1
\$\begingroup\$

"Feedback is a process in which information about the past or the present influences the same phenomenon in the present or future"

I find the idea of feedback quite fascinating and i could realize how universal it is ( applying not only to EE but to society, biological systems,etc ).

But for now, i'm focused on EE.
I'm interested in trying to capture all kinds of feedback used in electronical circuits ( of whatever kind, whether digital or analog ), so that i have the ability of looking a feedback in any electronical circuit and know which task its implementing there.

So far i got three different kinds of feedback.
1 - Positive feedback that tries to reinforce the value of the input as time passes. It is used in any circuit that implements hystheresis, in multivibrators, oscilators, etc ...
2 - Negative feedback that tries to drive the output value to desired/reference values. It is used in any control system.
3 - The kind of feedback used in sequential circuits, which acts as a way of the combinational circuit and the inputs to make the states evolve.
enter image description here

Now i'm curous ... My question is :
Are there more kinds of feedback other than those 3 being used in EE circuits or these three kinds would cover all kinds of feedback used in EE circuits ?

Thanks a lot in advance

\$\endgroup\$
2
\$\begingroup\$

Positive feedback circuits (e.g. a Schmitt trigger) are in unstable equilibrium, the same as a ping pong ball on top of an upside-down plastic cup. You only have to move the cup one way or the other a little bit and it wants to roll off. Assuming the cup has a large lip and the ball doesn't roll off the end, then you can make it go the other way by moving the cup the other way, but the ball wants to stay at the ends - it's hard to make it stay in the middle.

This applies to multivibrator circuits, that either oscillate from one side to the other (astable, like a clock) or wait for an input (monostable, like a pulse generator or bistable, i.e. a flip-flop). This feedback is all about reinforcing a state for constant inputs and switching it to the other state for changing inputs, which can be used to create memory and state machines (i.e. sequential logic circuits, like you mention). A flip-flop for example can be used to store 1 bit of information (you can check the last position (i.e. state) you left the ball in - left or right).

A negative feedback circuit (e.g. amplifier, active filter) is the opposite - it's a stable equilibrium system. Now the cup is turned round (the right way up) and the ball is in it. You can move the ball to the edges of the cup by moving the cup either left or right, but once you stop it will return to the centre, assuming you don't exceed the limits of the system (which would be dependent on the centre of mass in the ball example). This type is about control/processing.

In an amplifier, for example, the output wants to be zero. An increasing input (in either direction) causes a increasing output (with some gain possibly), which is fed back and taken away from the input (as opposed to adding it in the case of positive feedback), making the output smaller. You have to keep increasing the input or it will go back to zero.

Another example is automatic gain control (AGC), where you might have a radio receiver amplifier with a variable gain. The amplifier output is fed back to a control circuit that adjusts the gain of the amplifier depending on the strength of the received signal, thus maintaining a constant output. Other control examples include applications where there is fine-tuning of movement (like a robot hand that grips objects without crushing them by sensing the exerted pressure, feeding it back and adjusting the gripping force, converging on a constant value).

\$\endgroup\$
  • \$\begingroup\$ Thanks for all that input in describing the two first kinds of feedback i described ... My problem is whether we could perceive more kinds ( other than the 3 i listed ) of feedback in electronic circuits. \$\endgroup\$ – nerdy Jul 11 '14 at 19:51
  • \$\begingroup\$ Positive and negative are the only types. The third type you mention is an application of the first and is briefly discussed above. \$\endgroup\$ – user49118 Jul 11 '14 at 19:57
  • \$\begingroup\$ I think I get what you are saying now. You're asking what specific reasons there might be for the feedback loop in your general diagram? If so, the question is really about logic and you're actually asking what reasons one might have for knowing the previous state of an output when processing the current input. \$\endgroup\$ – user49118 Jul 11 '14 at 20:25
  • \$\begingroup\$ I just realized what you meant and i'm starting to see that just as the servo-mechanism is a negative feedback for changing inputs, the feedback in sequential circuits is jusst positive feedbackwith changing inputs... Is that it ? \$\endgroup\$ – nerdy Jul 11 '14 at 20:27
  • 1
    \$\begingroup\$ I can't see why negative feedback could not be implemented in digital systems, PID to mention it alone can be discretised and used in microprocessors. \$\endgroup\$ – Mister Mystère Jul 12 '14 at 0:03
2
\$\begingroup\$

Control systems:

