Bill Turner wrote:

You are speaking in *practical* terms, which is fine. It's true that at
relatively low frequencies, well below the self-resonant point, coils
appear to have constant inductance. No argument there. The discussion
came about because someone asserted that inductance was a constant,
REGARDLESS of frequency, and that is just not true.

I disagree. The inductive component of the impedance remains
essentially constant through resonance. What is non ideal about the
inductor is that it does not exhibit just inductance, but a parallel
combination if inductance and capacitance. Ignoring the capacitance
and calling the effect variable inductance is just not as accurate a
way to describe what is going on.

--
John Popelish

Bill Turner wrote:

That will work, no doubt. My point was that it takes some serious
engineering and careful testing; you can't just wrap some wire on a form
and expect it to work correctly across a wide range of frequencies.

This is a generality I can agree with.

--
John Popelish

Bill Turner wrote:

That will work, no doubt. My point was that it takes some serious
engineering and careful testing; you can't just wrap some wire on a form
and expect it to work correctly across a wide range of frequencies.

This is a generality I can agree with.

--
John Popelish

Bill Turner wrote:

On 07 Dec 2003 18:25:51 GMT, (Avery Fineman) wrote:

Write on the whiteboard 100 times: Inductance does not change
with frequency...reactance changes with frequency.

__________________________________________________ _______

Not true. Inductance and reactance are related by the formula
XsubL = 2 pi F L. If XsubL has changed, then so has the inductance, and
vice versa.

How could you possibly define it otherwise?

But the impedance of a coil near resonance is not well described as an
XsubL. It is a combination of XsubL and XsubC, including their
different phase shifts.

You cannot just measure the magnitude of impedance of a coil and
assume that you are measuring pure XsubL. You have to prove that this
is the case by some other measurement, like the phase relationship
between voltage and current.

--
John Popelish

Bill Turner wrote:

On 07 Dec 2003 18:25:51 GMT, (Avery Fineman) wrote:

Write on the whiteboard 100 times: Inductance does not change
with frequency...reactance changes with frequency.

__________________________________________________ _______

Not true. Inductance and reactance are related by the formula
XsubL = 2 pi F L. If XsubL has changed, then so has the inductance, and
vice versa.

How could you possibly define it otherwise?

But the impedance of a coil near resonance is not well described as an
XsubL. It is a combination of XsubL and XsubC, including their
different phase shifts.

You cannot just measure the magnitude of impedance of a coil and
assume that you are measuring pure XsubL. You have to prove that this
is the case by some other measurement, like the phase relationship
between voltage and current.

--
John Popelish

Bill Turner writes:

On Sun, 07 Dec 2003 15:54:05 +0000, John Devereux
wrote:

No, you are talking about the *reactance* ("reactive impedance"). We
have been talking about the *inductance* ! They are not the same
thing.

__________________________________________________ _______

No one ever said they were the same thing. They are related to each
other by the formula XsubL = 2 pi F L. That is a direct, linear
relationship.

The important thing here is the "subL". It applies only to the
inductive part of the overall reactance.

Are you saying that formula is correct as some (low) frequency but
incorrect at another (high) frequency?

No, it is always correct. It is practically the *definition* of
inductance so it had better be!

I'll say it another way: Inductance and reactance are directly related
to each other by the (2 pi F) factor. Given one (inductance or
reactance) you can calculate the other. There is no other way.

No. Because the "reactance" (without the sub-L) now has both inductive
*and* capacitive terms. When you measure the *overall* reactance of a
real life coil you are measuring the effect of *both* terms. You
cannot measure this combined reactance and then just plug the number
into a formula which ignores the capacitive part. You have to use the
general formula which include the self capacitance.

Ignoring the coil resistance (i.e. we have infinite Q) the correct
formula is something like:

Xtotal = 1
--------------
|1/Xc| - |1/Xl|

Where Xc = 1/(2 pi F C) and Xl = 2 pi F L.

Hopefully you can see how Xtotal behaves as you describe, even with
constant L.


--

John Devereux

Bill Turner writes:

On Sun, 07 Dec 2003 15:54:05 +0000, John Devereux
wrote:

No, you are talking about the *reactance* ("reactive impedance"). We
have been talking about the *inductance* ! They are not the same
thing.

__________________________________________________ _______

No one ever said they were the same thing. They are related to each
other by the formula XsubL = 2 pi F L. That is a direct, linear
relationship.

The important thing here is the "subL". It applies only to the
inductive part of the overall reactance.

Are you saying that formula is correct as some (low) frequency but
incorrect at another (high) frequency?

No, it is always correct. It is practically the *definition* of
inductance so it had better be!

I'll say it another way: Inductance and reactance are directly related
to each other by the (2 pi F) factor. Given one (inductance or
reactance) you can calculate the other. There is no other way.

