2
\$\begingroup\$

I have a hard time figuring out how to calculate the maximum power transfer.. Can anyone help?

I understand that you need to calculate the thevenin resistance by killing the sources and summing the individual resistances, but I'm unsure about the next few steps and formulars to follow.

enter image description here

Edit: After reading through the comments saying I should show more effort, I've tried to apply KCL at the three independent nodes and solving the equations for the unknown variables using wordmat: V_a, V_b, V_c. Assuming V_c is equal to the open circuit voltage V_oc, I've then used V_oc along with the calculated thevenin resistance in the maximum power transfer equation: P_max = V_oc^2 / 4*R_th.

However, I've most likely calculated V_oc wrongly.. Anyone who could tell me what I did wrong in my calculations?

btw, I accidentally mistyped the unit for V_c, which should be in volts.

enter image description here

\$\endgroup\$
  • 2
    \$\begingroup\$ You need to show more effort. Explain exactly what you are "unsure about". Do you know in general how to find the maximum power transfer from a Thevenin Equivalent Circuit? \$\endgroup\$ – Elliot Alderson Sep 26 at 20:28
  • 2
    \$\begingroup\$ And what did you get when you did that step? \$\endgroup\$ – Brian Drummond Sep 26 at 20:28
  • 3
    \$\begingroup\$ Just by the side, Fermat could have solved this without the use of the maximum power theorem or the use of derivatives (as such ideas hadn't yet been developed.) He might have found it enjoyable as math was just a hobby for him and not a profession. The point here is that "learning how to think about" such problems is more important than solving one. (A point too few will take too heart, or perhaps even understand well.) It is a good problem for sharpening one's thinking. \$\endgroup\$ – jonk Sep 26 at 20:34
  • \$\begingroup\$ well, after googling around, the maximum power transfer equation is: P_max = V_oc / 4*R_th. R_th I know how to find. I then tried to calculate V_oc by removing R_load and using nodal analysis at the top three nodes and trying to solve the equations with the unknowns, but I'm not sure I did it correctly, as I got some odd numbers. I'm not looking for the specific answer, just want to know the methodology so I can apply it myself. All the examples I've seen use simpler circuits and therefore fewer equations/unknowns. Is it true that in order to find V_oc I'd need to write 3 equations? \$\endgroup\$ – Timothy Sep 27 at 17:34
  • 1
    \$\begingroup\$ @Timothy Thanks! (You need to use our names [one of them] with an @-sign next to it if you want to get our attention. You'll have to pick someone or else write several comments.) I'll look this over to see if I can extend my answer further. \$\endgroup\$ – jonk Sep 29 at 23:54
4
\$\begingroup\$

Overview

I'm going to borrow (copy) from the Wikipedia page on maximum power transfer that you bring up:

The maximum power transfer theorem states that, to obtain maximum external power from a source with a finite internal resistance, the resistance of the load must equal the resistance of the source as viewed from its output terminals. Moritz von Jacobi published the maximum power (transfer) theorem around 1840; it is also referred to as "Jacobi's law". The theorem results in maximum power transfer across the circuit, and not maximum efficiency. If the resistance of the load is made larger than the resistance of the source then efficiency is higher, since a higher percentage of the source power is transferred to the load, but the magnitude of the load power is lower since the total circuit resistance goes up.

This also qualifies as a theorem since it fits the idea of "a general proposition not self-evident but proved by a chain of reasoning; a truth established by means of other more prosaic accepted truths." (Some may need to spend a little time parsing that sentence to get its fuller meaning.)

That Wikipedia page also provides a calculus-based proof, so I won't bother with it here.

The problem this leaves you with is in converting (somehow) the remainder of your circuit -- the part other than the load itself -- into its Thevenin equivalent, which in ordinary cases is just an ideal voltage source and its series resistance.

Approach Options

You can approach this in a variety of ways. One way is to use what you know about such conversions steps and apply them to the parts you can see in the circuit until you achieve the final equivalent. Another is to just ignore all of that for the moment and treat it as a black box within which you cannot yet see anything, but where just two leads protrude. In this case, the process would be to first measure the voltage with a voltmeter and to then short the two leads together with an ammeter and measure the resulting current through it. Obviously, the hidden Thevenin equivalent would present its voltage directly when measured with an ideal voltmeter, since there would be no current through its Thevenin resistance and therefore no voltage drop behind which the Thevenin voltage might hide. So that measurement gives you one part (that you don't care about here) for free. Also, when applying the ideal ammeter, you will now measure the current that is produced when the now-known Thevenin voltage source produces a current through its Thevenin resistance. Knowing that earlier voltage and now knowing this current, you can compute the resistance by dividing the current into the voltage. That's it.

In your comments, you mention nodal analysis and you correctly mention the likely need for three equations. So let's go with that. First, remove the load resistor from the circuit. Then, assuming you can correctly form the three equations in three unknowns (the three voltages of the unknown node voltages in your schematic, when you assume that the bottom wire is, by definition, \$0\:\text{V}\$), then solving for all three node voltages gives you the Thevenin voltage since this must be the same voltage as at the right-most unknown node. (Or you could simplify the circuit, first.) Now, if you place a wire across the point where the load resistance is currently located and work out the current in that wire, you'd have the two values you need to compute the Thevenin resistance of the circuit prior to the load and you'd have solved the problem by the now-trivial application of the maximum power theorem.

That said, I prefer something that is a little easier to do, if you have a handy tool that solves simultaneous equations for you. Just remove the load itself and generate the nodal equations (three of them) for the circuit. Solve it once in order to get the Thevenin voltage. Then insert a \$1\:\text{A}\$ current source or sink where the load was at (all you need to do is add one trivial term to one equation) and recompute the voltages, again. You now have two Thevenin voltage results -- one with no load and one with a very specific load. The change in the Thevenin voltage (their difference) divided by the change in load current, which goes from \$0\:\text{A}\$ to \$1\:\text{A}\$ and so is very easy to also figure out, gives you the Thevenin resistance, just as well as shorting the output did. Of course, all you've done here is to force a certain voltage drop through the unknown Thevenin resistance; and since you already know the unloaded Thevenin voltage it becomes very easy to work out what the Thevenin resistance must be in order to produce the observed change in voltage.

Summary

Does that help, sufficiently? Or do you need more?

| improve this answer | |
\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.