Commercial designs have a prime goal: Get acceptable results with as few parts as possible. CFA meets the criteria of fewer parts. There you go
For state of the art designs it could be the opposite. Quote from Krell FBI review:
The catch is that running in the current domain requires a lot of transistors—up to twice as many, watt for watt, as the voltage domain—which means that an amp like this must be heavier and more expensive, and run hotter, than it otherwise would.
I did try (still trying) to get Bruno to write an article on class D, but he 'feels no urge to educate my competition'. That's another factor with cutting edge technology.
jan
More from the same Krell FBI review:
Dan D'Agostino, Krell's proprietor, hit on this idea while designing A/V processors. The high bandwidth of video signals forced him to work in the current domain (voltage-based circuits aren't optimal for video's high bandwidth), which led him to wonder if the current domain might expand the bandwidth of audio signals too—and it did. According to Jim Ludoviconi, Krell's technical manager, the audio bandwidth in the current domain exceeds that of the voltage domain by "an order of magnitude."
It seems that the bandwidth matters in audio too.
Dan D'Agostino, Krell's proprietor, hit on this idea while designing A/V processors. The high bandwidth of video signals forced him to work in the current domain (voltage-based circuits aren't optimal for video's high bandwidth), which led him to wonder if the current domain might expand the bandwidth of audio signals too—and it did. According to Jim Ludoviconi, Krell's technical manager, the audio bandwidth in the current domain exceeds that of the voltage domain by "an order of magnitude."
It seems that the bandwidth matters in audio too.
I think they are talking here about the small signal coupling ('CAST') and not the power amplifier topology.
I worked in industrial instrumentation for about 15 years, and the 4-20 mA loop was and is ubiquitous for the very reason that it is extremely resistant to noise pickup (magnetic and capacitive). The bandwidths of these industrial loops are low - usually only a few Hz. However, I can see that this is easily configured for wide bandwidths and you will have none of the stability issues you get with driving capacitive loads. As a result, you can have much, much wider small signal bandwidths.
I worked in industrial instrumentation for about 15 years, and the 4-20 mA loop was and is ubiquitous for the very reason that it is extremely resistant to noise pickup (magnetic and capacitive). The bandwidths of these industrial loops are low - usually only a few Hz. However, I can see that this is easily configured for wide bandwidths and you will have none of the stability issues you get with driving capacitive loads. As a result, you can have much, much wider small signal bandwidths.
The reason is much simple, omho. There is not so much designers who are really interested in music (at least in their R&D department) or had noticed a difference due to CFA because they never try or correlate..Commercial designs have a prime goal: Get acceptable results with as few parts as possible. CFA meets the criteria of fewer parts. There you go
So, when they begin a new project, they just think VFA.
The second reason, THD is a marketing argument, VFA is a easier way to get impressive numbers.
I don't think CFA need more parts than VFA, on the contrary. You can remove two transistors of the LTP and add 4 resistance to turn a classical VFA amp in CFA.
Well to get equivalent numbers with CFA, you need a good knowledge on how design fast and near perfect CSSs, and care to filter the PSU ripples to cancel the lack of PSSR that a LTP offered.
That's the politic of the forum witch makes-you feel that way.
Andrej (Lazy Cat) had published several CFA projects for DIY. All as original as possible (nothing is really new in audio).
They all begin with a simple working schematic, resuming the idea, and you can follow all the developments states up to very complete final state, with printed boards and components list. Very instructive.
The last one was the VSSA. A perfect example on how to design a *top of the art* amplifier with very few parts.
After this project was published, a group Buy was organized upon the request of many members. And Andrej was nice enough to make all this boring work to list all the component's source etc...
Then, always upon requests, to produce the printed boards, then to get the SMD soldered on it, then to make a real Kit with all active device in the box.
He was, at this time, contacted by some commercial company, and, while he continued to work on the first samples, made, as usual, some little changes with a component value, change that he did not published, but offered in the group buy boards. The administration of the forum decided to move its topic to the commercial area, because 'non open source' (one component was not exactly the one on the schematic) !
For me, When it is about CFA, Andrej (L.C.) is 'the man', as Bruno for class D.
But we have the luck to have other people experienced in CFA, over here, like Nelson Pass, Richard Marsh, Bonsai, Manso etc. (non exclusive list)
Not the point of view of everybody here, remember how i was fired when i tried to demonstrate-it ! It seems young (and uncivilized ) people (always the same) just do not realize that phase or delays matters in a feedback path. And that, in practice, even with 10Mhz of closed loop bandwitch, you still are not fast enough to get 20 000hz feedback bandwidth flat (ratio equal to 200hz one).
Last edited:
In my case I suffered from anti placebo . That is I didn't beleive in the design I made as much as anything . My friend did believe and convinced me . I made an ultimate version of it . It was compared to the Lyra Connoisseur . It was not far off ( using Lyra Helikon ) and not always in favour of the Connoisseur . The tweaks were mono construction ( x 2 ) . 3 op amps in parallel to get statistical noise cancelling . Finding best gain per section is important . Polystyrene caps . I am told NPO ceramic are good ! S N was about - 75 dB typical with modern MC pickups . Overload margin about 32 dB . Passive input filter a good idea . Would like to make double inverting active version using 50 R MC load . Most MC are OK with that . Might win more than it looses .
In a very civilized manner, I will call you again for this utter nonsense. There is no delay to speak about in a feedback path, and the "feedback bandwidth flat" concept is techno babble.
This is about the feedback chasing his tail, not even an original story.
Don't you understand what i was talking about ? Please, make a response curve measurement at the output of your LTP. You will see (for constant level at the input) that the level begin to increase after (around)... 1Khz !!!
Why are you now singling out the LTP while you were talking closed loop phase flatness @ 20KHz?
My point is that I can achieve 20K phase flatness with an LTP, enhanced VAS and standard FET output no problem. So again, what's the point?
One thing we never can look at is pre-distortion inside a typical transistor amp . Or we seldom do .
When building a zero loop feedback tube amp it is a useful bonus sometimes to match the reverse phase sections for best compromise . This technique is not universally approved of ( triodes only religion , triodes = easy , doesn't always mean best ) . I like pre-distortion as we get a nice results with no loop delay . The loop being far less than ideal with a transformer in the way . Also I have found such amps accept feedback nicely and do not produce the expected results . They look like low distortion versions of the original ( 16 .5 db fb in the one in mind ) , that is not always true of the conventional way . If anyone has an interest in tubes I would say use cathode resistor bias to get repeatable results . Do not run your driver device too hard ( EF 184 at 4 mA will be fine ) . Many dismiss pre-distortion as being too variable ( also they saw a pentode ) . I was using tubes from my junk box when I designed my amp . Some were nearly dead at full power ( 80 % emission at a guess ) . At sensible power they were fine and the results changed by a fraction from new to ancient . I do not like LED's as cathode reference .
When building a zero loop feedback tube amp it is a useful bonus sometimes to match the reverse phase sections for best compromise . This technique is not universally approved of ( triodes only religion , triodes = easy , doesn't always mean best ) . I like pre-distortion as we get a nice results with no loop delay . The loop being far less than ideal with a transformer in the way . Also I have found such amps accept feedback nicely and do not produce the expected results . They look like low distortion versions of the original ( 16 .5 db fb in the one in mind ) , that is not always true of the conventional way . If anyone has an interest in tubes I would say use cathode resistor bias to get repeatable results . Do not run your driver device too hard ( EF 184 at 4 mA will be fine ) . Many dismiss pre-distortion as being too variable ( also they saw a pentode ) . I was using tubes from my junk box when I designed my amp . Some were nearly dead at full power ( 80 % emission at a guess ) . At sensible power they were fine and the results changed by a fraction from new to ancient . I do not like LED's as cathode reference .
Bonsai, thanks for clarification. It was not that clear (to me) from the test review which part of the amp is using it. Now I understand that they wanted better small signal transfer.
Good idea. Meantime, to avoid making again a fool of yourself, you may want to read a bit about phase shift and group delay and how are they related. Even Wikipedia will do.
BTW, I may give it up myself, I'm tired of sanitizing around.
Waly,
Your style is arrogant and dogmatic. You speak as if all issues in electronics are resolved and you know the right answers. I am amateur and I am bit sceptical that we could explain everything. In fact because I have background in humanities I know that all important questions do not have answers and some question never will. So I try to be open minded and gather different opinions. I think that the nature of feedback is one of those questions. It's very controversial issue where I see different people with different answers. Esperado offered his view and you just dismissed his view without giving convincing alternative view. I tried to understand why I have so similar listening experiences as Esperado (VFB vs. CFB) and I am open for explanations. Even veterans like Dick Marsh are to some extent surprised with perception (by the people that listened) that CFB is superior sounding in audio applications. Waly, try to be less young, pretend to be cool old gentleman. You know more math than me, but you are not Grisha Perlman nor the man who proved last Fermat's theorem.
It seems that some people do not understand the nature of the problem: to relate subjective perception of experienced listeners with exact electronic laws. This is very complicated issue.
Last edited:
Just away on holiday now to Rhodes ( yikes ) . Take up valves . 0.1 % is a dream . I did it once without loop feedback at 1 watt using MJE 340 and 350 current sources and sinks . The valve guys would hate me . Valves as voltage amps and transistors as current amps. I fell in love with those transistors as they always worked . The valve boys told me you must use FET's . Why on Earth would I ? Seems FET's are allowed as they confuse they for being silicon valve analogues . I used no internal loop feedback .
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.