  1. OPEN LOOP
  2. CLOSED LOOP (Feedback)

    2.1 Positive feedback

    2.2 Negative feedback

  3. FEEDFORWARD
\$\endgroup\$
  • \$\begingroup\$ So, you listed 2 things not related to feedback and one thing related to feedback but that just named the two kinds of feedback i had mentioned on my original post ... \$\endgroup\$ – nerdy Jul 11 '14 at 19:51
  • \$\begingroup\$ Exactly. Very often, instead of describing in detail what is already available, it is more useful to situate the context of the subject. \$\endgroup\$ – Dirceu Rodrigues Jr Jul 11 '14 at 20:09
  • \$\begingroup\$ We also should realize that such a classification ("positive/negative" feedback) is rather problematic because always there is a smooth transition between both types. In most cases (perhaps in all?) and for rising frequencies a desired negative feedback will turn into positive feedback, caused by unavoidable phase shifts. Hence, where is the "threshold" between positive and negative feedback? \$\endgroup\$ – LvW Jul 12 '14 at 8:32
  • \$\begingroup\$ Correct. The time delays (or its frequency domain equivalent - phase shift) can change the sign of feedback (positive/negative). \$\endgroup\$ – Dirceu Rodrigues Jr Jul 12 '14 at 14:32
1
\$\begingroup\$

If I understand your question properly then your idea of the different types of feedback is flawed. The idea that only certain types of feedback exist in circuits (schmidt-trigger, etc.) is a much higher-level perspective than it actually is. These are just circuits made to interpret a condition.

The types of feedback you listed are not anything special in that it is still just either a 0 or 1, just like everything else in digital systems. In regards to analog circuitry, it will still be below threshold or above threshold

Not only that, but the number ways that a device can react to different inputs is infinite. There are so many combinations of the foundational components that we could not list every conditional.

Instead of taking a top-down look at the EE systems, take the ground-up. It is all about 0s, 1s (digital) and passing thresholds (analog).

\$\endgroup\$
  • \$\begingroup\$ Feedback may also be analog, so your second point does not make much sense. Op amps and BJTs for example are used often with analog feedback loops. \$\endgroup\$ – sherrellbc Jul 11 '14 at 18:17
  • \$\begingroup\$ If you'll look at the last sentence, you'll notice I address this. \$\endgroup\$ – Funkyguy Jul 11 '14 at 18:23
  • \$\begingroup\$ Just because i can realize amplifiers in a reaaly large number of ways doesnt mean that they all stop sharing their intrinsic property of amplifying signals ... Just because we can realize feedbacks in a varied number of circuits doesn't mean we cant group circuits that used feedback in a similar manner, in the same category. It's all about abstraction. The negative feedback of a feedback amplifier is doing exactly the same thing as the negative feedback of a servo-mechanism, only the results are different because the circuits that implement the feedback is different. \$\endgroup\$ – nerdy Jul 11 '14 at 19:48
  • \$\begingroup\$ Also, i don't understand how you could say that all kinds of feedback i listed ( the negative feedback of a feedback amplifier or the positive feedback in a oscilator ) are not anything special in that it is still just either a 0 or a 1. This doesn't make sense, oscilators and feedback amplifiers in that case are analog circuits. \$\endgroup\$ – nerdy Jul 11 '14 at 19:48
  • 1
    \$\begingroup\$ Just accept ricks answer. He and I are saying the same thing, just in different ways. \$\endgroup\$ – Funkyguy Jul 11 '14 at 20:21
1
\$\begingroup\$

No. In fact your third is simply a subset and/or a combination of your first two depending on its function.

You could arbitrarily break positive and negative feedback up into smaller subgroups, but I don't see the point. Do you see a point?

http://en.wikipedia.org/wiki/Feedback#Types

\$\endgroup\$
0
\$\begingroup\$

Feedback is a phenomenon by which the signal at input of a system is affected by change in output. Based on the effect of this feedback on input, we can classify it into two classes:

  • Feedback that augment the cause - positive feedback (analog) or additive feedback (digital)

  • Feedback that lessen the cause - negative feedback (analog) or subtractive feedback (digital)

\$\endgroup\$
  • \$\begingroup\$ So, you just made short sentences describing the first two kinds of feedback i listed in my post ... and this has nothing to do with my question :( \$\endgroup\$ – nerdy Jul 11 '14 at 19:53
  • \$\begingroup\$ OOPS. My comment is not for this response (typo error). \$\endgroup\$ – Dirceu Rodrigues Jr Jul 11 '14 at 20:07
  • \$\begingroup\$ @nerdy Sorry about that. I just wanted to say that there are only two classes of feedback. The 3rd kind in your original post can be classified into one of the above two classes. \$\endgroup\$ – nidhin Jul 12 '14 at 3:47

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.