Some questions about negative feedback

[npn]

Hello!

I had a few questions about feedback amplifiers.

Why do the four types of amplifiers each have a specific type of feedback? For e.g. why is a current amplifier always said to use shunt series feedback? Is there any particular reason for this? I've searched various books for this, but all of them have implicitly stated the amplifier types.

The closed loop gain of a feedback amplifier is given by A/(1+AB). So for a trans-resistance amplifier, the gain A will be the trans-resistance gain, right?
Suppose the open loop gain of an Op-amp inverting amplifier is small. Naturally, virtual ground won't work here. Can I go ahead and use A/(1+AB) here? Or should I find the trans-resistance gain first?

 

[LvW]

Why do the four types of amplifiers each have a specific type of feedback? For e.g. why is a current amplifier always said to use shunt series feedback? Is there any particular reason for this? I've searched various books for this, but all of them have implicitly stated the amplifier types.

Amplifiers can have different types of input and output (current, voltage). For example, the feedback signal can be derived from the output voltage and added at the input to the input voltage. That is voltage controlled voltage feedback like in non-inverting opamp applications. For inverting opamp circuits the feedback signal is combined with the input current - thus we have voltage controlled current feedback. This explains the principle. For OTA's we normally have current controlled feedback loops. For transistors you can have both.

The closed loop gain of a feedback amplifier is given by A/(1+AB).

Not for all configurations. For example, not for inverting opamp circuits.

So for a trans-resistance amplifier, the gain A will be the trans-resistance gain, right?
Yes, the gain is given in ohms.

Suppose the open loop gain of an Op-amp inverting amplifier is small. Naturally, virtual ground won't work here. Can I go ahead and use A/(1+AB) here? Or should I find the trans-resistance gain first?

Such a formula applies also for small open loop gain values of A(s). However, as mentioned above, it is not correct for inverting applications.
For classical voltage mode opamps: A_closed=-A(s)*Hf/(1+A(s)*Hr) with Hf=forward factor R2/(R1+R2) and Hf=(B)=R1/(R1+R2).

Applying the feedback model for transimpedance amplifiers leads to forward and feedback factors given in (1/Ohm).

 

[npn]

What I don't understand is why a particular amplifier is associated with a particular topology. Is it just to make calculations simpler?
e.g. When represented by blocks, why does a series shunt configuration, have an amplifier block with a voltage gain? Why is it not shown as, say a current amplifier?

 

[LvW]

You always have two alternatives to describe an amplifier using the terms "series, shunt" or "voltage, current".
The meaning is as follows: Voltages are added if they appear in series and currents are added if they are combined in parallel (shunt).
For my opinion, the most logical way to desribe amplifiers is to use the phrase "voltage/current controlled voltage/current feedback".
Thus, all possible misinterpretations are eliminated.

 

[FvM]

What I don't understand is why a particular amplifier is associated with a particular topology. Is it just to make calculations simpler?

Feedback configurations are representing different properties of real circuits, they aren't only a matter of calculation. They are chosen, because they serve a purpose related to the amplifier functionality.

If you imagine an ideal OP with inifinite (or at least high) gain, that is intended to work as a voltage amplifier with high input and low output impedance, you'll notice that series-shunt feedback (or in LvW's terms, that are also more familiar to me, voltage-controlled voltage feedback) is the only reasonable feedback option.

In other words, the feedback topology follows the intended amplifier function. In my opinion, the Feedback Configurations paragraph in Gray/Huerst/Lewis/Meyer gives an almost exhaustive explanation of the topic. I can just suggest to study it thoroughly.

 

[mtwieg]

I had a professor in school who used the whole series/shunt current/voltage feedback terminology. It was incredibly confusing and often incoherent. I've seen it give demonstrably incorrect analysis on numerous occasions (mostly in cases where there aren't recognizable A and B terms in the circuit, such as in the inverting amplifier topology). Maybe my professor and textbook were both just awful, but I'm been convinced that this method of analysis isn't worth bothering with. I've had better success just algebraically solving small signal models.

 

[LvW]

I think the confusion (using the series-shunt terminology) results also from the fact that even this terminology is applied in two different forms:
For example: The term "series-shunt feedback", which means "voltage feedback controlled by a voltage" sometimes is named as "shunt-series" feedback.
In the latter case, the authors modify the sequence - they like to say: voltage controlled feedback in form of a voltage".
For my opinion, one reason more to disregard this confusing terminology.

 

[FvM]

I'm not completely aware of the different "schools" of feedback terminology in English text books. Gray/Huerst/Lewis/Meyer are using it as a to/from notation, series-shunt means series connected at the input and "shunted" at the output. I admit, that I need to translate it in my head and check twice before writing about it. It's clearly not my preferred description. Don't know, if someone uses the terms the other way around, this would be really weird...

Razavi prefers the (in my eyes) more intuitive voltage-voltage as from/to notation. Saying "voltage-controlled voltage" instead of "voltage-voltage" would indeed add some clarity.

In my opinion, distinguishing between the basic four topologies is basically helping to get an intuitive understanding of feedback operation. The differentiation is real in the same way as we can distinguish between voltage and current sources or voltage and current amplifiers. The description however abstracts from real circuit properties. In most cases, it's still useful, but in some cases you may face difficulties to determine a clear topology at all. The lastest, when the signal flow direction through the feedback network becomes questionable, because you have as much feedforward as feedback, you better go without it.

As mtwieg stated, network equations still apply.

 

[npn]

Thanks everyone!
They taught us control theory at college at around the same time as this, so i tend to visualise a feedback amplifier as its canonical form . The standard equation for a feedback system is A/1+AB. In books, to get a similar expression, they change the type of amplifier. That's what had been confusing me.