Interpretation and application of Poynting's theorem?

harrylin
In several recent threads the Poynting's theorem was brought up, and the discussion there became a distraction from the questions at hand without really being solved:
https://www.physicsforums.com/showthread.php?t=628896 (this post continues from the discussion starting with post #198)

Roughly, the theorem states (I think) that the energy that dissipates as heat inside a volume equals the energy that enters it minus the stored energy that is extracted from it.

A basic issue is the meaning there of the term E.j.
According to some, the energy transferred from EM fields to matter (work) is E.j.
However, the texts that I consulted specify that it is the Ohmic heating rate, or the dissipated energy inside the volume under consideration. Clearly there is some ambiguity here.

A paper that appeared a few years ago in Europhysics Letters 81 (6): 67005 (thanks Wikipedia!)

Clarifications are welcome!

Last edited:

Mentor
A basic issue is the meaning there of the term E.j.
According to some, the energy transferred from EM fields to matter (work) is E.j.
However, the texts that I consulted specify that it is the Ohmic heating rate, or the dissipated energy inside the volume under consideration. Clearly there is some ambiguity here.

A paper that appeared a few years ago in Europhysics Letters 81 (6): 67005 (thanks Wikipedia!)
In the reference you cited please see page 2, second column, first paragraph, in the description of equations 7, 8, and 9. E.j clearly refers to mechanical energy in that equation, which is appropriate at the microscopic level.

I believe that you are simply misreading your other texts. Most likely they discuss Ohmic dissipation, but don't even mention mechanical work in their derivation of the E.j term. You are interpreting that to mean that the E.j term applies only to resistive losses. The absense of an assertion about mechanical work should not be taken as an assertion of the absence of mechanical work.

Please see point 3 in the derivation of Poynting's theorem in the Wikipedia entry I have posted multiple times now:
http://en.wikipedia.org/wiki/Poynting's_theorem#Poynting.27s_theorem

harrylin
In the reference you cited please see page 2, second column, first paragraph, in the description of equations 7, 8, and 9. E.j clearly refers to mechanical energy in that equation, which is appropriate at the microscopic level.

[..] Please see point 3 in the derivation of Poynting's theorem in the Wikipedia entry I have posted multiple times now:
http://en.wikipedia.org/wiki/Poynting's_theorem#Poynting.27s_theorem
Sorry I forgot to include the Wikipedia reference to that paper:
http://en.wikipedia.org/wiki/Poynting's_theorem#Generalization

harrylin
Harald, there are better examples of the Poynting theorem that explicitly express reactive (i.e. 'stored' or 'static/quasi-static') field energy in the formula; e.g. "my.ece.ucsb.edu/bobsclass/201C/Handouts/Chap1.pdf" [Broken] see (1.31) on p9 there. Anyway, crucially one has to distinguish between purely formal results treating fictitious Amperian currents as classical real rho*v current densities, and the actual case in magnetic media involving intrinsic moments that definitely do *not* respond to an applied E as though real classical currents.
That looks interesting, and it also refers to "ambiguities".
I have in mind to check out the references and then I'll give feedback of how I understand those.

Last edited by a moderator:
harrylin

1. It appears to me that the distinction that http://arxiv.org/abs/0710.0515 makes between Eext and Eind could be essential for a good understanding. I assume that Eext refers to externally applied electric field strength and Eind to induced electric field strength.
It's the first time that I see that distinction being made, and it was also not brought up in the discussions from which this topic is a spin-off.

That same distinction is made in http://my.ece.ucsb.edu/bobsclass/201C/Handouts/Chap1.pdf [Broken] : There the author uses Einc for incident or applied field, and Escat for "a component produced by the induced currents". This agrees in essences with my interpretation of the meaning of the symbols that I saw earlier. And the second text adds the clarification:

"The distinction between impressed and induced currents is therefore a natural breakdown
in terms of “cause and effect”."

In the earlier threads in which Poynting was brought up for an analysis of "what does work", this distinction between Eext and Eind was not made. I think that such a distinction between cause and effect must be made if one wants to use Poynting's theorem for inferring what does work and what does not.

Does anyone disagree? And if so, why?

2. The particular question that came up in the thread on magnets, was how we should describe Lenz-Faraday magnetic induction in a zero resistance current carrying wireloop with the use of Poyntings theorem.

While definitely work is done under displacement, it suddenly strikes me that here the term E.j can only be zero. How does Poynting's theorem look like for this particular case?

At first sight, following eq.1.31 of "bobs" handout, it looks to me that the energy per volume that is extracted from the magnetic field for doing work equals M.H for the case without external source (like a permanent magnet).

Is that correct?

Last edited by a moderator:
Mentor
1. It appears to me that the distinction that http://arxiv.org/abs/0710.0515 makes between Eext and Eind could be essential for a good understanding. I assume that Eext refers to externally applied electric field strength and Eind to induced electric field strength.
I have been through the whole paper, but not in super detail, but I couldn't find Eext and Eind. I apologize if I need new glasses, but could you point out which part you are referring to. I did find a reference to jext and jind, could that be what you are referring to?

However, I don't understand your point in general. The paper you cited re-derives and confirms the usual microscopic Poynting formula. It is not a contradiction of the standard formula, but a confirmation of it. In particular, he explicitly contradicts your idea that the E.j term represents only Ohmic heating.

Last edited:
Q-reeus
...While definitely work is done under displacement, it suddenly strikes me that here the term E.j can only be zero. How does Poynting's theorem look like for this particular case?

