Tubes to Transistors: It's Not in the Numbers

Brannon Klopfer
June 11, 2012

Submitted as coursework for PH250, Stanford University, Spring 2012

Fig. 1: A vintage (circa late 1950s/early 1960s) Dynakit ST-70 push-pull vacuum tube amplifier photographed in the dark. It utilizes two EL34 tubes per channel, as can just be seen from the photograph (click to enlarge). Author's photo.
Fig. 2: Spectrum of a typical rock song (Sultans of Swing, by Dire Straits). The blue line is the original song; the magenta line is slightly clipped; the red line is severely clipped. Data were normalized to the center frequency. Clipping was done with software, not amplifiers. Plot created using Audacity and GNU Octave.


High fidelity audio amplifiers have a simple job description: take in an audio signal and make it bigger. It would seem to follow that high fidelity audio amplifiers could be quantitatively evaluated in an unambiguous fashion by considering their specifications - frequency response, total harmonic distortion, signal-to-noise, output power, etc. However, this is not the case, as it does not matter what an oscilloscope or spectrum analyzer says; listening to music is not a scientific experiment, but a physiological process.

Audio amplifiers today fall into two general categories: transistor-based ("solid-state") and vacuum tube-based ("tube") amplifiers. There is some overlap between the two categories, with certain "hybrid" amplifiers utilizing both technologies. Some use solid-state components in the power supply and vacuum tubes in the actual audio amplification, while others use both solid-state and tube technology in the audio amplification.

Overdrive and Clipping

All amplifiers will introduce distortion, to some extent. And pushed beyond their limits ("overdriven"), this effect becomes particularly salient. For high fidelity listening this is of course undesirable. Not only is the music not faithfully reproduced at this point, but such distortion can damage speakers. As power increases, waveforms tend to "clip," that is, near the peak of a sine wave the amplifier cannot provide any higher voltage, so the waveform plateaus, with solid-state amplifiers typically exhibiting sharper plateaus, or "harder" clipping. [1] The sine wave then begins to resemble a square wave. This in turn means that more power is dissipated by the speaker, since square waves carry a higher average power. In the limit of a perfect sine wave distorting into a perfect square wave, twice the power would be produced. (Of course, treating music as a pure sine wave is a gross idealization, but the basic idea remains.)

In addition to higher average power, a clipped signal can be thought of in terms of harmonic distortion - that is, a clipped sine wave can be described as multiple pure sine waves, at higher frequencies. See Fig. 2 for an example of a clipped song. In particular, note that the spectra of the clipped versions deviate substantially at the high frequencies, as described. The clipping was simulated in software, and as such it exhibits ideal hard clipping, in the sense that the waveform abruptly plateaus at a consistent point.

Unlike solid-state amplifiers, tube amplifiers tend to distort in a pleasing fashion when overdriven. [2] Although still undesirable in high fidelity systems, it can lead to much sought-after sounds for guitarists. [1]


The solid-state audio amplifiers tend to win on paper. Tube amps are "huge, hot, inefficient, and somewhat dangerous," and "offer no practical advantages over their solid-state counterparts." [2] Comparing specifications of a typical, if vintage, middle-of-the-road tube amp to a typical middle-of-the-road solid-state amplifier of the late 1980s corroborates this sentiment.

I have chosen to compare a Dynaco/Dynakit ST-70 tube amplifier (Fig. 1) to a NAD 7140 stereo receiver. These were chosen because they are of similar power output and were introduced approximately the same time after their respective amplification technology was invented: the Dynaco roughly 50 years after the vacuum tube, the NAD roughly 40 years after the transistor. The Dynaco is rated at <1% distortion (presumed to be total harmonic distortion, or THD, though the manual does not specify), whereas the NAD rated at <0.03% THD (smaller is better). The Dynaco's signal-to-noise is rated at "better than 90 dB" referenced to its rated power of 35 watts, whereas the NAD is rated at 116 dB (larger is better). And, at 32 lbs., the tube amplifier is almost twice the weight of the comparatively svelte NAD (16.5 lbs.). [3,4]

Despite the objective advantages of solid-state amplifiers, tube amps are still manufactured today - and it's not because they're cheaper. [1] Perhaps it's because the distortion (even far from overdrive) is more pleasing to the human's ear-brain system. Or perhaps it's just because, as Fig. 1 demonstrates...well, they glow!

© Brannon Klopfer. The author grants permission to copy, distribute and display this work in unaltered form, with attribution to the author, for noncommercial purposes only. All other rights, including commercial rights, are reserved to the author.


[1] E. Barbour, "The Cool Sound of Tubes," IEEE Spectrum 35, No. 8., 24 (1998).

[2] D. Sweeney, "Tools and Toys: Mixed Marriage," IEEE Spectrum 42, No. 6, 72 (2005).

[3] Dynakit Stereo 70 Instructions for Assembly Operation, Dyna Company, 1959.

[4] NAD 7140 Stereo Receiver Data Sheet, NAD Electronics Intl., 1987.