Yes Jan maybe I should get out more!
I do admit I am pretty feisty about my ideas...
But it is not personal I felt you were a delightful fellow over lunch.
I am convinced of my ideas as most committed designers - that is the reason the never agree!
Hugh
I do admit I am pretty feisty about my ideas...
But it is not personal I felt you were a delightful fellow over lunch.
I am convinced of my ideas as most committed designers - that is the reason the never agree!
Hugh
Member
Joined 2009
Paid Member
Funny thing is that I don't hear huge differences in sound with changes in bias of the output stage even though it does in theory and in simulation influence the distortion quite a bit. Yet changes in the value of Cdom have quite a noticeable impact on the sound I perceive. So this experience tends to align with your views Hugh.
Hmmm. Interesting.
Have you simulated your amp? If so did you look at these issues?
No problem if you haven't - there's only so much time in a day.
Jan
Have you simulated your amp? If so did you look at these issues?
No problem if you haven't - there's only so much time in a day.
Jan
Member
Joined 2009
Paid Member
Yes, I do remember doing simulations - of course both the bias and Cdom affect the distortion profiles but I don't have the experience to relate a simulated FFT to what I hear. Generally, I find that I prefer the sound of a clean (low distortion) amplifier and have no issue using gobs of local and global feedback - the amplifiers where I've used high feedback have none of the harshness or fatigue often claimed for high feedback. I've been unable to explain why some amplifiers sound harsh and others don't but I remember harshness when using CFP in the input and the VAS but not when used in the output. I've not had the patience to twiddle around with varying nested feedback and listening to the results. I've suffered more harshness from cable/speaker issues than I have from amplifiers.
CFPs are very linear as long as they conduct, when the current approaches zero they become very non-linear*. Therefor I would only use them in input or Vas stages and as (pre) driver where they can be run in class A.
But there are lots of other factors determining the overall result of course.
Jan
* Since a CFP relies on internal feedback within the pair they are ruler flat linear, but as the current goes to zero, the feedback collapses and all hell breaks lose. Bad!
But there are lots of other factors determining the overall result of course.
Jan
* Since a CFP relies on internal feedback within the pair they are ruler flat linear, but as the current goes to zero, the feedback collapses and all hell breaks lose. Bad!
Is so than in your opinion what can be done for max. linearization of a/B-class biased OPS before GNFB is applied ? , excluding pure A-class biased OPS .
For an AB output stage there is an optimum bias voltage which can be measured across the two emitter resistors. Both Douglas Self and Bob Cordell have written about it.
It is different for a triple emitter follower output stage than for a sziklai output stage. The nice thing is that it is independent of the value of the emitter resistors.
For instance, IIRC, not sure for which type of stage, but in one case the optimum value is about 25mV across the emitter resistors.
Of course the bias current then depends on the value of the emitter resistor. But it is the voltage across them that sets the optimum value, NOT the emitter resistor value or the bias current!
Jan
Member
Joined 2009
Paid Member
There's something going on that keeps them sounding great when used in the output. Perhaps it's minority carrier lifetime in the base.
Jan,
But CPF output stage can certainly be used in class A. If configured as an emitter follower, it even can have very low output impedance along with low distortion (and low order) on it's own. I even submitted an article about one to Linear Audio!
Bill
Response to post 63
Hugh my intention to use 47K shunt from VAS collector down to ground, it is to reduce the influence of impedance variations the drivers output stage. The idea is to avoid pre-drivers or buffer.
Regards
GE
Hugh my intention to use 47K shunt from VAS collector down to ground, it is to reduce the influence of impedance variations the drivers output stage. The idea is to avoid pre-drivers or buffer.
Regards
GE
Yes, I had thought this too..... the fixed resistance of 47k swamps the variable, complex impedance of the drivers, it works and sounds good.
