Yes, Analog devices Quad Core topology but this is far too complex for use as power amp. Work in progress to simplify. I should mention patented so its only for research and learning. Roy Gosser from analog Devices is a brilliant engineer.
I would still consider this a CFA Waly. Just like there are variations on a theme wrt VFA, you can expect the same on CFA. It meets the two tests for CFA as follows:-
1. The current into the TIS is not limited to the tail current source ( there isn't one anyway) but by the H bridge coupling resistance and the peak output voltage.
2. The amplifier closed loop bandwidth is not CLG constrained as in a VFA
It does not meet both of the two 'pointers' that I postulated, but those were simply included to help in the identification of classic CFA topology designs
1. + input is hi z and - input is low z. No, in the H bridge design, the - input is buffered and therefore hi z.
2. How many gain stages - H bridge still meets this pointer for CFA - there is only one gain stage
Bonsai --- I would not have any other way to know which schematics you are refering to... ones posted somewhere or something you have not posted before. I can go to your web site, though - if it is there. (?)
Some very good designs there BTW 🙂
Waly -- Its only natural to think of the VFA for audio as composed of the diif pair input with CCS. It is the standard which we think of for VFA -- such as D.Self's Blameless VFA amplifier topology.
I also think it is still much as a CFA in terms of core operation. High gains the OLBW seems to behave similar to VFA ... but for audio power amps, it is still within the CFA operating behavior characteristics.
Thx-Richard
Some very good designs there BTW 🙂
Waly -- Its only natural to think of the VFA for audio as composed of the diif pair input with CCS. It is the standard which we think of for VFA -- such as D.Self's Blameless VFA amplifier topology.
I also think it is still much as a CFA in terms of core operation. High gains the OLBW seems to behave similar to VFA ... but for audio power amps, it is still within the CFA operating behavior characteristics.
Thx-Richard
Last edited:
Waly, all I am doing here is proposing some clear tests on whether an amplifier is CFA or VFA. So, you can apply the two 'tests' and two 'pointers' to any schematic.
If you have a better way of differentiating the two - just share it with us.
(I do not agree with your 2nd point. The fact that amplifiers may behave similarly at specific gains is not the determinant of their topology)
If you have a better way of differentiating the two - just share it with us.
(I do not agree with your 2nd point. The fact that amplifiers may behave similarly at specific gains is not the determinant of their topology)
Waly, you've made your point. It's obvious you prefer extra complexity and more bits, cheap or otherwise, for the same performance. Old fogeys prefer "an engineer is someone who can do for 2 bob what any fool can do for a quid." Sorry for these old fashioned sentiments.
I take this to mean you HAVEN'T got any better solutions to high slew in VFAs compared to manso/Marsh & Stochino.
You could have said this without the 'yus guys are all idiots' sneering 😡
_________________
Now how about showing us how clever you really (??) are by expanding on
You could actually be useful ... instead of MikeK2 if you tried.
Last edited:
.. unclean, unclean!
Its rather sad that when someone posts something that might be of use in designing better amplifiers like Walys stuff about unconditional stability with CFAs, they are reluctant to take it further. 😡
However, the rants over semantic/pedantic issues and MikeK type 'yus guys are all idiots & deaf' posts go on forever.
Maybe its an infectious disease 😱
Its rather sad that when someone posts something that might be of use in designing better amplifiers like Walys stuff about unconditional stability with CFAs, they are reluctant to take it further. 😡
However, the rants over semantic/pedantic issues and MikeK type 'yus guys are all idiots & deaf' posts go on forever.
Maybe its an infectious disease 😱
I found that too many people ignore the fact that amplifiers should be seen as one package with the active load they drive. The fact that some of these so called CFA have unconditional stability could be the reason why the amp will perform well when given with arbitrary speakers, including cheap ones. I think it is one important direct or indirect benefit.
Those who offered ideas "to design better amplifiers" ....
Apex,the cat and RNmarsh.
More examples for the final output - big money OEM's .. ha ha.
I don't care where I get the "input" for my LT spice.
When I'm done with the "output" , I've have the result of a research
lab's hard work. Amps that work , and sound good , and last decades.
Good enough for me.