At first sight, following eq.1.31 of "bobs" handout, it looks to me that the energy per volume that is extracted from the magnetic field for doing work equals M.H for the case without external source (like a permanent magnet).

Is that correct?
Against my better judgement, dive into imo cut to the chase, and offer yet another reference: http://arxiv.org/abs/1208.0873 - see esp. sect. 8. Quite along the lines I argued 'back there'. (another paper by that author generated quite some flack awhile back, but that was a different paper.)

harrylin
I have been through the whole paper, but not in super detail, but I couldn't find Eext and Eind. I apologize if I need new glasses, but could you point out which part you are referring to. I did find a reference to jext and jind, could that be what you are referring to?
You are right - so where the one text makes the distinction in the J, the other makes it in the E!
I interpret E.J as a whole, in which case it doesn't matter. Does it matter? If so, why/how?
However, I don't understand your point in general. The paper you cited re-derives and confirms the usual microscopic Poynting formula. It is not a contradiction of the standard formula, but a confirmation of it. In particular, he explicitly contradicts your idea that the E.j term represents only Ohmic heating.
I just discovered that there are at two E.j terms that are distinguished; it's thus important to know what someone means with "the" E.j term, and it may well be that some discussions (+derivations) only consider one of the two. What does Ohmic heating correspond to?

PS. my point in general here is that "Poynting", at least the generalised version, must be consistent with any direct case analysis such as the one to which I referred in my second question. And, as I mentioned, this seems to work fine. If so, then there remain only a few details about Poynting's theorem to clarify for me.

Last edited:
Mentor
You are right - so where the one text makes the distinction in the J, the other makes it in the E!
I interpret E.J as a whole, in which case it doesn't matter. Does it matter? If so, why/how?
I don't know. In principle if you had Eind and Eext and jind and jext then you could have up to 4 distinct E.j terms. So it is just good to be clear what you are talking about.

I just discovered that there are at two E.j terms that are distinguished; it's thus important to know what someone means with "the" E.j term, and it may well be that some discussions (+derivations) only consider one of the two.
The author of this paper shows that "the" E.j term is the sum of the ind and ext terms. That is discussed in the paragraph immediately preceding eq 10. I think you are correct that some derivations only discuss one (implicitly assuming the other is 0), but it appears that the j in Poynting's (microscopic) theorem is equal to the sum of both in general.

harrylin
I don't know. In principle if you had Eind and Eext and jind and jext then you could have up to 4 distinct E.j terms. So it is just good to be clear what you are talking about.
Yes, it would be good if anyone can clarify if there are two or even four E.j to distinguish.
The author of this paper shows that "the" E.j term is the sum of the ind and ext terms. That is discussed in the paragraph immediately preceding eq 10. I think you are correct that some derivations only discuss one (implicitly assuming the other is 0), but it appears that the j in Poynting's (microscopic) theorem is equal to the sum of both in general.
I understood from the summary in Wikipedia that the generalization was done later; OK I'll have a second look at that (later!).

harrylin
Against my better judgement, dive into imo cut to the chase, and offer yet another reference: http://arxiv.org/abs/1208.0873 - see esp. sect. 8. Quite along the lines I argued 'back there'. (another paper by that author generated quite some flack awhile back, but that was a different paper.)
Thanks, that's an interesting paper.
Sect. 8: "a static E-field [..] acts on the loop current". That is very different from the ideal permanent magnet case that I discussed, but these examples may still provide guidance. So I'll have look at that (later).

Q-reeus
Thanks, that's an interesting paper.
My pleasure. And not only interesting, but more directly relevant than the other articles earlier cited.
Sect. 8: "a static E-field [..] acts on the loop current". That is very different from the ideal permanent magnet case that I discussed, but these examples may still provide guidance.
Depends what you mean by ideal permanent magnet. If you mean one composed of perfectly conducting loop currents (that was your model?), just remember a permanent magnet made from such could not be a permanent magnet! A real PM is necessarily of radically different nature. In addition to sect. 8, consider carefully the last two paragraphs in sect. 3, all of sect. 9, and sect. 11.
[It has occurred to me your main concern is possibly the term 'static E-field' - implying an electrostatic field. No - just a poor choice of words there. Further down it is made clear the field has non-zero curl owing to association with a time-changing B. So evidently the sense was 'a steady-state curl E', implying a constant ramp current dI/dt as source of applied E, which can be true only for a limited time span.]

The main value of that article is in it's confronting the issue of just how different in character is the 'bound current density' Jm = 1/μ0∇×M to that of the other two - free current Jfree and bound polarization current Je_bound. Latter two are real currents and respond as such to an applied E, while the former is not, and it shows as admirably demonstrated in the article. Something Jackson and I presume Griffiths fail to do. I do not agree with the author that hidden momentum is an absurdity. It's quite real but using his preferred formalism it is 'hidden from view'. 'Hidden energy' is an interesting issue all by itself and there is a further aspect not covered by the author there. But let's not go too far astray here.

Last edited:
harrylin
My pleasure. And not only interesting, but more directly relevant than the other articles earlier cited.

