John Curl's Blowtorch preamplifier part II

Status
Not open for further replies.
wow does that have all the markers of fanboyism wrapped up in one concise post!

I objected to and pointed out the repeated pattern of claims for CFA advantages that in fact were not exclusive to that topology

to a fanboy everything good under the sun is attributed to their object of admiration

to a fanboy unthinking Loyalty is a absolute requirement

any attempt to add perspective, place it in a larger context is an attack

Amen to that.

And to add insult to injury, what is herein called a "CFA" has very little of the CFA properties (as known from the IC industry), due to the high closed loop gain.

The high slew rates due to "current on demand" are not, by any means, a property of CFAs, it was clearly shown that VFAs with "current on demand" can easily be designed. In its probably simplest form, see the Stochino amplifier.
 
diyAudio Member RIP
Joined 2005
Amen to that.

And to add insult to injury, what is herein called a "CFA" has very little of the CFA properties (as known from the IC industry), due to the high closed loop gain.

The high slew rates due to "current on demand" are not, by any means, a property of CFAs, it was clearly shown that VFAs with "current on demand" can easily be designed. In its probably simplest form, see the Stochino amplifier.
I recall nonlinear slew rate enhancement circuits presented in nuclear science papers in the early 70's. I learned so much from the nuc sci folks, as soon as I stumbled on some of the journals. The instrumentation literature in those days for astronomy was pathetic, with guys even from Princeton getting into hard cover books with totally wrong theories of JFET noise, for example.
 
I really don't want to responsible gor tudning the topic into a car story. To me, bugatti veyron is just a VW howpiece. It's a car that mkes no sense to me, fast all right, highly driveable and well ahndling all right, but I don't need 16 cylilinders for that, it's a perfect wa of killing lost of perfectly good money in a show of self-self prootion (mine is bigger than yours). You can get all that from an AC Cobbr for a lor less money. Without being disrespectful, I don't think you'd last long in a Bigatti Richard. And if you drive at street speed, then what's the point of having such a car, you might as well buy a VW Polo, it's a lot cheaper and less prone to stealing.

It's been said many times that if you ever drive an Alfa Romeo, you'll be hooked for life. I understood tht when I once got an Alfa, a relaivrly modest model, for two weeks, IT's nothing in particular, rather it's all of it, how the endgine propels you, how the car behaves, how it reacts. This is comletely in accord wih my years of driving a hand made casr, custom or tailor made for me, the way I like it. Including a Yugo bodywork, but that was all that was Yugo's. 142 hp fro a naturally aspirated 4-in-line FIAT 1.6 litre engine. Custom fuel injection, the type Ford reccomends when your factory unit is no longer enough. Because of the small weight, just 820 kilos with full reservoir, it does 0-60 mph in just over 6 seconds. The first thing I had to do was to learn to drive it all over again. I can do awesome things with it, but at he cost of some ompromise, stiffness versus a semblance of comfort. Let's just say i have some good hands-on experinece with that sort of stuff. The days of such vehiles are numbered by seed limits, cameras and millions of cops.
 
Amen to that.

And to add insult to injury, what is herein called a "CFA" has very little of the CFA properties (as known from the IC industry), due to the high closed loop gain.

The high slew rates due to "current on demand" are not, by any means, a property of CFAs, it was clearly shown that VFAs with "current on demand" can easily be designed. In its probably simplest form, see the Stochino amplifier.

I'll drink to that. Way back in 1973, the Otala and Logstroh's low TIM amp had its VAS stge biased at 20 mA and an effetive slew rate od 100V/uS for a nominl output of 25750W into 8/$ Ohms.

I read somewhere on the net that CFA was more susceprible to PSU problems than VFA. Is this true?
 
Hehe, the magic wold "CFA" was pronounced. Count W... came out of his coffin ;-)

I read somewhere on the net that CFA was more susceprible to PSU problems than VFA. Is this true?
Nothing different between them but the input stage.
LTP null the PSU ripples because The + &- branches suffer the same ways from those noises, but in opposite phase. You just have to filter the rails of the CFA's input stage with some regulatior or Cap multiplier. While you are on this, you can power the VAS as well from the regulated rails and had even more PSRR ;-)
 
Last edited:
www.hifisonix.com
Joined 2003
Paid Member
wow does that have all the markers of fanboyism wrapped up in one concise post!