It also reduces the loop gain of the amp, robbing some of the global feedback network. It loads up the VAS and brings more low order harmonics, mostly musical, because it is single ended. I prefer to put in nested fb from the VAS back to the negative fb node because this way you reduce the THD inside the nested loop, the first and second stage, before you apply the output to the output stage. Then the global feedback is then largely working on the output stage, as Jan had mentioned.
But it works, and brings sonic improvement....... why change something it is works?
Cheers,
Hugh
It also reduces the loop gain of the amp, robbing some of the global feedback network. It loads up the VAS and brings more low order harmonics, mostly musical, because it is single ended. I prefer to put in nested fb from the VAS back to the negative fb node because this way you reduce the THD inside the nested loop, the first and second stage, before you apply the output to the output stage. Then the global feedback is then largely working on the output stage, as Jan had mentioned.
But it works, and brings sonic improvement....... why change something it is works?
Cheers,
Hugh
Yes of course, I was just commenting that when they are starved they get ugly. You can use them in an output stage but be sure that they keep on conducting through crossover, which contrary to population believe is no sinecure.
Jan
Jan,
Let me offer my experiences of CFPs where I spent many months and years testing between 2004 and 2008.
I found a bipolar/mosfet CFP outstanding in Class A in my 1994 28W Glass Harmony.
In my first Lifeforce I used a LTP comprised of bipolar CFPs. This brought very good resolution compared to a standard Lin/Bailey/Self topology and gave more dynamic range since the master devices were operating at near constant current. This flattened the S shaped transconductance curve, which bedevils most simple LTPs particularly for high amplitude signals, viz bass material.
In around 2005 I began to investigate nested fb, although I had built a Cherry amp in the mid-nineties but had problems with stability of the reactive elements. This pointed the way to resistive networks - more predictable and conventional lag comp - so I then addressed using a CFP in a VAS.
Here I found problems. In a perfectly functional Self with EFII OPS the sound quality (note that I do not mention linearity) was strident, unmusical. With different LG caps I restored it but needed to use degen on the master. I had discovered much higher loop gain but metering the compensation was difficult. BTW, I had noticed that with all the loop gain I could throw some of it away so I could then drive the OPS which more linearity. I also found that I could increase Zout and thus reduce damping factor. Somehow this improved my bass presentation, bringing moe decay to bass notes
I had very little funds and noting that thousands of very hard working designers had palpably and consistently better THD and rarely improving sound I thought a better methodology was needed. So I conserved my funds and bought used 475 Tek and the ears of good friends....
In the meantime I read, and tested some more tube circuits, which measure very badly but sound rather better than most SS amps with vanishingly THD.
My conclusion has been that linearity is not the whole story, the ear seems to enjoy some sorts of distortion which tube mavens call harmonics. This tells us that some types of harmonics are acceptable - but some harmonics are machine-like and destroy the human perception of good music.
Linearity is mandatory for accurate measurement but it may not fully apply to audio amps. What are the hallmarks of the JLH 1969 Class A, the early Nelson Pass Zen 10w and the Jean Hiraga 15w from 1966? All display high THD but most of the harmonics are H2, H3 and H4. Not much more....
This is food for thought and it has sent me on a long journey towards good, subjective sound. To my mind THD is meaningless unless the harmonics are identified and given measurement.
In closing (this is very slow on a cellphone!!) we base our thoughts on ideas and many alight on math and measurement. We have to think beyond the linearity of the present paradigm. It is not enough to explain how we hear music. There is something more to it and with good spice programs we can examine some of the harmonics. Good spectrum analysers are now reasonably cheap and this detective work is within grasp to the home hobbyist.
Yes, when CFPs turn off they become very ugly! My 2c on this matter.
HD
Let me offer my experiences of CFPs where I spent many months and years testing between 2004 and 2008.
I found a bipolar/mosfet CFP outstanding in Class A in my 1994 28W Glass Harmony.
In my first Lifeforce I used a LTP comprised of bipolar CFPs. This brought very good resolution compared to a standard Lin/Bailey/Self topology and gave more dynamic range since the master devices were operating at near constant current. This flattened the S shaped transconductance curve, which bedevils most simple LTPs particularly for high amplitude signals, viz bass material.
