high power tube amps vs ss

If I could make transistors do what I can do with ease with tubes I would make transistor amps.

But I can't and no-one else seems to be able to do so either, although a very few (less than 5) come close.

That thing that tubes do is they don't make odd ordered harmonic distortion the way that transistors do. Most solid state devices have a non-linear capacitive aspect that is inherent in the junctions of the device. This capacitance is multiplied by the amount of current through the device. This is one reason transistors make more odd ordered harmonic distortion than tubes.

The discussion about guitar amps does have relevance here. Guitar players prefer the smoother (lower odd ordered harmonic distortion) sound of tubes, especially when overdriving (clipping) the amplifier.

Because tubes (triodes in particular) are the most linear form of amplification known, it is possible and practical to built a tube amp with no negative feedback that has fairly low distortion. This is really hard to do with transistors, and it is notable that the best transistor amps made (in my book) are all zero-feedback designs. However, 99% or so of transistor amps use feedback, and feedback actually *increases* the odd-ordered harmonic distortion in any circuit.

Since the ear uses odd ordered harmonics to figure out how loud a sound is, by adding trace amounts to the signal the result will be brighter and harsher.

The bottom line is if you want it to sound real instead of just a really good hifi, you have to find a way to get rid of that negative feedback. Usually that means going to tubes, and then finding a speaker that works. BTW:

High efficiency speakers would make life a lot easier, but I thought those type of speakers sounded shrill and thin. It may be because I have only heard them with ss amps.

The last sentence here is the important bit.

My speakers are 98 db 1 watt/1 meter, and go to 20Hz. I can shake the building with a 60 watt amp, without concern of clipping the amp. Most 'high power' systems I hear cannot play with the volume, the impact or the clarity I hear at home. Even at +100 db sound pressures, it still sounds smooth and relaxed. IME most audiophiles won't listen at levels over 90db because their system is too harsh. Real music plays at +100 db levels, if you can't do that at home without discomfort, something is wrong with the system.

I'm a big fan of Nelson Pass, Ayre and Ridley Audio. That's my short list... I think there are a few other candidates out there but I've not been able to hear them.

I do not consider Dynaudio speakers and tube amps to be a natural match by design in general, and would not do that without hearing first but that's not to say you might not find something to like there if you heard it.

Agreed- in spades!

To that end I would like very much for either/both to cite some tube amps that are as fast as any ss amps they've heard so far. Imo, you can't get there without serious speed in an amp.

One way to measure speed is rise time, the voltage you can reach in a given amount of time (one microsecond). This is a test that is done with square waves as an input. How does 600V/usec sound? That is faster than most transistor amplifiers, but in this case we are talking about a tube amp.

With rise time there is also usually bandwidth as well. We get full power to about 200KHz and only down about 2 db at 300KHz- this with no feedback. The bandwidth is cut off above that- the output section has bandwidth to about 60MHz or so... Once I used modified version of the amp as a booster amplifier for a CB radio transmitter, years ago, just to prove that we could do it.

So yes, they are fast.

I have never tried tube amps and I am fearful that I would miss some SLAM and Impact of the presentation. Does your system belt it out pretty loud and non fatiguing?

When I put transistor amps in my system I find myself turning up the volume in order to try to get some slam out of the amp. They just don't have it compared to the (usually) much lower powered (tube) amps that I normally run. But keep in mind these amps do have full power right down to 2 Hz.

Its funny, I've heard for years that OTLs are supposed to be wimpy amps, but in practice I find that to be more true of transistor amps, by quite a stretch, at least as far as bass is concerned. That was part of the reason I began investigating what the implications of the amp/speaker interface was really all about. I'm old enough that I remember older audio products from the 50s and 60s, and I have collected texts from the old days too. I noticed that these days things are not done the way they did them in those older textbooks, but it was not because the books were so old the math was wrong. It was because of that Paradigm thing I've been harping on. Quite literally the Voltage Paradigm has its own set of charged terms- 'output impedance' is a good example, meaning something entirely different from what it means in most fields of electronic endeavor. The result is a whole lot of confusion, but we audiophiles are quite used to it (because this has been going on for decades) and so our solution is to take it home and listen to it, regardless of what preferences we might have.

That, IMO, is not very efficient, but until the industry talks openly about these paradigms or models, whatever you want to call them, that's the way its going to be.

But Ralph, I think you may have mentioned yourself that if a "pure" Power Paradigm tube amp that doesn't use negative feedback and that has a somewhat high output impedance is matched up with a speaker having a "bumpy" impedance curve, there could be coloration of the acoustic presentation. Ergo, the need and importance of matching this type of amp with a speaker having a relatively benign impedance curve. The problem, I think, is that those types of speakers are in the minority of what's out there in the market. As I may have mentioned, some of the "big name" speakers like Magico and the Revel Ultima Salon 2 have impedance curves that would make a "pure" Power Paradigm amp dizzy. Or am I missing something . . . again as usual??

