Forr :
Expressions VFA and CFA should apply to amplifiers where it is either the output voltage or the output current which is under control, and should not describe the way the feedback is achieved [if we admit that the feedback applied to the emitter(s) of the input BJT(s) is of current nature which is fundamentally wrong]
Bonsai's reply :
You are forgetting what the middle letter stands for: feedback
We are talking about the feedback quantity and not the output quantity that is being controlled.
"MFB" (Motion FeedBack in usual terminology for loudspeakers) means a servo mechanism controlling the motion of the cone, not the way it is done. So why not the same for VFA and CFA ?
When I was younger, a feature of feedback applied to the emitter(s) of the input BJT(s) which puzzled me was that the impedance of this input was low but did not enter in the equation of the output voltage.
Provided there is enough loop gain, in the non-inverting configuration, VFA or so-called CFA obey to the gain equation : CLG = (Ra + Rb) / Rb, (Ra connected to the output, Rb connected to ground), do they ?
Has anybody ever said that the controlled loop of the QUAD 405 was of a CFA topology ? Feedback is applied to emitter of Tr2 :
Expressions VFA and CFA should apply to amplifiers where it is either the output voltage or the output current which is under control, and should not describe the way the feedback is achieved [if we admit that the feedback applied to the emitter(s) of the input BJT(s) is of current nature which is fundamentally wrong]
Bonsai's reply :
You are forgetting what the middle letter stands for: feedback
We are talking about the feedback quantity and not the output quantity that is being controlled.
"MFB" (Motion FeedBack in usual terminology for loudspeakers) means a servo mechanism controlling the motion of the cone, not the way it is done. So why not the same for VFA and CFA ?
When I was younger, a feature of feedback applied to the emitter(s) of the input BJT(s) which puzzled me was that the impedance of this input was low but did not enter in the equation of the output voltage.
Provided there is enough loop gain, in the non-inverting configuration, VFA or so-called CFA obey to the gain equation : CLG = (Ra + Rb) / Rb, (Ra connected to the output, Rb connected to ground), do they ?
Has anybody ever said that the controlled loop of the QUAD 405 was of a CFA topology ? Feedback is applied to emitter of Tr2 :
An externally hosted image should be here but it was not working when we last tested it.
Last edited:
So if we're disallowed from calling a VFB amp that because in the limit, no voltage actually is fed back, and we're prohibited from calling a CFB amp that because ideally no current is fed back, where does this leave us in terms of useful nomenclature?
There are still practical differences between the two schemes in that we don't have as much freedom in selecting the fb resistor in the CFB case, and we also can't use a fb cap if we value stability. CFB also does not tie us to a particular GBW product.
There are still practical differences between the two schemes in that we don't have as much freedom in selecting the fb resistor in the CFB case, and we also can't use a fb cap if we value stability. CFB also does not tie us to a particular GBW product.
What's wrong with the accepted definition? Voltage feedback is when you return a sample of the output voltage, 'controlling' the output voltage. ANY power amp where the feedback lowers Zout is a voltage feedback topology.
Any amp where you return a sample of the output current, 'controlling' the output current, as in a current drive amp, is current feedback.
Then the other thing is how you add the returned signal to the input signal, either in shunt like in an LTP, or in series. So you have current derived-series applied, voltage derived-shunt applied etc, four cases. Clear, unambiguous, no confusion.
Of course the topology is different when you return the signal to a transistor base or a transistor emitter; there must be millions of different topologies, and they all have their advantages and disadvantages. Are we all going to give them trendy names??
jan
Of course that using a common base inputs scheme will
yield some differences in caracteristics but neverless
theses are stretched VFBs , and particularly the TI exemple
is not usefull for audio unless its signal input is bufferered ,
who wants an audio amp with 50R input impedance.?..
Besides , in the topologies at stakes here only the negative
input is at low impedance.
Doing some simulations using Siemens 5Ghz devices
with a VFB and a "CFB" biaised such that the critical
standing currents are the same there s not that much
difference in caracteristics , gain/frequency of course , but also ,
surprisingly , phase response that is not worse with the VFB ,
slew rate being 50% higher for the CFA, from 500V/us to 800V/us ,
both numbers being totaly useless in audio applications.
Linearity is about the same but as expected PSRR is 25db worse
for the CFA , a value that itself make theses designs irrelevant
for audio in their simplests iterations.
Attachments
Great investigation Wahab
It confirms what I said in the very beginning of this thread.
It is a short sumup of the differences:
Oh I see - you object to the accepted definition of CFB and VFB and now its me that's doing that? How funny
Sounds fine except I've not, as far as I'm aware in my 40-odd years in electronics, come across any example of the latter. Do you have any? If not then its only fine in theory to have nomenclature which is of no practical use. I should here point out that I have come across 'Ace Bass' type configurations which sense the voltage across a series resistor at the output, and feed back that voltage. That's still voltage feedback because a voltage is being sensed and the output stage delivers a voltage.
