Hans,
Sorry I have been slow. I have been reconsidering things. Also the with time difference I write while you sleep (or should be sleeping
). I haven't looked into your query. Does my analysis equations need changing? If so can you suggest a change?I think I should explain my "ulterior motives' for being involved in this thread on CFA issues. My belief is the CFA is one of many circuits that we don't really understand the underlying principles of these circuits. This problem has been highlighted indirectly (in this thread for example) by the wide range of often contradictory beliefs about how it works and how it compares to the VFA and how it works. It has been stated recently by Syn08 and forr (eg post 2327) and others.
To address this lack of fundamental understand of the underlying principles of our amplifier circuits I have a long term goal of solving the large signal equations for as many circuits as I can. I am now 6 years into this hobby retirement project and now on a second edition update to my first 2012 document (PAK links in sign below). The analysis is a new field of applied mathematics, very difficult and time consuming and academics don't seem interested in doing the maths themselves with their limited time and no identifiable pot of gold in sight to get the grants.
This thread has given me sufficient insights into the wide range of beliefs about how the CFA works and how it compares to the VFA to help me compile a section on the CFA in my future analysis of the CFA (if time permits).
BTW I have just completed the Early effect for the single transistor common emitter amplifier and you can see from this how complicated and difficult it is to add just one non-linearity like the Early effect to the most basic amplifier circuit (download the interim PAK209 PDF).
I should start a new thread on all this but haven't yet so I can focus on doing the maths and updating my site docs. Despite the difficulty I now know that it is possible to use this new approach to solve our circuits to explain how they work. But the complexity of the maths will need some software to automate the analysis and this is beyond my resources and best left to a bunch of academics and PhD students.
I'd love to hear from anyone interested in following my hobby, and maybe prod me start a thread "Better analysis of <xyz> amplifier circuits"
. Members can PM me else use Contact on my website.All the best for this thread forr and all the rest. I appreciate all your help. @ N101N thanks (to me you are so humorous) keep up the posts, and I agree: simulators don't tell us how circuits work - we need analysis for that.
Cheers,
IH
Last edited:
No I don't think so.
For whatever reason, you just denied above that Black called the output series feedback "current feedback" and that a CFA topology is sampling the output voltage of the open loop amplifier. If you need authoritative references on any of the about, just say so. Previously, you flatly dismissed the results of analyzing the CFA using the two-pole formalism on the grounds of "it was done before" (does this make the end results wrong?).
BTW, this discussion was/is not about merits, not about audio amplifiers, not about who's better and not about what you built and the quality thereof. At least to a point, it was about ways to characterize a CFA amplifier, and I believe it was clearly proven that the inverting input current flow method is, while entirely correct, a quick way to reach the right conclusions, and universally accepted way, not really required to identify the special CFA small signal properties. Each and every small signal CFA property can be mathematically determined using the canonical shunt output, serial input feedback mechanism. A fact that, for whatever reason, you also flatly refuse to accept.
It was also shown that what we call today the CFA circuit topology and its properties was known since before WWII, while the "CFA" concept was introduced in the early 80's by Comlinear. Although asked repeatedly, and once again for whatever reason, you did not pick up the question "what were CFAs and how did engineers manage the "CFA" before 1980". But that's ok, addressing/answering questions is optional, no fault here.
Sorry, then I must have missed the whole purpose of your reply/comment/expanse. I'm having a hard time tying COD on two concepts ("expansive" and "compressive") that don't make any sense to me in this context, probably a 'bout of ADD on my side.
Hans Polak,
It is not a matter of being stupid. Having infinite knowledge is an attribute of gods, but everybody can learn. What I am saying is that the repetitive mathematical tautologies do not give any factually new information, not more than what is implicitly contained in the initial data and premises. The measured outcome is predetermined.
Anything that is mutable is not definable and not representable. It is not possible to mathematically express a change, impact or effect. It is senseless to make calculations on signals, the Miller capacitance, bandwidth and impedance. A signal does not have a resistance, voltage and current property and should not be categorized as such.
I do not call steady state analysis a prediction. Finding the length of the hypotenuse is not predicting the future (nor deciphering the past).
Maybe it is inappropriate to introduce this kind of rigor here. I am aware that the purpose of discussion is not always factual exploration. False information can be preferable to no information. Although it is not my intention to hurt anybody's sentiments, it may occur as an incidental damage.
I will give my best. I have already provided a definition on more than one occasion, contending that an analytical solution does exist for this issue. I do not want to hide my disappointment that the discussions have been permanently stuck at square one. Yet there is something wrong somewhere.
Thank you.
It is not a matter of being stupid. Having infinite knowledge is an attribute of gods, but everybody can learn. What I am saying is that the repetitive mathematical tautologies do not give any factually new information, not more than what is implicitly contained in the initial data and premises. The measured outcome is predetermined.
Anything that is mutable is not definable and not representable. It is not possible to mathematically express a change, impact or effect. It is senseless to make calculations on signals, the Miller capacitance, bandwidth and impedance. A signal does not have a resistance, voltage and current property and should not be categorized as such.
I do not call steady state analysis a prediction. Finding the length of the hypotenuse is not predicting the future (nor deciphering the past).
Maybe it is inappropriate to introduce this kind of rigor here. I am aware that the purpose of discussion is not always factual exploration. False information can be preferable to no information. Although it is not my intention to hurt anybody's sentiments, it may occur as an incidental damage.
I will give my best. I have already provided a definition on more than one occasion, contending that an analytical solution does exist for this issue. I do not want to hide my disappointment that the discussions have been permanently stuck at square one. Yet there is something wrong somewhere.
Thank you.
You’re confusing Amplifier topology with the canonical feedback forms.
Black, IIRC wrote about f/back in the 1930’s (I have one of his papers up on my website somewhere).
There’s no way Black would have foreseen what Comlinear came to refer to as a ‘current feedback amplifier’ which is what we are talking about here.
Black, IIRC wrote about f/back in the 1930’s (I have one of his papers up on my website somewhere).
There’s no way Black would have foreseen what Comlinear came to refer to as a ‘current feedback amplifier’ which is what we are talking about here.
Single input device (no complementary) CFA have the "small signal constant bandwidth" property but not the COD.
Herve,
I did not refer to a DIT implementation but to the test with the CFA-VFA.asc test below.
V(out) for a VFA is the open loop gain and for a CFA the transimpedance divided by resp 500 Ohm and 5K.
The input current to V(in-) also clearly shows the differences between the two topologies.
Supply voltage is set at +/- 5Volt, but this can be easily changed for circuits needing more.
That's what I have done with one of your own circuits shown somewhere in this Thread.
So two lines is CFA, one line is VFA.
Hans
Attachments
Figures 9 and 10 and the surrounding text of https://www.edn.com/electronics-blo...ok-s-and-Rosenstark-s-Loop-Gain-Measurements- give a nice presentation.
I'll let Hans reply, but the circuit he proposes to discriminate between topologies does NOT employ DIT. It is effectively open loop from an AC point of view, but closes the loop at DC to ensure that the amplifier output is not sitting on one of the supply rails.
Running in circles again, yes it does. The positive slewing is limited only by the feedback network, you cannot expect symmetrical slewing from an asymmetrical topology.
Do I? Really, do I?
For the rest, no comments
. Only one quick question: why are you so grumpy when it comes to this topic? Snubbing legitimate arguments, one line dismissive comments, lack of any appetite for following up argumentation with an open mind, etc... This is not the Bonsai I used to enjoy 8 years ago.
Last edited:
Yes but it seems to me you turn the wheel.
I don't expect anything, but you can't claim all CFA have COD when asymmetrical have not.
Do we agree on the fact that single input transistor amplifiers are CFA ?
Yes.
Single input transistor amplifiers do have COD, for charging the compensation cap. Discharging would be at the same rate for a symmetrical CFA, but for a single ended it cannot happen that fast; there is simply no current path.
A classic Lin topology VFA will charge the Miller compensation cap at the max rate proportional to the long tail pair current source, but once again discharging will also be much slower, since once again there is no current path. A symmetrical VFA will discharge the Miller cap at the same rate as charging.
Otherwise, I don't know what is your understanding and expectations, regarding COD.
My understanding is that only diamond or similar and double diamond - H bridge- have the COD property.
These topologies can be implemented on CFA and on VFA.
So I don't understand why you claim that COD is a native property of CFA when it is not.
So:
a) what is your definition of COD?
b) why do you think a single ended CFA doesn't have COD?
Without clear answers to these questions discussion cannot advance. I suspect we have different views of a)