A typical 300B SET with zero negative feedback. Frequency response .The green line represents a simulated 8 ohm speaker.
@jtgofish FWIW the kind of speaker in the simulation is rarely used with SETs. They are low power and so speakers used with them tend to higher efficiency; also designed to expect a higher output impedance of the amplifier, with no expectation that the amp behave as a voltage source!
Most tube amps with zero feedback will act more like a power source than a voltage source. So the standard ’simulated speaker load’ is irrelevant if the user has his SET set up correctly.
In @charles1dad ’s case, the speaker he is using was designed specifically for tube amps with higher output impedance and no feedback.
For more on this topic see:
The Voltage and Power Paradigms
Since loudspeakers of all types typically have much wider frequency variation than amplifiers, the theory is that the tonality imparted by distortion figures more importantly than tonality induced by FR error. This is because the ear/brain system converts all forms of distortion to tonality and pays more attention to that as a result. SETs produce innocuous distortion so can be fairly neutral in this regard.
The brightness and harshness of traditional solid state is an example of this: you can’t measure it on the bench in terms of FR error, since the brightness is caused by distortion interpreted as a tonality. Its really hard to get away from this problem with amplifiers using feedback, since what usually happens is the the distortion vs frequency starts to rise somewhere near 1KHz due to a lack of Gain Bandwidth Product.
Ideally distortion vs frequency should be a flat line across the audio band, and with any zero feedback amplifier with sufficient bandwidth that’s exactly what you get. Alternatively this can be fairly easily accomplished with a self oscillating class D amplifier, since obtaining super high GBP values is fairly easy with them so feedback is properly supported across the entire audio band. See
The F word
for more information.
When distortion rises with frequency, its because the feedback is decreasing with frequency and so distortion is going up. This usually does not show in THD figures; in effect sweeping this problem under the rug. So the actual distortion at higher frequencies where the ear is more sensitive can be quite a lot higher than the THD values suggest!
So while I’m not an SET fan by any stretch, picking on them due to FR errors due to output impedance is not the way to expose their real problems!
