Excellent to see this, thank you.
Unfortunately the plots, are blurred and hard to read.
It should not be too hard to improve them, how did you pull them from LTspice?
There are some issues of loop probe placement that cloud the results a little.
If you post the ASC I can check this.
Best wishes
David
Last edited:
Hello Waly
I attached a cfa and I can't find a better solution than the shunt in therms of closed loop bandwidth,group delay via feedback line and phase deviation .Miller it's not better,pole splitting,etc .Of course all these with a phase margin.
I tried to point you to my CFA amp not using shunt compensation, but you missed my post or just ignored. The CFA with no shunt compensation is here. http://www.diyaudio.com/forums/solid-state/243481-200w-mosfet-cfa-amp-46.html#post3771044
BR Damir
Where is the loop gain, haw have you determined the group delay, what is "feedback line and phase deviation", why "Miller it's not better, pole splitting, etc...", what is "all these with a phase margin".
Your attachment has a schematic, but no results and their interpretation.
Name dropping is your distinction not mine.
Shunt compensation = suboptimal for VFA
Reason : Loads the vas when used as dominant pole compensation. Only advantage is when used as aid to stabilize outputstage and is added if needed.
Shunt compensation = optimal for CFA
Reason : 1 Also loads the vas but not detrimental as inputstage can supply any current needed.
2 Increase in open loop gain at high frequencies.
3 Loading of the vas can be of benefit to increase bandwith.
4 Shunt capacitance prevents the impedance seen at the vas collector from becoming inductive at high frequencies.
5 Its built-in when shunt compensation is used, no need to add anything.
If you need more details find Sassen´s analysis of shunt compensation on google as already recommended.
I have shown a design using shunt compensation that so far outperforms all other designs shown here with all kinds of different compensation schemes. I see designs posted with better results than mine but that only because they compromise PM/GM. When I compare by having the same PM/GM using shunt compensation the shunt again wins.The only scheme I see here that has promise which I havent had time to analyse is Cherry with which dadod has had very good results. You questioned with the argument that ULGF was too high for stability to be possible. Bonsai was the other member who questioned it. If you look at his latest posts youll notice he has changed his mind, he did proper analysis now. I did this many years ago, learnt through experience of designing and building CFA amps and not stuck in a chair doing simulation.
Last edited:
Damir, dont worry I didnt ignore, I just have little time to post on DIY. I am in any way going to abandon this thread and move on to that thread and the wire with gain thread.
Without having looked at your files yet, thanks for someone besides me doing proper shunt compensation analysis.
Seems no-one wonders why shunt compensation is always used by companies producing CFA opamps.
Incorrect. Lead lag compensation is used to compensate the Miller loop, not to stabilize the output stage.
Loading the VAS is bad because it draws lots of current, and hence distortions are increased. Or you need to bias the VAS very hot.
Nope. Shunt compensation is also a first order compensation, no advantage over Miller from this perspective. Not sure why open loop gain is relevant, anyway, not that it changes anything, but perhaps you meant the loop gain?
Please prove this extraordinary statement.
Maybe (I would not think so, and moreover, I don't see why this is specific to shunt compensation), how/why is this important?
Huh?
Name dropping. Sassen didn't claim anywhere that shunt compensation is optimal. If you think I'm wrong, please provide a quote.
Nothing relevant and/or on topic here, except unsubstantiated claims together with the common claim "I have experience".
The designers that decided to rely on shunt compensation in commercial amplifiers had, in my opinion, a pretty cynical approach. Although suboptimal, shunt compensation is simple and very stable when it comes to component tolerances and variability. This is at premium for any commercial design. With a very few exceptions, squeezing the last drop of performance is never the primary objective (otherwise said, "they can't hear it, anyways").
Not only designers in commercial amplifiers but Im talking here of commercial opamp manufacturers. Youre going to tell them they have a cynical approach, thats laughable. The primary objective of commercial opamp manufacturers is squeezing the last drop of performance from a part.
Experience - Wikipedia, the free encyclopedia ( Just to refresh your mind to its meaning)
BTW I much younger than you are. 😉😀
I have to conclude that any form of dialogue is impossible. You refuse to provide proof for any of your extraordinary statements, and constantly try to cloud the facts in a technobabble language. I don't speak that language, so I'm afraid this has to stop here.
When you will be in the mood of having an intelligent discussion about frequency compensation, VAS loading effects, etc... I will probably still be around. Meantime, you may want to reconsider if and why shunt compensation is optimal for CFAs. According to your own statement, quoting Sassen doesn't help.
Modern high performance CFA op amps are usually not shunt compensated. Don't take the simplified schematics in the datasheets, and the common stupid behavioral models, as having any resemblance to the silicon. Behavioral models are supposed to help in modeling the small signal performance (and then whatever single pole at the right frequency will do the job) and are not designed to model the large signal behavior (to which VAS loading, input stage current, etc... are belonging).
Waly, sorry but you're not the one in a position to demand proof. All you do is bash what others lay on the table, without bringing anything yourself. Plenty here has asked you to show your creations, nothing has surfaced.
That's not a good reason to ignore when other "experienced members" are spreading technical nonsense.
Indeed, I see there are members with their day jobs in the Audio business contributing here, what was I thinking?
Talking about a vested interest in revamping an audio fashion .
Last edited:
A message in here somewhere --
This all reminds me of a conference where a paper was presented by a well know Canadian mathmatics professor --- he went on and on for an hour about this particular approach he recommended was superiour to the others and presented his proof. Finally when it was over, a question was asked.... well, how many dB quieter is this method... because it never was mentioned in the talk. The answer was 1-2 dB.
When all heard this.. a spontaneous loud groan was heard from the audience and everyone got up and left. Optimum is fine but not if it is minimal improvement and at a large amount of effort to get it or extra parts/costs/ space. Then, it is just an intellectual exercise.
Thx-RNMarsh
This all reminds me of a conference where a paper was presented by a well know Canadian mathmatics professor --- he went on and on for an hour about this particular approach he recommended was superiour to the others and presented his proof. Finally when it was over, a question was asked.... well, how many dB quieter is this method... because it never was mentioned in the talk. The answer was 1-2 dB.
When all heard this.. a spontaneous loud groan was heard from the audience and everyone got up and left. Optimum is fine but not if it is minimal improvement and at a large amount of effort to get it or extra parts/costs/ space. Then, it is just an intellectual exercise.
Thx-RNMarsh
Last edited:
Thanks, Richard. Reading some inputs, in this thread, I sometimes have to pinch myself, afraid of some teleportation accident...
There's no problem with the loop probes Dave. It is what it is and clearly explains the reason why you can close the loop far higher on CFA's and still have perfectly stable amps.
I've compressed them a bit too much. I will put them up on my website over the weekend.
Last edited: