Funny, but too bad... it was a good link that can help to gain a better understanding as independent of its flaws. Thanks Hans, Jan and Scott.
Fig 1 and Fig 2 doesn't have feedback either. Calling the signal Vf doesn't make it so. The question became: With the In+ node connected to ground, what stimulus was there to have caused a Vf to begin to exist, or for that matter to be sustained? The so-called Vf isn't different from any other independent signal, being imposed on the In- node, making Fig 2 as derived from Fig 1 identical to Fig 3 using Thevenin equivalent voltages and resistances, and thereupon making the identification as Vf fictional.
Figure 4 is an example that can be simulated to illustrate the onset current waveform caused by a step function input stimulus without feedback. The frequency of the input square wave is selected as arbitrary. It should be noted that Ro1 is the output resistance of the input buffer, hence if any library devices are selected modelled, Ro1 should be eliminated as already internal to the device.
In an AD844 CFA device the Tz node resistance is approx. 3 M Ohm, meaning that a full scale output of +/- 15 volts requires 5uA to achieve, being far less than input stimulus current passing through R Thevenin to V Thevenin in this simulation. Cf and Cg have been included to show that in the R Thevenin equivalency both capacitors exist in effective parallel and therefore both contribute to peaking as a cause for potential instability that Scott commented on earlier.
The above is intended to support the conclusion that input stimulus current exists as a function of voltage in absence of feedback, preceding feedback return even if in the nano second region. The existence of delayed feedback can be simulated by increasing the value of the Tz node capacitance.
This is restate that in a process control there exists both voltage determined input stimulus currents flowing through the R Thevenin equivalent resistance and current limiting by feedback, being delayed in time, phase, or otherwise. It is the ratios that remain unclear, being extreme or otherwise, in any determination if voltage input stimulus is greater or lesser than current limiting feedback.
I look forward to understanding is there exists any relationship between the Middlebrook solution in the determination of a VFA or CFA and mine, particularly at the point of crossover.
Of course there is a time delay but the better question is whether this delay is important. I assume you are talking about speed of light sized effects?
Phase delay has no innate importance to me, at light speed or otherwise, rather only that it recognizes that two simultaneous effects exist in a control process as both contributing to a device being recognized as marginally or extremely a VFA or CFA device. How I think of a CFA is that of a current limiting feedback amplifier, this in recognizing that input stimulus exists in the overall control process of a CFA, not in absence of it.
'Phase delay' ??
You are desperately looking for straws to support your position, and now you even come up with imaginary straws.
Jan
You are desperately looking for straws to support your position, and now you even come up with imaginary straws.
Jan
I believe you are making a big mistake. Vf is an independent signal where it seems that you interpret it a Vf(eedback), but it is the stimulus itself.
Figure 4 is an example how to verifying open-loop transimpedance gain z(jf), feedback factor β(jf), and loop gain T(jf) in a CLOSED LOOP system.
In the PDF you have added, you are stimulating the In+ with a square wave. Totally different from fig 3 and 4 you are referring to.
As I suspected before, you are telling one thing but doing the opposite.
Hans
As a matter of fact, we have passed page 200 and posting #2000.
Isn't it time to celebrate this as a jubileum and take a beer ?
Hans
Isn't it time to celebrate this as a jubileum and take a beer ?
Hans
I think it was W.C. Fields who said, "“Everybody's got to believe in something. I believe I'll have another beer.”"
Attached is my writeup on this topic. I would apologize for the poor editing, hand scribbling and spelling (or lack thereof), it is the result of a couple of hours of quick brain-to-paper dump. I'm not in the mood of putting this in a pretty form, it's holiday season after all.
The algebra was reduced to a minimum, to ease the comprehension, it can be expanded by request. I believe if this approach will not persuade the unbelievers that, from a small signal properties, CFAs are of the same breed (feedback type) as your next long tail VFA, and can be analyzed (and their special properties determined), using the standard VFA toolset, then nothing ever will.
Please note that along these 10 pages no "current summing", "inverting input current flow", "gm" or other such specious arguments were used, but only abstract, general models (with the exception of the short implementation discussion).
The algebra was reduced to a minimum, to ease the comprehension, it can be expanded by request. I believe if this approach will not persuade the unbelievers that, from a small signal properties, CFAs are of the same breed (feedback type) as your next long tail VFA, and can be analyzed (and their special properties determined), using the standard VFA toolset, then nothing ever will.
Please note that along these 10 pages no "current summing", "inverting input current flow", "gm" or other such specious arguments were used, but only abstract, general models (with the exception of the short implementation discussion).
