Dude. That would be awesome! LM386's for everybody and it's finally settled.
Handy four item list though.
Last edited:
Bonsai, I understand perfectly your reasoning, and I can imagine how you get to it, it's just that it never has been that way until those CFA things came along.
What we have been taught, and what STILL is being taught, is that if you use feedback to control the plant output current, that's current feedback.
Look at the paper linked in the section at the bottom of this:
On Feedback, Feedforward and Error Correction
THAT is what is being taught.
So, with that, I bid you farewell -
jan
What we have been taught, and what STILL is being taught, is that if you use feedback to control the plant output current, that's current feedback.
Look at the paper linked in the section at the bottom of this:
On Feedback, Feedforward and Error Correction
THAT is what is being taught.
So, with that, I bid you farewell -
jan
Last edited:
It's current feedback only if the sampled current remains a current at the point of application of the feedback.
With a transadmittance amplifier (voltage in current out), the feedback is a transimpedance since the feedback network samples the output current and applies it as a voltage for algebraic summation with the input voltage.
In both cases feedback is used to control the output current, but the term current feedback is only used to describe the former.
See below.
http://www.diyaudio.com/forums/soli...current-feedback-amplifier-8.html#post3414948
100% wrong!
What you've described is series (current) derived feedback; which can be described as current feedback if the sampled signal is algebraically summed in parallel with the input current (shunt applied), or transimpedance feedback if the sampled signal is algebraically summed in series with the input voltage (series applied).
http://www.diyaudio.com/forums/solid-state/208416-pros-cons-current-feedback-amplifier-8.html
Last edited:
Let s say current sensing is sent to the inverting input of an inverting
amplifier.
This time , current sensing is also sent to the inverting input
but the input signal is sent to the non inverting input , the
amp being forcibly non inverting.
Let s admit that practicaly the results will be the same in both cases
and that the theses two classification are mainly done for ease
of modelisation of the respective configurations....
Jan, CFA has been around and described as that since the early 1980's and in IC form since 1987 . . . It has not just come along.
As I explained earlier, if I look at the topology differences between what is commonly known as a VFA and a CFA, I can see why they behave differently and how a CFA came be called that.
Clearly you are unhappy about the naming situation (I note the swipe at the bottom of the page on your website). Best thing you and the the other people who don't like the CFA name can do is get together and write to TI, ADI, LT, Maxim et al to tell them their CFA naming for the last 25 years is wrong and they need to change it.
Last edited:
That's an interesting statement, and I would appreciate if you could substantiate it.
DC precision? Who cares in audio, a servo is almost always present.
Noise? CFAs have large current noise, otherwise voltage noise can be as low as VFAs. So there might be a constraint for the source output impedance. But this is a usually a non-isue for audio power amps.
Distortions? VFAs may have lower distortions at low frequency (that is, higher loop gains compared to CFAs). Again, not a significant issue for audio, where LF distortions are usually not a big concern. I would agree though that the essential "single ended" behaviour of CFAs do not allow distortion cancellation as (potentially) in VFAs.
I would concede there's no good reason to claim that CFAs are anywhere superior to VFAs for audio, but to state that CFAs are inferior that's, to me, a biased stretch.
I like CFAs mostly for the (nearly) gain-bandwidth independence. The abiility to preserve low gain stability, while still pushing up the unity loop gain frequency is unique to this topology. It is very hard to build a wideband VFA while having it stable at low gains.
If you wonder if this matters for audio, yes it does. Foe example, I've seen amps that burst into oscillations at power up and power down, until the amp settles. It's ok, until such an amp blows in your face. You may claim such as incompetently designed, but correcting this behaviour in a VFA would necessary require sacrificing the otherwise great AC spec.
Think it's important to present here the info from the other thread where "blameless" fans are gathered, blindfolded to the other's opinions. Again my statement follows: current feedback amplifiers outruns VFA in all parameters important for audio signals amplifying. This statement is based on years of experiences, testing and listening ready made amps from known producers as well as from all kind of DIY amps made from both topologies. Attached three steps describing what VFB actually is - to some "upgraded" current feedback circuit, while logical thinking will lead you to use wire only (fig1.) instead of nonlinear part (fig2.) if you pursuit an audio quality in power amplifiers. Case closed. 
Attachments
Is the supposed CFA's higher distortions at low frequency because of non-linearities outside of the input stage (i.e. output stage distortions with lower loop gain) ?
That is in isolation does the CFA input stage (compared to an LTP with all else equivelent and ideal) still have higher low freqeuncy distortions.
Following along these lines then any class-A topology (even low power headphones) "might" be a good candidate for the CFA. Similarly if methods exists which can raise the CFA low frequency loop gain (baxendall super pair or some other high impendance current mirror) then wouldnt that be equivelent to some input inclusive compensation techniques (combinded loop gain and local feedback loop of the CFA input stage).
Thanks
-Antonio
That is in isolation does the CFA input stage (compared to an LTP with all else equivelent and ideal) still have higher low freqeuncy distortions.
Following along these lines then any class-A topology (even low power headphones) "might" be a good candidate for the CFA. Similarly if methods exists which can raise the CFA low frequency loop gain (baxendall super pair or some other high impendance current mirror) then wouldnt that be equivelent to some input inclusive compensation techniques (combinded loop gain and local feedback loop of the CFA input stage).
Thanks
-Antonio
Firstly, there is no need for an excessively large closed loop bandwith (say, greater than 2Hz~200KHz) in audio applications.
Secondly, the appearance of constant bandwidth in "CFAs" is an electronic sleight of hand.
It's due to the fact that the gain of the first stage is roughly equal to its load impedance divided by the effective impedance at its emitter.
Because the feedback network is connected to the input stage's emitter, it has a direct impact on first stage gain, and, by implication, on forward path gain for the same compensation element.
If the Thevenin impedance of the feedback network, as "seen" from the input stage's emitter, is reduced by, say, reducing the value of the feedback resistor connected to ground, this has a effect of simultaneously increasing first stage gain and the closed loop gain of the amplifier.
It is this increase in first stage gain, and, therefore, increase in forward path gain, that causes the bandwith to appear constant even as the user attempts to increase the amplifier's closed loop gain by reducing the value of the feedback resistor connected to ground.
To reiterate:
1) There is no need for excessive bandwith in audio applications.
2) "CFAs" deliver significantly poorer linearity than VFAs.
3) "CFAs" deliver poorer PSRR than well designed VFAs.
4) There is no need for servos with well designed VFAs; this not true of "CFAs".
There is no conceivable excuse for using "CFAs" in audio applications.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.