Does Kirchhoff's Law Hold? Disagreeing with a Master

[rfeecs]

What you are saying is equivalent to say: here is a silicon diode,  does Kirchhoff's law apply?  Or if you prefer: here is a transmission line, does Kirchhoff's law apply?  Don't you have to put the element in a circuit first?

Well it is a bandpass filter, so the complete circuit would be a generator at the input and a load at the output, both with 50 ohm impedance in this case.

Even without that, I would still call this a circuit.  Just like a lumped bandpass filter is called a circuit.  This microstrip filter has paths from one element to the next and from the elements to ground, forming loops, just like the lumped version.

The Lewin circuit is so simple, with just wire loops and lumped elements.  So it could be easy to say it is just a transformer plus some lumped resistors and voltmeters.

I was trying to show a circuit that obviously wasn't that simple.  I guess I went too far to the point where it doesn't even look like a circuit at all.

[ogden]

I was trying to show a circuit that obviously wasn't that simple.

Kirchoff's circuit Law do not apply to AC circuits that does not have DC circuit properties.  Look at the input of that microstrip filter - for DC current it is dead short. The same about output.

Wikipedia explains it well enough:

Both of Kirchhoff's laws can be understood as corollaries of Maxwell's equations in the low-frequency limit. They are accurate for DC circuits, and for AC circuits at frequencies where the wavelengths of electromagnetic radiation are very large compared to the circuits.

Other way to say it:

You do not apply Kirchoff's law to circuits where current in the conductor is not uniform.

[jesuscf]

I was trying to show a circuit that obviously wasn't that simple.

Kirchoff's circuit Law do not apply to AC circuits that does not have DC circuit properties.  Look at the input of that microstrip filter - for DC current it is dead short. The same about output.

Wikipedia explains it well enough:

Both of Kirchhoff's laws can be understood as corollaries of Maxwell's equations in the low-frequency limit. They are accurate for DC circuits, and for AC circuits at frequencies where the wavelengths of electromagnetic radiation are very large compared to the circuits.

Other way to say it:

You do not apply Kirchoff's law to circuits where current in the conductor is not uniform.

Kirchoff's circuit laws apply to anything that is modeled correctly for the range of frequencies you want to design/analyze.  The keyword in the Wikipedia quote above is 'accurate'.  For example, if you want to place a resistor in an AC circuit working at frequencies over 100MHz, it  will not behave just like a resistor.   Take a look at this EDN article:

https://www.edn.com/design/components-and-packaging/4423492/Resistors-aren-t-resistors

where they show a model of a 1k ohm carbon resistor used at such frequencies:



The same applies to circuits whose physical dimensions are similar to the wavelength of the applied AC signals.  In these cases even wires must be modeled accordingly, usually as transmission lines.  It is a compromise: how accurate do you want your results vs. how much you are willing to work.

As for the statement "You do not apply Kirchoff's law to circuits where current in the conductor is not uniform": what if you define a Gaussian surface surrounding the conductor with non-uniform current (or any part of a circuit by that matter) and compute the integral of all the currents crossing the surface?  You'll find that the integral is zero; same as Kirchoff's current law!
 

[ogden]

As for the statement "You do not apply Kirchoff's law to circuits where current in the conductor is not uniform": what if you define a Gaussian surface surrounding the conductor with non-uniform current (or any part of a circuit by that matter) and compute the integral of all the currents crossing the surface?  You'll find that the integral is zero; same as Kirchoff's current law!

I did not mean skin effect but transmission line effect - current uniformity trough the length of the circuit/loop. Sorry about wording.

Example: Speed of electric impulse in the cable (velocity factor) is ~65...85% speed of light. We just pick 195000 km/sec figure. So we have battery, lightbulb and 195km cable (twisted pair) to connect both. When we connect battery to the cable, current trough battery is flowing immediately but lightbulb is not lit yet, current will start to flow trough it (only) after 1ms. In this case Kirchoff's circuit law does not hold apply, sorry.

[bsfeechannel]

In this case Kirchoff's circuit law does not hold apply, sorry.

Kirchhoff's law doesn't hold. Lump modelling doesn't apply.

[rfeecs]

To use network analysis with transmission lines, the transmission lines are treated as components, not wires.  So, for example, a transmission line is treated as a two port.  At each port the voltage and current is uniquely defined.  All the retardation effects are contained within the two port.
[Edit] - Wikipedia entry for Two-port network:  https://en.wikipedia.org/wiki/Two-port_network

This has already been discussed in this thread.

