That "tube sound" and power ratings

Further, the topology of the circuit says a lot about what sort of distortion it will make. For example, a push-pull amplifier with a single-ended input will make primarily 2nd ordered harmonic distortion, but will also have some of the 5th. If the amplifier is fully differential and balanced, the primary distortion product will be the 3rd harmonic.

If negative feedback is applied whether tube or solid state trace amounts of the 5th, 7th, 9th and beyond will be added.

At any rate loudspeaker efficiency is important when dealing with tubes, as tube power is more expensive than transistor power! Its important with any system to avoid distortion but in fact distortion is the name of the game as the human ear translates distortion into tonality- this is why transistor amps tend to sound bright, even though the distortion is very low, what distortion is present tends to be the odd orders, to which the human ear is particularly sensitive.

A lot of push-pull amplifiers will actually make more distortion below a certain power level, so for those amplifiers its helpful to have a speaker that is a little less efficient, so the distortion is happening at a lower level and is hopefully less noticeable. But this is why some amps seem to lack detail at lower power levels.

There is definitely more to this than meets the eye!

We all know there are plenty of 8 ohm speakers. 16 ohms is a little harder to find. Some examples:

Classic Audio Loudspeakers
Audiokinesis (several models)
High Emotion Audio (Bella Twin- one of the best monitors I've heard)
Quad ESL57 and ESL 63
Lowther, PHY and other full-range drivers
Coincident Technology (not always 16 ohms, but definitely higher impedance)
Rogers LS35A
many vintage horns- Altec, JBL, EV, etc.
ZU Audio has made 32 ohm versions of some of their speakers
Some Pipedreams are set up as higher impedance

Its not important in a 16 ohm load that the impedance curve be ruler flat. In fact that is true of 8-ohm speakers as well. What does seem to be important is that the impedance variation does not fall below about 1/2 the nominal impedance, and when so doing is at a crossover point. ESLs seem to be an exception.

An example of a speaker that is difficult for a tube amplifier is the B&W 802D. This speaker is rated nominally 8 ohms but employs dual woofers, which are both 8 ohms and in parallel, so really this speaker should be rated at 4 ohms since the most power made by the amp will be seeing this load. If you see dual woofers in a loudspeaker, it is really good idea to inquire with the manufacturer about the impedance at the woofer frequency! If it is 4 ohms I would ask if its possible to make a version where the woofers are in series rather than in parallel. If you get a lot of pushback on this topic, move on!

Something that is really important when matching a tube amplifier to a loudspeaker is to understand that it can't double power as impedance is cut in half. Where this gets you in trouble with speaker designers is that many of them use the sensitivity specification when rating their speakers. Sensitivity is 2.83 volts into the speaker and measured at 1 meter. If the speaker is 8 ohms, this is one watt. But if the speaker is 4 ohms, this is 2 watts- IOW the speaker is actually 3 db less **efficient** than it would seem to appear.

Where the trouble spot is: many designers don't really see a difference between sensitivity and efficiency- so they think that by paralleling a driver, the efficiency of the speaker is increased by 3 db! IME when you try to correct them on this issue, often you get an angry response.

Now it happens that if you parallel a pair of woofers or put them in series, they will act exactly the same in the cabinet designed for them, the only difference is the values used in the crossover. I am going to use Kirchoff's Law here to make the point- if you have a single woofer and put one watt into it, it might make 92 db at one meter. If you have two of the same woofers in either parallel or in series and put 1 watt into the array, the result will be the each driver will have 1/2 watt on them- and the output will be exactly the same 92 db. If the efficiency were to actually go up, it would violate Kirchoff's Law, and because that is impossible (just like its impossible to violate Ohm's Law) it doesn't happen.

I can't tell you how hard it can be to get this fact across to speaker designers- its enough of a problem that I consider it an embarrassment (due to lack of engineering knowledge) when I encounter it. When you run into this, its guaranteed that the designer is using a transistor amplifier for design and testing and really does not understand the relationship between the amp and the speaker.

Now if you know all this above, and you know that higher impedances are good for tube amps as they will make less distortion and more power, then you are well ahead of many speaker designers and you now have a better chance of extracting all the performance out of the tube amp that otherwise is left on the table. Its your dollars in that amp- why not use it to best effect??

Hazyj, There is something to the idea that smaller amps sound better. In the case of push-pull tube amps, this has almost entirely to do with the output transformer itself. The reason for this is that the more power the output transformer can handle, the less bandwidth it has. So in the larger designs, usually the designer has made a tradeoff based entirely on what he thinks is important.

If the amp employs negative feedback, usually the high frequency bandwidth is the area that is sacrificed. This is because there is not a lot of energy at high frequencies so its possible to get away with this if you also run negative feedback to help output a little with the voltage response of the amp which will be degraded (another way of looking at this is the output impedance will be higher at the area of the rolloff and beyond).

In the old days, 60 watts was about the limit where the output transformer could make the bandwidth required to really be hifi. That bandwidth BTW is well past 20Hz-20KHz; its nice to get response up to 100KHz if you can so phase shift at 10KHz is not adversely affected. This really helps with the presentation of the soundstage, depth in particular.

These days designers seem to have pushed that power level up to about 100 watts before bandwidth really starts to suffer.

Now the problem with push-pull is often that there is a phase-splitter circuit that introduces some distortion. This is not true of all P-P amps, but it is true of most of them. This generally is not an issue until you get into the lower power regions of the amplifier, at which point the distortion of the phase splitter comes into play. For this reason a highly efficient speaker (+98 db or so) may not be the best choice with a plus-100 watt amplifier as you may never get the amp out of that region of higher distortion at lower power levels.

