Constant Power

[Atmasphere]

I couldn't disagree more. I contend that OTLs do not have the same power versus impedance. The measurements of Ralph's MA-1 showed how big the discrepancy is between 4-16 ohms at 1% THD. The power levels aren't close. If you can come up with some other measurements of other OTLs that show they put out the same power as impedance rises up to 16 ohms, please publish them. My comparison of the RM-9 and the MA-1 was spot on in my opinion because the RM-9 fit the definition of a power paradigm amplifier as described by Ralph in his paper. The MA-1 did not.

There were several problems with Bascomb King's measurements. Here they are:
1) the amp draws more power when driving 4 ohms. If the AC power sags, the output power goes down. He set that AC voltage on a variac when the amp was at idle. Being a class A2 circuit, it draws significantly more power at full power. The line voltage sag caused erroneous readings.
2) we found 2 tubes with bad sections in the amp when it was returned. This too would cause lower power output
3) it is evident from his measurements that Bascomb had one speaker terminal at ground. This caused the amp to have greater distortion at full output on account of uneven drive to the power tubes, and also less power.

The actual power output of the MA-1 is 135 watts into 4 ohms, 140 into 8 and 150 into 16. At power levels above 11 ohms the power output of the amp is decreasing, at 30 ohms it is still 145 watts. If you convert this to decibels, the differences at all impedances might be about 1 db. Please refer to the diagram in the manufacturer's response where this is pointed out.

That's your opinion and I don't agree. Frankly, since Ralph twists everything around including even his own words to suddenly mean something else other than what was originally written and he is relentless in his attacks on me and the mods aren't stepping in to stop the nonsense, I'm not posting anymore on this thread even though I started it because you can't have a decent discussion with someone who is bent on subterfuge and name calling. I'm sick of the nonsense and I started this thread to eliminate the nonsense of the other thread which started this one off. I'm off of this merry-go-round. All of this going around in circles is just way too much.

There have been no twisted words. I maintain that you have failed to read and understand the article I wrote. That article has been up for about seven years and has been read by thousands of people at this point. From time to time people have made suggestions to make it more clear. If you have some I am happy to entertain them.

So far I've not resorted to name-calling. I have called you out on Logical Fallacies. Its not personal- its just that a Logical Fallacy is an inherently false argument. It makes intelligent debate impossible. If you read the page I have now linked several times, familiarize yourself with how the fallacies work, and then refrain from them, then we can have a reasonable discussion. I implored you to do this at the very beginning of this thread.

OK - so power paradigm amps are those with damping factors roughly <20, not < 100. And these will offer roughly constant power to a varying impedance load. Then how do we explain andy_c's graph in post #33???

Andy's graph shows the response of amplifiers with varying output impedance. What we can see from this graph is that if the resonant point of the speaker is not set correctly, it could result in too much bass. This is a parameter that the speaker designer can adjust with cabinet size, port size and so on. Duke has elsewhere mentioned some of these techniques.

You haven't described this "black box technique", nor the formula from the Radiotron Designer's Handbook.

The black box technique is taught in engineering and technical school classes as a method of obtaining the impedance of an unknown circuit. In this case the circuit makes some power so we simply find out how much power it can make and into what impedance. As you then lower the load impedance from that point it will be observed that at some z impedance below that point the output power will be cut in half. The missing power is being dissipated in the output section itself- it therefore follows that the output section must be the same impedance as the load.

But measuring output impedance is done the same way regardless of the type of amplifier. You short the input of the amp, inject an AC current into its output, then measure the amplitude of the resulting AC output voltage. The ratio of the output voltage amplitude to the amplitude of the injected current is the magnitude of the output impedance. You just have to make sure the injected current is high enough so the measured output voltage is well above the noise, yet not so high that it causes the amp to operate in a nonlinear manner.

This is the standard method used if you are operating in the Voltage Paradigm. It is useful because the amplifier is able to respond to the injected signal via its feedback mechanism. However it does not actually measure the impedance of the output section itself- it is a measure of the servo gain. How is this different? Simple. If you remove the feedback you would get a higher impedance number. If you add more, it will go down. Now if the output impedance is in fact lower, adding more feedback will get you more power into lower impedance loads- Ohm's Law (and the Power formula) requires it. But that does not happen in reality. If I add 40 db of feedback to our M-60, we can get an output impedance that is about 0.1 ohms. But the output power into 4 ohms remains exactly the same. This is true of all tube amplifiers. Therefore the output impedance is unchanged. What is different is the way the amp responds to load variation- but it can only do that within the limits imposed by the output section itself. If you want it to really have a lower output impedance (at which point it will make more power into a lower impedance) you have to add more output tubes, bigger power supplies, more output transistors, bigger heatsinks, etc. You can't get something for nothing, Kirchoff's Law sees to that.

IOW, the Voltage Paradigm has some defined terms. 'Output Impedance' is one of them- and is confusing to many because they think it means what it says. It doesn't!

I remember Bascomb King struggled with this concept when I pointed it out to him many years back. The struggle comes from being couched in the terminology and practice of the Voltage Paradigm; it is very hard to see out of that box when you may very well have been taught these concepts in school!

Yes. Global feedback has a problem because the propagation delays of several stages compound. That is not a necessary condition for feedback, however. There can be local feedback to each stage, hence no accumulation of delay around the whole amp. This is not a new concept.

Yes- quite true. However if you have multiple stages each with its own local loop feedback, the problem remains- you will still get more odd orders (as they compound through the circuit) and it will still be audible.

And to whatever extent global feedback may raise odd-ordered harmonic distortion, it only matters if the level is high enough to hear. Same is true for any tube amp including OTL, which also have odd-ordered harmonic distortion. The question is how much, and how is it distributed spectrally.

