Actually, that's understandable. The negative feedback loop of the tube amplifier is weak. Therefore they are not good sources of voltage and drive the speakers partly with the current.
By the way, I found the model for the speaker here: http://tamivox.org/dave/speaker/graphs.html
Well given that crossover distortion can be brought below 0.001% in a good analog solid state design, the above is only relevant if comparing poor quality amplifiers, as 0.001% should be well below audibility (if tested objectively with double-blind A/B testing) I assert.
The whole history of "transistor sound" dates from the earliest transistor amplifiers which also used transformers for crude class-B as complementary devices weren't available. Distortion was often very high, especially cross-over. Things have changed.
Many tube amplifiers have far greater distortion than ven the ~0.01% of a bog-standard modern solid state amp, leading to significant intermodulation products. Proponents of valve amps seem to assume all signals are single tone and ignore intermodulation entirely.
But as I've mentioned under powered amps will clip, and hard clipping is bad news especially (that distortion of 0.001% suddenly becomes more like 50%).
Digital sources provide a guaranteed max envelope for signals, so you can set up an amp to never clip from such a source (though you rely on the mastering to have used enough headroom, which is a whole other kettle of (over-compressed) fish..)
I guess, the main reason therefore is the fact, that not all audible distortion (like very small spikes) significantly increases the THD value.
Thus is to note, that a THD value below 0.001% is clearly audible in one case (e. g. if very small spikes are present) and inaudible in another case a THD value of 1% or 2% (e. g. if there are only low order distortion like H2 or H3).
I start in this case this thread:
https://www.diyaudio.com/community/...st-possible-thd-n-really-the-best-way.367692/
Anywhere under
http://www.renardson-audio.com
was described, that a sufficiently low THD value does not guarantee that these are inaudible - especially in the case of crossover distortion - because the influence of spikes to the THD-N value is not very present.
If there were really perfect circuit approaches with a small quiescent current in the output stage (class AB) in the latest amplifier designs without disadvantages regarded residual audible distortion, there would no longer be a real reason to choice the huge and inefficient heaters (like Pass Aleph 1.2) in class A - sometimes even single Ended - for buy or to have - just like large tube power amplifiers.
The only Class AB power amp I know without audible crossover distortion is the Horch 3.0 (Germany) - go to
http://www.horch-gmbh.de/Produkte/horch-classic.html
Unfortunately, the effort involved in selecting the output transistors is very high and a usual repair service therefore impossible.
P.S.: An AB comparison isn't always helpful either, as there are differences in the character of the sound that you only notice when listening to music for long periods of time.
Current drive changes the frequency response of the transducer.
Something that the hifi world typically doesn't like.
But it can be used to extend the bandwidth of e.g. a guitar speaker.
SS guitar amps often use load current NFB to increase the amp's output impedance.
The broadband speaker boxes without crossovers fit perfectly. The woofers can be added for more bass. They are driven by the separate AB/D amplifiers... I've already tried that: works very well.
Here is the measurement with the real amplifier and speakers.
If you really want a tube sound, you can use the following trick: Connect a resistor parallel to the speaker. But I'll warn you straight away: the sound will deteriorate.
I also made the comparison with my tube radio SABA Wildbad. I adapted its audio input for the CD player and especially for my amplifier I bought the SABA loudspeaker successor Monacor SP 205/8. And... I'm sorry, but this time the tube sounded worse than the transistor.
If you really want a tube sound, you can use the following trick: Connect a resistor parallel to the speaker. But I'll warn you straight away: the sound will deteriorate.
I also made the comparison with my tube radio SABA Wildbad. I adapted its audio input for the CD player and especially for my amplifier I bought the SABA loudspeaker successor Monacor SP 205/8. And... I'm sorry, but this time the tube sounded worse than the transistor.
It won't go serial. The output resistance of this amplifier is already around 200 ohms. If that becomes 208 ohms, you probably won't notice the difference.
With 8 ohms in parallel, the output resistance is reduced from 200 ohms to just under 8 ohms. You will already hear the deterioration of the sound.
What the ear does is decisive. Supposed "overshoots" due to impedance humps are easy to get rid of by a room-adapted tuning of the speakers, placement, cabinet filling, cabinet construction, cabinet shape and so on. And also the choice of the wires has a significant influence. Not only their diameter. Minimal properties that can be added up to a superb sound;-)
In general, I wanted to show that bipolar transistors can sound good on their own, without global negative feedback loop, if they are driven in accordance with their physical properties.
But you are also right. This amplifier can be made even more linear, e.g. with 2SA1943 2SC5200 pair. But from my point of view this sounds stinck normal like ... hi-fi
MJ15025 MJ1524 pair sounds best to me.Well if you add a lot of local feedback, yes! Witness the surprizingly good performance of the CFP output stage, which has heavy feedback built-in, or the VAS with its heavy capacitive feedback at higher frequencies to linearize it, and the use of emitter degeneration.
True a single transistor as emitter follower with constant current load can be a good performer, but with no voltage gain its not enough to build a power amplifier.
The real oomph of a BJT is the high power gain available in common emitter - 100x current gain and potentially a large amount of voltage gain too (1000x is possible even due to the huge transconductances available). This raw gain is ripe for harvesting as linearity via feedback, both local and global, resulting in circuits with more modest gains but at good linearity (remember the raw device has an exponential transfer function, about the least linear starting point imaginable).
There are tradeoffs between local and global feedback to be made, and choosing the right trade off is part and parcel of low frequency amplifier design.
Much much better than a "complementary-transistors-pp-amp" (?) sounds a SE amp. The ear not only detects the half-wave different "sound" influence (all sorts of noise) of different complementary transistors...But you are also right. This amplifier can be made even more linear, e.g. with 2SA1943 2SC5200 pair. But from my point of view this sounds stinck normal like ... hi-fi
I actually only have local DC voltage negative feedback for T3. This negative feedback has no effect for an AC signal. This way there is no negative feedback for audio signal at all. This results in the pure current gain hFE(T1) x hFE(T4). No gain is wasted on negative feedback.