SET amplifiers get around this problem by not having a phase splitter (that's how we do it too although our amps are push-pull). This allows them to have a distortion characteristic that becomes unmeasurable as the power decreases to zero. To really take advantage of this, you really do need a high efficiency loudspeaker! The reason is two-fold with SETs- first, that bandwidth limit discussed with the OPT of P-P amps is more profound with SETs, the practical upper limit being only about 7 watts before bandwidth is degraded (for this reason many SET designers don't bother trying to get the bass right in the OPT as the speakers that have that kind of efficiency don't play the bottom octave either). The second reason is that you never want to push an SET past about 20% of full power or else the higher ordered distortion products come into play (when this happens, its usually on transients, and because the human ear uses the higher orders as loudness cues, the result is that the SET sounds a lot more dynamic than it has any right to because the loudness cues are occurring on the transients. You read about this 'dynamic' character with SETs all the time, but its an indication that the speaker used is not efficient enough).

Where is is all going is that if you want to really get the most out of your tube amplifier investment dollar, the selection of the loudspeaker is really important!

Something else to be aware of: Tube amps make more power and run more efficiently into higher impedances. You really don't want to make **any** amplifier (tube or transistor) work all that hard- the harder you make it work, the more distortion it will have. So you want to avoid 4 ohm speakers generally speaking- the amp will make less distortion into 8 or 16 ohms. This has nothing to do with whether the amp employs negative feedback or not- and it has everything to do with the fact that OPTs operate more effectively (wider bandwidth, lower distortion, less amplifier power converted to heat) into higher impedances. This is why you are seeing more 16 ohm speakers on the market these days- designers are slowly figuring this out.

^^ Charles, I think human nature is the culprit: the most powerful human motivator being to look good, the second most powerful not to look bad. You fail both of those if you don't understand the difference between power and voltage if you design speakers.

BTW, the Tonian and Ocellia have both used the PHY drivers that I mentioned. I think Tonian has moved away from the PHYs but still makes very easy to drive loudspeakers.

If anything, the move to tube amps in the last 30-40 years has been for the things that tubes do that are difficult or impossible for transistor amps. I don't see audiophiles that are into tubes doing it for any other reason then they want to get as close to the musical experience as possible.

I really don't think you have to know anything about tubes, how tube amps work or why people use them. All you really have to know to understand that they work really well for music, is that over half a century after they were declared obsolete, they are still around- and we still play them and even the tubes/transistor debate is older than many audiophiles! IOW the economics tells the story!

That's definitely true that in the studio it's processing the signal - but at home, tube amps are often doing the same thing. It is a form of distortion. Not all tube based systems distort this way, but I do see lots of listeners using underpowered amps for the speakers they have.

Tubes are a great way to emotionally get close to the music - I think of tubes as a way to create an illusion that performers are in the room with you - but I never think of this as accurate or neutral - but it can be very involving...

At home distortion is not the name of the game. Seriously. The ability of a tube amp to reproduce a more realistic soundstage (as depicted in your 2nd paragraph) is because they are better at low level detail which comes from the greater linearity that tubes offer. I've always been perplexed by the universal admission by both tube and transistor camps that tubes are inherently lower distortion, but then tubes are always getting docked for being 'distorted' simply because they sound better. The real reason they sound better is they have less of the distortions to which the ear is very sensitive (specifically higher ordered harmonics) although some designs may have more of the types to which the ear is not.

Many recordings are in fact made to present a true stereo image, which means a 3D soundstage with depth and Left-right extension, with everything in between.

20% of full power (I'm assuming that means average power) is a significant % but not outrageous if the speakers aren't crazy-sensitive. Am I to understand that SET owners go into it knowing they NEED to keep this contract if they want to enjoy their sound? If the "SET sounds a lot more dynamic" as it encroaches on this forbidden region, isn't it reasonable to think that SET owners (or dealers or wannabes) might poke through that limit just a wee bit to enjoy the subtle excitement of the extra dynamics? I need to hear this myself to know, but my guess is that doing such a thing might lead to one of those "wow, I never heard THAT before!" kind of moments.

I do see more people using speakers that are less efficient then the SET should really be used with, and they still enjoy the result. But its a simple fact that they are not getting nearly the performance out of their amplifier, simply because they are pushing it too hard. Most SETs don't employ feedback (this has a lot to do with the fact that you can't use feedback to extend bandwidth in a power amplifier, and if you are not really really careful with your timing constants in the feedback loop, the poor high end response of most SET OPTs will really get you in trouble- peaks before the roll-off, unstable operation, that sort of thing.

So since feedback is unwieldy and sounds wrong because it rarely can be applied properly in an SET, most designers don't do it. As a result, if you run it over 20% full power (on peaks, not average) the higher ordered harmonics will come into play, causing exaggerated dynamics and harsher sound. BTW once you know that distortion is responsible for a lot of the 'dynamic' quality of SETs, its easier to hear that it is in fact distortion doing it, so by reading this I may have ruined if for a lot of people. Sorry about that.

To take this a bit further, distortion is really a problem in all systems, not just SETs or tubes in general. As a result, I suspect that in about 90% of all audiophile conversations, the word 'dynamics' can be substituted with the word 'distortion' without altering the meaning of the conversation at all. For this reason I am careful to use a different word, like 'impact' when I want to talk about dynamic contrast. Too much unconscious stigma with the word 'dynamics'.