Yes. And here we are at the nub of it. The trace amounts of odd ordered generation are considered 'negligible' by most designers. The problem is that the human ear is more sensitive to these harmonics than it is to human vocal tones, in fact more than even modern test equipment. So what shows up as trace amounts in the noise is easily heard. Audiophiles have long had terms for this- bright, harsh, clinical, hard, chalky, etc- all referring to added odd orders that are very hard to measure. It might also help to understand that a lot of these harmonics are turning up in the band of frequencies to which the ear is most sensitive- bird song frequencies- between 2 and 10KHz. Now what the ear is tuned that way is the subject of another thread- here we don't have to know why, just that it is.

Could you describe the essential characteristics of distortion criteria (even and odd) that correlate with good sound?

To the best of my ability, yes. Such an amplifier would have no IMD and no odd-ordered harmonics. The amount of 2nd would not be very important (this does contribute to lushness, so such an amplifier might be described as warm, lush, having bloom, etc.) by comparison. Less than 5% of the 2nd at 4th at full output would be very hard to hear.

If you could have an amplifier with 0.1 ohm output impedance, and exactly the same distortion characteristics of your OTL amps, would you want it?

Sure! A long time ago I discovered that removing the feedback made the amp sound better. It took me a long time to figure out why. But one of the immediate consequences is that there were a lot less loudspeakers that the amp would sound right on. But I made the choice, since adding the feedback caused the sound to degrade- and this was consistent on amps that were not OTLs too. So I came to the realization that if the speaker required feedback of the amp, there was no way the speaker could ever sound like real music as a result. It might sound good for a stereo, just not real. That has limited the marketplace for us over the last 35 years. It would be nice to have a bigger market

I think you are talking about perceived loudness?

Yes. We know sound has a pressure because we can measure it. But I get the feeling from your response that you are on to what the problem is- if distortion is added, the perceived loudness will be higher. I think this is why so many audiophiles don't really want to play peaks in their system much over about 95db. It gets too painful. The reality is if the stereo is really working right, it will be relaxed and unopressive even at 115db. An orchestra can reach peaks of 120db. How are we to reproduce it accurately if we can't stand to be in the room at the same volume as the original performance? In addition, how is it helpful when others are screaming at us to turn it down? What should be happening is that the system is involving and inspiring such that others want in on the party.

 

[Roger Dressler]

Thank you, Ralph, for the extensive reply. Much appreciated. I understand the distinction between output impedance with and without feedback. I'd like to see if I can clarify some points about this.

If you remove the feedback you would get a higher impedance number. If you add more, it will go down. Now if the output impedance is in fact lower, adding more feedback will get you more power into lower impedance loads- Ohm's Law (and the Power formula) requires it. But that does not happen in reality. If I add 40 db of feedback to our M-60, we can get an output impedance that is about 0.1 ohms. But the output power into 4 ohms remains exactly the same.

Is this not an apples/oranges problem? It is the same error being committed as when a power paradigm amp is measured at max output for different loads, and comparing those power levels as an indication of output impedance. I reiterate, that is not how one determines output impedance, yet you seem to be doing so here based on the bolded part. I agree with you that the added feedback in your tube amp will not in crease the output power of the amp. But at some level well below max, the change from 8 ohm to 4 ohm will essentially double the output power in the load, as the output voltage will be prevented from sagging by the added feedback.

The max output power of the amp of course cannot change, as that is limited by the power supply and the output conductance of the driver devices. And this is equally true for SS amps.

So I think we could eliminate a lot of confusion (as apparently also befell our friend Bascom King) if we remove the issue of maximum power from the discussion of small signal behavior of the amp and speaker interface, which is in fact the relevant area affecting sound quality.

This is true of all tube amplifiers. Therefore the output impedance is unchanged. What is different is the way the amp responds to load variation- but it can only do that within the limits imposed by the output section itself. If you want it to really have a lower output impedance (at which point it will make more power into a lower impedance) you have to add more output tubes, bigger power supplies, more output transistors, bigger heatsinks, etc. You can't get something for nothing, Kirchoff's Law sees to that.

See what I mean? You are intermingling the discussion of max power output with the nature of how an amplifier interacts with the speaker at low power. Totally different issues. Are they related? Yes, indirectly as you say: the more output devices used, the lower the inherent impedance of the output stage. But that is not what is causing the confusion here.

IOW, the Voltage Paradigm has some defined terms. 'Output Impedance' is one of them- and is confusing to many because they think it means what it says. It doesn't!

Oh, but it does. It relates to how the amp controls the load below the onset of gross nonlinearity (i.e. clipping and the like).

I remember Bascomb King struggled with this concept when I pointed it out to him many years back. The struggle comes from being couched in the terminology and practice of the Voltage Paradigm; it is very hard to see out of that box when you may very well have been taught these concepts in school!

A tube amp with feedback and 0.1 ohm output impedance will control the speaker (damping factor) exactly the same as a SS amp with 0.1 ohm output impedance. They may run into overload at different points. That's a different problem, unrelated to output impedance.

Yes- quite true. However if you have multiple stages each with its own local loop feedback, the problem remains- you will still get more odd orders (as they compound through the circuit) and it will still be audible.

I'm not sure there is equivalency here, but I will go away and study.

Yes. And here we are at the nub of it. The trace amounts of odd ordered generation are considered 'negligible' by most designers. The problem is that the human ear is more sensitive to these harmonics than it is to human vocal tones, in fact more than even modern test equipment.

I would find it hard to believe that test equipment could not resolve sufficiently. Usually the problem is in knowing what to ask the equipment to measure, and how to interpret that data relative to human perception. Huge problem, that.

Sure! A long time ago I discovered that removing the feedback made the amp sound better. It took me a long time to figure out why. But one of the immediate consequences is that there were a lot less loudspeakers that the amp would sound right on. But I made the choice, since adding the feedback caused the sound to degrade- and this was consistent on amps that were not OTLs too. So I came to the realization that if the speaker required feedback of the amp, there was no way the speaker could ever sound like real music as a result. It might sound good for a stereo, just not real. That has limited the marketplace for us over the last 35 years. It would be nice to have a bigger market

Very glad to know that the positive sonic benefits are not due to high output impedance. Gives me something else to investigate.

