Is the CFB topology superior, and why?

[rcw666]

All you need to do Jay is to look up the course material for electronics that is taught in every tertiary institution on the planet, I have, and that is what it says, and if it doesn't match your prejudices well I am sorry but science is like that.

Don't shoot the messenger if you don't like the message is all that I can say, and I am just the messenger.
rcw

[Jay]

Originally Posted by rcw666

All you need to do Jay is to look up the course material for electronics that is taught in every tertiary institution on the planet, I have, and that is what it says, and if it doesn't match your prejudices well I am sorry but science is like that.

Don't shoot the messenger if you don't like the message is all that I can say, and I am just the messenger.
rcw

No, I'm not shooting you the messenger. It is not that I don't like the message either. I want complete answer, technical justification from your side, not the basics we all know. Of course, not everyone can think of complex things. For example, go in detail with what you think is a misconception. Or, explain about inter-modulation and other issues, show that every aspects have nothing to do with high slew rate or anything.

It is a nature of intelligent people to look beyond the normal boundaries. Sometimes, it is just a kind of intuition (which is of course impossible without underlying theories). Slew Rate = frequency times 2pi times Vpeak is nothing. It is too basic to be considered as "technical justification". You shoot others by saying they have assertions with not good enough justifications. I think they have already one or a few steps above accepting the basics/standards. And that really help people like me to find the justification itself. I want the "answer", and I feel that I can get the answer from/through people who can think beyond the normal boundaries.

[Jay]

Originally Posted by Elvee

I have no conclusions ready. I am freely reflecting on the subject, and I am sharing my thoughts.

I think there are a number of misconceptions about all this: firstly, in actual circuits the difference between the two is not as clear cut as many would like to believe, and secondly the virtues or flaws attributed to one or the other seem to disappear as soon as you normalize the conditions of the comparisons.

Bigun's opinion looks convergent with mine.

My opinion is that they are apple and orange. Like solid state versus tubes, like mosfet versus bipolar. Each has their own characteristics. It is a bit impossible for me to "compare" a bipolar to a mosfet to find out which one is superior. Because no one is superior than the other.

My interest is to compare the best implementation of CFB, and the best implementation of VFB for audio listening purposes.

I think CFB noise cause is dominate by current, not voltage as in VFB. This means that for a high power amp where (voltage) gain is high, (noise) performance at high power is optimal with CFB.

I don't know if F5 is CFB or VFB (and that's not my interest to know the difference), but I found that similar topology TSSA has preferable performance at high power. F5 is only good at first watts, driving high sensitivity drivers. At high power, F5 and F5 Turbo is less suitable than a TSSA.

[counter culture]

Originally Posted by rcw666

All you need to do Jay is to look up the course material for electronics that is taught in every tertiary institution on the planet, I have, and that is what it says, and if it doesn't match your prejudices well I am sorry but science is like that.

I tried to solve this by just looking it up, but I didn't get far.

Both Self and Cordell deal extensively with slew rate, generally in the terms expressed by rcw666, i.e. the slew rate necessitated by an amplifiers maximum power output at its highest frequency (20kHz). Self devotes some pages to methods of altering the slew rate.

Cordell, however, goes on to say this:- 'In practice a much larger value is required for low-distortion performance on high-frequency program content.'

He doesn't go on to justify this, however.

I thought initially that a simple way to get some answers as far as HF distortion is concerned, would be to set up an LTSpice sim of a conventional amplifier with global NFB, set it up to calculate THD20, and then fiddle with the Miller compensation cap to first set the slew rate to and then to increase it to above the value predicted as necessary by the max. slew rate required at a given 20kHz output amplitude, and see if it affects the THD20.

Unfortunately however, on reflection, the Miller compensation acts to linearise the VAS through local feedback, which has its own effect on the distortion performance. This interaction means that although the slew rate is affected by the Miller compensation, modifying it isn't going to show the effect (if any) of slew rate on HF distortion in isolation.

Perhaps someone else has some idea how to analyse this?

I do think, Jay, that it is really incumbent on you to 'show that every aspects have' something 'to do with high slew rate or anything'.

[rcw666]

I am afraid Jay that it is you who need to supply justification for your assertions, I already have supplied the facts as they are recognised by every authority on the planet, and that is that a pre-amp needs a slew rate of note more than around .5V/us. to cover the audio band and op amps such as the 5534 cover this requirement more than adequately and are still the industry standard because of this and other factors, another of which is much greater flexibility of feedback arrangements over CFB types.

