"Veritasium" (YT) - "The Big Misconception About Electricity" ?

[snarkysparky]

The line integral of the electric field around any closed path is equal to the voltage difference between the starting point and the stopping point.

Start at the battery positive terminal.  Travel along the wire of very low resistance.  The voltage drop along the wire is minimal.  So the Electric field vector gradient is minimal along the wire.  But get to the load with its resistance and the electric field that originated at the battery terminals is impressed there.   Work done on the load by current flowing along an electrical field gradient.

DC...  Energy transfer is ENTIRELY in the wire.

[Bud]

@EEVBlog I think your model is a fallacy.  You assume there is capacitive coupling between the wires. But lets eliminate capacitive coupling by inserting a shield between the wires, or use shielded wires with the shields soldered together all along the way and grounded. So ideally zero capacitive coupling. There will be no spike then at t=0, which proves the model is wrong.
Secondly, it was not necessary to go to the trouble of modelling a transmission line to demonstrate capacitive coupling, just draw two wires with a capacitor.

[bdunham7]

And mentioning that, the question is very deliberately set to 1m apart, because he knows damn well the capacitance will be practically zero so it's harder to justify his claim with that. When it's 1m apart you have some realistic numbers to work with.

I agree that he probably has been quite crafty in setting up the question.  And in addition, he has omitted all the values that we would need for any actual calculations, like the wire diameter and the characteristics of the 'light bulb' and the battery, although he does show those in the video.

However, there is a serious problem with modelling a transmission line as you have, with the capacitor first.  This seems to imply that the initial current is higher than the actual transmission line model will predict because the capacitor is intially a short circuit instead of the characteristic impedance of the line you are modelling.  And even in your model, if you extended it out with a larger number of elements, you should see the current taper off.  But you also appear to have modelled a 1-ohm transmission line, so it won't taper much.  1 pF would be a better capacitance value, and perhaps put a 1k resistor in series with the first capacitor.

In a model using a 1k transmission line and zero resistance wires, the initial current would be 1/2100A and that would continue for as long as the wires go.  But I still think we're missing something regarding the propagation time.

[ejeffrey]

With absolutely no discredit to Dave's response video, the transmission lines approach represents the length of the wires effectively and demonstrates the coupling from source to load is not relying on anything that cannot be modelled with lumped components and certainly doesn't require analysis of all the fields going on...

Well you use field equations to calculate the Ls and Cs. 

But... does it not highlight one weakness of the transmission line model in that it does not handle the lateral (1/c) delay properly? Only that of an infinitesimally thin structure that so happens to have the equivalent characteristics of something 1m wide? So it doesn't *actually* provide an answer to the multiple-choice, only that it's not 0.5, 1 or 2 seconds?

Yes, that is correct.  Lumped element models of transmission lines assume that the phase shift across the capacitor plates is zero. That's generally true enough for coax and twin lead for the intended excitation mode, but breaks down e.g. if you excite a non TEM00 mode of the coax or obviously for waveguides.  Waveguide modes still have characteristic impedance and propagation constants which you could turn into an LC circuit model but that model doesn't map to distinct conductors.

The other factor that I didn't see veritasium include is the common mode behavior.  For the geometry specified (1 meter separation), it's entirely possible that the capacitance to earth or other nearby conductors will be larger than the wire-to-wire capacitance at least until you get away from the planet.  In that case you have more like two weakly coupled microstrip waveguides and the lamp turn on will be delayed somewhat.

[ejeffrey]

In this example there is no steady state power flow in the space between the wires.   PERIOD.... 

Ver is completely wrong about this.

Poynting vector is for self propagating electromagnetic waves.   Not static waves.

Why would you say that?  You will definitely get the right answer if you integrate the Poynting vector across a surface, and the contribution from inside the wire will be nearly zero.  Obviously the transient behavior is determined by field equations, what is the point where that changes?  The answer is it doesn't.  It's just that KCL, KVL, and Ohms law (all derived from maxwell's equations with certain assumptions or approximations) are usually a lot more convenient for quasi-static analysis with no traveling waves.  But the field equations aren't wrong.  Good engineering is, as Dave said, knowing what models to use when, but models that neglect the field behavior will certainly break down in extreme situations such as presented in this video.
 

