Current feedback - Voltage feedback, how do I see the difference?

Status
Not open for further replies.
Jump to Latest
#261
millwood said:



if we had relied on textbooks for truth, the earth would have been the center of universe, no man would be flying and telephony wouldn't have been possible, etc.

and I am sure whatever professors you had in college wouldn't be worth his / her name tag's worth if s/he couldn't recognize the inherent and real difference between a CFB and a VFB.

Majority cannot assure you the truth but you have to wonder why the majority doesn't agree with you.

The bandwith gain product thing (which some mentioned a while ago) is real and you will not be able to get a VFB to function like a CFB in that department.

If you could, you would have defied the known laws of physics (not that it is impossible).
Click to expand...


Millwood,

My offer to send Cherry's study still stands.

He in fact did just the modeling work-out you mention, although you surely wouldn't like the result.

One conclusion he comes to is that a so called CFB obtains its seemingly constant bandwidth by having a lower open loop gain over a wider band than would be able with the same components in a classical VFB opamp. If you would cripple a classical VFB opamp the same way, that is, lowering the gain from what would be maximally possible, that classical VFB opamp also shows almost constant feedback versus closed loop gain. It's a little bit like putting a resister load on a Vas stage, it looks like the freq characteristic becomes flatter, but you are only lowering the open loop gain below the turn-over point.

Jan Didden
 
#262
Upupa Epops said:
To Jan : Jan, you are realy good theoreticaly thinking man, so I have for you this question : whatabout to use in differential input stage " supertransistors ", used in OPA 660 ( for example ), 'cos they have excellent transfer characteristic ?
Click to expand...


Yes, that would work I think. But you would be using only a unipolar part of it. presumably you would put a collector load in the collector lead and pick off the amplified signal there. Actually, it is an interesting idea. Can you simulate it?

If you don't have a model for the OPA660, try the AD844 or 846. That's almost identical in concept, the only difference being that the output of the "diamond transistor" (another marketing term if there ever was one, this circuit is used in almost all opamps!) is internally connected to the buffer.

Jan Didden
 
#263
janneman said:
My offer to send Cherry's study still stands.

Jan Didden
Click to expand...

I would love to see the cherry study. is it available on the web?

janneman said:
the "diamond transistor" (another marketing term if there ever was one, this circuit is used in almost all opamps!)

Jan Didden
Click to expand...

I would agree with that. But having fully symmetrical common emitter stages may have some merits.
 
#264
@ Jens,

You have a point here. Up to my knowledge CFB and VFB is not mentioned in textbooks. What is put up there are simply the basic 4 variations: series-shunt etc.

The point is what do we consider as a feedback signal. Is that the signal fed back to the input, which can be either current or voltage. Or is that a (scaled) copy of the output signal? Which can also be a current or a voltage. The phrases “Current feedback” and “Voltage feedback” do simply not define what is meant.

But let’s look how it is dealt with in an analogy in motion control. There we speak of “velocity feedback” and “position feedback” or a combination of both. Here we clearly mean the output of the system and not the feedback signal itself. That can be a mechanical signal or an electrical one like a voltage derived from the output velocity/position.

@ Jan,

Re-read the app notes on CFB op-amps. The higher gain-bw product at higher closed loop gain relies on the lower impedance of the feed back network then (driving the neg. input harder) and as such keeping closed loop bandwidth almost constant regardless of the closed loop gain setting. You set the gain by changing the resistor from input to ground and not by changing the resistor from output to input. The latter one is usually given by the manufacturer for optimum performance. But you can tweak the bandwidth of the amp by the total impedance of the feedback network as seen by the neg. input. You can’t do that with a VFB op-amp.

Cheers 😉
 
#265
Pjotr said:
.......But let’s look how it is dealt with in an analogy in motion control. There we speak of “velocity feedback” and “position feedback” or a combination of both. Here we clearly mean the output of the system and not the feedback signal itself. That can be a mechanical signal or an electrical one like a voltage derived from the output velocity/position.
Click to expand...

in all cases, including classical control, feedback is defined by the feedback networks transfer function....


Pjotr said:
..... The higher gain-bw product at higher closed loop gain relies on the lower impedance of the feed back network then (driving the neg. input harder) and as such keeping closed loop bandwidth almost constant regardless of the closed loop gain setting. You set the gain by changing the resistor from input to ground and not by changing the resistor from output to input. The latter one is usually given by the manufacturer for optimum performance. But you can tweak the bandwidth of the amp by the total impedance of the feedback network as seen by the neg. input. You can’t do that with a VFB op-amp.

Cheers 😉
Click to expand...

correct in all respects.....but this simply means that the feedback network is allowed to affect the foward path gain of the first stage in so-called 'current feedback' amps., which does not in anyway add or detract from the fact that the feedback networks transfer function is still a voltage in both cases.....

...which is probably why most textbook authors don't bother discussing 'current' feedback topologies as a separate feedback application....
 
#266
Pjotr said:
[B[snip]@ Jan,

Re-read the app notes on CFB op-amps. The higher gain-bw product at higher closed loop gain relies on the lower impedance of the feed back network then (driving the neg. input harder) and as such keeping closed loop bandwidth almost constant regardless of the closed loop gain setting. You set the gain by changing the resistor from input to ground and not by changing the resistor from output to input. The latter one is usually given by the manufacturer for optimum performance. But you can tweak the bandwidth of the amp by the total impedance of the feedback network as seen by the neg. input. You can’t do that with a VFB op-amp.Cheers 😉 [/B]
Click to expand...

Are you kidding? I can DREAM those app notes! Maybe you should read Cherry, I bet you haven't been within a mile of it.
THEN we can talk.

Jan Didden
 
#269
Pouring gasoline onto the fire.....

Here is an explanation on how current mode amps work. Written by Sergio Franco.

Notice the mention of series-shunt feedback.

So there!
 

Attachments

  • cmtext.gif
    cmtext.gif
    30.2 KB · Views: 225
#271
Re: Pouring gasoline onto the fire.....

Jocko Homo said:
Here is an explanation on how current mode amps work. Written by Sergio Franco.

Notice the mention of series-shunt feedback.

So there!
Click to expand...


yup....series-shunt.....in other words, voltage feedback.....

....the rest is rather dubious though...... :scratch2:

for thorough appreciation of this unfortunate topology, see Cherry....
 
#272
Pjotr said:


That was not so obvious from post #261




Are we bashing again? Very helpful Jan :no:
Click to expand...

No bashing intended Pjotr!
The app notes are there to sell opamps. Cherry's paper is there to explain feedback topologies. That's why.

And I can't see the connection between post # 261 and knowing app notes?? In that post I mentioned some results of Cherry, I thought that is clear.

Jan Didden
 
#273
Re: Re: Pouring gasoline onto the fire.....

mikeks said:



yup....series-shunt.....in other words, voltage feedback.....

....the rest is rather dubious though...... :scratch2:

for thorough appreciation of this unfortunate topology, see Cherry....
Click to expand...

Yes, isn't that curious. Cherry dissects so-called CFB opamps and comes to a rudimentary diagram which looks a lot like Fig 25-1, but his conclusion (reasoned) is also: voltage feedback. Maybe I have to read Cherry again. The mind starts to boggle...

Jan Didden
 
#274
Difference or not?

To all our discussion participants: Can you do this with any OP-Amp? Or only with "so called" CFB-Amps?

http://www.diyaudio.com/forums/showthread.php?postid=129558#post129558

I strongly believe that there are quite important differences between those amps.

And I do not believe that the engineers are chosing them based on "wrong" (according to some statements here) terminologies from the marketing guys. Selling Op-Amps to engineers is not the same as selling audio gear to consumers.

I share Christers un-emotional and pragmatical point of view about all this stuff.

Ciao
Tino
 
#275
Re: Re: Pouring gasoline onto the fire.....

mikeks said:



yup....series-shunt.....in other words, voltage feedback.....

....the rest is rather dubious though...... :scratch2:

for thorough appreciation of this unfortunate topology, see Cherry....
Click to expand...


of course it is voltage that's being fed back to the input: a current feedback amp produces a voltage signal.

so you couldn't have x-series feedback configurations on a cfb.

I thought that was clear from day one.
 
#276
Oh, brother.......

for thorough appreciation of this unfortunate topology, see Cherry....
Click to expand...

Unfortunate???? Just because you don't like it??

Give us a break.........

Can't you do anything other than genuflect at the altar of Cherry??

He may be some sort of academic god to you, but to some of us he is just another pompous university professor. Quoting him really impresses the hell out of us.

Jocko
 
#277
Nelson Pass said:
...the industry refers to as a current feedback op amp in that
it has a low impedance negative feedback input - the emitters
of the (buffered) input transistors.
Click to expand...

I don't know many times I should repeat myself, but the
differential input is inherently responding to a low impedance
voltage, and making a distinction as to whether this
constitutes response to current or voltage is meaningless,
as no input of any amplifier can really ignore either, although
they try.
 
#278
#279
Re: Oh, brother.......

Jocko Homo said:


Unfortunate???? Just because you don't like it??

Give us a break.........

Can't you do anything other than genuflect at the altar of Cherry??

He may be some sort of academic god to you, but to some of us he is just another pompous university professor. Quoting him really impresses the hell out of us.

Jocko
Click to expand...

Jocko,

Fully agree, this forum has more than it's share of genuflection. My reason for referring to Cherry's paper is that it is the only one I know that discussed feedback topologies quite wideranging and detailed. Which indeed does not make him a deity. Which goes for all of us, of course.
Feeling better?😉

Jan Didden
 
#280
Re: Re: Oh, brother.......

Pjotr said:


Yes indeed, let’s come up with somebody else:

Have a look here

And for a break you can read this:

“Theory and Practice -Thinking Styles in Engineering and Science”

Beware, it is rather philosophical but maybe it can give some insight on how this discussion goes on and on and on and maybe some idea why.

Cheers 😉
Click to expand...


Pjotr,

Good references. Especially # 1 seems interesting. His ref 6 is a thick tome I bought a few years ago. Above my head, mostly. Meant for IC designers.

Jan Didden
 
Status
Not open for further replies.
Top Bottom