The characteristics of a "current feedback" amplifier are due to the fact that the first stage is basically a common emmiter amplifier with major loop feedback applied to its emitter.
Recall that the gain of a common emmiter amplifier is simply Rc/Re, where Rc is the collector load and Re the effective emitter resistance or impedance as the case may be.
Because the major loop feedback network is connected to the first stage's emitter, it modifies the gain of the first stage and, consequently, the forward path gain of the amplifier.
In other words, reducing the thevenin impedance of the feedback network reduces the amplifier's forward path gain, and conversely.
Further, because "current feedback" amplifiers actually employ shunt derived series applied major loop negative fedback, they're, in fact, voltage feedback amplifiers and were refered to as such long before someone decided to call them "current feedback" amplifiers.
A true current feedback amplifier must necessarily possess series derived shunt applied mojor loop negative feedback.
On the contrary, Douglas is correct: "current feedback" amplifiers are inferior (in audio applications) to voltage feedback amplifiers which use a differential amplifier for their first stage.
Michael, stop talking about resistors and forward loop gain because that is what is causing the confusion.
Think about the current flows in the feedback network and up through the common base summing junction stage. Therein lies the correct answer.
You have an entirely different situation with a VFA. The feedback current in these topology amps does not directly drive the TIS and is only related to the TIS input current in as much as the LTP transistor beta effects it. In a VFA the feedback voltage is the key quantity and it steers the LTP current into the TIS input or away from it as the case may be.
You can test this proposition simply by changing the values of the feedback network in a VFA, but keeping the ratio the same. Within reason, you will not affect the performance of the VFA in frequency response, rise/fall times or bandwidth. In a CFA, the result will be very different - you will change the rise fall times and the bandwidth of the amplifier.
Think about the current flows in the feedback network and up through the common base summing junction stage. Therein lies the correct answer.
You have an entirely different situation with a VFA. The feedback current in these topology amps does not directly drive the TIS and is only related to the TIS input current in as much as the LTP transistor beta effects it. In a VFA the feedback voltage is the key quantity and it steers the LTP current into the TIS input or away from it as the case may be.
You can test this proposition simply by changing the values of the feedback network in a VFA, but keeping the ratio the same. Within reason, you will not affect the performance of the VFA in frequency response, rise/fall times or bandwidth. In a CFA, the result will be very different - you will change the rise fall times and the bandwidth of the amplifier.
Last edited:
There is no such thing as "forward loop gain", and, no, there is no confusion on my part.
The current flowing through the feedback network is irrelevant, because it is the output voltage being sampled (shunt derived negative feedback) and NOT the output current.
Last edited:
I'd be interested to hear some evidence to back up these assertions.
The fact that there is much talk on a topic does not mean a lot. I seem to recall much discussion of the Mayan calendar recently.
Can you point me to a CFA that gives Blameless performance?
Last edited:
Douglas, we could well ask you the same question about your assertions. . .
Show us a VFA that has the same performance as a CFA?
You want ultra low distortion - I want wide bandwidth
You say PSRR is Important - I think fast rise fall times are important
and so on it goes.
See where this will get us? No where - and we are not even talking about the subjective aspects.
Last edited:
at least one DSL driver "CFA" op amp that has been characterised for audio has really good specs: http://www.ti.com/lit/ds/symlink/tpa6120a2.pdf
and my few quick sims at power amp levels does suggest "CFA" front end linearity for large error signals can be usefully better than diff pair
this makes the topology a natural match to "low global feedback" designs which force large input error signals
but the possible advantage isn't unique to the TZ input differencing
and my few quick sims at power amp levels does suggest "CFA" front end linearity for large error signals can be usefully better than diff pair
this makes the topology a natural match to "low global feedback" designs which force large input error signals
but the possible advantage isn't unique to the TZ input differencing
Yes, I have looked at that TI part quite closely before and the specs are very good.
I would say 'in general' CFA may be easier to comp as well, since you don't have the integrating TIS that you see in a classic MC VFA - although MIC of course allows you to get to high SR's. I recently tried comp'ing a CFA power amp with just the feedback resistor values and it works remarkably well - something I may explore further in a future design.
I would say 'in general' CFA may be easier to comp as well, since you don't have the integrating TIS that you see in a classic MC VFA - although MIC of course allows you to get to high SR's. I recently tried comp'ing a CFA power amp with just the feedback resistor values and it works remarkably well - something I may explore further in a future design.
CFA input stage properly designed gives way better performance, practically in every aspect, than usual LTP. LTP has one nonlinear part in a feedback path too much, since it only introduce extra non linear distortions in a system that it has to be additionally corrected with even more complex error correction currents. CFA is commonly degraded because of non proper designs, usually never used CCS for the input bias current, only injecting resistor. That is distortion mechanism never eliminated from this input topology. Design properly and than compare.
Also the ratio between feedback and CFA input stage currents is commonly mismatched. CFA topology needs very low feedback divider impedance to operate in low distortions. So a lot of things to discuss and to teach in the book.
I would say that rather many CFB chips at least have very good specs and the TPA6120 is pretty nice and my all discrete CFB amp am I very pleased with. I'm not sure with Mr. Self means here but I'll guess we talk about different things.
Semantics: "The study of language meaning"
Try to resist the use of words before establishing their true meaning.
The principles I've communicated on so-called "current feedback" amplifiers are elementary Higher National Certificate electronics here in the UK.
Just had a quick look.
5 microvolts noise at Av of 2
Wouldn't an AD797 have less than 1 microvolt in that condition?
Planned to ask you about noise in your Sx amp thread but since you're here
Seems to me the input emitter follower pair add noise then the common emitters add their noise at the same level.
In a typical VFA LTP the first pair are common emitters and only they effectively add noise.
I assume this is why CFA is usually said to have worse noise but I haven't studied it much yet.
Maybe not important in your application (90 dB sensitive speakers IIRC) but in my 115 dB compression drivers...
Best wishes
David
Last edited:
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.