It is similar to the "fix-ups" used for inductors and capacitors.

If you want to say Kirchhoff's laws can't be used in circuits with transmission lines, then you must say they can't be used in circuits with inductors, transformers and capacitors.  Or that they can only be used at DC.

The reality is that RF engineers simulate circuits with all those components.  Those simulators rely on KVL and KCL, and they work.  There is no argument here beyond semantics.

[jesuscf]

As for the statement "You do not apply Kirchoff's law to circuits where current in the conductor is not uniform": what if you define a Gaussian surface surrounding the conductor with non-uniform current (or any part of a circuit by that matter) and compute the integral of all the currents crossing the surface?  You'll find that the integral is zero; same as Kirchoff's current law!

I did not mean skin effect but transmission line effect - current uniformity trough the length of the circuit/loop. Sorry about wording.

This is what I meant:

Example: Speed of electric impulse in the cable (velocity factor) is ~65...85% speed of light. We just pick 195000 km/sec figure. So we have battery, lightbulb and 195km cable (twisted pair) to connect both. When we connect battery to the cable, current trough battery is flowing immediately but lightbulb is not lit yet, current will start to flow trough it (only) after 1ms. In this case Kirchoff's circuit law does not hold apply, sorry.

Yes Kirchoff's circuit laws do work:

[bsfeechannel]

The reality is that RF engineers simulate circuits with all those components.  Those simulators rely on KVL and KCL, and they work.  There is no argument here beyond semantics.

We need to use this thread to end this confusion once and for all. And the problem is not related to semantics.

Kirchhoff implies lump-modelling. Inverting this logic is false. Let's see this better exposed with a truth table.

Kirchhoff holds	Lump-modelling possible	Implication	Comment
True	True	True	Every circuit for which Kirchhoff holds can be automatically lump-modeled.
True	False	False	If you're saying that you can't lump-model a circuit where Kirchhoff holds you're contradicting the line above. So, false.
False	False	True	If you find a circuit that is impossible to lump-model then Kirchhoff doesn't hold (Lewin's circuit).
False	True	True	Kirchhoff doesn't hold, but you can still lump model  the circuit under certain conditions. Feynman picture 22-9

Inside a transformer, an inductor, a generator, an antenna or a transmission line, Kirchhoff doesn't hold. But you can still lump model them.

People see that those components can be lump-modeled and think that Kirchhoff holds. No, it doesn't. It will be impossible to explain how those things work using Kirchhoff.

But they do worse. Because they see that the behavior of the components that can only be explained by Maxwell's equations can be lumped, they think that every circuit can be lump-modeled.

This intuition is false.

Lump-modelling is a technique for circuit analysis. Kirchhoff's laws are the description of a physical phenomenon. Let's not confuse them.

[Berni]

Long wires are again one of the gotcha's of circuit modeling.

Just like ignoring inductance in Dr. Lewins experiment, ignoring transmission times long enough to affect behavior causes KVL to break down if the wire is just drawn as a line in the equivalent schematic.

As we have found multiple times on this thread Kirchoffs circuit laws are not fundamental laws of the universe. But a lot of circuits are a special case where they can be directly turned into a schematic and they work fine with circuit analysis. Such as connecting a resistor across a battery with short cables. There still are inductive and capacitive parasitics, transmission line effects etc... that we omitted in the schematic, but due to the nature of the particular circuit they don't affect how the circuit works so it doesn't matter that they got ignored.

So for the 195km cable you can model it as a before mentioned 4 port device to lump the entire cable. Then Kirchhoffs circuit laws work again as they always work in fully lumped circuits. The transmission line inside that box can be modeled in various ways, but the chain of inductors and capacitors is a perfectly valid way of doing it. That way of modeling a transmission line is actually quite well suited for explaining them as you can actually see the waves traveling down the chain as they do over the cable and each node on the LC chain does correspond in voltage at a point along the cable.

So yeah Kirchhoffs circuit laws indeed don't apply to long cables in real life, but they can be lump modeled to make it work.

[ogden]

So for the 195km cable you can model it as a before mentioned 4 port device to lump the entire cable. Then Kirchhoffs circuit laws work again as they always work in fully lumped circuits.

Sure you are right - *IF* you model cable as a string of many lumped elements. My oversimplified example contained three elements: 1) battery 2) cable 3) light bulb. That was whole point of example - to show case where you can't apply KVL and what's even more important - *why*. No need to cheat to get around pre-requirement for KVL I was actually illustrating! During described 1ms you can't apply KVL but later when current is uniform through the loop, KVL appies.