 

[mep]

Thank you, Ralph, for the extensive reply. Much appreciated. I understand the distinction between output impedance with and without feedback. I'd like to see if I can clarify some points about this.
Is this not an apples/oranges problem? It is the same error being committed as when a power paradigm amp is measured at max output for different loads, and comparing those power levels as an indication of output impedance. I reiterate, that is not how one determines output impedance, yet you seem to be doing so here based on the bolded part. I agree with you that the added feedback in your tube amp will not in crease the output power of the amp. But at some level well below max, the change from 8 ohm to 4 ohm will essentially double the output power in the load, as the output voltage will be prevented from sagging by the added feedback.

I can't resist...First of all, OTL amps don't have a separate set of engineering rules they get judged and measured by in a testing laboratory just as their is no such distinction between speakers when they are measured with regards to the power paradigm and voltage paradigm words that Ralph uses in his paper (but no one else has adopted to my knowledge). Negative feedback decreases the gain of the circuit and it can't give you more power output than not using negative feedback. Positive feedback can increase gain, but using positive feedback in an audio amplifier would also increase distortion and to my knowledge it isn't used by audio amp manufacturers.

Second, to make the following statements: "So I came to the realization that if the speaker required feedback of the amp, there was no way the speaker could ever sound like real music as a result. It might sound good for a stereo, just not real" is a very self-serving argument if you are in the business of making amplifiers that use no negative feedback. The vast majority of amplifiers on the market use some amount of negative feedback, and it doesn't matter if they are tube or SS. And in case *anyone* wants to twist my words around again, I said the vast majority of amplifiers on the market use some level of negative feedback. Saying "vast majority" is much different than saying "all" amplifiers use negative feedback. How many speaker companies do you think would raise their hand and say "Oh please, don't ever use my speakers with an amplifier that uses negative feedback because you will ruin the sound?" More importantly, how many speakers were designed, built, and sold to be used with amplifiers with zero negative feedback in order to sound their best? Damn few would be my guess. Basically what Ralph is saying is that only the few speakers on the market that will work with an OTL or any other amp that uses no negative feedback are the "right" speakers that will sound like real music and all other speakers that require some level of negative feedback to sound their best are flawed (again, the majority of all speakers known to man).

The bottom line is that in order to buy into the beliefs espoused in the paper that I *never* read or understood, the "rules" of human hearing require different measurements than those that are used to measure non-OTL and non-negative feedback amplifiers in labs and audio publications that actually perform measurements. And you will need special speakers that will perform optimally with zero negative feedback. And as Ralph discovered: "A long time ago I discovered that removing the feedback made the amp sound better. It took me a long time to figure out why. But one of the immediate consequences is that there were a lot less loudspeakers that the amp would sound right on. " Is anyone surprised by that finding? I'm not.

 

[Atmasphere]

Is this not an apples/oranges problem? It is the same error being committed as when a power paradigm amp is measured at max output for different loads, and comparing those power levels as an indication of output impedance. I reiterate, that is not how one determines output impedance, yet you seem to be doing so here based on the bolded part. I agree with you that the added feedback in your tube amp will not in crease the output power of the amp. But at some level well below max, the change from 8 ohm to 4 ohm will essentially double the output power in the load, as the output voltage will be prevented from sagging by the added feedback.

The behavior of the amp will be the same if the output is only one watt. You don't have to have the amp at full power to see this.

With respect to the feedback comment, we are in complete agreement. I am not saying a tube amp cannot be a Voltage Source- certainly they can.

The max output power of the amp of course cannot change, as that is limited by the power supply and the output conductance of the driver devices. And this is equally true for SS amps.

So I think we could eliminate a lot of confusion (as apparently also befell our friend Bascom King) if we remove the issue of maximum power from the discussion of small signal behavior of the amp and speaker interface, which is in fact the relevant area affecting sound quality.

See what I mean? You are intermingling the discussion of max power output with the nature of how an amplifier interacts with the speaker at low power. Totally different issues. Are they related? Yes, indirectly as you say: the more output devices used, the lower the inherent impedance of the output stage. But that is not what is causing the confusion here.

I understand the difference between the full power measurement and lower power tests. We don't need to test the amp at full power to observe the behavior. I am guessing that you got into this discussion from this thread only (this thread is a spinoff of another thread), and so have probably not seen this link, so here it is
https://www.google.com/search?client=ubuntu&channel=fs&q=fisher+A-80+amplifier&ie=utf-8&oe=utf-8
This is a Google search on an old Fisher A-80 amplifier. The first hit these days has a YouTube image of the damping control on an old Fisher amp. The control is marked at its extremes and also at noon: Constant Voltage, Constant Power (noon) and Constant Current. It is a dual-ganged pot that controls current and voltage feedback. At noon the two cancel and one or the other predominates at the extreme settings.

I point this out because it shows that this idea was understood a good half century ago. Fisher was by no means the only amplifier manufacturer with a damping control like this, EV and others made amps like this too.

Oh, but it does. It relates to how the amp controls the load below the onset of gross nonlinearity (i.e. clipping and the like).

Absolutely! But we are not talking about output impedance, we are talking about the servo gain, which creates the appearance- reducing the sag as in your comment above. I think we both agree that we are talking about amps operating within their linear range, and also that we are not going to be playing at full volume all the time. The trouble is, the behavior of a Constant Power amplifier remains different even if we are only talking about 1 watt. Otherwise I have to take it that what you are suggesting is that a speaker that is compatible at one volume will be incompatable at another volume, but experience shows that if a speaker is incompatible with the amplifier on this account, it will be so at any power level.

A tube amp with feedback and 0.1 ohm output impedance will control the speaker (damping factor) exactly the same as a SS amp with 0.1 ohm output impedance. They may run into overload at different points. That's a different problem, unrelated to output impedance.

