4 ms·

Isn't non-linearity a desirable property for audio tubes?

hamiltonkibbe 12y ago
Yes but not all non-linearity. With a high plate voltage (and a reasonable bias point) you get gentle compression as the tube starts to break up. That's the good (T00B) sound, with the even harmonics and everything. When you run a low plate voltage you start to run into some of the undesirable things that happen when you get close to the point where the tube stops conducting entirely. you start to hard-clip the signal on one half-cycle and all those odd square-wave harmonics start to jump out, and that can sound even worse than some mediocre diode hard-clip/distortion stage designs. EDIT: If you look at the plate curves for e.g. a 12AX7 http://www.paleoelectronics.com/blog/wp-content/uploads/2009/01/12ax7-withpower.png http://www.paleoelectronics.com/blog/wp-content/uploads/2009... The way you pick your operating point is you draw a diagonal line from the plate voltage at the bottom to the saturation current on the left, and the output current moves along that line as the input signal changes (the input voltage is the voltage at the top of each curved black line). A good operating point is one where the intersections of the curves with the load line (the diagonal line you drew) are evenly spaced along the load line. If they are perfectly spaced, the amplification is exactly linear. As an example, if you plot a load line for 350V plate voltage and 2mA max plate current the intersections with the curves will be closer together on the right and move farther apart as you move left along the load line. That is the asymmetrical compression that gives a nice tube amp distortion tone. If you do the same thing with a 12V plate voltage (and any plate current) you start running into that 0V grid curve and it's hard to reason about what's happening in between the two data points (curves) you have, because at some point the curves change and start looking like the 0V one, but it's hard to tell where. Once you start driving the grid positive the tube will stop conducting entirely and hard-clip your signal, which is what I was referring to before. The TL;DR; would be something like look at how the 0V curve goes the opposite way as all the others, that's not where you want to be.
- kazinator 12y ago
  Under-voltage is only bad if the tube is designed for a high voltage and you don't give it what it is designed for. If the tube needs 200V, you give it 200V. If it needs 30V, give it 30V. Some snobs will say that nothing which runs at only 30V will have the proper headroom. ... and then they drive their amp with chip-and-diode based overdrive pedal (e.g. Ibanez Tubescreamer), running on 9V, haha.
  - hamiltonkibbe 12y ago
    I agree with your first point entirely, but there is precedent in industry for putting starved-plate tube stages in things just for marketing. A "great" component in a shitty design can sound worse than a "shitty" component in a great design. I just have a feeling that the marketing push to "Just put a tube in it" will lead to more of these "token" tube applications because it's cheaper to run it at whatever voltage rail is already there than putting some hulking transformer just to run a 300V B+ rail for a single preamp stage. The headroom thing is all relative, I don't think your amp would be too happy if you fed it a 300V P-P signal from your 3kV Audiophile-Grade-Headroom pedalboard... I also can't wait for the 64-bit digital audio revolution. Nothing with less than 384dB of dynamic range is worth listening to! /s
    - kazinator 12y ago
      If you're only getting 384 dB out of 64 bits, you're doing it wrong by using integers. 64 bit double floats go from, what, around E-308 to E308 in the exponent. So that is around 616 Bels, or 6160 dB. :)

Yes but not all non-linearity. With a high plate voltage (and a reasonable bias point) you get gentle compression as the tube starts to break up. That's the good (T00B) sound, with the even harmonics and everything. When you run a low plate voltage you start to run into some of the undesirable things that happen when you get close to the point where the tube stops conducting entirely. you start to hard-clip the signal on one half-cycle and all those odd square-wave harmonics start to jump out, and that can sound even worse than some mediocre diode hard-clip/distortion stage designs. EDIT: If you look at the plate curves for e.g. a 12AX7 http://www.paleoelectronics.com/blog/wp-content/uploads/2009/01/12ax7-withpower.png http://www.paleoelectronics.com/blog/wp-content/uploads/2009... The way you pick your operating point is you draw a diagonal line from the plate voltage at the bottom to the saturation current on the left, and the output current moves along that line as the input signal changes (the input voltage is the voltage at the top of each curved black line). A good operating point is one where the intersections of the curves with the load line (the diagonal line you drew) are evenly spaced along the load line. If they are perfectly spaced, the amplification is exactly linear. As an example, if you plot a load line for 350V plate voltage and 2mA max plate current the intersections with the curves will be closer together on the right and move farther apart as you move left along the load line. That is the asymmetrical compression that gives a nice tube amp distortion tone. If you do the same thing with a 12V plate voltage (and any plate current) you start running into that 0V grid curve and it's hard to reason about what's happening in between the two data points (curves) you have, because at some point the curves change and start looking like the 0V one, but it's hard to tell where. Once you start driving the grid positive the tube will stop conducting entirely and hard-clip your signal, which is what I was referring to before. The TL;DR; would be something like look at how the 0V curve goes the opposite way as all the others, that's not where you want to be.

kazinator 12y ago
Under-voltage is only bad if the tube is designed for a high voltage and you don't give it what it is designed for. If the tube needs 200V, you give it 200V. If it needs 30V, give it 30V. Some snobs will say that nothing which runs at only 30V will have the proper headroom. ... and then they drive their amp with chip-and-diode based overdrive pedal (e.g. Ibanez Tubescreamer), running on 9V, haha.
- hamiltonkibbe 12y ago
  I agree with your first point entirely, but there is precedent in industry for putting starved-plate tube stages in things just for marketing. A "great" component in a shitty design can sound worse than a "shitty" component in a great design. I just have a feeling that the marketing push to "Just put a tube in it" will lead to more of these "token" tube applications because it's cheaper to run it at whatever voltage rail is already there than putting some hulking transformer just to run a 300V B+ rail for a single preamp stage. The headroom thing is all relative, I don't think your amp would be too happy if you fed it a 300V P-P signal from your 3kV Audiophile-Grade-Headroom pedalboard... I also can't wait for the 64-bit digital audio revolution. Nothing with less than 384dB of dynamic range is worth listening to! /s
  - kazinator 12y ago
    If you're only getting 384 dB out of 64 bits, you're doing it wrong by using integers. 64 bit double floats go from, what, around E-308 to E308 in the exponent. So that is around 616 Bels, or 6160 dB. :)