I objected to and pointed out the repeated pattern of claims for CFA advantages that in fact were not exclusive to that topology

to a fanboy everything good under the sun is attributed to their object of admiration

to a fanboy unthinking Loyalty is a absolute requirement

any attempt to add perspective, place it in a larger context is an attack



in discrete audio power amps compensation is able to be tailored - CFA op amp let you do it externally with feedback Z - in discrete design VFA compensation can be varied however justified

CFA do cut a pole out of the forward gain path - discrete power audio power amps are generally limited by output device speed, packaging, wiring
so the faster small signal stage is of limited advantage


the CFA thread mixed in the minimalism rhetoric as well - but when you want low audio distortion you need high loop gain which requires compound transistors, cascodes, "Super Pair"...


and of course there is Richard Clark's, Carver's "challenges" - Geddes using $200 chip amped Pioneer home theater amps...

given all of that maybe CFA fanboys should just chill - we are handily past audibly critical choices in audio power amp design, into "conceptual art" designs, builds just to please our eyes and brains with topology, parts choice, artistic execution


My point proven. Thank you jcx.
 
Not setting my hopes to high, but perhaps this time you would care to provide an answer: what exactly do you mean by "easier to compensate for stability"? From your previous attempt, my impression was that you mean "compensating without having a clue of what I am doing".

I can try to answer that. I don't claim I am a CFA expert just beginner but I would like to say a few things.
To get high loop gain at 20 kHz both type VFA or CFA need some kind of multiple pole compensation mostly used TPC or TMC. In the VFA I used TMC as I've found a bit simpler to get a good result. The compensation components value were quite critical, not much fluctuation allowed or the stability could be compromise. My best VFA result were, PM about 60 degree, GM hardly higher then 10 dB if I wanted to get the loop gain about 60 dB. I built that amp and it sounds and behaves very nice. The THD is very very low up to 1 kHz but not so good at 20 kHz.
I tried different CFA IPS and BJT or MOSFET OPS and came to conclusion that, if I don't want to overcomplicate IPS, the best is to use complementary diamond or complementary supper pair(I prefer a supper pairs as is very simple to cascode) and enhanced complementary VAS/TIS( EF enhanced) very similar I used in my VFA(but not complementary). What surprise me is how easy is to compensate it, components value not critical at all. I started with TPC and added a kind of OIC. You Waly should remember the schematic, I ask for your opinion but you said is to complicate to analyze. I could easily get the PM more then 80 degree and the GM more then 20 dB and all that with the loop gain in excess of 80 dB at 20 kHz. And I could change capacitance value with no big change in the compensation. I would call it easy to compensate. The same was valid if I used MOSFET or BJT OPS, with mosfet a bit easier as it needs no predrivers.
The best thing I've found about my CFA is similar distortion at all frequencies and all output power. With VFA THD fluctuates much more specially with frequency( if the loop gain is not lowered at LF with local NFB around VAS or by shunt compensation)
Damir
I have built the mosfet version and it is very stable
 
www.hifisonix.com
Joined 2003
Paid Member
Stochino is not current on demand. The IPS is run at high tail current and the comp scheme supports high SR.

I agree that if you raise the loop gain on a CFA, you end up with the same compensation design issues you have on a classic VFA - I discuss this point in the write up comparing the two topologies on my website. However, in their basic forms, the CFA is wider bandwidth and the slew rates are higher. The phase accumulation in VFA's is greater.

Finally, if OS and whoever else decide their CFA's sound better, who are you, or anyone else to criticize them? If you think it's nonsense, stick with a damn VFA.
 
Last edited:
www.hifisonix.com
Joined 2003
Paid Member
I don't think a blanket 'CFA's are easier to comp' statement is necessarily correct.

More correctly, classic CFA's - low(er) loop gain and HF poles that fall closer to the ULGF - mean only light compensation is required for stability - you do not need to 'pole split' which is the case with classic VFA's. .

On my sx-Amp, the loop gain -3 dB is 60 KHz (c. 35 dB) and a single comp cap of 220 pF is used.

On VFA's, because of the higher loop gain, phase accumulation proceeds more quickly, and more of the HF poles fall above 0 dB. So, in classic Miller comp, you have to use pole splitting to get them stable. That said, you can of course use MIC or related comp schemes to improve SR performance in VFA's, which is where most modern VFA designs have headed.
 
Last edited:
Here is an example of an 8 transistor core CFA config (my head phone amplifier of 2-3 years ago); The test signal (yellow) is just below the amps distortion:

View attachment 508549


.... faster as heard on brushes, finger snaps or other transient signals. Esperado here has been saying the same thing.

However, if there is any audible correlation to a measured parameter, the higher slew rate amp is better sounding. CFA excel in easy to get high SR and still have low distortion as well. That is the only parameter that is significantly difference from VFA and CFA of OS's maxed out designs. The DADoD CFA amp is also up around 450v/usec SR with ultra low distortion.

Is there a 450v/us SR VFA of same power output etal and as low THD to compare with 450v/us CFA?

Same Same.
Case closed?

THx-RNMarsh

On most music , nearly impossible to hear ... that is why I mentioned
Parsons. Only the better mastered recordings highlight the CFA
superiority. Listen to Metallica on the CFA/VFA - impossible to differentiate.

I had one VFA that could do >600V/us (Infidel) , but it was picking up
radio broadcasts in Greece (Fail !! ).

A more accurate loudspeaker system might also "highlight" the speed.

PS - good to see your AP will read >-110db. At 0db , it needs to.
My test setup can not do >-100 and only at 1K (cheap hardware). :(

OS
 
Dadod, your point would be well taken if you could explain how you arrived at your amp compensation solution (set aside that you are always showing the results far from the critical areas, like output close to the rails, that the ULGF is unusually high, etc...). Given that there is no way that I see to analyze the loop gain other than by simulation, I'm suspecting you used the same tool and lots of error and trials until you reached some sort of optimum. Should you invest the same amount of effort in a VFA, you could reach the same level of performance ( which is definitely possible). Now, if you are talking about the sensitivity of the compensation, I haven't seen a shred of proof that it could be higher for a VFA.

To save me a separate response to a fan club member, the Stochino amp is very current on demand. A little cap makes the whole difference in the world, compared to a standard LTP.
 
www.hifisonix.com
Joined 2003
Paid Member
nor does it justify CFA fans launching ad hominem attacks, claiming examination of ideas is "persecution"

Agree, but we have hundreds on this forum who claim their amplifiers (of whatever persuasion you care to nominate) sound better because of x or y.

But, let anyone say 'my amp sounds better because it's CFA' and we have the anti-CFA hounds giving chase (just picture it - hilarious!).

And you are one of them!
 
Last edited:
Member
Joined 2014
Paid Member
Holley and other American carbs had accelerator pump circuits, never needed to add fluid damping to make those work. You could change the shot size and the shot timing very easily. Why make anything more complex than necessary and why need the constant maintenance is the real question? I guess it is as always the American designs are meant to be simple and just work, European designs just seem to want to add some elegance to a simple solution, added complexity for no real gain in final usage. Don't get me wrong, I can appreciate an elegant design and nice looking parts but many times simplicity wins in the end with less maintenance and service life. Yes a row of Weber carbs looks nice but having to synchronize them is a pain, just like a row of carbs on a motorcycle, same process of looking at equal vacuum drop across multiple carbs.

I never found balancing webers much of a problem (but that was only 3 barrels). Now when carbs are argued the problem is that the starting point is very different on different sides of the pond. for a 250+CI V engine with a shared manifold a 4-bbl with progressive throttles is a really really good solution. On an 850cc 4-pot its not. If you can't fit a twin choke setup in then a variable venturi or constant depression setup works really really well, at least until the rubber perishes and the oil leaks out.

DCOE webers. Well power, economy, drivability: pick 2. Didn't help that weber never let out the secrets of how the designed their jets. Took me a while to get my head around it. A wideband lambda sensor helped, but there is still no real explanation how to modify a main jet to give you the fuel curve you want . BUT nothing beats the sound of a DCOE on full chat.

Each has its place. vive la difference etc.
 
I don't think a blanket 'CFA's are easier to comp' statement is necessarily correct.

But the CFA requires rail conditioning to get VFA PSRR. In the end
, circuit complexity evens out between the two.

Using the same multipliers with a VFA symasym swapped out for the CFA
leaves me with massive PSRR (-120db+).
I forgot to mention , the VFA symasym can reproduce my DSP hall reflections
"better" - where you say to yourself " it is this designs strength".

I've had to sit right between the speakers and really listen hard to hear
the mid and highs to access which amp does what better. I still have
30YO hearing up to >15khz - still real hard with both amps <30ppm.

Both the CFA and VFA settle instantly , no heat - even no issues when
the rails collapse at shutdown . Both have 2 compensation caps , MIC/lead
for the VFA , symmetric MIC-CFA.
If I put a 22pF in the VFA , I would have a oscillator and blow my Zobel.
Even if I put 10pF's in the CFA , it might make the zoble warm with a Squarewave
... but not self oscillate.
"Easier to compensate" could also mean harder to compensate wrong.
Much easier to make a VFA oscillator.
OS
 
Bil,
I think we can agree that electronic fuel injection is just so much better than a carburetor that I don't miss having to set float levels and changing jets and idle circuits. A couple of oxygen sensors and mass air flow and all the other sensors and you just have such a superior system. Yes it is much more complex but a major improvement in every sense. No you can't fix it with a screw driver and a couple adjustments to limp home but you can never match the power output or fuel economy of fuel injection. I've been a fan since 1974 on my Mercedes 450SEL, never had a problem with the injection system in 500K miles besides a new fuel pump and filter.
 
Status
Not open for further replies.