Power output of tube amps compared to solid states

Unsound, you are right, a lot of ESL designers do work with solid state. I am of the opinion that they have a particular challenge- break out of the niche that they are in by coming up with an ESL that actually works with transistors...

The problem is two-fold. First, the impedance decreases as frequency increases, meaning that a transistor amp will make more power, causing brightness. Bass is an issue, as there can be some pronounced impedance peaks in the impedance curve. This prevents a transistor amp from making power. This is why a 200 watt tube amp can keep up with a 600 watt transistor amp on a set of Sound Labs, as the 600 watt amp may only be able to make 75 watts in the bass, where the tube amp can be capable of nearly full power.

The second is of course that the impedance curve has nothing to do with driver or box resonance, something that is fundamental to the operation of the Voltage Paradigm. In fact ESLs prefer to see flat power response out of the amp rather than flat voltage response.

To limit these issues a lot of ESL guys keep the speaker impedance very low- 4 ohms in the bass and 0.5 ohms at 20KHz is common. You still have the 8:1 change in impedance, but many transistor amps cannot make much in the way of additional power into 0.5 ohms and at that impedance, the speaker cable itself is a huge limiting factor. Its a band-aid approach, and when you see this its an ESL manufacturer that wants to cash in on the extra market share that they see in transistors.

You may have noticed that this is an entirely different example of how a tube amp with less power can be more powerful than a transistor amp; whenever you are dealing with high impedances this can be the case. Sound Labs have a peak of over 40 ohms in the bass. The 600-watt transistor amp above driving that peak might only make 75-100 watts.

Atmasphere wrote: "What tubes bring to the table is the ability to build a low-distortion amplifier without loop feedback. With no loop feedback, time-domain distortions are 100% eliminated. With feedback, time-domain distortions become the name of the game."

My understanding is that the reason time-domain distortions are of audible significance has to do with the human auditory system. The ear has a characterstic called "masking" by which it ignores a low-level signal that is near (in frequency) to a high-level signal. Audio data compression algorithms (such as MP-3) take advantage of this and simply omit signals that would likely be "masked".

Masking works great in the frequency domain, but guess what - it fails miserably in the time domain! Unless the loud and soft signal happen at exactly the same time, the soft signal is not masked. Distortions that arrive slightly later in time, even if they are much lower in amplitidue, are far more audible than the same distortion which arrives simultaneously with a masking signal.

Duke

Atamasphere, again I think your argument is sound, but, once again, it's interesting that for example, that J. Gordon Holt found the 160 WPC ss Threshold SA 1's to have better bass than the 225 WPC tube VTL 225's on his Sound Labs.

Duke, perhaps a bit off topic, but in a previous thread Atmasphere offered a link to a 55 year old paper by a speaker manufacturer's engineer that within the context of that paper, regularly suggests the use of feedback to provide appropriate critical damping factor. Most of the speakers referenced in that article appear to be of higher impedance i.e. 16 Ohms, which I suppose was typical of the times, as was probably the limited availability of high powered amplifiers. While I generally agree with the thinking behind not using feedback and IME the proof is there in the listening. My point being, that it might be hard to have and use absolutes in designing audio gear. There often seems to be a need for appropriate trade off to make the best complete package.

Unsound, amplifier/speaker synergy can definitely be used to advantage. This is how it's done with a high output impedance amp: The speaker designer uses impedance peaks to get the amp to deliver more power where he wants it. We almost always see impedance peaks in the bass region, so by playing with the enclosures's tuning frequency the designer can use those impedance peaks to extend the bass deeper than it otherwise would have gone. However if the amp's output impedance is too high, the bass will boom no matter what the tuning - so there is an "optimum" for a given speaker.

The reason this type of amp doesn't give good results with all speakers involves more than just the bass region. The speaker's impedance curve usually has peaks and valleys above the bass region, and a high output impedance (or current-source approximating) amp will tend to deliver more power into the peaks and less power into the valleys. A low output impedance (voltage source approximating) amp does the opposite. If a speaker has a smooth impedance curve above the bass region it can work well with both types, provided the bass tuning is adjusted accordingly. With Ralph's S-30, most of my speakers will exhibit roughly one-third to one-half octave greater bass extension than with a solid state amp, but I have to change the tuning frequency. That extra bass is pretty much a "free lunch". In practice I would say Ralph's amps are closer to a "constant-power source" rather than a "constant-current source", but that's still different enough from "constant voltage" to present unique challenges and opportunites for the speaker designer.

Now a designer can also take advantage of the "free lunch" to be had from a solid state amp, by dropping the impedance in the region where he needs more output. In that case, I'd parallel a second woofer in the bass region to drop the nominal impedance to 4 ohms, increasing the amp's output in that region. That calls for a second woofer and a larger enclosure, so it's maybe not as much of a "free lunch" as the first case.

Duke