Depends what you mean by ideal permanent magnet. If you mean one composed of perfectly conducting loop currents (that was your model?),
As you certainly remember, I calculated the classical case of two single zero Ohm current carrying loops as it is the minimal configuration of two permanent magnets.
just remember a permanent magnet made from such could not be a permanent magnet!
I cannot remember that for the simple reason that I have never seen such a claim in a classical physics textbook (and not even in Quantum physics either).
A real PM is necessarily of radically different nature. In addition to sect. 8, consider carefully the last two paragraphs in sect. 3, all of sect. 9, and sect. 11.
[It has occurred to me your main concern is possibly the term 'static E-field' - implying an electrostatic field. No - just a poor choice of words there. Further down it is made clear the field has non-zero curl owing to association with a time-changing B. So evidently the sense was 'a steady-state curl E', implying a constant ramp current dI/dt as source of applied E, which can be true only for a limited time span.]
Once more: I found that an E-field in a circle-symmetric closed loop configuration is an impossibility. If you disagree, please try to sketch one.
The main value of that article is in it's confronting the issue of just how different in character is the 'bound current density' Jm = 1/μ0∇×M to that of the other two - free current Jfree and bound polarization current Je_bound. Latter two are real currents and respond as such to an applied E, while the former is not, and it shows as admirably demonstrated in the article. Something Jackson and I presume Griffiths fail to do. I do not agree with the author that hidden momentum is an absurdity. It's quite real but using his preferred formalism it is 'hidden from view'. 'Hidden energy' is an interesting issue all by itself and there is a further aspect not covered by the author there. But let's not go too far astray here.
I'll certainly look at it in more detail, thanks!

Q-reeus
As you certainly remember, I calculated the classical case of two single zero Ohm current carrying loops as it is the minimal configuration of two permanent magnets.
Thanks for clarifying - yes I do remember that one. I disagree that it is a proper representation of two PM's - rather that of two electromagnets, and there is an important difference. As I explained in effect beginning with the #5 post back there and as Mansuripur does in the linked article - energy exchanges are totally different in the two cases. In the electromagnets case, in order to simulate genuine PM's, a source of emf in series with each loop current must act to keep the currents constant - and that means pumping energy into or our out of those coils, something PM's do 'for free'. I want to be clear here the series emf's (via batteries or similar) are exactly countering that owing to the -dA/dt emf induced in each coil owing to relative motions between them. We assumed perfect conductivity so no emf is consumed overcoming ohmic resistance.
Q-reeus: "just remember a permanent magnet made from such could not be a permanent magnet!"
I cannot remember that for the simple reason that I have never seen such a claim in a classical physics textbook (and not even in Quantum physics either).
Possibly because they are not concerned with the issue as it has arisen here and particularly in that other thread. We are distinguishing between acting as a source of EM field (and a current loop will do for that purpose - provided unrealistic constraints are imposed), and responding to such fields (very different between classical loop current and intrinsic moment). I made the point explicitly re impossibility of perfect loop currents being able to respond as though intrinsic moments in a number of posts, but this one is good enough. Lenz's law and thus intrinsic diamagnetism applies to classical loop currents - it does not and cannot do so in the case of genuine intrinsic moments. Putting this around - are you aware of any serious text on magnetism that claims it would be possible to create a PM out of classical loop currents? Also, have a look at:
sect'n. 12.2 under 'What spin is not' here: "folk.ntnu.no/ioverbo/TFY4250/til12eng.pdf" [Broken]
Sect'n. 23.1 here: "web.mit.edu/sahughes/www/8.022/lec23.pdf" [Broken]
Once more: I found that an E-field in a circle-symmetric closed loop configuration is an impossibility. If you disagree, please try to sketch one.
I'm frankly amazed you believe that! No need to sketch. Transformer action! That would be ∇×E=-∂B/∂t, or it's close cousin E = -∂A/∂t. You are aware lines of A tend to hug the contours of a conducting wire current-carrying circuit? A little closer to the issue here; if one moves a cylindrical bar magnet along it's axis, there is induced perfectly circular closed-loop lines of E, centred on the bar axis. That's why an emf appears in a coil when a bar magnet is moved in-and-out of the coil. I'm starting to see why my earlier inputs, over in that other thread were not particularly impressing you!
[this is the best link re intuitive feel I can find just now, on the link between lines of E associated with motion of a magnet and associated B field: http://hyperphysics.phy-astr.gsu.edu/hbase/electric/farlaw.html]

Last edited by a moderator:
harrylin
Thanks for clarifying - yes I do remember that one. I disagree that it is a proper representation of two PM's - rather that of two electromagnets, [..]pumping energy into or our out of those coils [..]
I called case 1 an electromagnet (again the simplest one that I could imagine); an electromagnet has open windings with an external source connected to them. However, the added complexity of pumping energy from outside into the magnetic fields could distract from the basics of "Poynting" that I brought up here. It's irrelevant for the discussion what you call case 2, as long as you understand what case 2 is - but even that appears not to be the case, see next!
[..] are you aware of any serious text on magnetism that claims it would be possible to create a PM out of classical loop currents? [..]
Yes of course: assuming that you are like me discussing classical physics, I already linked to Ampere's classical model (no Quantum here please!). And even your references pretend that they are magnetic, which is all that matters here.
I'm frankly amazed you believe that! No need to sketch. Transformer action! [..]
I only needed to make the sketch in my head - it took a mere 2 seconds. A transformer has open loops like an electromagnet, which breaks the symmetry. Why did you do such a big effort to not make a simple sketch of an E-field along a perfect loop without openings? You'll see that you can't do it, just as you can't sketch an ever increasing time zone shift around the Earth, or stairways with the top connecting to the bottom (you could if your name is Escher!).

And in order to reduce noise, for your sake I'll rephrase my post #11 as follows (this discussion did not progress for me since that post!):

Thanks, that's an interesting paper.
Sect. 8: "a static E-field [..] acts on the loop current". That is very different from the stand-alone closed current loop case 2 that I discussed, but these examples may still provide guidance. So I'll have look at that (later).

Last edited:
Q-reeus
I called case 1 an electromagnet (again the simplest one that I could imagine); an electromagnet has open windings with an external source connected to them. However, the added complexity of pumping energy from outside into the magnetic fields could distract from the basics of "Poynting" that I brought up here. It's irrelevant for the discussion what you call case 2, as long as you understand what case 2 is - but even that is not the case, see next!
Went back to that #216, to the case 2 there, and honestly it never quite made sense then or now - possibly just the grammatical construction. Here it is again:
2. For a true permanent magnet there is no external current source...
No argument here. But then...
Following Lenz-Faraday-Maxwell we find that the motion will induce an electric field in each wire that opposes the motion by reducing the currents and the corresponding magnetic fields...
Where the hell do wires suddenly enter? Aren't we talking about PM's?! Or have you switched back to case 1 unannounced? I guess so. In which case, yes to that point.
Thus the magnetic fields deliver the energy of the magnetic forces that drive the external application.
Mechanical equivalent: A spring that gives off stored energy to drive a clock.

Consequently for the case of a real magnet, not only according to [1] and [2] but also according to my personal definition of "work" the magnetic field "does work", as it loses energy by providing the acting force that does work."
Getting the hang now of your casually switching back-and-forth between Cases 1 and 2, I get your picture sort of: Work is magnetic because of the magnetic field energy changes. True to a point but only part of the picture. As said last time, energy exchanges are different between electromagnet case (genuine E.j work is involved), and true PM's (no or negligible actual E.j work involved).
[..] are you aware of any serious text on magnetism that claims it would be possible to create a PM out of classical loop currents? [..]

Yes of course: assuming that you are like me discussing classical physics, I already linked to Ampere's classical model (no Quantum here please!). And even your references pretend that they are magnetic.
But the use of Amperian loop currents in all such serious texts make it plain they are not to be considered what really applies. I refer you to the links supplied last time.
I'm frankly amazed you believe that! No need to sketch. Transformer action! [..]

I only needed to make the sketch in my head - it took a mere 2 seconds. A transformer has open loops like an electromagnet. Why did you do such a big effort to not make a simple sketch of a perfect circle without openings? You'll see that you can't do it, just as you can't sketch an ever increasing time zone shift around the Earth (well maybe you could if your name is Escher, but that's cheating!).
Did you check out that last HyperPhysics link I gave? Please think about the examples shown there - in particular involving axial motion between coil and bar magnet. What then is your idea of the E field generated when a cylindrical bar magnet moves along it's axis of symmetry? How could there be other than concentric circular E field lines?

On the matter of transformers having 'open' windings; sure there must be an opening somewhere in the primary if a current is to be driven through it. And an opening somewhere in the secondary if voltage/current/power is to be extracted. But that matters? Lets' suppose for the moment there is something special about that opening that makes it different to the rest of the circuit. There can be 100, 1000, 10,000 turns or any number N turns in principle. As N goes to a large value, how relevant is that there is some 'opening'? Clearly it's relative importance diminishes towards zero. But beyond that, the 'opening', unless a genuine open-circuit (irrelevant because then there is no current to consider) is in fact a continuation of the circuit. Current flows through the opening (as source - generator etc., or load - resistor etc.) to complete the circuit. If there is any current, there is always a complete circuit. Well there is the exception of open geometry conducting objects driven at or near self-resonance being able to carry appreciable currents, but that is far from what we are talking about here. At low frequencies, and moving magnets certainly qualify there, all current-carrying circuits are closed ones. [What's more, even a single turn coil can be easily formed in such a way that the input/output feed wires have no appreciable EM effect - use a twisted-wire configuration or similar; e.g. coaxial.]
[just caught your edit - yes, please give that article by Mansuripur some further thought.]

Last edited:
Looking at this in a possibly simplistic way I see E.j as being a more detailed variation of the power equation (VI) but being representative of power per unit volume rather than just power.By considering it as a power equation and one which expresses the conservation of energy I find it easier to apply it to real situations.
Consider the power equation (VI) as applied to a basic motor.At switch on and just before the motor starts to turn Ohms law applies and we can write:

Vs=IR (Vs= supply voltage)

As the motor turns and picks up speed an increasing back EMF(Vb) and reducing current(I') results and for any value of Vb and corresponding current I' we can write

Vs-Vb=I'R (R= circuit resistance) From this:

VsI'-VbI'=I' squared R

Ignoring other energy losses eg those due to friction,the above equation expresses the conservation of energy quite nicely.VsI' represents input power VbI' represents mechanical output power and VsI'-VbI' represents resistive power losses.
If the equation is reformulated in terms of E.j it is easy to see that E and j need to be properly defined.Taking E,for example,there would be an E value which corresponds to the supply voltage,an E value which corresponds to back EMF voltage and an E value which corresponds to the difference between the two and which is instrumental in heating losses.
At present I look at E.j as being an equation which is just one link in an energy chain.I don't consider it as being a prime source of energy,if there is such a thing,because energy changes can be traced back in time.

Last edited:
Q-reeus
At present I look at E.j as being an equation which is just one link in an energy chain.I don't consider it as being a prime source of energy,if there is such a thing,because energy changes can be traced back in time.
Yes of course that is true and Poynting theorem makes that explicitly and implicitly plain. But the seemingly tricky part of interest here arises in interpreting the meaning and application of the 'J' in E.J that applies to the case of so called Amperian currents in magnetic media, in particular permanent magnets. Although the character of the E part has also become under scrutiny of late, but hopefully not for long.

Per Oni
Q-reeus
He he he. Do we all pine for the days of childhood when playing with magnets brought a sense of wonder and awe, if not outright pleasure? How it all changes.
How true. Now it appears that those funny little pieces of metal are least understood of all. They probably took us to study magnetism/electricity in the first place! I also realize how little I know about them and how little there’s on the net.

For example say we have 2 permanent magnets a good distance apart. We then let them collide. The only thing I know for sure is that the combined energy content of the magnets has gone down. But eg what happened to: the flux φ, field strength H, flux density B, inductance L, possible current I, possible spin magnetic moment?

On the other hand, I am fairly confident I know what happens to their electrical equivalents in case of 2 opposite charged metal plates. I suppose lack of understanding is why harrylin started this thread in the first place.

Q-reeus
For example say we have 2 permanent magnets a good distance apart. We then let them collide. The only thing I know for sure is that the combined energy content of the magnets has gone down. But eg what happened to: the flux φ, field strength H, flux density B, inductance L, possible current I, possible spin magnetic moment?
Good questions. A few quick takes: Maximum net flux, as defined as the maximum number of field lines moving in a given general direction within the total system of two magnets, will be greater after colliding than before. As an equivalent solenoid lengthens = magnets join together, so the maximum internal field grows. So therefore greater maximum values of H and B also. But this is over a reduced effective net volume, hence the total magnetic field energy has diminished, the difference chiefly being in mechanical energy extracted, with a generally tiny additional eddy current component. If there are just permanent magnets involved, inductance doesn't apply. The only appreciable 'induced currents' will be of the fictitious 1/μ0∇×M kind - and that assuming each magnet was less than fully magnetized to start with. As for spin magnetic moments, intrinsic spin moments are invariant and rigidly aligned within each of the many magnetic domains comprising each magnet. I'm no expert on the complexities of all possible spin-orbit interactions, but empirically, spin contribution is overwhelmingly dominant in ferromagnets, and once saturation (full domain alignment) occurs, no more can be appreciably wrung-out. Must go. :zzz:

harrylin
Went back to that #216, to the case 2 there, and honestly it never quite made sense then or now - possibly just the grammatical construction. Here it is again: [..]
Now I understand what went wrong. The context is a wire loop. I cite (with added emphasis in bold, and corrections):

"[..] For the following analysis I will restrict myself to mere high school classical physics.

Perhaps the simplest case to model two permanent magnets is two single current loops 1 and 2. Let's assume current loops of 1 m circumference, made of 1 mm diameter wire; distance d between the loops ca.1 cm (diameter << d). As pictured, the currents flow through the wires (---) into the screen on the left (x), and flow out of the screen on the right (0):

1 x----0

2 x----0

Here the wires are attracted to each other by the Lorentz force: F = IxB . L

However, for calculation we can approximate this particular example with two straight wires of 1 m length by opening the loops above the screen and straightening them out:

1 --------->--------- I1

2 --------->--------- I2

Now we have the basic configuration of the definition of the Ampere:

F = 2.10-7 I1 I2 L / d

For this case I obtain F= 2.10-7 * 100 * 1 / 0.01 = 2 mN.

Suppose that the top magnet is connected to an external mechanism that is actuated by it. We allow the top magnet to move towards the bottom magnet over a small distance, for example from 11 mm to 9 mm. We will assume that the average currents are approximately 1 A (more follows).

Then the work done by the Lorentz force on the top magnet and whatever it is driving, according to [1] and [2]:

F.d ≈ 2.10-3 * 2.10-3 J = 4 μJ.

[..]

In that context, we can distinguish two main cases for the current flow.

1. A common electromagnet, let's say with a current source [...]

2. For a [STRIKE]true [/STRIKE]sourceless [STRIKE]permanent [/STRIKE]non-fading magnet there is no external current source.
Following Lenz-Faraday-Maxwell we find that the motion will induce a[STRIKE]n electric field [/STRIKE]EMF in each wire that opposes the motion by reducing the currents and the corresponding magnetic fields.[..]
"

Note the mistake which I now corrected: contrary to case 1, in case 2 the loop must be closed. I'll repeat once more: I found that an E-field along a circle-symmetric closed loop configuration is an impossibility. If you disagree, please try to sketch one with an E-field.

I checked out the Hyperphysics site 10 days ago while writing my above example. And I hope to find time in the coming week for reading more related to Poynting.
sure there must be an opening somewhere in the primary if a current is to be driven through it. [..]
Sorry, a closed wire loop has no opening, and "a current driven through" corresponds to case 1. Making it more complex adds noise to the discussion, and in this discussion I strive for a high S/N ratio.

BTW I see that you want to discuss a more complex matter related to J, but I think that that is contra productive as long as this simpler case related to E is not understood first.

Last edited:
Q-reeus
Note the mistake which I now corrected: contrary to case 1, in case 2 the loop must be closed. I'll repeat once more: I found that an E-field along a circle-symmetric closed loop configuration is an impossibility. If you disagree, please try to sketch one with an E-field.
Ah; at last it becomes clear (I think). Sorry about previous confusion - a case of words getting in the way of what was meant. So you were not denying the possibility of generating a circular E field, just saying the emf in a shorted perfectly conducting loop is necessarily zero - right? Well sure then, no argument! And if you look at my #61 from back there, that same point was made clearly then - Lenz's law applied and one has perfect diamagnetism. But it's from that point on that we seem to fundamentally differ. I went on to argue there that such diamagnetic behavour is the very antithesis of permanent magnetism, which has no respect for Lenz's law at all!

