For those who are into tube hardware, here is an interesting obervation.
It is often said that if a tube starts displaying a purplish hue due to some inpurities in the vacuum, it is harmless. I had a situation like this in my AudioNote clone preamp, and on occasion the same tube started to motorboat, but at other times it was fine without any apparent audible issue.
So I hooked it up and made a few measurements to see whether this was really the case. Since the tube did occasionally motorboat on it admittedly was a more severe case, but the results were interesting nevertheless.
Attached are the 1kHz spectra of two different tubes at +6dB gain with a 1Vrms input. The same channel in the preamp was used i.e. both tubes were used in the same socket in order to prevent possible differences due to wiring, component or PSU variations etc - any difference on the spectra is therefore attributable to differences between the tube itself.
The first tube displays near-perfect, almost textbook results. The 50Hz ripple and its harmonics are acceptably low, especially considering its high internal gain (the pot is at the input) and low PSRR [Ignore the low-frequency stuff, it came from improper equipment setup].
We get a very low noise floor, and only a predominant 2nd harmonic as we would expect and desire from tube gear since even harmonics, 2nd in particular, are sonically pleasing. This spectrum would be commendable even for a solid-state design with lots of negative feedback, which means that it is a very well-designed product and that the active voltage regulator I added is doing a fine job.
But now look at the other tube. The higher-order harmonics and thus THD has risen above the noise floor. It's still highly unlikely to be perceived, but I won't listen with a clear conscience if I knew the two channels were not closely similar. Additionally, I had to turn up the volume a little with the second tube to get the same gain, so its gain has reduced some that will lead to channel imbalance - and it's so seldom for quality pre/power amps to have balance controls..
In conclusion we can see that any tube that shows any sign of age or wear will indeed exhibit higher distortion and reduced overall gain. How this would affect power tubes under load I'm not sure, but I'm reasonably confident that it will share the same characteristics.
Sidenote: Just because certain harmonics aren't visible it doesn't mean they are not there, they're merely buried under the noise floor. This is exactly the principle exploited by the addition of dither in ADC's and DAC's: adding random noise to raise the noise floor and thus reducing SNR, but swamping the harmonics. This is desirable since harmonics are more distracting to the ear than noise of the same amplitude. A simple; exaggerated example would be a subwoofer playing a 20Hz tone. This should be almost inaudible, but as the volume is increased (for cheap subs often from the start), the driver will eventually start to produce a very audible 60Hz 3rd harmonic. The amplitude of the 20Hz fundamental is actually far higher, but since our ears act as a highpass filter we only hear the harmonic. This ghastly harmonic is much more irratable and easier to identify than white noise of the same level would be. Of course we are talking of levels far, far lower than this with the true fundamental we're listening to at several dB's higher.
Whether the 2nd harmonic of such an amplitude as depicted by this circuit would be audible is debatable.. it's 60dB below the fundamental and quite unlikely to offer any real sonic difference to a solid-state equivalent. This substantiates my listening experience of it: it's fast and detailed and very much like a solid-state device, which begs the question as to why bother with tubes if a well-designed one sounds the same as a transistor model that would cost a lot less to build and maintain? Isn't it easier to merely add a DSP to simulate a more profound distortion effect? Crude software-based versions already exist, and Bob Carver added a "2nd-harmonic injector" to some of his products. An accurate model of some of the better-known tubes and implemented with FIR filters in a DSP or FPGA is a more elegant approach and is increasingly investigated. Especially in modern electronics where we stay much longer in the digital domain, these effects can be implemented without any other sonic penalty as we would have 10 years ago by going the ADC-DSP-DAC route.
It is often said that if a tube starts displaying a purplish hue due to some inpurities in the vacuum, it is harmless. I had a situation like this in my AudioNote clone preamp, and on occasion the same tube started to motorboat, but at other times it was fine without any apparent audible issue.
So I hooked it up and made a few measurements to see whether this was really the case. Since the tube did occasionally motorboat on it admittedly was a more severe case, but the results were interesting nevertheless.
Attached are the 1kHz spectra of two different tubes at +6dB gain with a 1Vrms input. The same channel in the preamp was used i.e. both tubes were used in the same socket in order to prevent possible differences due to wiring, component or PSU variations etc - any difference on the spectra is therefore attributable to differences between the tube itself.
The first tube displays near-perfect, almost textbook results. The 50Hz ripple and its harmonics are acceptably low, especially considering its high internal gain (the pot is at the input) and low PSRR [Ignore the low-frequency stuff, it came from improper equipment setup].
We get a very low noise floor, and only a predominant 2nd harmonic as we would expect and desire from tube gear since even harmonics, 2nd in particular, are sonically pleasing. This spectrum would be commendable even for a solid-state design with lots of negative feedback, which means that it is a very well-designed product and that the active voltage regulator I added is doing a fine job.
But now look at the other tube. The higher-order harmonics and thus THD has risen above the noise floor. It's still highly unlikely to be perceived, but I won't listen with a clear conscience if I knew the two channels were not closely similar. Additionally, I had to turn up the volume a little with the second tube to get the same gain, so its gain has reduced some that will lead to channel imbalance - and it's so seldom for quality pre/power amps to have balance controls..
In conclusion we can see that any tube that shows any sign of age or wear will indeed exhibit higher distortion and reduced overall gain. How this would affect power tubes under load I'm not sure, but I'm reasonably confident that it will share the same characteristics.
Sidenote: Just because certain harmonics aren't visible it doesn't mean they are not there, they're merely buried under the noise floor. This is exactly the principle exploited by the addition of dither in ADC's and DAC's: adding random noise to raise the noise floor and thus reducing SNR, but swamping the harmonics. This is desirable since harmonics are more distracting to the ear than noise of the same amplitude. A simple; exaggerated example would be a subwoofer playing a 20Hz tone. This should be almost inaudible, but as the volume is increased (for cheap subs often from the start), the driver will eventually start to produce a very audible 60Hz 3rd harmonic. The amplitude of the 20Hz fundamental is actually far higher, but since our ears act as a highpass filter we only hear the harmonic. This ghastly harmonic is much more irratable and easier to identify than white noise of the same level would be. Of course we are talking of levels far, far lower than this with the true fundamental we're listening to at several dB's higher.
Whether the 2nd harmonic of such an amplitude as depicted by this circuit would be audible is debatable.. it's 60dB below the fundamental and quite unlikely to offer any real sonic difference to a solid-state equivalent. This substantiates my listening experience of it: it's fast and detailed and very much like a solid-state device, which begs the question as to why bother with tubes if a well-designed one sounds the same as a transistor model that would cost a lot less to build and maintain? Isn't it easier to merely add a DSP to simulate a more profound distortion effect? Crude software-based versions already exist, and Bob Carver added a "2nd-harmonic injector" to some of his products. An accurate model of some of the better-known tubes and implemented with FIR filters in a DSP or FPGA is a more elegant approach and is increasingly investigated. Especially in modern electronics where we stay much longer in the digital domain, these effects can be implemented without any other sonic penalty as we would have 10 years ago by going the ADC-DSP-DAC route.