Actually if they are the same power the overload will be at the same point too. But this is a Red Herring- that is to say its neither here or there with this conversation, right?

I would find it hard to believe that test equipment could not resolve sufficiently. Usually the problem is in knowing what to ask the equipment to measure, and how to interpret that data relative to human perception. Huge problem, that.

Agreed! and a breath of fresh air in your saying it I am of the opinion that if we figure out the right thing to measure, we could all relax and write about the fish we caught at the lake... but it does seem to me that as an industry, we don't measure what we should, and further, there is a lot more error out there than we like to admit. In this thread there is an example of it- Bascomb King made several errors in measuring output power and distortion on one of our amps. But you don't see him admitting that in the review.

The point here though is that the ear/brain system is very sensitive in this regard, such that the word 'negligable' should not be used when discussing the remaining odd-ordered harmonic distortion in any amplifier.

Very glad to know that the positive sonic benefits are not due to high output impedance. Gives me something else to investigate.

Right- the higher output impedance is a side effect, not the goal, but if you know about it you can design a speaker to take advantage of it rather than treat it as a liability.

The more I find out about our hearing/perceptual rules the more interesting it gets! I find it really fascinating that we are most sensitive to bird song frequencies. Birds are the first warning indicator in the forest that there is a predator nearby. It seems that they have influenced our hearing on that account.

 

[Atmasphere]

I can't resist...First of all, OTL amps don't have a separate set of engineering rules they get judged and measured by in a testing laboratory just as their is no such distinction between speakers when they are measured with regards to the power paradigm and voltage paradigm words that Ralph uses in his paper (but no one else has adopted to my knowledge). Negative feedback decreases the gain of the circuit and it can't give you more power output than not using negative feedback. Positive feedback can increase gain, but using positive feedback in an audio amplifier would also increase distortion and to my knowledge it isn't used by audio amp manufacturers.

You are correct- and as I mentioned earlier and as it says in the article itself: The problem here is human nature. We tend to stay within the limits thus set by the existing paradigms and to resist changes that threaten one's viewpoint of the world. When someone else creates challenges to the paradigms, it is normal also to try to protect one's world view by preventing the new idea from gaining ground.

IOW the word 'paradigm' is not used for engineering reasons. It is used to point at human nature.

Second, to make the following statements: "So I came to the realization that if the speaker required feedback of the amp, there was no way the speaker could ever sound like real music as a result. It might sound good for a stereo, just not real" is a very self-serving argument if you are in the business of making amplifiers that use no negative feedback.

That is true, but does not alter the truth of the statement in any way. It might however affect your ability to accept it.

The vast majority of amplifiers on the market use some amount of negative feedback, and it doesn't matter if they are tube or SS. And in case *anyone* wants to twist my words around again, I said the vast majority of amplifiers on the market use some level of negative feedback. Saying "vast majority" is much different than saying "all" amplifiers use negative feedback.

Again correct. We are however talking about high end audio which is a subset of the entire industry. Within this industry the idea of zero feedback has taken hold and there are many such examples, SETs being a very good one. Unknown 25 years ago but easy to find now.

How many speaker companies do you think would raise their hand and say "Oh please, don't ever use my speakers with an amplifier that uses negative feedback because you will ruin the sound?" More importantly, how many speakers were designed, built, and sold to be used with amplifiers with zero negative feedback in order to sound their best? Damn few would be my guess.

Again correct, but here are a few: Lowthers, most horns, most single-driver full-range loudspeakers (in addition to Lowther, PHY, Feasterex, etc.), plus Coincident Technology, Classic Audio Loudspeakers and more. You are right- only a damn few, but this is high end audio.

Basically what Ralph is saying is that only the few speakers on the market that will work with an OTL or any other amp that uses no negative feedback are the "right" speakers that will sound like real music and all other speakers that require some level of negative feedback to sound their best are flawed (again, the majority of all speakers known to man).

Incorrect- you have misinterpreted the comment. The speakers themselves might be capable of the task. But they have no chance as the amplifiers connected to them are simply not up to it. So the speakers can only reproduce what is fed to them and that signal will be colored. There is no way they can sound real.

The bottom line is that in order to buy into the beliefs espoused in the paper that I *never* read or understood, the "rules" of human hearing require different measurements than those that are used to measure non-OTL and non-negative feedback amplifiers in labs and audio publications that actually perform measurements. And you will need special speakers that will perform optimally with zero negative feedback. And as Ralph discovered: "A long time ago I discovered that removing the feedback made the amp sound better. It took me a long time to figure out why. But one of the immediate consequences is that there were a lot less loudspeakers that the amp would sound right on. " Is anyone surprised by that finding? I'm not.

(Once again), this has nothing to do with our amps being OTLs, it has to do with their having no loop feedback. From the article: "Ideally amplifiers under this paradigm have little or no feedback." That is why SETs, the BAT VK-60 and other tube amplifiers that are zero feedback are also Power Paradigm technologies. Human hearing/perceptual rules are the same regardless of the amp (not sure what you were saying in that part of the paragraph so I included that comment just in case to cover the bases.)

Here in the 21st century there are plenty of speakers to choose from. Some of this is due to the proliferation of SETs which have nearly the same requirements as our amps, although we can drive speakers they can't because our amps make a lot more power. But at any time in the last 60 years there have been speakers that work fine with our stuff- all the horns and ESLs of the 1950s and 1960s for example plus a good number of 'ordinary' box speakers like the Advents.

What you might think about is not that it is unsurprising that the number of usable speakers is smaller, but that a manufacturer would intentionally choose this path even though it was/is a simple matter to add the feedback to the amps. I'm not sure if you are aware of this, but Atma-Sphere is the only manufacturer of OTLs that are zero feedback.

