Group delay (GD) vs. slew/stability
CFA's have a very low GD. Don't need to play any compensation "tricks"
to get them to have uber high UG freq./slews.
My VFA "leach" was only able to reach 200V/us by the careful addition
of lead compensation. Without it , a higher value Cdom was needed
to prevent ringing on SW tests. The "leach" also simulated at 4X the
GD , (as compared to the typical CFA).
That extra phase delay requires the lead comp. to retain adequate
margin at UG.
The CFA "bypasses" all this with its utter simplicity.
In the end , I did get the leach and NAD/NX to nearly match each
other's performance .... it "blurred" the basic CFA/VFA differences.
OS
CFA's have a very low GD. Don't need to play any compensation "tricks"
to get them to have uber high UG freq./slews.
My VFA "leach" was only able to reach 200V/us by the careful addition
of lead compensation. Without it , a higher value Cdom was needed
to prevent ringing on SW tests. The "leach" also simulated at 4X the
GD , (as compared to the typical CFA).
That extra phase delay requires the lead comp. to retain adequate
margin at UG.
The CFA "bypasses" all this with its utter simplicity.
In the end , I did get the leach and NAD/NX to nearly match each
other's performance .... it "blurred" the basic CFA/VFA differences.
OS
Hey OS, have you seen the original super leach published in the
AUDIO magazine in the 80's? it used lead compensation on the emitters of
the input long tail pairs, any chance you can visit it for a simulation?
Bob, I hope you call the topology as it is meant to be and not something ridiculus like D Self did in his latest book. 😉
I cant quite remember the term he used right now, something like series connected LTP input amp or something in that line. 🙁
Last edited:
Bob, that's a very thoughtfull and sensible viewpoint, thanks for articulating it this way!
Jan
The input is made up of two buffers. One is the input for the signal, the other is input for feedback node. Both of them have have high impedance, just like in a Ltp based input section. Whoever decided on the definition didnt mean to involve the feedback resistor at this point but understandable as this is also not required to define a LTP based amp. See LM6172 datasheet or the patents involved, its a good reference as to the workings of this topology.
The J Curl circuit ?? I hope we talking of same circuit here.
The J Curl circuit is a VFA.
But if you took 0ne quarter of the circuit and let the FB resistor connect to GND it would be a CFA...?
Either something's wrong with your definition Or you're simply wrong..🙂
Either something's wrong with your definition Or you're simply wrong..🙂
My joke was because it is complex calculations with poles... and models... The most simple is simulators, i wonder if it is 5% accurate ;-).
BTW, simulators use (not simple) maths.
Good question/point. It may be the case that the CFA topology is one that is most often implemented with a push-pull VAS, and that may account for much of the SQ improvement. I'm a strong believer in superiority of PP VAS, and they are not as "natural" to do in a conventional VFA (apartt from the fully symmetrical VFA's using paralleled complementary input LTPs.
The CFA is just a natural, straightforward fit to a push-pull VAS topology.
Just a thought.
Cheers,
Bob
I get very annoyed by Doug renaming terms to his own use. His ditching TMC in favor of his own term, just because someone else (Edmond) came up with it is a good example.
I cringe when people use his term "Blameless" to describe a topology as old as dirt.
Cheers,
Bob
This I agree completely with. That's why I said in one of the first posts in this thread that Esperado was cheating a bit by having a PP vas integrated as opposed to the basic blameless VFA 🙂 The circuit of mine you may have seen around here manages to do just that, having a PP vas connected to a single LTP input, happily married.
Edit: oops sorry to have made you cringe 😀
While I think this description is correct, I'm afraid it is though incomplete without discussing what is the current diverted into the inverting input (whatever % of the feedback resistor current) doing. It is easy to devise an example in which that current is doing nothing (AC wise) - and such an example doesn't make a CFA.