Which amplifier sounds better -- tube or transistor?
It is a commonly accepted a well designed transistor amplifiers have better parameters: low THD reaching below 0.001%, wide flat bandwidth (DC to 500kHz), excellent transient response, low hum, noise and microphonics, near zero output impedance. On the other hand, tube amplifiers typically have high distortion (0.1...3%), limited bandwidth (15Hz...30kHz), may exhibet infra-low frequency peaking, transient overshoot, microphonics, noise, maybe hum, finite output impedance (2...20% of the load).
Yet, in spite of apparently inferior specifications, the sound of a tube amp is preferred in certain cases. Why? Because of various distortions introduced by a tube amplifier.
For music creation, distortion of a tube amp becomes an integral part of a musical instument. Tube amplifiers provide soft clipping, depending on the circuit, amount of negative feedback and type of the output tubes used. For example, a beam tetrode 6V6GT would give more agressive compression than a pentode EL84. An amplifier changes sound of a guitar the way a musician desires. Indeed, different tubes "sound" differently for a guitar player. In comparison, a transistor amp would give very harsh, sandy, noisy, unpleasant clipping, and below the clipping level the sound would be too sterile and plain. Also a relatively large output impedance might emplasize resonance of a speaker, giving more low frequency juice and boom and bass to the sound.
Thus for creation of music, undoubtfully, a tube amp has lots of merits.
Reploduction of music is a different story. A goal here is to listen purely to what recording engineers have mixed onto a CD and not add any more distortion on top of that. Low distortion solid state amp would be perfect for that. In this case it would be ridiculous to argue about whether 300B or 2A3 tube "sounds" better, while the signal has already passed through digitisation, digital signal processing, interpolation, compression, digital-to-analog conversion and numerous solid state operation amplifiers before reaching the Holy Grail of a tubed output stage.
However, even for sound reproduction some claim valve amps are better. To understand why it is so, let us ask a question: "Why do we use tone controls and equalizers?" The answer is to make sound better. Tone control introduces distortion (in this case frequency response distortion) which appears to improve the sound. We add distortion to make sound better!
Similarly, tube amplifiers also introduce distortion which makes the sound subjectively better. For example, low feedback phase margin at low frequencies might result in some pleasant bass boost. More likely, insufficient damping of the speakers by a tube amp will accentuate speaker resonance which might also sound "good". Inductive output impedance might somehow interplay with Ls and Cs of the speaker cross-over filters, pleasantly boosting highs... These distortions are subtle and produce mild and unique sound coloration which is almost impossible to simulate by an equaliser in front of a distortionless solid-state amplifier. This is why only way to compare valve amps is to listen to them. Music pieces are commonly used for listening tests. However, listening test results would be different on speech programs. A crispy, transparent, dry and detailed sound of a transistor amp would appear more natural for speech reproduction because of good speaker damping, than a rich, but boomy valve sound.
In other words, a tube amplifier in conjunction with the speakers becomes a tone control which in certain cases helps. Another thought is that perhaps a small amount of low order harmonics added by a tube amp better imitates mechanical nonlinearity of our ear-drums... But this is only a passing thought and perhaps too far-fetched...
Below is an example of a basic "Chinese" tube amplifier sold on the E-bay. It is based on FU-50 (GU-50, LS-50, ГУ-50) tubes in triode with 6SJ7 preamplifiers.
Fig. 1. Top view of GU-50 valve audio stereo amplifier.
Fig. 2. Bottom chassis view of GU-50 tube stereo amplifier.
While the amplifier is advertised to have 10W+10W output, it barely produces 2W per channel, because effective plate voltage is low. Due to losses in the rectifier 5Z3G tube, only 330V is available on the plates of GU-50. Given that about 50V is lost for the self-bias, it leaves 280V cathode-to-plate, which is not much for a triode connected tube running without grid current. Output transformer (3.5K to 4R/8R) is quite poor with 1.2 Ohm resistance of the secondary (8 Ohm) and 360 Ohm resistance of the primary. Thus the transformer has only about 75% effeciency.
Frequency response is mediocre.
Fig. 3. Frequency response of a GU-50 (FU-50) amplifier (8 Ohm load).
Harmonic distortion is quite high.
Fig. 4. Harmonic distortion of a GU-50 (FU-50) audio amplifier, 2dB below clipping, 8 Ohm load.
Distortion is about 2%, rising at low frequencies probably because of the drop of plate impedance and transformer magnetisation non-linearity. Reduction of distortion at high frequencies is probably caused by general response roll-off and consequently harmonics suppression.
The most surprising, however, is that notwithstanding poor specifications it "sounds good" (if not clipping) -- reasonably deep and rich bass, caressing non-irritating highs, etc. Focusing intensely and deliberately on the sound quality as such, you may notice some intermodulation flaws and near-clipping sonic artefacts, occasional "sand", but when just listening in a relaxed way, it gives, surprisingly, a good experience. That probably is the mystery of valve sound...
Therefore, it might be counter-productive attempting to reduce distortions (THD, frequency and damping factor) in a tube amp to the extra low levels. Then the tube amp will most likely sound as "sterile" as a quality reference solid-state amp -- disgustingly perfect. The whole charm and virtue of a valve amp -- is in its imperfect performance.
1) For music creation a tube amplifier is definitely better;
2) For sound reproduction an ideal, sterile, "reference" distortionless solid-state amp is generally preferred, especially for speech programs, e.g., in a broadcast receiver. For music, tube amps might sound better because of sound coloration they introduce.
3) Subtle distortion and imperfections of a tube amp -- that is what makes the valve sound rich and pleasant. A tube amp not only introduces its own distortion, but due to its finite output impedance, modifies the tone of the speakers, sometimes favourably.
4) It is questionable whether there is any point in designing an ultra-low distortion tube amplifier. Probably it will sound as sterile as a solid-state one, but most likely will be more expensive.