 

[mep]

Ralph-I'm glad to see that for once you agreed with the majority of statements I made. The one you said I was incorrect on baffles me though. You made the following statement: "So I came to the realization that if the speaker required feedback of the amp, there was no way the speaker could ever sound like real music as a result. It might sound good for a stereo, just not real."

So I said: "Basically what Ralph is saying is that only the few speakers on the market that will work with an OTL or any other amp that uses no negative feedback are the "right" speakers that will sound like real music and all other speakers that require some level of negative feedback to sound their best are flawed (again, the majority of all speakers known to man)."

And you replied: "Incorrect- you have misinterpreted the comment. The speakers themselves might be capable of the task. But they have no chance as the amplifiers connected to them are simply not up to it. So the speakers can only reproduce what is fed to them and that signal will be colored. There is no way they can sound real."

How could I have possibly misinterpreted your comment?: "So I came to the realization that if the speaker required feedback of the amp, there was no way the speaker could ever sound like real music as a result. It might sound good for a stereo, just not real." We are both talking about speakers that you said if they require feedback of the amp, they are never going to sound like real music. We aren't talking about speakers "that might be capable of the task." So now you are saying that if a speaker is being powered by an amp that uses negative feedback "that the signal will be colored"?

The bottom line here is that you are saying the majority of speakers on the market are flawed because they only sound their best with an amplifier that uses negative feedback and that all amplifiers that use negative feedback which also happens to be the majority of amplifiers on the market are also flawed. Good luck with that paradigm.

 

[mauidan]

The bottom line here is that you are saying the majority of speakers on the market are flawed because they only sound their best with an amplifier that uses negative feedback and that all amplifiers that use negative feedback which also happens to be the majority of amplifiers on the market are also flawed. Good luck with that paradigm.

Maybe Ralph can give us a list of speakers that aren't "flawed" and will work with amps that don't use negative feedback.

 

[Duke LeJeune]

The bottom line here is that you are saying the majority of speakers on the market are flawed because they only sound their best with an amplifier that uses negative feedback and that all amplifiers that use negative feedback which also happens to be the majority of amplifiers on the market are also flawed. Good luck with that paradigm.

Ralph isn't trying to build the amp for everyman, nor for everyspeaker. He's been building his power paradigm amps for thirty-seven years, so someone has been buying them. Last time I checked, he had the largest collection of Golden Ear Awards from The Absolute Sound of any manufacturer. Any manufacturer. Either he has indeed been incredibly lucky, or he's doing something right.

I first heard about Ralph and his amps when I was a fledgling SoundLab dealer back in early 2000. I called SoundLab and asked them what amplifier(s) they recommended. They wouldn't commit to anything, just responding in generalities that didn't help much. So I called them back and asked my question a different way: What amplifier manufacturers have you shown with at audio shows? Well suddenly all they could talk about was Atma-Sphere and how that was the best sound they'd ever had at a show. Aha.

 

[amirm]

Ralph isn't trying to build the amp for everyman, nor for everyspeaker. He's been building his power paradigm amps for thirty-seven years, so someone has been buying them. Last time I checked, he had the largest collection of Golden Ear Awards from The Absolute Sound of any manufacturer. Any manufacturer. Either he has indeed been incredibly lucky, or he's doing something right.

Bose sells more equipment in one minute than Ralf will do in its entire lifetime of its product. Surely we don't use that to say they are doing something more right than Ralf with respect to fidelity .

I first heard about Ralph and his amps when I was a fledgling SoundLab dealer back in early 2000. I called SoundLab and asked them what amplifier(s) they recommended. They wouldn't commit to anything, just responding in generalities that didn't help much. So I called them back and asked my question a different way: What amplifier manufacturers have you shown with at audio shows? Well suddenly all they could talk about was Atma-Sphere and how that was the best sound they'd ever had at a show. Aha.

I don't know much about Soundlab speakers so I did a search and there is a review of their speaker in stereophile from back in 1992: http://www.stereophile.com/content/sound-lab-1-electrostatic-loudspeaker-measurements

It apparently has different settings but one of them produces this for output impedance and phase:

"To investigate the answer to Dick's question, I measured the A-1's impedance magnitude at three settings of the Brilliance Control (BC) with DRA Labs' MLSSA system. With the BC set at Maximum (pot wide open), I obtained the curves in fig.1. Although the magnitude of the impedance is above 10 ohms from the upper bass to 2kHz, and well above 30 ohms in the bass, note the drop in the mid-treble and above. The cursor indicates a punishingly low 1.3 ohms at 20kHz with an equally punishing phase angle of –75.5°, dropping even further to just a small fraction of an ohm at ultrasonic frequencies. This measurement necessarily includes the resistance of the speaker cables DO was using (TARA Labs Rectangular Solid Core); as the margin of error in my measurement is probably around a fraction of ohm, it's quite possible that the A-1 presents amplifiers with a complete short circuit above the audio range. Fundamentally, therefore, in electrical terms, the A-1 with its BC set to maximum is nothing more than a large capacitor. If this isn't hell for a power amplifier, I don't know what is (footnote 1). No wonder DO's amps were choking."

Can you please walk us through whether this is the type of speaker that is supposed to do well with Ralph's amps?

 

[microstrip]

(...) I first heard about Ralph and his amps when I was a fledgling SoundLab dealer back in early 2000. I called SoundLab and asked them what amplifier(s) they recommended. They wouldn't commit to anything, just responding in generalities that didn't help much. So I called them back and asked my question a different way: What amplifier manufacturers have you shown with at audio shows? Well suddenly all they could talk about was Atma-Sphere and how that was the best sound they'd ever had at a show. Aha.

Sometime in the middle 90's a good friend of mine who distributed several high-end brands exposed a system including the Atmasphere MP1-MA2-Soundlab A1 and a Well Tempered Lab turntable at the Lisbon AudioShow. The room was large, and the sound of several LPs, including mostly Sheffield direct cut LPs was really impressive and lifelike - this system was a milestone in audioshow's in my country and is still object of conversation sporadically.