The basic fact about the CFB topology is that it allows greater bandwidth with a fixed internal pole splitting capacitor, and the advantage this occurs at frequencies that are far above the audio band. Why one would use such a device that has the basic disadvantage that it has very restricted configurations that can be used for feedback to gain a benefit which is of no use at all in a particular application I am not sure.

It is you who needs to tell us why it is better to do this, so please do so, giving actual technical reasons rather than just bald assertions.
rcw

[MagicBox]

When you don't use any form of active feedback, then your slew doesn't need to be more than what's needed for the signal.

But when you use a feedback loop, you do want the output to settle fast on startup transients and any other transients caused by the signal-to-reproduce (harmonics). A high bandwidth achieves that. A high slewrate in this perspective is nothing but a derivative of the quest for bandwidth.

[rcw666]

If you are talking about the so called TID then as Leach showed this is really a function of Gain bandwidth product.

His low TID power amplifiers had a GBP in the 8-10MHz. range and do does the 5534 type op amp, there is no need for any more than this, and people who claim there is can provide no technical reason for it other than some vague notion that more must be better.

in the end most people can't hear the difference between a 741 op amp and a 5534 one except for low level amplification in which the former is a lot noisier.

In forums such as this you of course get plenty who claim you can, but these assertions are based upon a lot of subjective assertions and not on double blind tests, and are thus of very dubious credibility.
rcw

[Jay]

Originally Posted by rcw666

If you are talking about the so called TID then as Leach showed this is really a function of Gain bandwidth product.

His low TID power amplifiers had a GBP in the 8-10MHz. range and do does the 5534 type op amp, there is no need for any more than this, and people who claim there is can provide no technical reason for it other than some vague notion that more must be better.

in the end most people can't hear the difference between a 741 op amp and a 5534 one except for low level amplification in which the former is a lot noisier.

Initially you had bold assertion that someone like Juma couldn't hear the difference between 741 and any of his pet op-amp. Now you add one exception. Funny thing is that you use the word "LOT" which implicitly means something obvious.

Do you know what is happening here? What you need is actually a few more practical experiences then may be you can at the end say with full certainty that YOU yourself can hear the difference so easily.

I know that there must be somebody else have higher intelligence than I am. I know that there are people who knows better than me. I know that there are people who can hear better than I do. I know that there are many things that I don't know. It is a skill that you seem don't have.

And don't try to be safe. Mention what is the limit of the frequency to be considered in audio. 20KHz? 8-10MHz? Nobody has ever said that more is better. There is always limit for everything. You just mention your number now, so we can compare with your number next year, and the next 5 year

[Bonsai]

Originally Posted by counter culture

I tried to solve this by just looking it up, but I didn't get far.

Both Self and Cordell deal extensively with slew rate, generally in the terms expressed by rcw666, i.e. the slew rate necessitated by an amplifiers maximum power output at its highest frequency (20kHz). Self devotes some pages to methods of altering the slew rate.

Cordell, however, goes on to say this:- 'In practice a much larger value is required for low-distortion performance on high-frequency program content.'

He doesn't go on to justify this, however.

I thought initially that a simple way to get some answers as far as HF distortion is concerned, would be to set up an LTSpice sim of a conventional amplifier with global NFB, set it up to calculate THD20, and then fiddle with the Miller compensation cap to first set the slew rate to and then to increase it to above the value predicted as necessary by the max. slew rate required at a given 20kHz output amplitude, and see if it affects the THD20.

Unfortunately however, on reflection, the Miller compensation acts to linearise the VAS through local feedback, which has its own effect on the distortion performance. This interaction means that although the slew rate is affected by the Miller compensation, modifying it isn't going to show the effect (if any) of slew rate on HF distortion in isolation.

Perhaps someone else has some idea how to analyse this?

I do think, Jay, that it is really incumbent on you to 'show that every aspects have' something 'to do with high slew rate or anything'.

One option here is to take a wide band closed loop amplifier (a good fast opamp will do) and feed this through a low pass filter. It's very easy ten to set the RC time constant and hence slew rate independent of the amplier and to test the results both subjectively and quantitively.

[homemodder]

Originally Posted by Elvee

In theory, any CFB amplifier can be converted into a VFB one having at least identical performances.

Could you post a schematic, I dont believe this can happen, at least not at same operating conditions as I mentioned in previous post.