vzaichenko,
strength is associated with force, pressure, amplitude, intensity. The low amplitude (feedback) signal at the source / emitter node is the distinguishing feature of the circuit topology in which the term "current feedback amplifier" has its origins, in contrast to the differential amplifier (op-amp) that exhibits very high gain required for voltage regulation.
Your halting definition fails to capture the notion.
strength is associated with force, pressure, amplitude, intensity. The low amplitude (feedback) signal at the source / emitter node is the distinguishing feature of the circuit topology in which the term "current feedback amplifier" has its origins, in contrast to the differential amplifier (op-amp) that exhibits very high gain required for voltage regulation.
Your halting definition fails to capture the notion.
Hi,
This is simply not the case. In both VFA with its "classic" LTP input stage where the feedback signal comes to the base of "-" input (high impedance node), and CFA with its "classic" complementary input stage where the feedback comes to the emitter(s) of "-" input (low impedance node), the amplitude of the feedback signal is the same. You've got an impression the amplitude is lower in CFA configuration by some reason - in fact, it's not. It's practically the same, assuming the feedback network impedances are chosen correctly in each case (it has to be significantly lower in the case of CFA configuration).
See the document at the first link in Post #655 by Walt Jung - SECTION 1-2, pager 1.23, 1.24.
Cheers,
Valery
Large signal amplitudes cannot develop over small resistances and within small voltages. As a consequence of this fact, current feedback inflicts less harm on the input signal.
In the diametrically opposite case of VFA, the feedback signal is applied to the highly sensitive input terminal of an additional gain stage producing a large distorted amplitude.
Given the topological differences, the overall impact of feedback is like like night and day.
You impose a significant restriction on yourself by seeing the world only in terms of impedance.
Please... read some good material about the feedback theory. As I mentioned, the amplitudes of the feedback signal in CFA and VFA cases are the same. The currents are different, but the amplitudes are the same. This is the fact. If you think otherwise - please read the theory. Please get me right - I'm not trying to be arrogant or something like that.
Same wrong assumption. Sensitivity of the inputs is the same. In the case of CFA we need more current to drive the "-" input (because of its low impedance by the way), so we use higher power rating resistors for the feedback network (same amplitude * higher current = more heat).
See one of my CFA front-end designs HERE. With +/-50V rails, I have to use 4 x 1W resistors in parallel in the feedback network.
There may be even more interesting configurations, like an injection of the feedback signal at the output of the VAS stage (ODNF). See the other design HERE.
Yes and no. That's actually the point of discussion in this thread
This is not true - some people here know I'm not the kind of guy who uses to restrict himself
However, as I mentioned earlier, for understanding what you drive the node with - voltage or current - impedance ratio is just a good technique. Although there are tons of other factors to consider when designing a good amplifier
Jan, that's surely an even simpler explanation. But I'm not sure the person we address it to is ready for it
(no offense).If I'm wrong or I'm not sure in something - I go study the subject. Then I'm ready for discussion. Some people continue pushing the wrong statement without trying to listen
It's like:
1) I know that I know;
2) I know that I don't know;
3) I don't know that I know;
4) I don't know that I don't know.
Option 4 is the most dangerous one
Definitely an offtopic here - sorry about that
Cheers,
Valery
Even that can be put in perspective. In the equilibrium state, assuming no gross distortion, the input signal and the feedback node signal are almost identical. The differences are often in the single-digit mV range (though increasing with high frequencies).
So even with a lower impedance at the feedback node, the current into the feedback node is still quite small because of the low signal level.
This is easy to see with the input stage model of a CFA, see attached. Even if the V- input is low impedance, if V+ = V-, no current will flow.
Jan
Attachments
This is actually a good point. The signal difference between the "+" and "-" inputs is rather small, however, in order for the feedback to control the "-" input (which is actually an output), the feedback network has to be rather high-current (low-impedance).
By the way - you know it better than me - this kind of feedback approach is well-known by the tube amps designers for very long time - I mean the configuration where a feedback signal comes to the cathode of an input tube
In this case, the bias and the feedback networks are often combined, using a capacitor to split the cathode resistor and separate the DC (bias) and AC (feedback) operations.
Cheers,
Valery
And here we have some interesting opportunities observed, like subtracting those signals in a separate circuit, slightly amplify the resulting error signal, and then subtract it from the full high-swing mail signal at the VAS output, for example (ODNF approach, mentioned above). The advantage of this arrangement is that all the error detection and processing circuit is small signal low power class A, so it's pretty easy to make it precise, fast and low-distortion.
Cheers,
Valery
By the way - you know it better than me - this kind of feedback approach is well-known by the tube amps designers for very long time - I mean the configuration where a feedback signal comes to the cathode of an input tube
In this case, the bias and the feedback networks are often combined, using a capacitor to split the cathode resistor and separate the DC (bias) and AC (feedback) operations.
Cheers,
Valery
Yes, cathode feedback is very, very old. But it was never called current feedback because that was defined as feedback depending on the output current rather than the output voltage.
The confusion started by some guys at Elantec who might not have been aware of the existence of the definition of current feedback as feedback from output current.
Chris Paul in his article has explained this very nicely.
Jan
I don't know; if the error voltage (at the feedback node) is say 10mV, and the input impedance is 50 ohms, the feedback current is just 200uA. Not really a reason for wirewound resistors ;-)
I think the reason for low impedances in the feedback of a CFA has more to do with bandwidth and stability (in fact the Rs is used to tune these parameters).
Jan
I see this discussion is still going on! 5 years since the first round
Good to see you back
I'm not aware of any substantial difference between CFAs and VFAs in the voltages across feedback resistors. Because of its high gain, certainly the magnitude of the current flowing into the (-) CFA terminal is typically so small as to dissipate little power in the feedback network. Am I missing some context here?
I'm a little concerned about statements to the effect that feedback signals in VFAs and CFAs are the same, but perhaps I've misunderstood. Due to both circuits' high gains, yes, the differences between the voltages at their (+) and (-) terminals are very small. But there are some basic differences between the feedback signals in the two that make it difficult to argue that they are the same. Please consider the following.
The signal portions of the outputs of both input stages are currents. In the VFA, those currents are sourced substantially from a constant current bias source in the input stage. These currents are steered by the difference in the voltages applied between the (+) and (-) terminals. In the CFA, the signal currents are derived through the feedback network which is driven by the amplifier output.
Yes, you are missing something. The feedback factor, or loopgain, for the CFA amplifier, is Zt/Rf
Zt ... transfer impedance
Rf ... feedback resistor
Zt is usually about 3Mohm. So, to get 80dB feedback, you need Rf = 300 ohm. If you use only 60dB feedback, the Rf = 3k, but the higher the Rf, the lower the bandwitdth. So, there is an optimum for Rf, usually less than 1k.
If your amp gives 30Vrms, then feedback resistor has to handle 30*30 / 300 = 3W in case that Rf = 300 ohm.
These are CFA very basics and I strongly recommend to study the mentioned Kester's AD publications or just google to find Intersil documents etc.