Mostly because of these great sessions even today if I refer I own SoundLab A1's the local audiophiles will show some interest or even respect - much more than for any other type of expensive speaker!

 

[Duke LeJeune]

Maybe Ralph can give us a list of speakers that aren't "flawed" and will work with amps that don't use negative feedback.

A speaker manufacturer looks at the amplifier's output impedance (and obviously also at its power output) rather than directly at the amp's amount of negative feedback when considering amplifier compatibility.

Here is an incomplete list of manufacturers who make speakers that work well with amps whose output impedances (damping factors) put them into the power paradigm category, which is generally the case with amplifiers having little or no global negative feedback. Don't take this as the final word in anything; you still need to investigate the particular amp/speaker pairing you have in mind: Wilson Audio, SoundLab, Klipsch, Coincident, Merlin, DeVore, Reference 3a, Silverline Audio, Classic Audio Reproductions, High Emotion Audio, Omega Loudspeakers, Edgarhorn, PiSpeakers, Tonian Labs, Hawthorne Audio, anything with a Lowther, PHY, Feastrex, AudioKinesis, and even Magnepan (ask Wendell Diller about Atma-Sphere MA-1s driving Maggie 3.6's).

Bose sells more equipment in one minute than Ralf will do in its entire lifetime of its product. Surely we don't use that to say they are doing something more right than Ralf with respect to fidelity

If quantity of sales is the criterion and Bose is the yardstick, then I guess we're all abject failures!

On the other hand, a wise man once told me that success is doing what you love and still being able to pay the bills. So maybe it's not so elusive after all.

I don't know much about Soundlab speakers so I did a search and there is a review of their speaker in stereophile from back in 1992: http://www.stereophile.com/content/sound-lab-1-electrostatic-loudspeaker-measurements

It apparently has different settings but one of them produces this for output impedance and phase:

Can you please walk us through whether this is the type of speaker that is supposed to do well with Ralph's amps?

Excellent question! Hang on, this is going to be fun - especially when we start talking about that nasty-looking nose-dive north of 10 kHz.

But let's start with what's happening south of 10 kHz. Ralph's larger amps (the MA-1 and MA-2; the monster MA-3 didn't exist back then) will deliver approximately constant power into that impedance curve up to 10 kHz, at which point their power output starts to roll off.

So, what happens between 10 kHz and 20 kHz? Obviously Ralph's amps are putting out decreasing power up there, so the high treble region is going to be rolled off, weak, dull-sounding. Right??

Nope!

The missing piece of the puzzle is what the speaker's frequency response curve is like up there. Take a look at the second graph in this link, which shows the effects of the brilliance control:

http://www.stereophile.com/content/sound-lab-1-electrostatic-loudspeaker-measurements

With the brilliance control max'd out we get the impedance curve that you posted, along with a 10 dB peak centered on about 17 kHz (when driven by a voltage paradigm amp)!

In this case, power paradigm amp + very low impedance in the high treble + 10 dB peak in the high treble = beautiful high treble!!

You see, the speaker's impedance curve zigs where the frequency response curve zags, and that's a recipe for potentially wonderful synergy with a power paradigm amp.

Isn't that virtual dead short at 20 kHz going to damage Ralph's amp? Nope, not even with a 20 kHz sine wave! Instead of frying its output stage trying to dump nearly infinite wattage into a dead short, like a "constant voltage" amp would, Ralph's amp delivers essentially no wattage into a dead short and so nothing overheats. In other words, into a very low impedance load, it starts to behave like a "constant current" amp.

(A voltage paradigm amp would put out far more power in the high treble, so we'd have to dial back the brilliance control accordingly. No problem, that's what it's there for. But down in the bass region and midrange region, where we see major impedance peaks, the voltage paradigm amp delivers reduced power. That would be fine if the speakers were optimized for constant voltage, but they aren't [see post number 150 by microstrip where he talks about SoundLabs driven by Atma-Sphere amps], so we have somewhat reduced bass power and midrange presence with a solid state amp.)

I don't think that impedance curve is representative of current-generation SoundLabs. My understanding is that nowadays the M-60 can probably drive them as well as the MA-1 could back in the day.

A bit of trivia - Dick Olsher bought the review pair of Sound Labs, and when he finally decided to sell them (some ten years after the review), he asked me to sell them for him.

 

[mauidan]

A speaker manufacturer looks at the amplifier's output impedance (and obviously also at its power output) rather than directly at the amp's amount of negative feedback when considering amplifier compatibility.

Here is an incomplete list of manufacturers who make speakers that work well with amps whose output impedances (damping factors) put them into the power paradigm category, which is generally the case with amplifiers having little or no global negative feedback. Don't take this as the final word in anything; you still need to investigate the particular amp/speaker pairing you have in mind: Wilson Audio, SoundLab, Klipsch, Coincident, Merlin, DeVore, Reference 3a, Silverline Audio, Classic Audio Reproductions, High Emotion Audio, Omega Loudspeakers, Edgarhorn, PiSpeakers, Tonian Labs, Hawthorne Audio, anything with a Lowther, PHY, Feastrex, AudioKinesis, and even Magnepan (ask Wendell Diller about Atma-Sphere MA-1s driving Maggie 3.6's

Thanks Duke for trying to answer my question for Ralph.

Lucky for me, I have speakers on your list and they're powered by an amp with no negative feedback.

 

[Duke LeJeune]

Lucky for me, I have speakers on your list and they're powered by an amp with no negative feedback.

You mean... I got it right??

Ha! The blind squirrel finds a nut!!

 

[Duke LeJeune]

To mep:

I came across a thread elsewhere on this forum wherein you describe switching from tubes to solid state... specifically, a tweaked-out Phase Linear amp, as I recall. Have to admit there's a definite retro-coolness to the Phase Linear thing.

Okay my question for you is, what were your speakers at the time? I'm just curious as to whether that might have played a role in your preference.

