> Middlebrook's original method requires that the injected loop test signal be inserted between a low impedance point
> (usually the output of the power supply) and a high impedance point (usually the feedback amplifier network).
> There is no backwards loading with such a insertion point.
This quote is not from my webpage. I don’t think that it’s correct either. In his original paper from 1975, Middlebrook did take impedances into account, but not backwards transmission. The conditions given above are for voltage loop gain.
But why use "V2 = AC 1" on the feedback path ?
For the voltage loop gain analysis, you can use any AC value, because you are calculating a ratio of two voltages and AC analysis is linear.
Why no backwards loading ?
Why must the test signal located between low impedance point and high impedance point ?
To understand the conditions for voltage loop gain, please read
https://www.omicron-lab.com/fileadmin/assets/Bode_100/Documents/Bode_Info_LoopGain_V1_1.pdf (which is also linked from my webpage).
> Tian's method allows for backwards loading so that the signal insertion point may be anywhere along the loop.
> To do this, it requires two measurements and some simple calculations to combine them.
How does Tian's method accounts for reverse feedback ? Is backwards loading equivalent to reverse feedback ?
Tian’s formula is symmetrical. He basically adds forward and reverse loop gain in his formula as you can see in his article.
What does it mean by the inserted probe elements result in smaller, sparser circuit matrix ?
You get a smaller matrix because you don’t need two copies of the circuit as in the original LoopGain.asc example.
> The General Feedback Theorem allows for multiple influential feedback loops all of which may have backwards loading.
Why GFT allows multiple feedback loops while Tian's Method and Middlebrook's method do not ?
The basic version of the GFT also allows for only one feedback loop. However, it can be extended for multiple loops, see
https://web.archive.org/web/2018100...l-network-theorem/gnt-integration-in-virtuoso