In around 2005 I began to investigate nested fb, although I had built a Cherry amp in the mid-nineties but had problems with stability of the reactive elements. This pointed the way to resistive networks - more predictable and conventional lag comp - so I then addressed using a CFP in a VAS.
Here I found problems. In a perfectly functional Self with EFII OPS the sound quality (note that I do not mention linearity) was strident, unmusical. With different LG caps I restored it but needed to use degen on the master. I had discovered much higher loop gain but metering the compensation was difficult. BTW, I had noticed that with all the loop gain I could throw some of it away so I could then drive the OPS which more linearity. I also found that I could increase Zout and thus reduce damping factor. Somehow this improved my bass presentation, bringing moe decay to bass notes
I had very little funds and noting that thousands of very hard working designers had palpably and consistently better THD and rarely improving sound I thought a better methodology was needed. So I conserved my funds and bought used 475 Tek and the ears of good friends....
In the meantime I read, and tested some more tube circuits, which measure very badly but sound rather better than most SS amps with vanishingly THD.
My conclusion has been that linearity is not the whole story, the ear seems to enjoy some sorts of distortion which tube mavens call harmonics. This tells us that some types of harmonics are acceptable - but some harmonics are machine-like and destroy the human perception of good music.
Linearity is mandatory for accurate measurement but it may not fully apply to audio amps. What are the hallmarks of the JLH 1969 Class A, the early Nelson Pass Zen 10w and the Jean Hiraga 15w from 1966? All display high THD but most of the harmonics are H2, H3 and H4. Not much more....
This is food for thought and it has sent me on a long journey towards good, subjective sound. To my mind THD is meaningless unless the harmonics are identified and given measurement.
In closing (this is very slow on a cellphone!!) we base our thoughts on ideas and many alight on math and measurement. We have to think beyond the linearity of the present paradigm. It is not enough to explain how we hear music. There is something more to it and with good spice programs we can examine some of the harmonics. Good spectrum analysers are now reasonably cheap and this detective work is within grasp to the home hobbyist.
Yes, when CFPs turn off they become very ugly! My 2c on this matter.
HD
Last edited:
Hugh,
Can't say I disagree with your post; I am fully well aware that there is/can be a difference between what a specific person prefers (I would say 'prefer' rather than 'is better' as the latter has no clear reference).
In my previous posts that lead to the present exchange I never referred to 'better' or 'worse' sound - I was only referring on the linearity of certain circuits. That's a great part of my interest.
Now you may ask, if I know that many people prefer equipment that has worse linearity or Zout or whatever than the best that an engineer can do, why bother? Because I believe that your target of reasonable low 2nd, 3rd 4th, while very low or absent 7th, 9th etc is a matter of engineering. Now I would like to have AL harmonics very low or non-existing. You indicated that our preference and that of many others is different.
But there is more to it than that. I have a lot (well, lets say several) friends that love the sound of tube amps, listen to horn speakers and almost exclusively play vinyl.
But when I ask them over to my listening space, and play mine (low distortion high damping ss) next to their fave stuff, switching between the two with my new toy (a Van Alstine ABX box), their reactions are interesting.
We often all agree that there are large differences, but they also comment that they can enjoy listening to either system! I believe but cannot prove that one's character has much to do with what one likes to do or to enjoy in life and that extends to audio reproduction.
There is an interesting analogy with the so called 'pyramid effect'; I won't bother you here with it but for me it lies at he root of the often vehemently defended positions here, even if when one started with this hobby or business, it might have fallen either way.
Jan
Can't say I disagree with your post; I am fully well aware that there is/can be a difference between what a specific person prefers (I would say 'prefer' rather than 'is better' as the latter has no clear reference).
In my previous posts that lead to the present exchange I never referred to 'better' or 'worse' sound - I was only referring on the linearity of certain circuits. That's a great part of my interest.
