Sometimes (don't take it as a law, please) they are prejudicial to transparency, while they measure better
We all try to make amps as simple as possible, it is intuitive. The Lazy Cat VSSA's success is a good example.
Of course, i don't count current sources circuits or PS rails ripple filtering as part of the complexity. Notice that, when CFA use CCS, you need *very fast* ones, to comply with the high slewrates, a point often neglected.
As nobody seems to be interested to make comparative simulations with the two files i provided for this purpose, i propose to simulation experts to compare VSSA with a VFB version of this amp (so easy, changing just the sigle input stage for a double in LTP and feedback path.)
Any one interested to make this comparative study ?
It should be more interesting that all those abstracts assertion like CFA (Or VFA) is better bla bla followed by all those "+1".
To waly, i agree on your post, of course. But note as I'm very maniac with overkill slew rates, i use CFA OPAs for my preamp, everywhere i can.
I had to modify my poor sounding DCX2496. It was easier in the filter output stage to use VFA instead of CFA. So, i tried to compare very fast VFA HF OPAs with equivalent closed loop bandwitchs than my favorite CFAs. They are not far, but i still find an audible difference in easy and transparency, in favor to the CFAs. i'm really interested to find out the technical reasons of this.
Last edited:
I'm an old scool man, and, as my first use of simulators is recent i'm still in the learning process of the tool.
As i already said, i'm an not an expert with simulators and tired to be fired with "your models are not good.. blabla".
Did you took in consideration, with practical impedance values, the parasitic capacitances in the base of the transistor, when you apply feedback to the -input of an LTP to the one of the emitor where Feedback is applied in CFA mode ?
Last edited:
as simple as possible
Me too. The number of gain stages of the super TIS (cacodes not counted) is just three (LTP, TIS and pre-driver), just the same as in any other simple mainstream amp (LTP, emitter follower and TIS).
BTW, how many gain stages were passed by the signal before it gets printed on CD.
(10, 100, 1000???)
Cheers, E.
Hi Esperado,Sometimes (don't take it as a law, please) they are prejudicial to transparency, while they measure better
We all try to make amps as simple as possible, it is intuitive. The Lazy Cat VSSA's success is a good example.
[..]
Me too. The number of gain stages of the super TIS (cacodes not counted) is just three (LTP, TIS and pre-driver), just the same as in any other simple mainstream amp (LTP, emitter follower and TIS).
BTW, how many gain stages were passed by the signal before it gets printed on CD.
(10, 100, 1000???)
Cheers, E.
BTW, how many gain stages were passed by the signal before it gets printed on CD.
(10, 100, 1000???)
Cheers, E.
Is the really true? It's what the sound engineers hear from their monitors that counts. They would effectively cancel the effects of a lot of these extra stages. We want to reproduce what they heard as closely as possible as this is how the music was intended to sound.
I know there's the mastering process to consider. But these are the only stages that can really be added.
Please, Waly, develop.
We are talking about > 10MHz bandwidth where 10K (minimum acceptable input impedance for a consumer VFA) makes a low pass FT with ~1.6pf.
On my point of view, it is, on the contrary, the *only* relevant difference between the two topologies: this added delay in the feedback loop with VFA witch makes a greater phase difference between signal and feedback at HF when they are mixed together.
With CFA, we can keep the impedance of the feedback a lot lower, and the parasitic impedance of the emitter is lower too. In practice: this FC can be set higher by a 10x factor in CFA. Everything else in the CFA:VFA amps can be kept identical, same devices, same currents.
Last edited:
Re he brutal comp thing, you can use Alexander compensation to very elegantly comp a CFA - the comp cap is connected from the TIS back to the feedback summing node similar to MIC. If you select the cap correctly, the front end loop is closed well above the audio band and you retain the same feedback factor at 20k as at LF. On my nx-Amp I was able to use a circa 75 pF cap IIRC vs a 220pF using standard shunt comp and retain very wide bandwidth and high slew rate performance.
You are right Waly, at high closed loop gains, the CFA does degenerate to VFB, but then this is not the normal operating mode for a power amplifier, so the point is moot in my view.