The electric field is preserved along the wire by its low resistance and shows up at the load to push electrons through the load.  Work done is  E*J     ( * is vector dot product   E is electric field and J is current density vector )

Exactly.  The electric field inside an ideal conductor is zero so the work done on the charges in the wire (E*J) is zero.  It's a different matter inside the load -- the load has a voltage drop across it so there is work done on electrons traveling through the load.  But for sure if you simulate this with a FEA solver and plot the energy flow it will show that it flows from the battery terminals mostly through free space to the load and then dissipates in the load with a negligible energy flow through the wires, even at DC.  Good engineering practice would generally not be to solve a problem like this with FEA, but it will certainly work.

[bsfeechannel]

Err, last I checked, engineering is an applied science.
It's why ohms law, Kirchhoff's laws, and countless other practical theorems were developed, so we didn't have to "go back to basics" and use Maxwell and Poynting for everything. Even conventional current flow is a thing for a reason.

As technology is being pushed to its limits, the traditional dumbed-down physics for the practical engineer is showing signs of exhaustion.

All the unanswered questions Mehdi poses in his interview on the corresponding Amphour episode for example can be readily answered with Maxwell’s equations and its corollaries like the Poynting Theorem.

We need to stop treating Maxwell’s equations and field theory as something to be avoided.

[bdunham7]

Exactly.  The electric field inside an ideal conductor is zero so the work done on the charges in the wire (E*J) is zero.

You're simply exchanging one imperfect, incomplete model for another.  The electric fields inside a conductor, even a superconductor, are not zero.  Your model just says they are because that mostly works out in the macro domain that it is intended for.  What goes actually goes on is more complex.  Quantum mechanics aside, I fail to see how you can continue to maintain the the energy flow (whatever that is) in the DC case is due to fields outside of the conductor without stating what those fields are. Poynting vectors are not fields. There is the magnetic field which is unchanging for DC current and then a static E-field where the conductor has a net charge.  Neither of those can do work on charges.  Now you can have a wrong or simplified model that still predicts at least some things correctly, so if you can have a battery on one end of a pair of wires and a load on the other and you have some diagram with some arrows that shows the E-field of the battery being magically transferred to the other end to do work on the load--whether it is Poynting vectors or monkeys with wheelbarrows--you still haven't explained how it came to be that the charge density is what it is at the load end of the wires.

[bdunham7]

We need to stop treating Maxwell’s equations and field theory as something to be avoided.

I'm not suggesting that we avoid or forget them, but what important advances in science or technology have been made recently using field theory?  Unless you continue on into the realm of quantum mechanics and so forth, you've long lost sight of anything all that new in the modern technological and scientific context.  The stuff you talk about is important to know and understand, but you also need to realize that as a model it, like every other model, eventually becomes either incorrect or inapplicable.  This is especially evident when we start making 'theoretical' arguments with simplified models that are physically impossible.  At least Lewin went to the trouble of demonstrating a theory with a model that could actually be built.  I predict that any effort to do so with the 'Big Misconception About Electricity" will not fare so well.

[besauk]

The capacitance / length and inductance / length do matter, in terms of getting a meaningful amount of power to the bulb.  For wires a meter apart, you'd be looking at best a couple of pF/m and a fraction of uH/m.  Since you are looking to see what happens 1m away, your simulation should use values for C and L accordingly (i.e. a couple of pF and a few hundred nH).  If you do so, the power reaching the bulb in approx 3 ns is *tiny*.   Certainly not enough to consider the bulb lit by any conventional definition.  Is there a signal at the bulb 3ns later - yes, but very small.

[besauk]

@EEVBlog I think your model is a fallacy.  You assume there is capacitive coupling between the wires. But lets eliminate capacitive coupling by inserting a shield between the wires, or use shielded wires with the shields soldered together all along the way and grounded. So ideally zero capacitive coupling. There will be no spike then at t=0, which proves the model is wrong.
Secondly, it was not necessary to go to the trouble of modelling a transmission line to demonstrate capacitive coupling, just draw two wires with a capacitor.

There is capacitive coupling between the wires - but far smaller than what Dave put in his model.  It would at best be a couple of pF/m.  Had he used realistic C and L parameters, the energy reaching the bulb at 3ns would be *tiny*.

[aneevuser]

I seem to have been thinking about an essentially different type of problem, having watched Dave's video, and read the last few replies.

My assumption was that we were considering a geometry where there was negliglible capacitive coupling between the bulb-side and battery-side, but where we would have to consider wire-to-ground coupling, and where the bulb-side would be considered to be in the far-field from the POV of the battery-side. So I wasn't really thinking about a transmission line scenario as Dave has modelled.