Hopefully you did not get sick with "putting unsaid words into debate opponents mouth disease", diagnosed for some here in this thread?

[edit] Removed incorrect statement about engineering, thanx to @jesuscf

[ogden]

If you want to say Kirchhoff's laws can't be used in circuits with transmission lines

Thing is I did not say that. Cool down your imagination please Also I did not say that KVL applies only to DC circuits. After all Lewin's experiment is AC circuit, just transmission line effects can be ignored for circuit and frequencies used.

[edit] Actually I can say that KVL cannot be applied to transmission line having transmission line effects (current is nonuniform through it) unless it is properly "lumped down".

then you must say they can't be used in circuits with inductors, transformers and capacitors.

I can't imagine how KVL can be applied to (across) transformer or microstrip filter shown here. Could you explain it please? - I am curious.

[jesuscf]

Sure you are right - *IF* you model cable as a string of many lumped elements. Engineering is exact science and circuit of my oversimplified example contained three elements: 1) battery 2) cable 3) light bulb. That was whole point of example - to show case where you can't apply KVL and what's even more important - *why*. No need to cheat to get around pre-requirement for KVL I was illustrating. During described 1ms you can't apply KVL but later when current is uniform through the loop, KVL appies.

First of all, I am sorry to break these news to you but engineering is neither a science nor exact!  Second, a very long 'cable' is far from an ideal conductor.  Similarly, a battery is far from an ideal source.  Finally, a light bulb is far from an ideal resistor.  Assuming you have idealized 'real' components and then proclaiming that KVL can not be applied is a straw man fallacy.

Circuit theory requires the correct modeling of all the physical devices used to describe the circuit.  If the circuit solution doesn't behave as the measurements in the circuit, it is not the fault of circuit theory, it is either the fault of the models used or the measurements themselves.

[ogden]

Second, a very long 'cable' is far from an ideal conductor.  Similarly, a battery is far from an ideal source.  Finally, a light bulb is far from an ideal resistor.
Assuming you have idealized 'real' components and then proclaiming that KVL can not be applied is a straw man fallacy.

Components are ideal until it is explicitly said that they are not. As you are not on the same page and fail to comprehend whole idea of dumbed down example so generic public can understand described transmission line effects, I have nothing to further discuss with you.

[Berni]

Sure you are right - *IF* you model cable as a string of many lumped elements. Engineering is exact science and circuit of my oversimplified example contained three elements: 1) battery 2) cable 3) light bulb. That was whole point of example - to show case where you can't apply KVL and what's even more important - *why*. No need to cheat to get around pre-requirement for KVL I was actually illustrating! During described 1ms you can't apply KVL but later when current is uniform through the loop, KVL appies.

Hopefully you did not get sick with "putting unsaid words into debate opponents mouth disease", diagnosed for some here in this thread?

Did i claim that a user said something? If so then that was not the intention sorry.

And yep it is a good example that shows that KVL can't simply be directly slapped onto a real circuit by approximating everything as being ideal, including wires.

Yep since KVL works in lumped circuits and attributes the voltage to an inductor while with Maxwell its just sums it in as a whole. Just two different ways of going about the same thing that come to the same result when used correctly.


So if you take yourself as being so knowledgeable in how to analyze circuits, have you decided yet what is the correct way to analyze the behavior of this circuit?
https://www.eevblog.com/forum/chat/does-kirchhoffs-law-hold-disagreeing-with-a-master/msg2189216/#msg2189216


I can give a trip that gain degradation on the transistor plays no role in the operation so a ideal transistor with a fixed Hfe can be used, but the parasitic capacitance in the transistor are important to the operation, all of them are stated in the datasheet for that transistor (Tho a ballpark figure for most small signal transistors is close enugh)

I'm guessing those loops are supposed to form a transformer and the circuit is an oscillator?  So we replace the loops with a lumped transformer?  I'm trying to read your mind, maybe not successfully.

The other way to read the schematic is a wire is a zero length connection in a netlist, so that circuit doesn't do much.

Yes the circuit has dimension that's why the size of the loops is marked down.

It is indeed an oscillator. Functions much like the classical blocking oscillator, but i modified the circuit to help keep the transistors operating point at a spot where it has lots of gain (sort of self regulates its collector bias current). This helps it keep oscillating with such a low inductance leaky "transformer" over a decent supply voltage range.