It's also a pity that this whole discussion has seemed to revolve around one parameter: distortion. Over on the Pass from, people are building simple, wonderful sounding amps that produce 0.1% distortion from mainly 2nd and 3rd harmonics, yet I defy anyone to call one of his amps crap sounding (they are mostly VFB by the way and based on very simple circuits).
What is not going to go away in this very long winded discussion is the fact that whether Doug, Bob or any other guru likes it or not, CFB is here to stay and I say that's jolly good.
And now, I am going to listen to Pratum Integrum on my very fine sounding CFA nx-Amp.
You are right Waly, at high closed loop gains, the CFA does degenerate to VFB, but then this is not the normal operating mode for a power amplifier, so the point is moot in my view.
It's also a pity that this whole discussion has seemed to revolve around one parameter: distortion. Over on the Pass from, people are building simple, wonderful sounding amps that produce 0.1% distortion from mainly 2nd and 3rd harmonics, yet I defy anyone to call one of his amps crap sounding (they are mostly VFB by the way and based on very simple circuits).
What is not going to go away in this very long winded discussion is the fact that whether Doug, Bob or any other guru likes it or not, CFB is here to stay and I say that's jolly good.
And now, I am going to listen to Pratum Integrum on my very fine sounding CFA nx-Amp.
Re he brutal comp thing, you can use Alexander compensation to very elegantly comp a CFA - the comp cap is connected from the TIS back to the feedback summing node similar to MIC. If you select the cap correctly, the front end loop is closed well above the audio band and you retain the same feedback factor at 20k as at LF. On my nx-Amp I was able to use a circa 75 pF cap IIRC vs a 220pF using standard shunt comp and retain very wide bandwidth and high slew rate performance.
You are right Waly, at high closed loop gains, the CFA does degenerate to VFB, but then this is not the normal operating mode for a power amplifier, so the point is moot in my view.
It's also a pity that this whole discussion has seemed to revolve around one parameter: distortion. Over on the Pass from, people are building simple, wonderful sounding amps that produce 0.1% distortion from mainly 2nd and 3rd harmonics, yet I defy anyone to call one of his amps crap sounding (they are mostly VFB by the way and based on very simple circuits).
What is not going to go away in this very long winded discussion is the fact that whether Doug, Bob or any other guru likes it or not, CFB is here to stay and I say that's jolly good.
And now, I am going to listen to Pratum Integrum on my very fine sounding CFA nx-Amp.
Sounds like you would like me to touch on CFA in my Second Edition.
Cheers,
Bob
I don't know what you are talking about, sorry. 10MHz bandwidth is not for audio.
Only if you scrap the Class D pages. I'm not paying again for that crap.
You seem to have a real talent to drop atomic bombs without war declaration
Look here how closed loop in audio bandwidth is affected by what happens between 1 Mhz (vfa) and 10 Mhz (cfa).
http://www.diyaudio.com/forums/soli...-self-wants-your-opinions-11.html#post3585406
Absolutely. A good way for you, too, to evaluate your political courage
There is no way in the known universe that a difference in closed loop bandwidth of 1Mhz and 10Mhz could be audible or measurably significant in the audio band. This is just voodoo electronics.
Provided you recognize in print that so-called "CFAs" are in fact deeply compromised VFAs.
Bob-
I have a suggestion. We now have two up to date versions of VFA for audio and many in the past literature. That's pretty much done now except for novel topologies.
I would like to see the beginings of a new book devoted to the CFA circuits only and treated in the same careful way as you have your current book. The first to do that book will generate a lot of interest for future editions. There is plenty to write about and plenty of interest and more good books for audio with CFA would be a great win-win.
Thx-Richard Marsh
I have a suggestion. We now have two up to date versions of VFA for audio and many in the past literature. That's pretty much done now except for novel topologies.
I would like to see the beginings of a new book devoted to the CFA circuits only and treated in the same careful way as you have your current book. The first to do that book will generate a lot of interest for future editions. There is plenty to write about and plenty of interest and more good books for audio with CFA would be a great win-win.
Thx-Richard Marsh
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.