No. Because the "reactance" (without the sub-L) now has both inductive
*and* capacitive terms. When you measure the *overall* reactance of a
real life coil you are measuring the effect of *both* terms. You
cannot measure this combined reactance and then just plug the number
into a formula which ignores the capacitive part. You have to use the
general formula which include the self capacitance.

Ignoring the coil resistance (i.e. we have infinite Q) the correct
formula is something like:

Xtotal = 1
--------------
|1/Xc| - |1/Xl|

Where Xc = 1/(2 pi F C) and Xl = 2 pi F L.

Hopefully you can see how Xtotal behaves as you describe, even with
constant L.


--

John Devereux

On Sun, 07 Dec 2003 04:31:46 -0800, Bill Turner
wrote:

On Sun, 07 Dec 2003 13:55:35 +0200, Paul Keinanen
wrote:

One can still argue that the inductance and inductive reactance are as
well as the capacitance and the capacitive reactance are still there
as separate entities, but we can not measure them separately from
terminals of the coil. Thus, this is an artefact of the measurement
method.

Not only can you *not* measure them separately, they can not be
physically separated either, since the parasitic capacitance is always
present between adjacent windings. I would not call it an artifact of
the measurement method, but rather an artifact of the coil itself.

The problem with circuits containing both inductances and capacitances
is that in one kind of reactance, there is a +90 degree phase shift
and the other with -90 degree phase shift. Thus, when these are
combined, they partially cancel each other, producing different
magnitudes and some phase shift between -90 and +90 degrees. If only
the resultant magnitude is used (and the resultant phase is ignored),
this would give the false impression that the inductance changes with
frequency.

Instead of using the resultant reactance on some specific frequency,
the inductance could be measured in a different way.

When a DC current I is flowing through and inductance L, the energy
stored in the inductance is W = I*I*L/2. This could be used to
determine the inductance L.

One way to measure the energy W would be to cut the DC current through
L and after disconnecting I, dissipate the energy in some kind of
integrating load across L. Even if there is a significant capacitance
across L, no energy is initially stored in C, since during the steady
state condition, the current I would be flowing through L, but there
would be no voltage difference between the ends of L (assuming R=0),
thus all energy in this parallel resonance circuit is stored in L.

After disconnecting the DC current I, the energy would bounce back
between L and C, but finally it would be dissipated by the external
load. The same energy would be dissipated in the external load even if
C did not exist (assuming zero losses).

Thus using this measurement method, the value of L would be the same
regardless if C is present or not.

Thus, getting a frequency dependent L, is a measurement artifact in
the method that you are using.

Paul OH3LWR

On Sun, 07 Dec 2003 04:31:46 -0800, Bill Turner
wrote:

On Sun, 07 Dec 2003 13:55:35 +0200, Paul Keinanen
wrote:

One can still argue that the inductance and inductive reactance are as
well as the capacitance and the capacitive reactance are still there
as separate entities, but we can not measure them separately from
terminals of the coil. Thus, this is an artefact of the measurement
method.

Not only can you *not* measure them separately, they can not be
physically separated either, since the parasitic capacitance is always
present between adjacent windings. I would not call it an artifact of
the measurement method, but rather an artifact of the coil itself.

The problem with circuits containing both inductances and capacitances
is that in one kind of reactance, there is a +90 degree phase shift
and the other with -90 degree phase shift. Thus, when these are
combined, they partially cancel each other, producing different
magnitudes and some phase shift between -90 and +90 degrees. If only
the resultant magnitude is used (and the resultant phase is ignored),
this would give the false impression that the inductance changes with
frequency.

Instead of using the resultant reactance on some specific frequency,
the inductance could be measured in a different way.

When a DC current I is flowing through and inductance L, the energy
stored in the inductance is W = I*I*L/2. This could be used to
determine the inductance L.

One way to measure the energy W would be to cut the DC current through
L and after disconnecting I, dissipate the energy in some kind of
integrating load across L. Even if there is a significant capacitance
across L, no energy is initially stored in C, since during the steady
state condition, the current I would be flowing through L, but there
would be no voltage difference between the ends of L (assuming R=0),
thus all energy in this parallel resonance circuit is stored in L.

After disconnecting the DC current I, the energy would bounce back
between L and C, but finally it would be dissipated by the external
load. The same energy would be dissipated in the external load even if
C did not exist (assuming zero losses).

Thus using this measurement method, the value of L would be the same
regardless if C is present or not.

Thus, getting a frequency dependent L, is a measurement artifact in
the method that you are using.

Paul OH3LWR

On Sun, 7 Dec 2003 19:13:36 +0000, John Woodgate
wrote:

Low-frequency iron-cored coils are quite another matter; the
inductance varies with frequency, voltage, temperature, previous history
and the state of the tide on Europa.

I assume that you are referring to DC biased iron cores (without an
air gap) or some high permeability ferrites with a strong DC bias
current. These do indeed show a variation of inductance depending on
the DC bias current.

Paul OH3LWR