I have built the circuit for real on the bench and built it in LtSpice. Worked fine in both cases.

I'm just looking forward to how bsfeechannel explains how one should analyze this circuit correctly since all of it is inside the so called "KVL dies here" zone. So if classical circuit analysis theory supposedly can't be used here, then there must be some other way to calculate its behavior.

[ogden]

Hopefully you did not get sick with "putting unsaid words into debate opponents mouth disease", diagnosed for some here in this thread?

Did i claim that a user said something? If so then that was not the intention sorry.

Indeed sorry - that many including you, ignored or missed purpose of my "195km transmission line" example: to illustrate how transmission line works and how it can be that current is not uniform trough the length of the circuit/loop. Obviosuly to apply KVL, we have to do something with that line  - either lower working frequency or use it's lumped elements model. Kind of whole thread is about "lumping to apply KVL"

Those who become agitated with feelings "how dare you explain such a simple things to *me*" can rest assured - this is forum, not chat. Here are plenty of readers who just read w/o posting. Also in case I use Layman's terms to explain something it does not necessarily mean that I think in Layman's terms.

Sorry about this offtopic.

[bsfeechannel]

I have built the circuit for real on the bench and built it in LtSpice. Worked fine in both cases.

I'm just looking forward to how bsfeechannel explains how one should analyze this circuit correctly since all of it is inside the so called "KVL dies here" zone. So if classical circuit analysis theory supposedly can't be used here, then there must be some other way to calculate its behavior.

If I tell your LTSpice that what you have between the base and the collector of your transistor is a dead short, it will refuse to simulate your oscillator.

Please refer to my previous message where I show the origin of the confusion between Kirchhoff's law and circuit analysis. They are not the same. I've been pointing that out along the whole thread.

[jesuscf]

Second, a very long 'cable' is far from an ideal conductor.  Similarly, a battery is far from an ideal source.  Finally, a light bulb is far from an ideal resistor.
Assuming you have idealized 'real' components and then proclaiming that KVL can not be applied is a straw man fallacy.

Components are ideal until it is explicitly said that they are not. As you are not on the same page and fail to comprehend whole idea of dumbed down example so generic public can understand described transmission line effects, I have nothing to further discuss with you.

As soon as you say a 'cable' is 195 km long it stops being ideal.  Forget about modeling it using the telegrapher's equation, just the fact that the resistance per meter will accumulate to a value significantly larger that the load after such length, renders your example as useless.

Note: The fact that you have nothing to discuss with me will not stop me from pointing out your fallacies.  After all, this is public forum.

[jesuscf]

Berni, may I ask what is the resonant frequency of your circuit?   Also, what would happen if you put a piece of metal close to the loops?  I am asking because your circuit could make a fine metal detector...

[Berni]

Berni, may I ask what is the resonant frequency of your circuit?   Also, what would happen if you put a piece of metal close to the loops?  I am asking because your circuit could make a fine metal detector...

I think the real circuit ran at about 17MHz while in spice somewhere around 21MHz. Tho i was using just a regular passive scope probe so the capacitance of it has likely affected the frequency too. As one might expect putting a lump of ferrite in there does drag down the frequency slightly. Haven't tested if non ferrous metal around the coil affects it, i would guess not since that doesn't change the inductance, just introduces extra loss in the form of eddy currents, large enough aluminium plate could probably stop the oscillator if it sucked away enough power from the coil.

Could probably also be used as a weak CW radio transmitter, but the waveform is not a sinewave and is not even symmetrical so there would be plenty of even and odd harmonics, but the power is really low as the whole circuit only uses 1 or 2 mA from the power supply.

[jesuscf]

Berni, may I ask what is the resonant frequency of your circuit?   Also, what would happen if you put a piece of metal close to the loops?  I am asking because your circuit could make a fine metal detector...

I think the real circuit ran at about 17MHz while in spice somewhere around 21MHz. Tho i was using just a regular passive scope probe so the capacitance of it has likely affected the frequency too. As one might expect putting a lump of ferrite in there does drag down the frequency slightly. Haven't tested if non ferrous metal around the coil affects it, i would guess not since that doesn't change the inductance, just introduces extra loss in the form of eddy currents, large enough aluminium plate could probably stop the oscillator if it sucked away enough power from the coil.

Could probably also be used as a weak CW radio transmitter, but the waveform is not a sinewave and is not even symmetrical so there would be plenty of even and odd harmonics, but the power is really low as the whole circuit only uses 1 or 2 mA from the power supply.

