage ratio, and now current ratio.

How do I measure short circuit current? Is an AC
milliammeter a good enough short circuit to make the
measurement? It will have some resistance and hence
some voltage drop. That's not a true short circuit.

 

[John Larkin]

"Watson A.Name
"Phil Allison"



** Strange then that a one turn loop could be up to a metre in diameter
with no change in voltage.

Faraday's Law is not strange.

John

 

[Phil Allison]

Faraday's Law is not strange.

** Not the point.



............ Phil