OS
Apex,the cat and RNmarsh.
More examples for the final output - big money OEM's .. ha ha.
I don't care where I get the "input" for my LT spice.
When I'm done with the "output" , I've have the result of a research
lab's hard work. Amps that work , and sound good , and last decades.
Good enough for me.
OS
I have been interested in your claim. In passive speaker crossover design I came to believe that there's nothing wrong with complexity (often it is a must) as long as it is done "right". The problem is, I will not accept the general measure of rightness. I have been worried about the "unmeasurables" that will have audible effect in term of fatigue and enjoyment. There are two things. First is the audible effect of phase or group delay. Second is the "stored energy" (of the coils) that will kick back the amp (now we see how feedback will affect the system).
I have no control on the stored energy issue, except that I can choose amplifiers with features I could find in my CFA.
So my only focus is in phase/delay issue. As I found it is true in case of speaker crossover design, I believe it holds the same with amplifier design.
Many of us has to choose "simplicity" as part of the subjective parameter in amplifier quality. This is because we have difficulty in explaining what the extra gain stage may bring to the table, sound wise.
I strongly believe that you are correct that the quality has nothing to do with the number of transistors (gain stages). But the assumption is that everything else is "equal". Problem is, I found that many assumption is wrong regarding audibility thresholds. So you cannot say for example that because our ears cannot hear anything above 20kHz so it doesn't matter what happen beyond that.
So it is the phase/delay behavior that I expect to be of similar quality when you say that complexity has nothing to do with quality.
So now, do you have certain detailed design (minimum) objective related to phase?
Last edited:
Thank you Damir. You too a Marry Christmas.
PS: As for your latest schematic, why you are wasting 6 trannies?
edit: To get a lower input offset?
Cheers, E.
PS: As for your latest schematic, why you are wasting 6 trannies?
edit: To get a lower input offset?
Cheers, E.
Last edited:
Member
Joined 2009
Paid Member
This is pointless and helps drive the arguments. We've already agreed that the CFA label is part marketing. There is also confusion because of a break from the past where it was used in a different context. Far better to focus on topology and design.
+1
Last edited:
I've used MOS level 11 at work in PStar and in Spectre RF. For hobby projects I usually try to avoid simulations as much as possible (I spend too much time simulating at work to have any desire to continue with it at home). If inevitable, I use LTSpice, but I've never simulated a complete transistor-level circuit in LTSpice so far.
Hi Damir,
I've seen it, but it still doesn't make me happy. Admittedly, as an op-amp it is a useful configuration, fast and a high Z inverting input. But for a power amp it is a waste of trannies, because we have lots of power to feed a low Z inverting input. IOW, we don't need this additional diamond buffer. You could make better use of these six additional trannies, for example a CMCL for the two TISes.
Cheers, E.
Edmond,
Have been using the CMCL (MCP version) in some prototypes along with experimenting with your DTMC. Both ideas work well.
Does the CMCL loop itself require any specific compensation? My skills in loop analysis are not quite up to the job yet, to answer that question independently...
Paul
Hi Edmond,
I agree with you, but small trannies are so cheap, specially those in simulator.
BR Damir
PS. sometime trannies are cheaper then high power resistor needed in CFA feedback circuit.
Last edited:
I agree with Edmond, A Diamond buffer is not sonically invisible, I have evaluated several buffers by inserting them into the chain as a replacement for a RCA-RCA joint. It's clear to me that it robs the music for some life clarity and focus. (I know this is not very scientific and can be debated) but for me personally, my ears are good enough to guide my selections
So having two diamond buffer in one amplifier, one for the input and one for the feedback can't be beneficial compared a simpler circuit that can be built without.
So having two diamond buffer in one amplifier, one for the input and one for the feedback can't be beneficial compared a simpler circuit that can be built without.
I agree on the two points. The second, for me, is the most important. We can believe most of the evils of a circuit in the open loop can be canceled by feedback, but anything (and even resistance's imperfections) in the feedback loop have a huge influence, reduced by nothing.
Compared to a CFA, the LTP of a VFA adds an active device in the feedback loop. may-be this is the only thing witch makes a difference.
A diamond buffer there ? Oh, my God !