My understanding is that non ferrous metals should decrease the inductance, therefore increasing the resonant frequency.  Anyhow, nice simple circuit with useful practical applications...  Did you use magnet wire?  What gauge?  AA battery?

[ogden]

As soon as you say a 'cable' is 195 km long it stops being ideal.  Forget about modeling it using the telegrapher's equation, just the fact that the resistance per meter will accumulate to a value significantly larger that the load after such length, renders your example as useless.

It does not need to be ideal. I just picked long enough line to bring timing into intervals most will not be afraid of. If I would talk about > 2GHz frequencies and according wavelengths, many would not understand how transmission lines work, what is velocity factor. "Telegraphers equation" indicates that you do not understand transmission line physics as well.

Second, a very long 'cable' is far from an ideal conductor.

So what? Fact that cable has some significant resistance somehow makes my example of transmission line inoperable or what? Don't blame me if you can't wrap something into your linear logic.

[Berni]

My understanding is that non ferrous metals should decrease the inductance, therefore increasing the resonant frequency.  Anyhow, nice simple circuit with useful practical applications...  Did you use magnet wire?  What gauge?  AA battery?

I still had the circuit so i hooked it up and played some more with it.

To power it i use a capacitor with twisted wires coming up to it from my lab PSU so that i can easily adjust the voltage while the capacitor ensures a low impedance at high frequency. For the transistor i used (random 2SC2705 i had a bag of in TO92) the circuit works from 0.6V to 3.5V so i picked 1.5V as somewhere in the middle and conveniently can be a alkaline cell. For the wire i just used some wire from a CAT5 cable. You can use a twisted pair to get both coils in one loop.

In terms of being a metal detector its not so good as shown since its very sensitive to stray capacitance. It does make a great theremin tho since the frequency wanders as you wave your hand around the loop. The 1K resistor makes the whole thing high impedance.

But you can fix that if you add a 1uF capacitor in parallel with the resistor and battery. This gives the transistor a low impedance supply voltage that it regulates on its own. Using this the operating voltage range becomes 0.8V to 60V (Probably even higher if the 1K resistor handles the power) tho it does seam to shift into a different oscillation mode above 5V. With that the circuit is no longer sensitive to stray capacitance as much. Placing a ferrite core inside again lowers the frequency but placing a aluminum heatsink inside raises the frequency so it is indeed a metal detector. Tho the frequency can also shift with changing the supply voltage or the transistors temperature.

[jesuscf]

It does not need to be ideal. I just picked long enough line to bring timing into intervals most will not be afraid of. If I would talk about > 2GHz frequencies and according wavelengths, many would not understand how transmission lines work, what is velocity factor. "Telegraphers equation" indicates that you do not understand transmission line physics as well.

Really?  I would recommend you some enlightening reading from chapter 10 of "Engineering Electromagnetics" by William H. Hayt and John A. Buck where the figure below comes from:

[ogden]

It does not need to be ideal. I just picked long enough line to bring timing into intervals most will not be afraid of. If I would talk about > 2GHz frequencies and according wavelengths, many would not understand how transmission lines work, what is velocity factor. "Telegraphers equation" indicates that you do not understand transmission line physics as well.

Really?  I would recommend you some enlightening reading from chapter 10 of "Engineering Electromagnetics" by William H. Hayt and John A. Buck where the figure below comes from:

What exactly you imply by pointing to this common model of transmission line? - That you understand it contrary to my trolling? If you understand - good for you.

Even if 195 km twisted pair line is built using copper conductor, it does not make my example useless as you say. After all battery voltage nor light bulb current were specified. By arguing about nonessential properties of my oversimplified example you are just wasting electronic ink of internet and to be honest - making yourself fool.

[jesuscf]

What exactly you imply by pointing to this common model of transmission line? - That you understand it contrary to my trolling? If you understand - good for you.

Even if 195 km twisted pair line is built using copper conductor, it does not make my example useless as you say. After all battery voltage nor light bulb current were specified. By arguing about nonessential properties of my oversimplified example you are just wasting electronic ink of internet and to be honest - making yourself fool.

Wow, ad hominem fallacy!

Non-essential properties you say!  Go tell that to the electrical engineers of your local power utility and watch their reaction.  If reading the book I recommended is out of your reach, try watching a PBS documentary from the series American Experience about the first trans Atlantic cable.  I think YouTube has it.