Controlling the voltage in a resistive sense referenced to ground is controlling the sense current, and the load current.
Personnally I never saw this circuit called as a CFA.
I don't know what you call "modern" or "classic" but I think there are current on demand VFA amp for more than 35 years.
According to your criteria, the attached amp scheme is a VFA or a CFA ?
Attachments
Single device input stages have not this feature but retain the one of output current equal to input current which you adopted as the main criterium of CFAs.
He-he...
Yea.
If the freq response greatly depends of the feedback/source resistance like in common base stage - then feedback are mostly current.
Hi Chris, Oops. You are right. I can't break the loop and apply DIT. Sorry. My bad. I need to think again.
.
I did it in #1338, did you miss that one ?
As long as the inverting input can source or sink a current In, in this case into 100K, the output of the CFA will respond with a voltage In*Z(s).
So even in the absence of Rg, DIT classified it as CFA.
Hans
According to the DIT test with Rf=Rg=1K, it certainly is a very convincing CFA.
Hans
Attachments
Yes. "Modern VFAs" have been said to be enhanced CFAs. The aim of the schematics is to show that the three circuits are VFAs.
Again...
Transistor and circuit topologies strictly denote mode of connection. The nomenclature CFA / VFB pertains to input node for feedback without attributes.
No concealment - no revelation.
Hiraga Le Monstre comprises several current feedback connections and can be seen as the apotheosis of CFA.
Transistor and circuit topologies strictly denote mode of connection. The nomenclature CFA / VFB pertains to input node for feedback without attributes.
No concealment - no revelation.
Hiraga Le Monstre comprises several current feedback connections and can be seen as the apotheosis of CFA.
Ohm's Law. At least we agree on that.
Sometimes called the "Vacuum Tube Bible", see http://www.tubebooks.org/Books/RDH4.pdf top of page 307.
The uA709 is over 50 years old. It and its immediate successors had no current on demand.
My criteria would first deem this to be "amusing." If a practical use for it were found and it entered wide circulation, it might deserve its own name.
Agreed.
He was talking about removing Rf, not Rg. Middlebrook's DIT method requires a global loop to have a return signal.
Gentlemen,
As I’ve mentioned, I’ve not understood the relevance of the rebuttals you’ve supplied to Post 1028. Would you be so kind as to summarize the assertion(s) of mine that you object to?
Regards and thanks in advance for your attention to this.
- Chris
Ian and Chris,
Sorry, it was late and I accidently used the wrong names for the two resistors.
Looking at the image in #1338 it should have been obvious that I only changed names but did the right thing.
So Rf was 10Gig and Rg was 100K.
And the prove is here that the DIT method also works without a return signal.
I would suggest not to keep restricted to too stringent rules.
The DIT method only means Dual Injection Technique, and that is for the full 100% what I do, but I can still play with Rf and Rg as much as I like.
Hans
Could you please state the rules for using Middlebrook's DIT outside of a loop? Certainly there are valid tests that can be applied without global loops. But let's be careful that we don't start another (at least partially) nomenclature-based dispute!
Last edited:
I don't object to anything, but as you say I don't see its relevance. When you summarize the small signal behavior of a conventional op-amp nothing is lost by dropping the Aol i.e let it go to infinity. The same is true here the detailed behavior at near DC is simply not important.
Hi Chris,
Let me try to answer this from my side.
You found an (unexpected ?) discrepancy in the transimpedance.
Next step made by you was to suggest to make the Early Effect responsible for that.
To start with, I invested quite some time checking all your findings, to which I agreed.
Then I changed your model by replacing Q1 to Q4 to ideal transistors with a parametrised resistor in parallel to simulate Ro.
With a value of ca. 140K for Ro, I could exactly replicate your measurements with the 2N3904/3906 model.
Although I agreed with you on your findings and your conclusions, I failed to see the relevance of this phenomena.
Thats one of the reasons why I also made the DIT model, to investigate to what degree this Early Effect had any importance on the working of the CFA.
The DIT simulation with your very model in completeness with all transistors, confirmed my suspicion that apart from some insignificant effect on loop gain, the Early Effect could be put aside as a complete unimportant issue.
But despite all my efforts, there were still many postings from you where you kept asking "I ask again for the umpteenth time: Why is ic/vbe different from I q / (k T)?"
The question was already answered with Ro but you kept on asking.
It must be quite obvious that I invested a significant amount of time in checking, building models and so on to answer your question and to show the irrelevance of the matter.
But for some reason it keeps popping up over and again.
That's why I suggested to close this book, without the intention of being unfriendly.
Hans
Thanks for the explanation, Scott. I agree that the balance of gm vbe to vce / Ro operation is not important to the operation of the amplifier. I hope you know that I never claimed it was.
I also understand that some of the folks on this post depend on a purely or at least mostly transconductance-based operation of the input stage transistor to support their claims of voltage feedback. Especially with the audio frequency high loop gain of a device like the AD844, these opposing currents are almost equal in magnitude. The Ro-related portion of the phenomenon is effectively a resistor between the inverting input and the low impedance current mirror input. The Ebers-Moll / Early and the Hybrid Pi model evince this, and we can also infer it from the behavior of a sim'd transistor in a feedback loop as the loop gain increases without bound.
Making it clear that the transistor's operation is not even close to that of a pure transconductor is the only point I was trying to make, targeting those who assert it is.
I also understand that some of the folks on this post depend on a purely or at least mostly transconductance-based operation of the input stage transistor to support their claims of voltage feedback. Especially with the audio frequency high loop gain of a device like the AD844, these opposing currents are almost equal in magnitude. The Ro-related portion of the phenomenon is effectively a resistor between the inverting input and the low impedance current mirror input. The Ebers-Moll / Early and the Hybrid Pi model evince this, and we can also infer it from the behavior of a sim'd transistor in a feedback loop as the loop gain increases without bound.
Making it clear that the transistor's operation is not even close to that of a pure transconductor is the only point I was trying to make, targeting those who assert it is.