Sorry I got to this late.
It seems there are some things that need clarifying. If I am not beating a dead horse too much...
Bifwynne, by adding negative feedback to an amplifier you do indeed move the amp towards the Voltage Paradigm. This is because as you surmised the 'output impedance' is lowered. I paraphrased the term because it is misunderstood however, and that is where the clarification comes in. Unsound, you might want to pay attention to this.
Under the Voltage Paradigm, you have the term 'output impedance'. The term has a definition which is not intuitive. It refers to the amount of servo gain that the amplifier has which allows it to react to a load. It does not refer to the actual output impedance of the amp, as measured by any other field of endeavor in the world of electronics.
How can we know this? The answer is simple. If the output impedance were indeed lowered, the amp could drive a progressively lower and lower impedance. It might even make more power. But we see by adding feedback to an amplifier that the output power into lower impedances does not change.
IOW, what is happening is that the feedback gives the amp the ability to adapt to its load within certain limits by taking samples of its performance and using that as an error correction. The only way you can really get a lower output impedance is with bigger output transformers, more tubes or more transistors. The extra ability to drive a lower impedance does not come out of thin air or feedback- to do so would violate a fundamental rule of electronics known as Kirchoff's Law.
The difference between the Voltage and Power Paradigms has more to do with feedback then tubes/transistors.
With regards to Current Source, Power Source and Voltage Source:
If the amp has a lot of voltage feedback it could become a Voltage Source. Most Transistor amps attempt to be Voltage Sources (and most of them run a lot of feedback). What this means is given a sine wave input signal, and a load of variable impedance (like a typical box speaker), the output voltage will always be the same. Note that with this model (the Voltage Paradigm) the *current* will vary as the load impedance varies. This is how such an amp can double power as impedance is cut in half, as with half the impedance the current will be double.
Under the Power Paradigm the amplifier will attempt to make constant power into any load rather than constant voltage (there is a little math of note here, Power = Voltage X Current). IOW a Power Source. This is the behavior of a tube amp without voltage feedback. Depending on the load variation, both output voltage and output current will vary.
Under the Power Paradigm, the output impedance of the amplifier is in fact the actual impedance of the output section of the amp.
An amplifier can also be a Current Source, although there has never been a paradigm of design, test and measurement that has evolved around it. This type of operation is done by using current feedback rather than voltage feedback. Constant Current amps tend to have a very high output impedance by any measure, often well above 50 ohms. This does not mean that they cannot work with conventional speakers. Nelson Pass as done a lot of work in this area in recent years. |
People like Atmasphere & Berning do that. And, back in the old days there was Harvey Rosenthal (?) who came up with the 1st zOTL amp (I think I'm remembering this correctly??) No. Harvey Rosenberg had a contract for a while with David Berning (which never went anywhere), at or near that time he coined the 'ZOTL' term, which refers to David's amplifiers (which employ an unusual output transformer, one lacking the usual limitations of normal output transformers; IOW not an OTL, and a brilliant design regardless). The output transformer is put in place between the tube power output stage & the speaker input to buffer the tube amp from the wild swings in the impedance & phase of the speaker.
On the primary side of the output transformer, the tube power output stage sees a constant impedance. By working into a constant impedance, there is optimum power transfer from the tube output stage into the output transformer primary windings load impedance. So, the waters (if you may) are calm/serene.
On the secondary windings side of the power output transformer, the waters are rather choppy due to the speaker impedance & phase variations vs. freq. Several output taps are provided to match the speaker impedance such that there is more optimum power transfer between the secondary windings & the speaker input. This set of comments are incorrect. The transformer does not buffer impedance or phase at all. It *transforms* it (hence the term). So whatever swings of impedance seen in the load are translated to much higher impedances which are what the tubes see. If the impedance of the load is too low, the tubes will make distortion and less power; if too high the transformer will ring (distort). That is why taps are provided. This is also why a lot of designers see the need for negative feedback, to tame the distortion of mismatched loads on the tubes due to impedance variation in the load, as well as the distortion of the transformer itself. You can have impedance variations in the load and have it work perfectly fine with an amplifier that has no feedback; that is to say that feedback is not required for a neutral presentation on a speaker that has a variable impedance curve. It turns out that the ear has a tipping point where it will favor tonality due to distortion over actual frequency response variation (and FWIW, just look at the charts of any speaker- **no-one** in the world has actual flat frequency response from any speaker; this is why I see the Voltage Paradigm as an entirely failed concept, not the least of which it ignores human hearing rules). IOW its often far more important to have low distortion in many cases than perfect voltage response. As Bifwynn found out, it is the interface between the amp and speaker that is far more important than the cost of either the amp or the speaker. When you understand the Voltage and Power Paradigm concepts essentially you take a lot of the guesswork out of matching amps and speakers. Imagine the money saved if the industry actually talked openly about this! |
Why are there speakers designed with widely fluctuating impedances and steep phase angles in the first place? A woofer in a box will have a resonance that is expressed as a peak in impedance. If there is no accommodation for this fact either in the amp or the speaker, the result will be too much energy at the impedance peak. In the Voltage Paradigm the amplifier power is throttled back. This effectively insures flat frequency response. In the Power Paradigm the box design puts the peak at a lower frequency to take advantage of the extra energy- but again netting fairly flat frequency response, but wiht the additional benefit of bass extension, which might well be up to half an octave. Some speakers don't have impedance curves based on box resonance. With such speakers, the Voltage model falls apart. A good example is an ESL, whose impedance curve is based on a capacitor. It really works a lot better if the amp makes constant power rather than constant voltage. That way the amp can make power in the bass regions where the impedance is high. This is why transistor amps tend to be bright and bass shy on ESLs- and typically why owners of ESLs and transistors put the speaker too close to the rear wall, to get bass reinforcement. They are not realizing the full potential of the speaker. Anytime a Voltage Paradigm product is used with a Power Paradigm product, a tonal aberration will occur. We audiophiles call that 'equipment matching'. One could also ask why amp designers choose not to build amps that can deal with such loads? (This question is posed in the context of 'difficult' loads with low impedance or high phase angles) The answer here is quite simply, such amps that can drive such loads are usually incapable of sounding like real music, as they have design features that violate human hearing/perceptual rules. One common example is the application of negative feedback, which, due to propagation delays in the amplifier circuit, causes ringing distortion, typically odd ordered harmonics (5th, 7th and 9th) which are used by the ear to sort out how loud a sound is. This is a pretty fundamental hearing rule. Amps that violate it have the coloration of brightness and also come off as harsh. If you want to get away from that you have to figure out how to make a low distortion amp that uses no feedback. As stated earlier, distortion often take precedence over actual frequency response errors by the human ear, IOW its more important to have low distortion than it is to have perfectly flat frequency response (which can't be counted on in the best of circumstances anyway). |
The box resonance peak is where it is. How does the Power Paradigm amplifier move that peak to a lower frequency?
If the resistance in the bass region goes up (due to a box resonance) then the current into that higher resistance goes down. the voltage must go up to keep constant power. Where does the bass extension come from? This is a matter of design. The peak exists as a relationship of the driver and the box. Change the relationship, change the peak. The bass extension comes from designing the box so the peak occurs at or below the cutoff of the driver, in such a way that the peak maintains the bandwidth. Not hard to do if you understand speaker design. I believe that a traditional SS amp (one that was not designed for an ESL in mind) will likely sound mediocre 'cuz of the high capacitance load that will tend to make that amp oscillate & eventually fail. OTOH, if a SS amp is designed with an ESL in mind (& a few names come to mind that are being used successfully with ESLs & planars) then these SS amps will be effective. Belief and reality are usually two different things. In this case I invite you to do the math. How much power will an ideal 400 watt SS amp make into 30 ohms and into 3 ohms? If you answer 'about 100 watts and about 900 watts' then you have some grasp of the problem. to be fair, when the SPL get's really high, my ears hurt. So what's the point of that exercise?? ;>') If that happens with less than 105 db, then its an indication that the system has some sort of odd ordered harmonic generation associated with its operation. A sound pressure level meter might be handy to sort out the actual volume level. |