Its seeming from what you've written that you want Zout lowering feedback to be called 'voltage feedback' and Zout raising feedback to be called 'current feedback'. But then which one is it that makes Zout negative?
Jan, so a typical CFA input stage can be thought of as a complimentary emitter follower stage driving the equivalent internal high impedance gain node (i.e. 100% feedback for this stage).
This leads me to believe that then the changes in the base emitter voltages for these transistors changes symmetrically (while the collector currents change exponentially but do not change equally, with their difference equal to the required additional load current). Compared to an LTP where the currents change symmetrically and the base emitter voltages change differently because of the non-linear iv-curve. So for an LTP of increasing signal the much lower rate of change of the side turning off leads to a larger error signal (balanced currents) where as the CFA or emitter follower does not have this constraint.
THanks
-Antonio
"What's wrong with the accepted definition? Voltage feedback is when you return a sample of the output voltage, 'controlling' the output voltage. ANY power amp where the feedback lowers Zout is a voltage feedback topology.
Any amp where you return a sample of the output current, 'controlling' the output current, as in a current drive amp, is current feedback."
No it's not- it's current output. And as already noted by someone earlier, you can use voltage feedback (derived across a small current sense resistor) to control the output current.
You are mixing up output controlled quantity with feedback quantity and they are not the same thing.
;-)
Any amp where you return a sample of the output current, 'controlling' the output current, as in a current drive amp, is current feedback."
No it's not- it's current output. And as already noted by someone earlier, you can use voltage feedback (derived across a small current sense resistor) to control the output current.
You are mixing up output controlled quantity with feedback quantity and they are not the same thing.
;-)
So would you like to scrap the CFA concept from the last 50 years of technical literature, as being only a bastardized son of VFA? Hmmm... Wouldn't it be simpler to accept CFA as a topology that may have some advantages in certain applications?
BTW, I think I missed this, why would you think CFA is bad for audio? There are a number of highly regarded (technically and subjectively) models that are using a CFA topology. As far as I can tell, both CFA and VFA have pros and cons. I don't buy Self's opinion about amplifiers with "too much slew rate" or "to much speed" as being more prone to oscillations. High speed and high slew rate are in fact great features of CFA's, without sacrificing any Bode stability margins.
Ha interesting observation
I rather like Bob Cordell's observation that a designer designs according to what he fears the most. So when there's over-engineering (as there most certainly is in the case of chasing vanishingly low THD) and ultra-low noise (when most people's source material is 16bit) there's presumably fear of less-than-blameless measurements. A kind of arms race between the distortion analyser designer and the amp designer?This is not entirely correct.
The type of feedback is defined by the feedback transfer function:
1. If the feedback is shunt (voltage) derived and series (voltage) applied, then the feedback transfer function is defined as a voltage.This is voltage feedback, and the closed loop amplifier is a voltage amplifier.
2. If the feedback is shunt (voltage) derived and shunt (current) applied, then the feedback transfer function is defined as a transadmittance. This, therefore, is transadmittance feedback, and the closed loop amplifier is a transimpedance amplier.
3. If the feedback is series (current) derived and series (voltage) applied, then the feedback transfer function is defined as a transimpedance. This, therefore, is transimpedance feedback, and the closed loop amplifier is a transadmittance amplier.
4. If the feedback is series (current) derived and shunt (current) applied, then the feedback transfer function is defined as a current. This, therefore, is current feedback, and the closed loop amplifier is a current amplier.
Note that category 2 above also lowers the Zout of the amplifier, but is not voltage feedback.
Clearly, so-called "CFAs" belong to the first category above, and should never have been called "CFAs".
Last edited:
Neither do I: slew rate and stability are not related
But if going after very high slew compromises linearity, PSRR, DC offset, then it's not worth it as far as audio is concerned. I can't think of an audio application where "CFAs" are better than VFAs.
Last edited:
Have you done any listening to both kinds of designs by any chance? If so how would you describe the SQ differences?
<edit> I agree with your point about 'if going after slew rate' - but it rather begs the question. Is that what CFB proponents are really 'going after' or is the increased slew rate a side effect of some other benefit - like SQ?
Last edited:
well done that is how an engineer should proceed...but still not consistent.
First to do is to define the transfer characteristics that you want the amp to perform. Next is transfer function and that is realized by type of feedback.
The type of feedback is defined as exactly as possible by the deviation (or error) of transfer function to transfer characteristic.
This yields error pickup and error feedback.
Shorthand for Sound Quality.
Incidentally I wasn't meaning 'scientific' listening tests as those I've come across have only tended to focus on differences between two components. I'm interested in describing the differences (if any) not in identifying if a difference may exist.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.