Now you may ask, if I know that many people prefer equipment that has worse linearity or Zout or whatever than the best that an engineer can do, why bother? Because I believe that your target of reasonable low 2nd, 3rd 4th, while very low or absent 7th, 9th etc is a matter of engineering. Now I would like to have AL harmonics very low or non-existing. You indicated that our preference and that of many others is different.
But there is more to it than that. I have a lot (well, lets say several) friends that love the sound of tube amps, listen to horn speakers and almost exclusively play vinyl.
But when I ask them over to my listening space, and play mine (low distortion high damping ss) next to their fave stuff, switching between the two with my new toy (a Van Alstine ABX box), their reactions are interesting.
We often all agree that there are large differences, but they also comment that they can enjoy listening to either system! I believe but cannot prove that one's character has much to do with what one likes to do or to enjoy in life and that extends to audio reproduction.
There is an interesting analogy with the so called 'pyramid effect'; I won't bother you here with it but for me it lies at he root of the often vehemently defended positions here, even if when one started with this hobby or business, it might have fallen either way.
Jan
Jan,
This has taken me by surprise, actually! I had expected a 'vehemented position' from you, perhaps rather the opposite of my own view, but you clearly do appreciate the subjectivity in this matter. Hats off, Jan, I appreciate your point of view.
I too have found that some people love tubes, and some really like a sharp, incisive, resolving low THD situation. But there are many practitioners who exploit these phenomena, as, for example, recording engineers like George Martin who always liked the H2, H3 mix of a tape recorder, or Rupert Neve, who liked to design with high H2 and H4 in his famous mixing desks. The fact is, the majority of people like a 'warm' sound, particularly for popular music, and I am attracted to this group as they comprise the majority, to whom I engineer my amps.
Ciao,
Hugh
This has taken me by surprise, actually! I had expected a 'vehemented position' from you, perhaps rather the opposite of my own view, but you clearly do appreciate the subjectivity in this matter. Hats off, Jan, I appreciate your point of view.
I too have found that some people love tubes, and some really like a sharp, incisive, resolving low THD situation. But there are many practitioners who exploit these phenomena, as, for example, recording engineers like George Martin who always liked the H2, H3 mix of a tape recorder, or Rupert Neve, who liked to design with high H2 and H4 in his famous mixing desks. The fact is, the majority of people like a 'warm' sound, particularly for popular music, and I am attracted to this group as they comprise the majority, to whom I engineer my amps.
Ciao,
Hugh
Member
Joined 2009
Paid Member
I have come to be more skeptical about historical preferences and historical data because tastes and equipment have changed a lot over time. Some of the things that have changed and may be relevant include - speakers. Tube amps were low power, they needed very sensitive speakers and treble horns were common. Such beasts could be harsh if not tamed by some sweet tube distortion. Harsh speakers are still a problem today at times. I think there are modern speakers with better accuracy in the bass and modern amplifiers with low damping factor are more capable of producing strong bass through these speakers where low distortion of a SS amp is often preferred. There are many other changes I could refer to, not least the style and quality of recordings - which can drive strong preferences for the sound reproduction system.
fyi - I've not seen any evidence that the 'young generation' prefer 'warm' sound with their popular music, actually I'm not sure that they prefer any particular sound, it all seems to be about convenience and portability - almost exclusively SS.
fyi - I've not seen any evidence that the 'young generation' prefer 'warm' sound with their popular music, actually I'm not sure that they prefer any particular sound, it all seems to be about convenience and portability - almost exclusively SS.
Last edited:
I often marvel how bad people read what I write - the fact that I own 3 oscilloscopes apparently defines me already.....
In general, people are always quite surprised that 'engineers' actually listen to music! What a concept, as JC would say.
Jan
Actually that is a reason why I like more transparent equipment. It's up to the mastering engineer how he wants the music to sound. I don't feel the urge to second-guess him and/or to modify his work.
Imagine an art lover taking a brush to the Mona Lisa because he likes his colors 'warmer' 😎
Jan
Last edited: