The characteristics of a single-ended triode depends a huge amount on the circuit around it. A simple cathode resistor acts as local negative feedback. Bypassing it with a capacitor removes the local negative feedback for a range of frequencies. Then there’s the choice of load, resistor or transformer. All have different electrical characteristics, all fall under the “single-ended triode” umbrella.
Personally, I’m quite skeptical about the idea that second-order harmonics are “warm” or “round” sounding. Whether or not distortion is “musical” depends on whether you are making music or adding noise to something that is already music; if you are playing a guitar then the distortion is music, if you are listening to a song, the distortion in your audio equipment is by definition noise.
I understand that sometimes distortion is inevitable even when it’s not desirable. Your microphone preamplifier will sometimes get overloaded, or you might be listening to a song at high levels with transients beyond the power capabilities of your system. So for microphone preamplifiers, go ahead and stick a tube in there. But for listening, it is usually cheaper to fix your distortion problems by buying a more powerful solid-state amplifier with more headroom, and more expensive to fix your distortion problems by switching to a tube amplifier with better distortion characteristics.
That is simply incorrect. I think there’s a fundamental misconception about what distortion is, and how it works, at play here.
Any distortion produces intermodulation distortion.
A sine wave will produce exactly an octave harmonic if you distort it with only second-order distortion. However, nobody listens to pure sine waves. So if you feed anything more complicated than a pure sine wave, you will get something other than harmonics at octaves.
Perhaps I should have been more specific; 3rd order IMD. The IMD3 can’t be produced by the ideal JFET drain current (Vgs-Vt)^2; tubes are the same. Of course there are other distortion mechanisms too. This is in the RF power amp world (been designing them for 20+ years), the IMD3 are the main concern as those fall around the carrier; we generally don’t care about 2nd order since those are easily filtered (but not in audio).
Maybe it’s a stretch to apply this to audio. If you play a flat on piano, and IMD3 falls on a sharp, then that sounds bad. The distortion one octave higher sound fine. Even the sharp one octave higher sounds better than the adjacent sharp.
I bought a tube amp kit for Christmas, so it will be interesting to measure the two-tone IMD3. I only have measured audio THD once, and that was using a spectrum analyzer, so only looked at harmonics. Even if you used two-tones, the IMD would fall within the RBW of the spectrum analyzer.
Single ended triodes of low to no degeneration are not the same current profile as a JFET.
The Child Langmuir law has a 3/2's exponent, so you get 2nd and 3rd for the single ended case. Further, virtually all guitar amps have a push pull amplifier output (of questionable symmetry) operating into the speaker, so the even order terms are cancelling to some degree here.
The general character of many tube amps is to make a smooth transition from predominantly 2nd order distortion to a mixture of odd order harmonics. This ends up as primarily the 3rd order as a guitar amp speaker makes a heck of an underdamped lowpass filter.
Between the cathode bias bypass caps, the dc blocking caps, and the RC behavior of grid stopper resistors combined with the Miller capacitance, guitar amplifiers have a lot of internal bandwidth filtering that reduces IMD amplitudes.
---
JFET current behavior in the saturation range is as you say, but not necessarily so in the ohmic region. Generally people using JFETs in these circuits use at least some of their ohmic range behavior.
Since you have skill and experience in this area, and also have a tube amp kit, I suggest googling for "Tubes 201." This is an excellent summary article for the behavior of vacuum tubes.
Also, Aiken amplification website, and Merlin Blencowe's book I mentioned yesterday you would find to be good reads.
> Maybe it’s a stretch to apply this to audio. If you play a flat on piano, and IMD3 falls on a sharp, then that sounds bad.
Do the math, your distorted spectrum will contain sum and difference frequencies. (For me, what makes the math easy is breaking a real sine wave into a sum of its complex components, which are easier to work with.)
This is, as you say, what makes 3rd order bad in RF, because if I have two frequencies F1 and F2, F1+F2 or F1-F2 can be filtered out no problem, but 2F1-F2 or 2F2-F1 is a problem. But this intuition about distortion does not extend to music.
In music, it is not so much the individual frequencies that are relevant, but the harmonic spectra, and the spectrum is within spitting distance of baseband, so filtering won’t do you any good (at least not in eliminating distortion).
And since music contains harmonic spectra to begin with, for a given F in the signal, you probably already have 2F, 3F, 4F. So when you calculate the spectrum for 2nd order or 3rd order distortion, you get the SAME frequencies in the output either way, just in different quantities.
As a result, as electric guitarists know, you have to be very careful about what harmonic content you feed into distortion. If you feed in a note, you get the same note out with different texture. If you feed an octave, it sounds good. Perfect fifth, great, the distortion adds overtones of the octave below the root. A major third is iffy and something you’d use with less distortion. As you get farther away from simple intervals, like octave 2:1, fifth 3:2, fourth 4:3, or major third 5:4, it gets worse and worse. (Eddie Van Halen reportedly retuned his guitar, tuning the second string down 14 cents, to get the major third closer to a harmonic third.)
And this is why distortion for a single instrument sounds good, but distortion for an entire song with multiple instruments (some of which are usually inharmonic) sounds bad. Even with 2nd order distortion, do you really want sum and difference frequencies between overtones of the guitar and overtones of the snare drum? No.
Personally, I’m quite skeptical about the idea that second-order harmonics are “warm” or “round” sounding. Whether or not distortion is “musical” depends on whether you are making music or adding noise to something that is already music; if you are playing a guitar then the distortion is music, if you are listening to a song, the distortion in your audio equipment is by definition noise.
I understand that sometimes distortion is inevitable even when it’s not desirable. Your microphone preamplifier will sometimes get overloaded, or you might be listening to a song at high levels with transients beyond the power capabilities of your system. So for microphone preamplifiers, go ahead and stick a tube in there. But for listening, it is usually cheaper to fix your distortion problems by buying a more powerful solid-state amplifier with more headroom, and more expensive to fix your distortion problems by switching to a tube amplifier with better distortion characteristics.