Superposition with Dependent Sources

[smk037]

Almost every EE I've talked to claims that you need to leave dependent sources in when using superposition to solve a circuit.

I had a professor in undergrad who used to claim that statement is invalid. I've used his method since I learned it without any concern. I never saw any problem with it.

Here is a paper describing his method/argument. At the end of the paper, someone presents a circuit that would contradict what Dr. Leach is saying. I agree with Dr. Leach that the circuit presented has a floating node and can not be solved by any technique.

I'm curious to hear some other opinions on this.

http://users.ece.gatech.edu/mleach/papers/superpos.pdf

 

[smk037]

bump...
would love to hear some opinions on this

 

[Ratch]

smk037,

That paper looks like good stuff, provided you use the technique within its limitations. That means no floating nodes. Thanks for bringing this to my attention.

Ratch

 

[smk037]

I have never encountered any circuits with floating nodes in my college career, so maybe I don't know exactly what I'm talking about (Although I don't think I have ever encountered floating nodes since I used this method throughout every circuits course I'd ever taken).

BUT, it seems to me that the circuit presented by the man who rejected the method can not be solved by any conventional circuit analysis techniques. I have tried it myself using nodal analysis and mesh currents and came up with answers that do not work. I've also tried using multisim to solve the circuit, but again with no luck.

 

[Ratch]

smk037,

OK, I reviewed the rejection letter of Professor Leach's method, and here is my analysis.

1) In the first paragraph, the reviewer makes a flat out statement that Professor Leach's method does not work, and avers that two given examples just happen to produce the correct answer.

2) In the second paragraph, the reviewer pontificates about a resistor being considered as a voltage source or current source. I see no relevance about this to Leach's method not working.

3) In the third paragraph, the reviewer gives a counter example where Professor Leach's method does not work, and avers that the problem can be solved by deterministic methods.

My analysis indicates that the reviewer is wrong. Here is why. The two independent current sources I1 and I2 can be any value whatsoever. And the dependent source has to equal I1 and I2. Yet, the dependent source also has to equal g*(v1). That makes two condradictory conditions that prevents any solution. For instance, suppose I1 = 10 amps and I2 = -10 amps. Then the dependent source would not have to supply any current, but g*V1 would surely be some finite value. Therefore no solution method will be successful because the problem has no solution. So, as far as I can determine, Professor Leach's method works for all cases under the conditions he specifies.

Ratch