The ear has a tipping point where it will favor colorations due to distortion **over** actual frequency response errors. On top of that you can't count on flat frequency response from any speaker. The Power Paradigm favors the idea that the equipment be designed to obey human hearing perceptual rules (as opposed to arbitrary specs on paper- the Voltage Paradigm); with this in mind, it can be seen that it might be more important to keep distortion down over trying for a perfectly flat frequency response.

Of course, if the frequency response is really messed up that won't do either, so you are correct that some care must be taken. In general though, if the speaker is a Power Paradigm device as is the amp, there are no worries. The problem comes in when you try to mix the two technologies- in that case you can always count on a tonal aberration.

Just to be clear- its not important that the speaker have a flat impedance curve- it just has to be designed to work with the power response of the amp. If I can quote Duke Lejeurne of Audiokinesis. Duke starts out his comments by quoting another poster on that thread:

Drew Eckert wrote:

"Some audiophile amplifiers have silly high output impedances which interact with the speaker's varying impedance to change the frequency response and this is exacerbated with low load impedances. Output Transformer Less Tube amps are especially bad although single ended triodes without global feedback can also have problems.

"For instance an Atmasphere M-60 Mk.II.2 has a 4.1 Ohm output impedance.

"Driving a 3-way speaker with impedance varying from 16 to 64 Ohms this would cause a 1.4dB output difference between the minimum and maximum impedances.

"With 4 to 16 Ohm impedance the difference would be 4dB. This is not atypical for a 3-way...."

In reply, I'd like to point out that, with the same 4 to 16 ohm impedance difference described here, the power that a transistor amp puts out varies by 6 dB, because it is putting out constant voltage rather than constant wattage. Why does the audio world accept this without a blink, and yet think there's a problem when a tube amp exhibits less variance in power output into the same load??

It is because the audio world is accustomed to the way transistor amps behave, and most speakers are designed to work well with transistor amps. The designer designs the speaker to sound right when driven by an amplifier that puts out 1 watt into the 8-ohm portion of its curve, 2 watts into the 4-ohm portion of its curve, and 1/2 watt into the 16-ohm portion of its curve, all at the same time (2.83 volts).

Now, what if the designer's goal was a speaker that works great with an amplifier that puts out approximately constant wattage, regardless of the impedance curve (within reason)? Well that can be done just as easily, but there are fewer amplifiers designed that way out there, so his potential market is smaller.

These two approaches to amplifier and speaker design have a name: Voltage paradigm, and power paradigm. You can read more about the subject here:

www.atma-sphere

At this point Duke put a link to the Power Paradigm article. Continuing:

Okay, what kind of amplifier sounds best? Boy that's a long debate for another day, but a lot of people familiar with many types of amps prefer the sound of a good OTL or SET amp, assuming a good speaker pairing, and that includes yours truly. Some speaker manufacturers give priority to building speakers that will work best with power paradigm amps because they believe that combination sounds best, and just accept that they are fishing in a smaller pond.

Is it possible to build a speaker that works well with both types of amps?

Yes, by keeping the impedance curve as smooth as possible, the speaker will work well with both types of amps. And with such a speaker, you can really make an apples-to-apples comparison of the different amplifier types, rather than actually evaluating whether a (typically roller-coaster impedance) speaker synergizes best with a voltage paradigm amp or a power paradigm amp.

Drew's numbers above illustrate an argument in favor of using a high impedance speaker with a low-output-impedance amp: The amp's output is approximately constant-power when the speaker's impedance is varying between 16 and 64 ohms, but the amp's power output changes significantly when the speaker's impedance varies between 4 and 16 ohms (same 4-to-1 variance in both cases). Also note, in both cases the amp's power output change vs speaker impedance is in the opposite direction of what happens with a voltage-paradigm (solid state) amp, and this is probably the main reason why simply dropping a specialty tube amp into your current system is a roll of the dice (and the odds are against you) unless you already know your current speaker is a good match.

Back to the original question, what's with 4 ohm speakers if they're harder to drive, well in general they can play louder with a solid state amp, and most people have solid state amps, so they get more sound per dollar with 4 ohm speakers (quantity outsells quality). Also, most woofers are 8 ohms, so if the designer wants to use two such woofers, he has to choose between series connection (16 ohms) and parallel connection (4 ohms). Most choose parallel connection because 4 ohm speakers outsell 16 ohm speakers. I believe that most amps - tube or solid state - sound better into a 16 ohm load, so my home audio two-woofer designs are 16 ohm loads (whereas my prosound two-woofer designs are 4 ohm loads, because there we're trying to maximize available SPL with solid state amps). What do I do about the typical halving of maxium power from solid state amps when driving a 16 ohm load? I start out with speakers that are about 3 dB more efficient, and of course pay a corresponding price in box size vs bass extension. However there is one "free lunch" to using power paradign amps: They generate equal or even increased, rather than reduced, power into the speaker's virtually inevitable bass impedance peaks, and if we keep this in mind and design our box accordingly, we get back most of the bass extension we otherwise would have lost when we traded off in the direction of higher efficiency.

So I take the position that, all else being equal, 16 ohm speakers sound better than 4 or 8 ohm speakers, and that power-paradigm amps are well worth seeking out matching speakers for.

Duke
dealer/manufacturer/power paradigm groupie

this is taken from the speakers forum: http://forum.audiogon.com/cgi-bin/fr.pl?cspkr&1368259003&openflup&13&4#13

Regardless of design paradigm, isn't it the frequency response and distortion measured in the end that matter? These are components designed with certain criteria in mind that are obviously not interchangeable and must be matched together somehow. Its important to be aware of the technical details that matter, like impedance characteristics, to have the best chance of getting best results, but in the end, I do not think either paradigm can be measured as definitively better, although I suspect that the way these things are usually determined, via certain accepted distortion measurements, etc., that the common voltage paradigm measures better when done correctly.

Mapman, Put in a nutshell, to answer the first question above, 'no' is the short answer. The longer answer is that the ear cares about certain distortions and others not so much. In addition, the ear will interpret (as I have mentioned previously) some distortions as tonality, and will favor them over actual frequency response errors or accuracy.

An excellent example is how some amps can sound bright, but measure the frequency response and they are perfectly flat. This is because trace elements of odd ordered harmonics are interpreted by the ear as brightness even though it does not show on the instruments.

Another way of looking at this is that the Voltage Paradigm for the most part ignores human hearing rules, opting instead for arbitrary figures on paper. In essence, an example of the Emperor's New Clothes. I do not think that this was done on purpose, its just how things have worked out in the history of the last 45-50 years or so.

You have to understand that back in those days, there was very little that was understood about how the ear actually perceives sound. So the Voltage model was set up around low distortion and flat frequency response.

In the interim, we have learned a lot about human hearing, but one thing I find amusing is that one of the earlier facts we discovered was that the ear uses odd ordered harmonics as a gauge of sound pressure. That was known by the mid-1960s. Yet the industry ignored it.

The Power model rests on the idea that if we build the equipment to obey human hearing rules, the result will sound more like music. Well, if we are to obey one of the most fundamental hearing rules, we have to get rid of negative feedback, otherwise the result will always sound brighter and harsher than What Is Real.

The evidence that this is not a topic of debate is all around us- probably the easiest to understand is that, over half a century after being declared obsolete, tubes are still with us (and we are still having these conversations). If the Voltage model was really the solution, it would have eclipsed all prior art and would be the only game in town. That it failed at that speaks volumes.

Now in saying this I am not trying to make you or anyone else wrong. I would love the Voltage model to actually work, but IMO there are only a few examples that do; they represent a tiny minority, much smaller than the pond I am fishing from, to use Duke's apt expression.

Its more a matter of what collectively we choose to ignore, things like the fact that the stereo can sound loud or shouty at times. Its been my experience that if the system is really working, you will never have any sense of volume from it, that is to say its relaxed even with 110db peaks. If you cringe at the thought of running the volume that high then you know exactly what I mean. Yet real music hits peaks like that all the time and we don't cringe.

The huge unflatness of the ear sensitivity chart would also seem to debunk any claims one might make about being able to hear flat frequency response. If you hear it as being flat, it in fact cannot be. Significant equalization would have to be applied to the source to have any chance. At that point, what you hear as flat would no longer be natural, rather "enhanced" to make it sound that way to compensate for lack of flat response with our hearing.

Its more complex than that, our ear/brain system can recognize acoustic environments and compensate for them... BTW I hope you are not suggesting that we need to compensate our ears with EQ.

I have trouble understanding how the ear hears something as "bright" that does not evidence itself somehow when measured.

I've always taken that as some resulting frequency anomoly in one of those frequency ranges where the ear is most sensitive, but how serious can it be if not even measurable? Where is the evidence that the effect exists, much less the cause?

If we can't measure it can it exist? Sure! Our instruments have limits of their own- noise being an excellent example (another being the tendency to quantify a phenomena as a reading on a meter...). When an amplifier has low harmonic distortion measurements, its often described as having such low distortion that its "buried in the noise of the instruments".

The simple fact is that in regards to sensitivity to odd ordered harmonics, our ears are **more** sensitive than instruments. This is not hard to understand if you also know that the ear is that sensitive because it uses odd orders to gauge sound pressure- look at it as a survival trait. If you can't tell how loud a tiger is growling, you may well soon be dead. The ear needs to be pretty sensitive as a result. There are other things that the ear sucks at compared to instruments; this simply isn't one of them :)

General Electric did the studies of this phenomena back in the 1960s. It was perhaps one of the first real forays into the hows and whys of human hearing perceptual rules. We have learned a lot more since then.

Since I have set up my system to be absent of loudness cues, it frees me to set the volume to the level that seems appropriate for what I am listening to. A solo guitar being quieter than a full orchestra for example.

For example, do the harmonic loudness artifacts you describe commonly fall into the frequency ranges depicted that ears are most sensitive to?

The answer to this question is usually 'yes'. The reason is that the amplifier has to work harder to reproduce the lower frequencies. Their harmonics will often show up in the 'birdsong' region.

I can provide a simple test that shows how important this odd ordered thing really is. It can be done with simple equipment and is quantifiable. I've offered it before. If anyone here is interested, just say so.