Anyway if I now get your main point of 'zero emf in Amperian loop current' to rephrase it, you are saying there is thus zero E.J type work being done owing to E being zero everywhere in those loop currents. And thus all the work must be magnetic. That's it in a nutshell? Well I agree that is the case if such Amperian loop currents actually existed in permanent magnets. But in actuality the overwhelmingly dominant source of B field in ferromagnetism is intrinsic moments, and their character is opposite to the diamagnetism of an Amperian loop current. Ferromagnetism and diamagnetism are opposites!

I dug up those links in #14 for a reason. Let me now quote from both. First, from 12.2, p10 in http://folk.ntnu.no/ioverbo/TFY4250/til12eng.pdf
What the spin is not
Uhlenbeck and Goudsmit based their spin hypothesis (in 1925) on the classical notion of a rotating electron, with a certain mass and charge distribution. Modern scattering experiments have shown, however, that the size of the electron, if it differs at all from zero, must be smaller than 10-18m. It has also turned out to be impossible to construct a classical model with a mass and charge distribution that reproduces the spin and the magnetic moment of the electron. Thus the electron behaves as a point particle, and we have to state that the spin and the magnetic moment (with ge ≈ 2) of this particle can not be understood as the result of any kind of “material rotation” which can be pictured classically and which can be described in terms of a wave equation and a wave function. The latter is only possible for the orbital angular momentum, for which l can take only integer values, while the spin quantum number s can also take half-integral values, depending on which particle we are looking at. Note that for a given particle species s is completely fixed. Thus for the electron he intrinsic angular momentum has no choice, it has to be |S| = h/(2π)√(3/4), in contrast to the orbital angular momentum which can vary, even if it is quantized. Again we see that the spin does not behave as we would expect for an ordinary rotational motion.
And from 23.1, p214, and 23.1.2, p214-5 in http://web.mit.edu/sahughes/www/8.022/lec23.pdf
One reason we have avoided covering this subject is that it is not really possible to discuss it properly within getting into a detailed discussion of the quantum mechanical description of matter. The way in which matter responds to magnetic fields is totally determined by the quantum mechanical nature of their molecular structure, particularly their electrons...
Intrinsic magnetic moment of the electron
One other quantum mechanical property of electrons plays an extremely important role in this discussion: electrons have a built-in, intrinsic magnetic moment. Roughly speaking, this means that each electron all on its own acts as a source of magnetic field, producing a dipole-type field very similar to that of current loop. Because this field is associated with the electron itself, it does not exhibit the Lenz's law type behavior of the field that we see from the orbits.
And if you care to study it, that article by Mansuripur only reinforces and expands on this greatly. I emphasized all this from the start back in that other thread, but it collectively went in one ear and out the other. So with this emphasized contrast between intrinsic moments and classical loop currents - given by authority figures - it's a case of accept or reject, but don't complain the issue has not been plainly stated for all to see. :grumpy:
Sorry, a closed wire loop has no opening, and "a current driven through" corresponds to case 1. Making it more complex adds noise to the discussion, and in this discussion I strive for a high S/N ratio.
We now have your intended meaning straight I believe, so that bit is now moot. But please Harald - in the interests of that low S/N ratio objective, strive for clarity!
BTW I see that you want to discuss a more complex matter related to J, but I think that that is contra productive as long as this simpler case related to E is not understood first.
Said my piece on that above. Your simpler case is fine - just as long as there is no confusing that diamagnetic loop(s) scenario with the situation for real PM's - which as per quoted text - in respect of magnetic response are wholly governed by quantum mechanics. That has to be faced.

harrylin
Oops I made a colossal blooper here:
[..] contrary to case 1, in case 2 the loop must be closed. I'll repeat once more: I found that an E-field along a circle-symmetric closed loop configuration is an impossibility. If you disagree, please try to sketch one with an E-field. [..]
As I realize now, that challenge was based on a secondary definition which does not hold here; the electrical potential approach with a field from high to low potential simply isn't valid for such a case. In other words, my correction was wrong! Sorry for the confusion
Ah; at last it becomes clear (I think). Sorry about previous confusion - a case of words getting in the way of what was meant. So you were [..] just saying the emf in a shorted perfectly conducting loop is necessarily zero - right? Well sure then, no argument [..]
Regretfully I brought you off track, as there certainly is not only an EMF but consequently also an E-field according to the fundamental definition of electric field: https://en.wikipedia.org/wiki/Electric_field

Once more my excuses - but making that blooper and correcting it was useful for me, and may be useful for some others.

Last edited:
harrylin
Looking at this in a possibly simplistic way I see E.j as being a more detailed variation of the power equation (VI) but being representative of power per unit volume rather than just power.By considering it as a power equation and one which expresses the conservation of energy I find it easier to apply it to real situations.
[..]
If the equation is reformulated in terms of E.j it is easy to see that E and j need to be properly defined.Taking E,for example,there would be an E value which corresponds to the supply voltage,an E value which corresponds to back EMF voltage and an E value which corresponds to the difference between the two and which is instrumental in heating losses.
At present I look at E.j as being an equation which is just one link in an energy chain.I don't consider it as being a prime source of energy,if there is such a thing,because energy changes can be traced back in time.
Yes, thanks for the elaboration. I also read an article about VI which highlighted the complexity of it all, and helped to understand it better.

harrylin
[..] I suppose lack of understanding is why harrylin started this thread in the first place.
That is very right. In the thread on magnets, some claims were made concerning Poynting's theorem that appeared to contradict my analysis there. However, that analysis was based on the fundamental equations on which also Poynting's theorem was based. I suspect that the disagreement was due to a misapplication of Poynting, and I would like to get this straightened out.

Q-reeus
Regretfully I brought you off track, as there certainly is not only an EMF but consequently also an E-field according to the fundamental definition of electric field: https://en.wikipedia.org/wiki/Electric_field
OK, thanks for that further correction - I do value such honesty much :!). That last link is a good one but deficient in one respect that is crucial to our consideration of Amperian loop currents. It considers the effect of changing threaded flux on a loop whose self-impedence is essentially resistive, not inductive. That means induced loop current is directly in phase with the driving emf, and there is a net dissipative power εI generated (ε being the emf in the loop). That cannot apply in the case of an ideal zero-resistance Amperian loop, which must be governed by a purely inductive self-impedence. Then it is the time-rate-of-change of induced current that is in phase with driving emf. Then induced emf and driving emf exactly cancel, satisfying the condition that E.J in any such loop is always zero. Which in that setting is equivalent to saying the loop exhibits perfect diamagnetism. Moving magnets always generate emf's, and the static/quasi-static E field owing to the scalar potential -∇phi is not relevant to our considerations if permanent magnet interactions are the focus. (an emf can certainly give rise to electrostatic fields - e.g. time-changing flux threading an almost-closed conducting loop terminated by a capacitor.)
Once more my excuses - but making that blooper and correcting it was useful for me, and may be useful for some others.
If only everyone here was so honest. YouTube can be a great resource in getting some 'visual intuition' in things like basic EM theory. Take your pick from this smorgasbord:
[Apologies for the atrocious intro to some of those lectures - my turn to be honest and admit to not checking those first ones out properly. Still - the content is accurate.]

Last edited by a moderator:
PhilDSP
If there are just permanent magnets involved, inductance doesn't apply. The only appreciable 'induced currents' will be of the fictitious 1/μ0∇×M kind - and that assuming each magnet was less than fully magnetized to start with. As for spin magnetic moments, intrinsic spin moments are invariant and rigidly aligned within each of the many magnetic domains comprising each magnet.

Hmmm now. First of all I need to apologize for not having the time to have followed this thread or to examine the articles mentioned (and likely for not having time henceforth either).

But I'm curious about what you're considering "fictitious". It looks like the term 1/μ0∇×M might represent displacement current. If so, then it probably shouldn't be considered "fictitious".

Q-reeus
Hmmm now. First of all I need to apologize for not having the time to have followed this thread or to examine the articles mentioned (and likely for not having time henceforth either).

But I'm curious about what you're considering "fictitious". It looks like the term 1/μ0∇×M might represent displacement current. If so, then it probably shouldn't be considered "fictitious".
That term does not imply time variation and is present in a purely magnetostatic situation. Say we have a uniformly magnetized cylindrical bar magnet. The only non-zero Curl M here occurs at the cylindrical surface, where as a function of radius from the bar axis, M plummets step-wise from it's constant interior value to zero just outside the surface. This represents the so-called Amperian surface current, which purely formally then provides the justification for claiming E.J type 'work' can be done on a permanent magnet. However, there is no actual current, and what's more it does not respond to an induced E = -dA/dt as would a physically real current. I refer you to the linked article given back in #7, and as further expanded in #12. To avoid the absurdities pointed out there, we must distinguish physically unrealistic implications from indiscriminately applying formalisms, to the actual situation applying.

PhilDSP
Okay, thanks. Will need to dig into this in more detail some time later.

Per Oni
Good questions. A few quick takes: Maximum net flux, as defined as the maximum number of field lines moving in a given general direction within the total system of two magnets, will be greater after colliding than before. As an equivalent solenoid lengthens = magnets join together, so the maximum internal field grows. So therefore greater maximum values of H and B also. But this is over a reduced effective net volume, hence the total magnetic field energy has diminished, the difference chiefly being in mechanical energy extracted, with a generally tiny additional eddy current component.
I’m not sure about the net flux statement, but I’d like to start a new thread in a while about this subject. For now it’s perhaps diverting attention a bit too far away from the op.

Q-reeus
I’m not sure about the net flux statement, but I’d like to start a new thread in a while about this subject. For now it’s perhaps diverting attention a bit too far away from the op.
Fine. And I should add here that some of my statement there is not strictly correct - conflicting with my #233 back in that other thread. But to elaborate fully get's into some vexed territory. Let's just say that the strictly accurate statement involving energy changes is that (mechanical+any heat) net energy change = negative of net change in -m.B magnetic potential. Field energy change considerations introduce certain problems!

Darwin123
In several recent threads the Poynting's theorem was brought up, and the discussion there became a distraction from the questions at hand without really being solved:
https://www.physicsforums.com/showthread.php?t=628896 (this post continues from the discussion starting with post #198)

Roughly, the theorem states (I think) that the energy that dissipates as heat inside a volume equals the energy that enters it minus the stored energy that is extracted from it.

A basic issue is the meaning there of the term E.j.
According to some, the energy transferred from EM fields to matter (work) is E.j.
However, the texts that I consulted specify that it is the Ohmic heating rate, or the dissipated energy inside the volume under consideration. Clearly there is some ambiguity here.

A paper that appeared a few years ago in Europhysics Letters 81 (6): 67005 (thanks Wikipedia!)

Clarifications are welcome!
Once again with feeling!
The word heat is ambiguous. It has different meanings in different contexts.
The "Ohmic heating rate" is the rate at which the internal energy of the system is changing. It has nothing to do with "heat" as defined by the "energy carried by entropy".
The energy transferred from EM field to matter is work. When work is done on the resistor, the internal energy of the resistor is changing.
If the resistor is kept at a constant temperature, heat conduction and work are being done at nearly the same time. The energy transferred by E.j is work, which is immediately converted into internal energy. This internal energy is immediately conducted out of the resistor. The energy conducted out of the resistor is heat, in the sense that it is associated with energy.
The word heat sometimes means entropy time temperature, and sometimes means internal energy. You were fooled by the phrase "Ohmic heating rate". They really meant "rate that the internal energy would be changing in there was no thermal conduction."
The four laws of thermodynamics are usually expressed in such a way that "heat" means "energy carried by entropy". In the usual statement of these laws, "heat" does not mean "internal energy". However, the word "heat" is often misused to refer to types of work.
Consider the mathematical expression E.j. In terms of the laws of thermodynamics, E.j is work. In terms of the laws of thermodynamics, E.j is not heat. "Ohmic heating" is sloppy physics jargon which can be confusing. "Ohmic heating" is defined as E.j.

Darwin123
In several recent threads the Poynting's theorem was brought up, and the discussion there became a distraction from the questions at hand without really being solved:
https://www.physicsforums.com/showthread.php?t=628896 (this post continues from the discussion starting with post #198)

Roughly, the theorem states (I think) that the energy that dissipates as heat inside a volume equals the energy that enters it minus the stored energy that is extracted from it.

A basic issue is the meaning there of the term E.j.
According to some, the energy transferred from EM fields to matter (work) is E.j.
However, the texts that I consulted specify that it is the Ohmic heating rate, or the dissipated energy inside the volume under consideration. Clearly there is some ambiguity here.

A paper that appeared a few years ago in Europhysics Letters 81 (6): 67005 (thanks Wikipedia!)

Clarifications are welcome!
I don't think the word "heat" even belongs here. Maxwell's equations don't explicitly mention either temperature or entropy. In fact, there is nothing in Maxwell's equations that make it necessary for a system to reach equilibrium.
There is no scale length parameter in electromagnetic theory that can distinguish between internal energy and kinetic energy. Maxwell's equation are deterministic equation that don't involve any random variable. So there is no "heat" in Poynting's theorem.
The phrase "Ohmic heating" is deceptive because it implies that there is some random variables associated with
Thermodynamics comes into electrodynamics through the constitutive relations. Constitutive relations describe the properties associated with the substances. The constitutive relations have constitutive parameters like dielectric constant, permeability constant, and conductivity. The microscopic dynamics of the material determine the constitutive equations. It is here where thermodynamics is coupled to electrodynamics.
The article link mentioned by the OP tries to separate the mechanical energy of the particles from the electromagnetic field energies. However, it doesn't even try to include entropy and temperature. Without these quantities, it makes no sense to separate heat from work.
The microscopic parameters of the materials determine both the equation of state and the constitutive relations. Therefore, "heat" has to be discussed in terms of the constitutive relations. If you want to figure out how much the system was heated, you have to be careful about how you define the constitutive parameters. If you are not careful, your analysis of electrodynamics could violate a Law of Thermodynamics.
For example,the relationship between the electric field and the electron current is given by Ohm's Law, which is,
j=σE,
where σ is the conductivity of the material, j is the current density and E is the electric field.
The work done by the electric field is j.E. So the work, W, done by the electric field is,
W=σE^2
Note that in this case the work is rate of change of the internal energy. So,
dU/dt=σE^2
where U is the internal energy.
Basically, the atomic level dynamics of the electromagnetic system is characterized by the constitutive relations and the Lorentz force Law. These have to expressed in forms that don't violate the equations of state for your materials.

harrylin
[..] That cannot apply in the case of an ideal zero-resistance Amperian loop, which must be governed by a purely inductive self-impedence. Then it is the time-rate-of-change of induced current that is in phase with driving emf. Then induced emf and driving emf exactly cancel, satisfying the condition that E.J in any such loop is always zero. [..]
Good find - but I don't follow your conclusion. Probably with "driving emf", you refer to the emf that is induced by the changing magnetic field, and with "induced emf" you refer to the back-emf from self-induction. I don't think that those can be equal in magnitude, for what else but an electric field could drive the necessary current decrease? I recall that the magnetic field energy is proportional to I2, and this has to diminish when energy is taken from it.

Last edited:
Q-reeus
Could find - but I don't follow your conclusion. Probably with "driving emf", you refer to the emf that is induced by the changing magnetic field, and with "induced emf" you refer to the back-emf from self-induction.
Yes.
I don't think that those can be equal in magnitude, for what else but an electric field could drive the necessary current decrease?
Yes It is the applied E field (owing to relative motion of other magnet in our preferred scenario) that drives the current, but equally it is the ramping current that creates the back emf exactly cancelling applied E. For a perfect conductor - and that is the necessary model for an Amperian loop, the necessary boundary condition is zero tangent component of E at the surface or interior. Which can only be satisfied if Eapplied + Eback-emf = 0 everywhere in the conductor, including the surface. And as stated before, this is completely compatible with Faraday's law applied to a purely inductive loop -> perfect diamagnetism.
I recall that the magnetic field energy is proportional to I2, and this has to diminish when energy is taken from it.
Yes, but where is the conflict?