Then it is an industrial name of their specific topology. Each industries with their own version with that name. You may also make your own version, no one could stop an industry or a person to name their topology as CFA, since there are no agreement about this, but most of them has similarities and differences.
Then it is OK if someone has his own classification.
Then it is OK if someone has his own classification.
Last edited:
Member
Joined 2009
Paid Member
It's very easy to classify whether it's CFA or VFA . . . I've must have repeated it here about 5 times.
Yes, but not easy to make all agree with that. It is also better if in this thread we use Bonsai classifications to prevent some disagreement, but some time some people need their classification from their books or their factories to be accepted too.
I don't think there's an argument any more about how to define CFA vs VFA. 2 little tests and 2 little pointers:-
Tests
1. The TIS peak input current is set by the feedback resistor value
2. The CL BW is independent of CL gain
Pointers
1. + input is hi Z and - input is Lo Z
2 typically only one active gain stage
Bot the designs OS has put up above meet these criteria and are therefore CFA
Tests
1. The TIS peak input current is set by the feedback resistor value
2. The CL BW is independent of CL gain
Pointers
1. + input is hi Z and - input is Lo Z
2 typically only one active gain stage
Bot the designs OS has put up above meet these criteria and are therefore CFA
Member
Joined 2009
Paid Member
So VSSA would not be a CFA ?
You can add gain to the first stage, or every where you want inside the loop else than the feedback path, on my point of view.
A "CFA" is a "VFA" with the open loop gain and bandwidth controlled/modulated by the negative feedback network resistors.
The "current on demand" property (responsible for the very high slew rate) is not specific to a "CFA". It can be implemented in a VFA as well.
The CL bandwidth independent of the CL gain is a derivate result of the above property. Accordingly, for a non inverting amplifier, you calculate the feedback network resistors ratio depending on the CL gain, then you chose a feedback resistor value to choose the CL bandwith.
1. is not a direct requirement, although it's true for the most common implementations.
2. is definitely not a requirement. Two gain stages "CFAs" can easily be implemented.
Following the above definition, most audio CFAs are not "true CFAs", since the feedback resistor is much larger than the (negative input) input impedance (*). Hence, the GBW modulation by the feedback network is minimal. They are VFBs with "current on demand" providing high slew rate. They don't have much of two degrees of freedom as a true CFA.
Also, given a compensation network order and an open loop GBW product, CFAs cannot provide more loop gain than VFAs (hence more linearization properties), or more phase/gain margins, they are both subject to the Bode maximum loop gain theorem. In a CFA, you can manipulate the open loop bandwidth by changing the feedback resistor value, but this will result in a corresponding change in the open loop gain, so that the GBW product is ultimately constant.
(*) and that's because CL gain in audio amplifiers is rather high: 20-30dB. True CFAs, like the DSL drivers, are usually used at CL gains of only 6-10dB.
Last edited:
Lets not get into an argument about semantics again.
I said 'typically' and proposed some basic tests and pointers to help differentiate a CFA from a VFA.
Practical power CFA's do seem like VFA's if you look at the loop gain (although it's usually lower at LF) above the -3 dB roll off frequency, but that has nothing to do with it being a partial VFA - it simply reflects the aggressive comp needed to deal with the output stage pole.
Re current on demand, if you talk about a classic VFA with LTP fed from current source there is no CoD. As Richard Marsh has noted, the front end stage is compressive, but in a CFA you can go to 4 times the standing current (I simmed this) ie it's expansive. There may be clever circuit techniques in VFA to get around the LTP current limitation - certainly MIC which is a comp technique rather than some form of CoD helps because Cdom can be made much lower.
I said 'typically' and proposed some basic tests and pointers to help differentiate a CFA from a VFA.
Practical power CFA's do seem like VFA's if you look at the loop gain (although it's usually lower at LF) above the -3 dB roll off frequency, but that has nothing to do with it being a partial VFA - it simply reflects the aggressive comp needed to deal with the output stage pole.
Re current on demand, if you talk about a classic VFA with LTP fed from current source there is no CoD. As Richard Marsh has noted, the front end stage is compressive, but in a CFA you can go to 4 times the standing current (I simmed this) ie it's expansive. There may be clever circuit techniques in VFA to get around the LTP current limitation - certainly MIC which is a comp technique rather than some form of CoD helps because Cdom can be made much lower.
Last edited:
Can you give us a 'practical' example (SPICE world will do) of this mythical beast? 😱
No, it's a textbook VFA. Just fold up/down the schematic and see what you get. Hint: series-shunt feedback, as in pp. 280-3.
I would not rant again about the sub mediocre electrical performances of the VSSA.
No, I can't spend more time on this topic. Use your brain and do your own homework, I'm not here to do it for you. Worst case, get a good book like this one.
Thank you Waly for your generous & helpful comments which I'm sure are as useful as MikeK's.
I think you are trying to classify CFA and VFA using classic feedback theory modes. Doing this always gives the same answer: there is no such thing as CFA, only variations on VFA.
Suspend the feedback mode analysis because in this case it's the wrong tool for the job and look at the circuit. Apply the tests and the pointers I listed above and you should be able to arrive at the right answer.
By the way, just take a look at the transfer equation for a CFA and a VFA - although the final gain formula is the same, the underpinnings are very different. This should tell you you are dealing with something different.
And, don't forget the Wurcer test !
😉
Suspend the feedback mode analysis because in this case it's the wrong tool for the job and look at the circuit. Apply the tests and the pointers I listed above and you should be able to arrive at the right answer.
By the way, just take a look at the transfer equation for a CFA and a VFA - although the final gain formula is the same, the underpinnings are very different. This should tell you you are dealing with something different.
And, don't forget the Wurcer test !
😉
Really
Anyway, I don't want to re-start a useless debate, so I'll stop here. I've already posted in this thread all the underlying math to justify my interpretation, those interested can search for and draw their own conclusions.
I'm sorry to disappoint you, but not everybody is a "jurassic" "beach bum" with unlimited time budgets.
Speaking of Budgets, @ a dollar a page, this is some costly reading, maybe Santa can find one of these in the coal bin.