As I disagree with your CFA definition, I wouldn't say that the opposite direction of I(Rf) and I(In-) variations is prohibitive to the appartenance of the CFA family.
My point is that the input device current - the In- current- is voltage dependant ( V(in+ , V(in-) , Vce or Vds) and not current dependant.
Do you agree with this point or not ?
Once again, I disagree with your definition of current feedback, so I can't agree with your conclusions.
That is only to say that properties as current on demand, high slew rate, high bandwidth, etc .. are not exclusive to so called CFA.
May be I missed it but is it possible to have the .asc file, please ?
The name "Unbuffered VFA" summarize all the differences between VFA and so called CFA, as it describes exactly what so called CFA are : a VFA where the buffer between feedback network and emitter (source) of input device is removed.
Scott Wurcer showed how to introduce a delay by inserting a shielded wire of some length ahead of the feedback network.
Sorry I meant time delay. What do you perceive as the object of my desperation Jan? A desperate desire to be loved or wanted in the group, by anyone in the group? By you perhaps? Where do get this from, some book on psychology? Have you considered that in psychology the existence of an adversarial agent or collective can cause bonding within a cooperative collective? It seems like a collective hugging is going on that I haven't been invited to attend. This is genuinely amusing as certainly not a cause for any desperation.
Desperation is defined as rash or extreme behaviour that implies an emotional response seemingly suggestive of some desire to satisfy an egocentric self, to attach oneself to some form of will to power or protection. My personal (as opposed to universal) moral obligation is not to lie, meaning that I have no intention to deceive. This is in respect of myself, not in seeking the respect of others whose value judgments cannot be trusted coherent or rational. What I express is perceived true at the time being expressed, notwithstanding that moments later this can be considered untrue.
My motivation is purely to seed a differing view in anyone, not to change the minds of the universal collective. The fundamental principle of a collective, to continue as a collective as potentially in conflict with other collectives, has at its core untruth.
No I haven't ... and I haven't yet considered the arguments presented by many others Hervé.
The question had nothing to do with membership in the CFA family. Let me repeat it:
Are you among those who say that there is no current feedback because the electrons flowing through the input stage do not flow through the output stage? If so, your claim leads to the conclusion that negative current feedback is impossible, as I have made clear in prior posts. Are you willing to state this?
I'm not sure what you mean by "dependent." There is a set relationship between voltage and current in the transistor. A portion of the small signal "signal" current in the inverting input transistor in a high loop gain CFA is due to the transconductive effect of gm vbe. This looks like a voltage controlled current source. Another portion is due to the Early effect of vce / ro, which looks like a resistor. Also, the signal current flowing through the collector flows through the emitter and through the feedback network.
It's not "my" definition of current feedback - it's R.A. Middlebrook's. So you disagree with R.A. Middlebrook's Dual Injection Technique and the conclusions he draws from it, as described in the article I cited?
That is correct - you can go to the so called enhanced VFAs to get these features, which are innate to the basic CFA topology.
Certainly. Please see attached.
I beg to differ: it does not. There is no guarantee of current on demand, high slew rate or high bandwidth with a VFA. But these are inherent in a CFA.
Well from your posts I get that your explanation or position depends on a delay existing between input signal and feedback signal. So it is absolutely imperative that this delay exists, otherwise your house of cards falls down.
Jan
Single device input belong to the CFA category and have not these features which are specific of diamond input CFA's.
I did, more precisely I tried to do it. The foot of the stimulus emerging from noise with a low slope prevents a precise measurement with my apparatus.
Last edited:
As CFA means current feedback amplifier, I thought my answer was clear. The opposite direction of these current variations allow current feedback.
By the way, as audio signal doesn't contain DC, electrons which are moving around the input device emitter are not the same which are moving at the output, audio electrons don't flow.
If I resume what you write, we have :
ic = A*vbe + vce/B where A is a transconductance, and 1/B is the inverse of a resistor, so a particular case of transconductance.
It appears that the input device current is completly defined by voltages and input device parameters.
As input device output is the input of the rest of the amplifier, it is not so difficult to establish a relation between vc and output.
Then we have a input device current wich is completly defined by : V(in+), V(in-) and C*V(out) where C is part of the open loop transfer of the amplifier.
I call that voltage feedback.
I disagree with the actual definition of CFA, and I am not sure to understand why we need this MDIT to classify amp in the subcategory of UVFA.
These features are not innate to the basic CFA topology which is single device input.
I have to recall here that -for example- current on demand appears in VFA schematics before the diamond input CFA.
Thanks.
These are not inherent to CFA and some appears in VFA before in the so called CFA.
Last edited: