Well
Ed, the quote below is from a real engineer not generalizing (page 5 of MC article)
George
rather disingenuous coming from people unwilling to accept the engineering generalization that degeneration is negative feedback, gives the same "harmonic multiplication" - even Nelson Pass can accept that, measure the "new" error components in degenerated circuits
who is ignoring the evidence in front of them in that case?
who is ignoring the evidence in front of them in that case?
OK Scott,
For the slow group! (
) We are talking about fashionable audio designs. Not practical mass produced parts. I will not get into showing which audio designs have been produced that do not really meet those requirements.I will however mention clipping behavior and recovery is important in those systems where signal levels are not preset or limited. (Think live microphone use.)
ES
P.S. If you ever look at what I build you might actually find some IC's used when they fit the design. I think I have shown by clear measurements some IC's are limited by the feedback parts!
Last edited:
Pavel,
The testing protocol we are working on here is simple. We set up two of the same type loudspeakers in a stereo listening environment. The tweeters are aimed at the subjects ears. The room is NC 30. All amplifiers (Mono blocks) are set for the same gain and input impedance. Typically the left amplifier stays the same for all tests. The right amplifier is changed. It can be the same model as the left unit or a test model.
Right now the difference in amplifiers is just the feedback network.
The current issue is selecting music that our "Skilled" listeners can use to correctly identify "Something is wrong" or similar comments.
The next step will be to use amplifiers where distortion is deliberately introduced. Right now I am considering 9th harmonic, even orders nulled and thermal delays.
Then we will see if random folks can identify errors at a higher level and then find them at "Skilled" listener levels.
But as this is just playing around to see what pops up there are no promises made. (Or axes to grind, we send them out to a journeyman machinist with the rest of the tools to be sharpened.)
ES
shouldn't take anyone too long to compare, out of our relative number of posts on diyAudio the fraction we respectivily spend on feedback, theory, attempts to illustrate the theory in sims, providing references, actual engagement with other's technical arguments - responding to criticisms of the presentation, modifying arguments, looking for alternative presentations/viewpoints to persuade with technical arguments
can I anticipate?
you will want to claim Market Success justifies anything you want say
or you will beat you chest over your graduate EE Courses a Berkley, working with Spice developers
or you will go on and on about personalities you have known, appeal to "people you trust"
I actually do respect your accomplishments in audio, but don't see any reason to "give it a break" when you spout fuzzy, misguided or just plain wrong characterizations of EE knowledge, understanding of signals, circuits
I dearly wish you would engage in straight forward technical arguments
Last edited:
you had an opportunity in the "discussion", my sims of the JC-3 preamp inner loop to show you understand Bode Sensitivity analysis, its application to determining how differing feedback affects circuit performance
I missed the part where you did anything but assert your intention for your "inner loop" - did you indicate how I am wrong in the sim, its interpretation?
Your rational for why you included the "inner loop" MegOhm R doesn't mean the circuit actually works the way you claim - where does the Bode analysis go wrong with this circuit?
I missed the part where you did anything but assert your intention for your "inner loop" - did you indicate how I am wrong in the sim, its interpretation?
Your rational for why you included the "inner loop" MegOhm R doesn't mean the circuit actually works the way you claim - where does the Bode analysis go wrong with this circuit?
I really don't know. I designed that circuit in 1974, as a horn tweeter (T-350) amplifier for a 3 way all-horn loudspeaker system with John Meyer, in Switzerland. It worked quite well, at the time.
Mark Levinson wanted a slightly 'improved' version in 1975, for a small amplifier, which he built.
What is important is that: The amplifier is very fast considering the lowish Ft bipolar transistors available at the time, about 100V/us.
The design followed Otala's recommendation of reducing the open loop gain and increasing the open loop bandwidth.
As the original circuit was operated in Class A, heroic efforts for output stage linearity were not too important.
It was a successful experiment, as it worked very well for the task, sounding better than any other commercial solid state amp that we could find to compare it with. We could use commercial amps up to about 5,000Hz without too much trouble, but above 5KHz, they all sounded lousy, until the Otala prototype arrived within the year, and I bought it immediately, especially for electrostatic headphones. One of the best purchases that I ever made.
Now what does this all mean?
Does it really matter IF the open loop bandwidth is 20KHz? I think so, but it is difficult to get it without compromising the distortion level at near full output. It is just a matter of how much global negative feedback is available. Today, NO Parasound power amp has internal feedback resistors that reduce the open loop gain, it just is not practical.
So where is the argument? I compromise, like just about everybody else, to get 'specs'. I just don't 'sell out' to make a 'spec' at the lowest cost possible, trading any potential sound quality for it.
The CTC Blowtorch is the ONLY product that I have yet produced, that is not 'compromised' by commercial concerns, and I have not made one for at least 5 years and never expect to make another. Yet, I will say, it is the best piece of electronics I have ever made, and I expect it to last me my lifetime. Everything else is a compromise that I would replace, if I had the time and money to do so.
Mark Levinson wanted a slightly 'improved' version in 1975, for a small amplifier, which he built.
What is important is that: The amplifier is very fast considering the lowish Ft bipolar transistors available at the time, about 100V/us.
The design followed Otala's recommendation of reducing the open loop gain and increasing the open loop bandwidth.
As the original circuit was operated in Class A, heroic efforts for output stage linearity were not too important.
It was a successful experiment, as it worked very well for the task, sounding better than any other commercial solid state amp that we could find to compare it with. We could use commercial amps up to about 5,000Hz without too much trouble, but above 5KHz, they all sounded lousy, until the Otala prototype arrived within the year, and I bought it immediately, especially for electrostatic headphones. One of the best purchases that I ever made.
Now what does this all mean?
Does it really matter IF the open loop bandwidth is 20KHz? I think so, but it is difficult to get it without compromising the distortion level at near full output. It is just a matter of how much global negative feedback is available. Today, NO Parasound power amp has internal feedback resistors that reduce the open loop gain, it just is not practical.
So where is the argument? I compromise, like just about everybody else, to get 'specs'. I just don't 'sell out' to make a 'spec' at the lowest cost possible, trading any potential sound quality for it.
The CTC Blowtorch is the ONLY product that I have yet produced, that is not 'compromised' by commercial concerns, and I have not made one for at least 5 years and never expect to make another. Yet, I will say, it is the best piece of electronics I have ever made, and I expect it to last me my lifetime. Everything else is a compromise that I would replace, if I had the time and money to do so.
Well, I guess I scared everyone away! Sorry about that. I hope it is understood that I simply attempted to describe the design philosophy behind the JC-3. It was not a 'tour-de-force' of advanced mathematics, but a mating of a successful line driver with an output stage, to make a small power amplifier. It worked.
We're *all* guilty of hubris, of believing that the map is the world (our map! and not somebody else's), of believing our linear models of a non-linear world, and of curve-fitting from the safe mechanical known to the dark perceptual unknown. What else can we do? We're a modeling species - it may not even be possible for there to be any other kind of intelligent species.
But sometimes it's also valuable to accept that we don't really know *everything*, and that in the remaining twilight areas the "lore" of experienced trackers is most valuable. That's why I read this thread, for the lore. Thank you for keeping it going.
There's a Tarkovski movie whose title is translated poorly into English as _Stalker_ that comes to mind somehow.
Ruminations, sorry. But thanks for the thread,
Chris
But sometimes it's also valuable to accept that we don't really know *everything*, and that in the remaining twilight areas the "lore" of experienced trackers is most valuable. That's why I read this thread, for the lore. Thank you for keeping it going.
There's a Tarkovski movie whose title is translated poorly into English as _Stalker_ that comes to mind somehow.
Ruminations, sorry. But thanks for the thread,
Chris
If you can do a better one, why not do it? You will be remembered by what you do. And if jfets are not good enough any longer, why not do tubes?
- Status
- Not open for further replies.