Is this really the setup that the Veritasium guy was proposing? If so, it seems a little boring - it's hardly earth-shattering to note that two long wires close to each other can be capacitively coupled. I'd be more interested in a setup where we have the whole wire arrangement isolated in free space (so no ground coupling to consider), with the 0.5 light-second wires, but with say 1 km along the short sides.

[sandalcandal]

I think it says that at a point in a static field the energy into the point and out of the point are equivalent.   So no actual energy flow.

No, that would mean there's no accumulation if "flow" into and out of a point is equal.

If you have a box with equal water flowing into the box and out of the box then is there "no actual water flow"?

You might need to think it through a bit more rather than trying to make extraneous logical jumps to try support your conclusion.

[Bud]

I seem to have been thinking about an essentially different type of problem, having watched Dave's video, and read the last few replies.

My assumption was that we were considering a geometry where there was negliglible capacitive coupling between the bulb-side and battery-side, but where we would have to consider wire-to-ground coupling, and where the bulb-side would be considered to be in the far-field from the POV of the battery-side. So I wasn't really thinking about a transmission line scenario as Dave has modelled.

Is this really the setup that the Veritasium guy was proposing? If so, it seems a little boring - it's hardly earth-shattering to note that two long wires close to each other can be capacitively coupled. I'd be more interested in a setup where we have the whole wire arrangement isolated in free space (so no ground coupling to consider), with the 0.5 light-second wires, but with say 1 km along the short sides.

Right, it is weird everyone latched to a "transmisison line" approach, which to me is a wrong one and not what Veritasium meant. The wire could be spaced 1 or 1000kM part, shaped in a spiral or in a circle, etc.

[Terry Bites]

It’s not the fault of audiophiles for a change.
EM fields are where they make love to their livestock. Allegedly.

It’s a challenging set of ideas presented by young tassium, but nothing new. For most practical engineering purposes yellow balls rolling along conductors and tadpoles in the stream will do. But a soon as you get into transmission lines, antennas, EM propagation, photonics and semiconductor physics etc, the wheels will fall off.

Back in the day Mr Faraday scandlolously suggested that electrically charged bodies and magnets did not give rise to fields but created disturbances in universal fields. Much like mass distorts spacetime –but it doesn’t give rise to it. See, that was easy!

Maxwell went on to demonstrate the validity of this view. He also explained the difference between conventional current and displacement current. Conventional current does not flow through capacitors. The Equations of Maxwell were made more comprehensible and applicable by Oliver Heaviside (inventor of phasors). Poynting worked out how EM energy density can be calculated. This was in the 19th century and has proven to be correct. Of course, these processes of energy transmission are not lossless. The imaginary field line loops will not all be completed, fringing if you like.Its old physics and not Vertassium’s latest fantasy. (Not that he doesn’t get it wrong: That cross polariser stuff is not quantum magic.)


This view of reality in which all particles are localised disturbances of all pervasive fields is the foundation of Quantum field theory (QEF) Particle physics is not the study of billiard balls but field interactions- That what CERN does all day.  QEF is crucial to today’s technology in electronics, spintronics and quantum computing. It’s misleadingly called particle physics so it doesn’t upset joe public and the funding bodies. They give us the balls so we can sleep at night.
I can see slather accumulating around the mouths of tube fans so I'd better leave it there.

Watch some Walter Lewin lectures on Maxwell and see QEF made simple

I said I wanted to work in TV and they said; what as, a valve?

[SiliconWizard]

 

[bdunham7]

There is capacitive coupling between the wires - but far smaller than what Dave put in his model.  It would at best be a couple of pF/m.  Had he used realistic C and L parameters, the energy reaching the bulb at 3ns would be *tiny*.

That's the issue--this discussion doesn't go anywhere unless you posit physically impossible ideal parameters.

[EEVblog]

But... does it not highlight one weakness of the transmission line model in that it does not handle the lateral (1/c) delay properly? Only that of an infinitesimally thin structure that so happens to have the equivalent characteristics of something 1m wide? So it doesn't *actually* provide an answer to the multiple-choice, only that it's not 0.5, 1 or 2 seconds?

The whole is not the detail of the answer, it's the fact that *some* energy can get there in 1m/c seconds. And the (essentially trick) question is specifically set up that way to allow that answer.

[EEVblog]

I seem to have been thinking about an essentially different type of problem, having watched Dave's video, and read the last few replies.

My assumption was that we were considering a geometry where there was negliglible capacitive coupling between the bulb-side and battery-side, but where we would have to consider wire-to-ground coupling, and where the bulb-side would be considered to be in the far-field from the POV of the battery-side. So I wasn't really thinking about a transmission line scenario as Dave has modelled.

Is this really the setup that the Veritasium guy was proposing? If so, it seems a little boring - it's hardly earth-shattering to note that two long wires close to each other can be capacitively coupled. I'd be more interested in a setup where we have the whole wire arrangement isolated in free space (so no ground coupling to consider), with the 0.5 light-second wires, but with say 1 km along the short sides.

Right, it is weird everyone latched to a "transmisison line" approach, which to me is a wrong one and not what Veritasium meant. The wire could be spaced 1 or 1000kM part, shaped in a spiral or in a circle, etc.

He meant a transmission line, as he mentioned impedance in the video, he just didn't say it, because that gives the game away for engineers watching.
You will not ge the 1m/c answer if the lines are not 1m apart. He actually got that bit wrong and said the answer is 1/c, it's not, it's 1m/c. He doesn't even get the units analysis right, and maybe that's deliberate, because once again putting in the 1m/c instead of 1/c is giving the game away.

[EEVblog]

@EEVBlog I think your model is a fallacy.  You assume there is capacitive coupling between the wires. But lets eliminate capacitive coupling by inserting a shield between the wires, or use shielded wires with the shields soldered together all along the way and grounded. So ideally zero capacitive coupling. There will be no spike then at t=0, which proves the model is wrong.
Secondly, it was not necessary to go to the trouble of modelling a transmission line to demonstrate capacitive coupling, just draw two wires with a capacitor.

Their is no shield in the question, so I'm analysing it as presented.
It IS ultimately a transmission line problem when you have a step response.

[EEVblog]

Well so much for my 10min video.
Just the simulation part turned out to be 8 minutes 

It was always going to be a challenge

Pretty nice video on that one aspect regardless. I suspect there will be confused and angry comments from physicists without prior knowledge of transmission lines but not too much you can do about that I suppose. I think it would be nice if you could show how (roughly) a transmission line model can be set up using the given problem parameters to produce the same 1/c delay predicted by the "physics", that would really help push the point that the transmission line based model is accurate and equivalent; similar to the slides provided to Derek by the professors but explained more eloquently.

I haven't seen the slides.
I don't want to get bogged down in the actual quantitative details of the answer, I just want to explain how you can arrive at the 1m/c answer and that's it.

[EEVblog]

Whereas you seem to see "how much power is flowing in this point in space" merely as a potential means to an end, thinking about problem solving/engineering and applied, applied, applied. I'm not making a judgement call here, just trying to get the root of our subjective difference of opinion.

I'm an engineer, sue me 
Again, if even a physicist like Feynman can essentially say "meh", then that's good enough for me.

[ejeffrey]

Exactly.  The electric field inside an ideal conductor is zero so the work done on the charges in the wire (E*J) is zero.

You're simply exchanging one imperfect, incomplete model for another.  The electric fields inside a conductor, even a superconductor, are not zero.  Your model just says they are because that mostly works out in the macro domain that it is intended for.  What goes actually goes on is more complex.  Quantum mechanics aside, I fail to see how you can continue to maintain the the energy flow (whatever that is) in the DC case is due to fields outside of the conductor without stating what those fields are. Poynting vectors are not fields. There is the magnetic field which is unchanging for DC current and then a static E-field where the conductor has a net charge.  Neither of those can do work on charges.  Now you can have a wrong or simplified model that still predicts at least some things correctly, so if you can have a battery on one end of a pair of wires and a load on the other and you have some diagram with some arrows that shows the E-field of the battery being magically transferred to the other end to do work on the load--whether it is Poynting vectors or monkeys with wheelbarrows--you still haven't explained how it came to be that the charge density is what it is at the load end of the wires.

You can include the electric field inside the wire if you wan't it won't make a huge different to the result.  If you do a full E&M analysis you will find that the vast majority of the energy density is outside the wires.  There is very little energy density inside the wires and very little power flow by normal formulas, such as qE or E x B.

It's certainly a bit of a matter of semantics.  "Power flow" is not really an observable property.  The power produced by the battery and dissipated in the resistor/lamp are easily defined but the power flow requires a bit more care.

The value of a field-centric approach is that everything is well defined locally.  I can look at a tiny volume of space have some parameters defined there (E, B, Q, v), and assign that location an energy density (E^2  and B^2) and a power transport vector (E x B) without referring to the rest of the system.  Then if I want the total, I just add up the local values over my chosen volume or surface.

If you try to assign the power flow to inside the wires, you can't come up with a local, consistent model of the energy transport.  The energy transport I*V, but for that equation to make sense you need to define a point with zero voltage.  If the "0V" point is the negative battery terminal all the power will be transferred via the positive lead and no power through the negative lead.  Define the + terminal as zero volts and all the power goes through the negative wire.  You can even have more than 100% of the power flow through the positive lead, but some of the power return via the negative lead.

This all doesn't make much sense, so we tend not to do it.  Instead we look at the total circuit, analyze it, and say "1 watt flows from the battery to the lamp, lets just say it flows through the wires"  That's all fine, but not very quantitative.

[hagster]

The experiment is a bit of a red herring and doesn't really help with the main point of the film that the energy is transferred in the fields. At a certain level of abstraction the wires act as a capacitor and loosely coupled transformer and cause some energy to transfer to the build almost instantaneously. But probably not enough to heat up the filament and make a visible indication. Any change would cause some increased black body radiation at a longer wavelength.

Many people have pointed out that this all assumes  superconductive wires. Yes true, I think the example is simplified and this is not considered. Any resistance creates a voltage drop and hence an electric field. Fairly sure if you analysed the poyting vector there would be a net inflow of energy into the wires also as it absorbs energy.

The main point he is making in the video is totally correct however. The energy is in the fields not the electrons.

As an example, in a PCB we try to keep loop areas small so the fields can be effectively constrained along a designed path without loosing energy via radiation. This occurs because the energy is in the EM fields and hence any separation of these into propagating waves removes energy.

At the end of the day though, all science is basically an abstraction of reality. There are often multiple abstractions that make sense to use in the correct situation. If we dig deeper, we would discover that the electrons don't really have a set position and there existance is only the peak of probability distribution with infinite extent. If it makes sense for a particular engineering problem to consider the energy being transferred by the wire then there is no harm in doing that. But for AC signals the loss in fields(normally in a dielectric) should not be ignored.

Another physics fact to blow people's mind is that 'temperature' doesn't really exist. It's just a measure of the average momentum of the particles in a volume. At the quantum level there is no need for such a concept.

[ejeffrey]

I seem to have been thinking about an essentially different type of problem, having watched Dave's video, and read the last few replies.

My assumption was that we were considering a geometry where there was negliglible capacitive coupling between the bulb-side and battery-side, but where we would have to consider wire-to-ground coupling, and where the bulb-side would be considered to be in the far-field from the POV of the battery-side. So I wasn't really thinking about a transmission line scenario as Dave has modelled.

Is this really the setup that the Veritasium guy was proposing? If so, it seems a little boring - it's hardly earth-shattering to note that two long wires close to each other can be capacitively coupled. I'd be more interested in a setup where we have the whole wire arrangement isolated in free space (so no ground coupling to consider), with the 0.5 light-second wires, but with say 1 km along the short sides.

Right, it is weird everyone latched to a "transmisison line" approach, which to me is a wrong one and not what Veritasium meant. The wire could be spaced 1 or 1000kM part, shaped in a spiral or in a circle, etc.

It could be anything but the geometry presented is a transmission line and the standard transmission line circuit model is good enough to get the main effect even if it isn't quantitatively correct at the nanosecond level.

You could indeed do the thought experiment with a weird spiral shaped wire and you wouldn't be able to fine a good transmission line model.  If you wanted to simulate that you would need a field solver like HFSS.  But for the example presented it is enough to say that the transmission line model shows that the lamp lights immediately and then you can figure that it takes 3 ns because of the finite speed of light.

[ejeffrey]

He meant a transmission line, as he mentioned impedance in the video, he just didn't say it, because that gives the game away for engineers watching.
You will not ge the 1m/c answer if the lines are not 1m apart. He actually got that bit wrong and said the answer is 1/c, it's not, it's 1m/c. He doesn't even get the units analysis right, and maybe that's deliberate, because once again putting in the 1m/c instead of 1/c is giving the game away.

I think the purpose of leaving out that was just to avoid jargon to only the most salient points because most of his audience are not electrical engineers.  The fact that it is a transmission line is not necessary for the effect but it is a good way to model /simulate it.