And if you'd rather not say, that's okay.

Thanks.

 

[Phelonious Ponk]

This one strikes me as an (even more) esoteric installment of the old musical vs accurate argument. One extremely small subset of the high end, based in mid 20th-century technology, seems to be claiming musical superiority to all. The "superiority" doesn't show in the numbers. The "technology" (so unique it requires its own language) has failed to achieve broad acceptance (indeed we seem to have the one manufacturer here, in this discussion), even in the highest of the high end where the bizarre is often lauded. And arguments in its favor are put forth with statements that dismiss the best products in the business as sounding like "good hifi" vs. "real music."

Remove the great technical detail put forth to explain the objectively poor performance but failing to explain why all these distortions and limitations would sound more like "real music," and we could be talking about five-figure turntable feet. This hobby has sold bits of pottery you stick to the walls to make "good hifi sound like real music," why not an amp? Even if you are required to play it through Lowthers or Kilpsch?

There will always be a few who will embrace this stuff, wrap themselves in the fuzzy blanket of THD and call it musical. There will always be the many designers, engineers, music makers, music lovers and audiophiles who will prefer "good hifi."

No one ever changes sides, and if I'm the only one who found the "good hifi vs. real music" remark too arrogant, ridiculous and predictable to even be effectively offensive, few are even paying attention. Carry on.

Tim

 

[microstrip]

(...) I don't think that impedance curve is representative of current-generation SoundLabs. My understanding is that nowadays the M-60 can probably drive them as well as the MA-1 could back in the day. (...)

Surely it is not - the speakers changed a lot along three decades. Just look for data I measured on a ten years old A1 - the version that was not OTL friendly . Happily later Roger West released an high impedance modification to please his customers using tubes! The current one's are different - but I have not been able to find the time to measure them with the Toroid II . One think is sure - it is more more full bodied and efficient.

The changes in panel capacitance due to several improvements carried along the years and the crossover modifications have produced wide changes in impedance. And as you said the position of the bass, medium and treble controls can make an enormous difference.

Please ignore the spikes at 50Hz and harmonics - they are just mains induced noise that was picked during the measurement (carried using the SpectraPlus demo version).

 

[Atmasphere]

How could I have possibly misinterpreted your comment?: "So I came to the realization that if the speaker required feedback of the amp, there was no way the speaker could ever sound like real music as a result. It might sound good for a stereo, just not real." We are both talking about speakers that you said if they require feedback of the amp, they are never going to sound like real music. We aren't talking about speakers "that might be capable of the task." So now you are saying that if a speaker is being powered by an amp that uses negative feedback "that the signal will be colored"?

mep, if the amplifier imposes a coloration (in this case, brightness and harshness) then if the speaker is worth its salt it will also reproduce that. So if there is such a coloration the speaker will never sound like real music no matter how good it is. This is simply a matter of understanding that a speaker can't sound like real music if the amp driving it doesn't either. It really is that simple. Go back and read my comments again and you should be able to see that this is what I was saying all along.

The bottom line here is that you are saying the majority of speakers on the market are flawed because they only sound their best with an amplifier that uses negative feedback and that all amplifiers that use negative feedback which also happens to be the majority of amplifiers on the market are also flawed. Good luck with that paradigm.

Thank-you. However I'm not sure that I would say the speakers are outright flawed, so much as I do maintain that the approach itself has a flaw (as I pointed out in the article)- that being the coloration of brightness and harshness, which is an inescapable artifact of the Voltage Paradigm. There is no question though that Voltage Paradigm is the huge majority- I pointed that out in the article itself. And there is no question that this works fine for the masses (although I contend in the article that one of the reasons audio in general is a dying market is the fact that in general it fails to inspire the populace with the possibility of music; this brightness coloration is one of the main reasons why) that are on a beer budget. But in high end audio the aspiration is to get the technology to sound like real music, cost is no objection. The article points out as a result a general trend that has been going on in audio for some decades- looking back into the past for something we missed. This is why SETs (SET stands for Single-Ended Triode amplifier, a Power Paradigm technology) are proliferating right now, why horns have re-emerged as a serious high fidelity option, and in general why tubes are still around half a century after being declared obsolete.

Sure, we aren't going to be the next Bose nor would I want to be. But if I can improve the enjoyment I get out of playing the stereo because it sounds more natural- count me in, and I have the great fortune of being able to do just that for a living.

Surely it is not - the speakers changed a lot along three decades. Just look for data I measured on a ten years old A1 - the version that was not OTL friendly . Happily later Roger West released an high impedance modification to please his customers using tubes! The current one's are different - but I have not been able to find the time to measure them with the Toroid II . One think is sure - it is more more full bodied and efficient.

The changes in panel capacitance due to several improvements carried along the years and the crossover modifications have produced wide changes in impedance. And as you said the position of the bass, medium and treble controls can make an enormous difference.

The newest version of the Sound Lab, shown in Las Vegas this last January, has changes that makes it even easier to drive. The interesting thing is it does not seem to matter what kind of amp is used- the speaker is simply easier to drive. There were some issues in the transformer crossover that had nothing to do with that operating principle that got corrected. A nice side benefit is that the speaker sounds even better as well.

 

[ack]

I am guessing that you got into this discussion from this thread only (this thread is a spinoff of another thread), and so have probably not seen this link, so here it is
https://www.google.com/search?client=ubuntu&channel=fs&q=fisher+A-80+amplifier&ie=utf-8&oe=utf-8
This is a Google search on an old Fisher A-80 amplifier. The first hit these days has a YouTube image of the damping control on an old Fisher amp. The control is marked at its extremes and also at noon: Constant Voltage, Constant Power (noon) and Constant Current. It is a dual-ganged pot that controls current and voltage feedback. At noon the two cancel and one or the other predominates at the extreme settings.

I have a number of questions:

1. If I understand all this discussion so far correctly, you can have a constant power amplifier if you use no negative feedback for optimal sound (as you claim yours to be) and the damping factor is roughly < 20, but apparently, per the quote above, if you also have both voltage and current feedback loops in one's design, at the point where one would "cancel" the other??? If so, why would one do this (use both types of loops), and what happens to the damping factor as you adjust those?
2. It feels like that by adjusting the damping factor you achieve constant power??? How else might that be achieved?
3. It all sounds like there is a theoretical limit with amplifiers, beyond which they cannot perform any better; how would you define that limit in your own terms, or terms that we users care about (e.g. output power, distortion, damping factor, rise times, slew rate)??? Basically, what's the optimal combination of all parameters involved?
4. If negative feedback is so bad, why do most designers use it in their amp designs? One answer might be to better control the bass, by lowering the output impedance. Another might be lowering THD, but if GE showed that humans can tolerate high THD (you mentioned something like 40%), why would a designer strive to lower THD?

Thanks

 

[amirm]

Thank-you. However I'm not sure that I would say the speakers are outright flawed, so much as I do maintain that the approach itself has a flaw (as I pointed out in the article)- that being the coloration of brightness and harshness, which is an inescapable artifact of the Voltage Paradigm.

What if that harshness/brightness is what the recording is really like? That the improvement that is heard is a modification of the frequency response, boosting other parts of the spectrum and lowering the others, and thereby acting as an ad-hoc EQ? How do we separate the two? Clearly if constant power amp changes the frequency response as it drives real speakers, then it is doing far more than just reducing some distortion. That distortion change is far lower in audibility than changes it makes to frequency response. No?

There is no question though that Voltage Paradigm is the huge majority- I pointed that out in the article itself. And there is no question that this works fine for the masses (although I contend in the article that one of the reasons audio in general is a dying market is the fact that in general it fails to inspire the populace with the possibility of music; this brightness coloration is one of the main reasons why) that are on a beer budget. But in high end audio the aspiration is to get the technology to sound like real music, cost is no objection.

Are you saying non-mass market amps are not "voltage paradigm?" I am pretty sure there are a ton of audiophile amps that work this way. I may actually own one or two .

The article points out as a result a general trend that has been going on in audio for some decades- looking back into the past for something we missed. This is why SETs (SET stands for Single-Ended Triode amplifier, a Power Paradigm technology) are proliferating right now, why horns have re-emerged as a serious high fidelity option, and in general why tubes are still around half a century after being declared obsolete.

I think as a society, we have decided what is old, is once again cool. It is also the case that every audiophile seems to wake up in the morning and decide to set up shop making speakers or amps. Whether they have any design expertise, or resources to properly measure and evaluate what they have built or not. So such proliferation to me is not evidence of some approach being right from technical point of view. I had a local designer bring some tube amps to us to sell. I was horrified how distorted and bad they sounded. They looked very pretty and such with glowing tubes but there was nothing euphoric about the distortions it produced. Even when they offered to leave them at our shop for free I refused to take them up on the offer. Stuff like this should not be in the market. Anyone seemingly can get a schematic for a tube amp, build a few with fancy components and call it the day. Bad idea all around.

 

[Atmasphere]

I have a number of questions:

If I understand all this discussion so far correctly, you can have a constant power amplifier if you use no negative feedback for optimal sound (as you claim yours to be) and the damping factor is roughly < 20, but apparently, per the quote above, if you also have both voltage and current feedback loops in one's design, at the point where one would "cancel" the other??? If so, why would one do this (use both types of loops), and what happens to the damping factor as you adjust those?

Voltage feedback decreases output impedance (using the Voltage Paradigm term) and current feedback increases it. When you have equal amounts of both they cancel out.

It feels like that by adjusting the damping factor you achieve constant power??? How else might that be achieved?

To the first part, maybe, to the second, by simply not using feedback and also having a moderate output impedance.

It all sounds like there is a theoretical limit with amplifiers, beyond which they cannot perform any better; how would you define that limit in your own terms, or terms that we users care about (e.g. output power, distortion, damping factor, rise times, slew rate)??? Basically, what's the optimal combination of all parameters involved?

The limits are we cannot achieve infinite power or infinite bandwidth and zero distortion is impossible. However we can get a practical amount of power, bandwidth can be had over 200KHz (which IMO is about what is needed if you want to reproduce 20KHz correctly) and distortion can be controlled by the amplifier topology itself. The problem is you can't use feedback without incurring the bad kind of distortion, odd ordered harmonics, so you don't really design for a particular damping factor, the amp will have what it has based on how much power, what tubes you use, etc. I like to keep the output impedance down as much as possible and I don't like output transformers, so I use power tubes that have a low plate resistance and low voltage operating points. The bottom line here is that there probably is not a simple answer to your question.

If negative feedback is so bad, why do most designers use it in their amp designs? One answer might be to better control the bass, by lowering the output impedance. Another might be lowering THD, but if GE showed that humans can tolerate high THD (you mentioned something like 40%), why would a designer strive to lower THD?


Thanks

Distortion can mask the presence of signals. It is also interpreted by the ear as tonality. So the less distortion, the more detail and more neutral presentation. So feedback is often used for that reason (although it introduces odd orders), plus it is used to correct the voltage response of the amp so its more 'impervious' to the load under the Voltage rules.

GE showed that humans have a high tolerance for even ordered harmonic distortion, not odd ordered. Keep that distinction in mind. Many transistor designs will not run at all without feedback to prevent them from locking up against a rail so for them it is often essential. A lot of tube designers like to use pentodes or tetrodes because they can get more power without a lot of tubes. But these tubes are less linear and would have obvious distortion problems if there were no feedback. So feedback gets used a lot but it must always be understood to be a compromise. The Power Paradigm is a compromise too, but it is one method of getting around the problem of coloration due to brightness caused by negative feedback. If you can get a Power Paradigm amplifier matched with a speaker that gets you flat frequency response, you are home free. IOW its a bit tricker, but can be well worth while.