I wonder what happens if one listens to nicely produced music, say a big band with lots of brass and high frequency energy live at 105DB? Is it as comfortable as listening to the no NF tube amp at the same level? Its a fair question, I think. It is. If you sit close up in a concert hall it will easily be that loud. So? The amps I'm thinking of are capable of generating these kinds of wattages & sounding musical at the same time. Indeed that's what I was thinking in the 1st place when I wrote that post..... Its rather obvious from those numbers that the amp is going to be challenged at the least, challenged in the regard that it is likely to make too much power at high frequencies and not able to make enough power at low frequencies. And in practice, that is what you hear with them as well. The feedback is supposed to help the amp with this, but because the model is incorrect (IOW the speaker's impedance curve has nothing to do with box resonance) it does not work. Often as the speakers impedance changes so does the the sensitivity, and that should correspond to flat frequency response from such a ss amp. This is true for some box speakers- those that are built with this expectation. But as we saw in the ESL example, the model falls apart. When the Voltage Paradigm was being developed (in the late 1950s and early 60s EV and Macintosh led the way on this) ESLs were not a significant part of the market (and they still aren't). So there was not and still is not a concern to make them work right. Folks, if you've not picked up on this yet, the Voltage Paradigm is all about marketing and the almighty dollar. The ability to double power as you cut impedance in half has little to do with sound quality. It has a lot to do with looking good on paper. And a lot of audiophiles buy with their eyes rather than their ears. OTOH the Power Paradigm is based on the rules of human hearing, which makes it a lot trickier to execute. The Industry wants something convenient, so you can just plug things in (easier to sell). But our ears are more complex than that. Its the conflict of these two approaches in audio that generates the equipment matching conversations, the tubes vs transistors conversations, and the objectivist/subjectivist conversations. Its really aspects of the same issue. If the Industry was open about talking about this, audiophiles could save a pile of cash and have better sound at the same time. But very little in audio occurs for the sake of better sound. Mostly its about cashing in. |
Bombaywalla, actually that example is excellent. I have yet to hear a transistor amp that can play bass on a full range ESL, assuming the ESL is set up properly (at least 5 feet into the room). Horns are of course another example with the only exception I know of being the Avantgarde Trio, which is (or was) designed for transistors. One of the reasons horns got a reputation for being 'honky' is the fact that a crossover designed for an amplifier with a high output impedance will not work right with an amp that has a low output impedance. As a result the horn is trying to play material that is out of its passband. Full range single drivers are yet a third example. You will find that the users of such speakers are usually using a low power tube amp, usually one that has no feedback. A 4th example is the very first acoustic suspension loudspeaker ever made, the original Acoustic Research AR-1. This speaker was designed for an amplifier with a 7-ohm output impedance, and would not play bass right with most solid state amps as a result. With such amps a simple solution was to put a resister in series with the output of the amp, simulating a higher output impedance. I can also refer you to this: http://paulspeltz.com/tomcik/index.htmlThis article shows that just simply having a high damping factor does not always do the trick. Some speakers want a 10:1 damping factor, and others want 0.1:1 (and yes you read that right- such a speaker might be designed for open baffle operation). I did not make this stuff up BTW, it was well-known back in the 1950s when the industry was trying to sort out what to do about it. Tomcik, in his article above, proposed a solution that was also used by Fisher. Here is a google search on the Fisher A-80 amplifier: https://www.google.com/search?client=ubuntu&channel=fs&q=fisher+A-80+amplifier&ie=utf-8&oe=utf-8Take a look at the first hit. Its a YouTube image of a Fisher A-80 tube amp from the 1950s. Take a look at how the damping knob is labeled: "Constant Voltage" at fully counterclockwise, "Constant Power" at 12:00 and "Constant Current" at fully clockwise, where the amp is using current feedback. |
Powered speakers are like an albatross around your neck. If you want more power, you can't. You have to sell the amps *and* the speakers to upgrade. I have to agree with Bombaywalla. Some manufacturers of ESL's specifically made/make ss amps for use with their speakers, Acustat and Sanders come to mind. Many ESL manufactures use ss amps when demonstrating their speakers. I remember a particularly fine demonstration of Martin Logan's with Threshold amps. 'Many ESL manufacturers'. A bit cumbersome to be a real oxymoron, but nevertheless it is an oxymoronic phrase. And a bit of a red herring, there are not 'many' ESL manufacturers :)... There is more going on here than meets the eye. ML and Sanders (the Acoustat powered speaker used a tube amp) are both trying to make their speakers work better with transistors by reducing the impedance of their speakers (unfortunately, Quad has been treading this path in recent years too). To this end you encounter some very low impedances (0.5 ohms to 4 ohms) with them. However if you can get around the impedance problem (with a set of ZEROs or the old Atma-Sphere Z Music autoformer) what you find is that the tubes sound better even on those speakers. Sure, such and such a speaker might have sounded 'fine' with a transistor amp, did you then compare it with a tube amp in such a way that that the tube amp was not having a problem with the load? Just because a speaker sounded 'fine' is not the same as it sounding its best. The old Quad 57s and 63s, as well as most Sound Labs and some Acoustats, have impedance curves that vary from about 2 to 30 ohms or more. Usually if someone with such a speaker and a transistor amp claims that it makes bass, its because the speaker is too close to the wall. They are not winnowing out the performance of the speaker in that fashion. Paul Speltz, who make the ZEROs, has a letter from Steve McCormick, who makes (or made) the McCormick amplifiers like the DNA1. In the letter, Steve states that while the amp can easily double power into 4 ohms, it actually sounds better driving the 4 ohm load through the ZEROs (meaning that it is seeing 16 ohms). If you interview solid state amp manufacturers, you will find that they will pretty much agree on this point- just because they can double power into 4 ohms does not mean that they are sounding their best, and you can see that in the specs on any solid state amp- the distortion is higher as the load impedance is decreased. Essentially, if the goal is the audio *quality*, there is no argument for four ohms or less. If sound *pressure* is the goal, then 4 ohms has a weak argument. Looked at another way, a simple means of making your speaker sound smoother and more detailed is to increase its impedance, not because the speaker will work any differently, but because the amplifier will. |
Al, that is why I came up with the 'Paradigm' part of it. In investigating this topic years ago what I found was just as you pose in your last post, that the terms are slung loosely and freely, resulting in a lot of confusion.
On top of that, I used the word 'paradigm' since a paradigm is a platform of thought, outside of which exists only blasphemy. This is why you see such staunch argument on this topic!
I also saw failure. I say this because whenever a new technology comes along, if it is really better it will supplant the prior art. In the case of audio, the Voltage Paradigm and all that it entails failed to supplant the Power Paradigm which was the only game in town before 1956 or so. The proof of that? Tubes are still in production 60 years on, audiophiles still can't tell how an amp sounds by looking at the spec sheet. The tubes/transistor debate and the objectivist/subjectivist debate have raged on the internet as long as the internet has existed.
If you look at internal combustion, the overhead valve supplanted the side valve in the early 1950s with no-one looking back except collectors. That should have happened in audio too, but it didn't, for the simple reason that the Voltage Paradigm was not in fact better. It was just different. |
Historically speakers with lower impedances are more likely to be able to produce wave form fidelity than speakers with higher impedances. [facepalm]I can't let that one pass[/facepalm]. Its well known for decades that the Quad ESLs have had very low distortion (re.: 'wave form fidelity'[sic]), some of the lowest out there. Yet the ESL57 and 63s both have a fairly high impedance. In fact impedance has nothing whatsoever to do with waveform fidelity in a loudspeaker. That is entirely a matter of design. There are many good reasons for speakers to have a low impedances and the market place has proven this for a long time now. The 'good reasons' are sound *pressure*, not sound *quality*. I recommend you reread my previous comments. Its one thing if I say it- I own a small company that makes tube amps. But its another thing entirely when transistor amp manufacturers say that, and especially when that is re-enforced by the distortion measurements. We are talking about distortion that is audible too. Go ask Paul Speltz to show you that letter from Steve McCormick. |
I just do not buy into the assertion that power paradigm and no negative feedback is the only way to make music sound real. That's fine but it will not change the reality: The problem here is that the human ear is more sensitive to odd ordered harmonics than almost anything else. Add on top of that that the ear is most sensitive at higher frequencies (FWIW the human ear is tuned to be more sensitive to bird song frequencies, something that comes from our forebears as a survival trait); the result is we can hear odd ordered harmonic content that is hard to measure on the bench. We constantly hear how the ear is insensitive to this or that (for example we cannot detect the phase of a sine wave at all) but this is certainly an exception, and for a very good reason: the ear uses odd ordered harmonics to figure out how loud a sound is. This is vital to our survival- if you can't figure out how loud certain sounds are, you might soon be dead! The thing about negative feedback is it does two things- one thing we like, the ability to servo-control the output of the amp so it will produce flat frequency response on *certain* speakers. The other thing it does is reduce distortion overall while actually adding odd ordered harmonic distortion. When you add odd ordered harmonics even in trace amounts, it is audible because of what I have already explained. The result is it won't/can't sound real. You don't violate a fundamental human hearing perceptual rule without a price! Get rid of the feedback get rid of this problem! But now you have to sort out how to get flat frequency response and low distortion without feedback. There are ways to reduce distortion, but how do you get the flat frequency response? Use Power Paradigm design rules. Many speaker designers do this, whether consciously or not, depending a great deal on the sort of amplifier that they like to listen to. The Voltage Paradigm will work fine if you can build an amplifier that is simultaneously free of odd ordered harmonics and can also behave as a voltage source. That is the leading edge of the envelope in amplifier technology; so far no-one has been able to do it. There are some very notable amplifiers IMO that point the way- Ayre, Pass and I am also a fan of Berning, although the latter might be considered more of a hybrid approach. I suspect class D has something to offer here as well. |
Plus Ralph's paradigm (based around avoidance of use of NF as the best means to avoid those nasty odd order harmonics + the rest seems to stem from that), FBOFW, is more unique and "radical" and a huge minority relative to the norm these days, so I think he has more work to do to gain mind share overall accordingly. A lot of people think that this Power Paradigm thing is something of my doing. Its not. The Power Paradigm is what existed as the means of design test and measurement in the old days, pretty much everything before about 1958 or so, which EV and Mac lead the charge to move to a Voltage model. For evidence of that I direct you to Google: https://www.google.com/search?client=ubuntu&channel=fs&q=fisher+A-80+amplifier&ie=utf-8&oe=utf-8The Search is for 'Fisher A-80 amplifier'. The first hit has a YouTube image, a photo of the Damping (feedback) control on the amp. If you look at it closely, you can see the range of control goes from 'Constant Voltage' to 'Constant Power' and finally 'Constant Current'. IOW I didn't make this stuff up. FWIW it appears that there never was a current drive paradigm, although Nelson Pass has done a lot of experimentation in that area. I also found that the means of measuring output impedance has changed- the Radiotron Designer's Handbook, which dates from the early 1950s shows a very different means of doing so, with different math, than the techniques typically used today. It did take me a while to sort out why, and the why is that as an industry we went from using power measurements to voltage measurements. You can see this in the recording studio; older studio equipment is based on the power model, for example older microphones are often set up to drive 150 ohms, while modern mics are set up quite a bit differently! Another way of looking at this is output impedance- many SET designers are accustom to building amps with higher output impedances and choose speakers accordingly. This is that equipment matching conversation that all audiophiles hear about sooner or later. All I did, FWIW, was to gather these tidbits from all over and put them in the same place at the same time, and gave them a name so it would be easier to have a conversation about them. But many people have confused this with it being something I came up with, which is obviously not the case, if you look at that YouTube image. The fact that negative feedback does what it does is also not a matter of debate. It is knowledge that has been documented since the 1950s- I refer you to Norman Crowhurst, who outlined many design parameters of audio equipment in his many writings. (If we were able to build amplifiers with absolutely zero propagation delay as the signal propagates through the amp, then the feedback would work correctly with no ill effect. FWIW, most amplifier design theory that involves feedback actually assumes that this is the case. Unfortunately the real world does have have such examples, and being pragmatic, I figure since that is the case maybe we should look elsewhere for fidelity, since the feedback model clearly fails.) I persist in this is only because I want to forward the art. I don't see any amplifier technology as having the answers, all amplifiers have weaknesses and tradeoffs. I am simply pointing to What Is; what you want to do with the resulting knowledge is another matter. Now Mapman tries to make it look like I might be tilting at windmills; the evidence that I am not is very simple: If the Voltage Paradigm really worked, there would **be no vacuum tube industry**. Its really that simple. But here we are, over half a century after tubes were declared obsolete, and tubes are still being made (they are the easiest way to create an amplifier on the Power model). FWIW normally how it works is when a superior technology replaces the prior art, the prior art dies out. How many of you still use a rotary dial on your phone? |
Mapman, thanks for your comments. I think there probably is some sort of point where the odd ordered distortion is not detectable. The problem is, we can hear that which is very difficult to measure! So how do we tell?? Test instruments don't seem to be able to help us. Odd order harmonics from NF might be quite nasty relative to others as suggested, but that does not mean there is no way to get the negative effects into a range that is not significant, at least for most. No amp is perfect, I think we would agree on that? We do indeed. I think it can be cured too... BTW you might want to read Nelson Pass's article on distortion: https://passlabs.com/articles/audio-distortion-and-feedbackI think this article is a wonderful tutorial in what the amplifier designer is up against. Most of us here have been listening for decades. I've been into audio since the 7th grade (1970), and playing in bands and orchestras since the 6th grade (string bass). I run a recording studio, have three solo albums and play keyboards with a band on two more albums. I've cut LPs for commercial release on our own LP mastering lathe. Regarding the lathe: we cut a lacquer when testing the lathe that used the same cut repeated 4 times on one side, each cut with its own lockout groove. The way the cuts differed was by the kind of amp that was used for each cut. The first was using the stock transistor amps for the cutter head, the rest were using three different tube amps including our own. What was nice about this lacquer is we could play the differences between the amps for anyone, on any system (FWIW the solid state amp being obviously brighter and less detailed than the other amps...). Several people have suggested putting this out as an audiophile demo LP. I'm not so sure about that- might be a small market... |
