A seldom discussed virtue of vacuum valves is that they can comfortably run hot. Almost everything musically interesting happens in the first Watt or so. This requires that the amplifier be monotonic down through and past the noise floor. It's inherently wasteful, and valves accept the burden without drama.
All good fortune,
Chris
All good fortune,
Chris
Child's Law as mentioned above.
Oh, to be sure, one can (analytically) source a sine wave of some voltage, drop in a couple of well-chosen k1 and k2 values, a suitable B+ supply, a suitable RA anode resistor, and then capture its (RA) voltage response to the sine wave. And use the fabulous Fourier Transform to convert that from time-domain to frequency domain … to plot out the “harmonics”.
And sure enough, there is a spectrum that has a strong fundamental, a 2nd harmonic, a 3rd, a 4th, 5th, 6th … n-th. Voilá!
Yet, if you are doing that sim, try changing the sine VRMS to ¹⁄₁₀ its first value. Now what to the harmonics look like in the frequency plot?
Ah… quite a bit different.
Notably, much less compared to the fundamental.
Likewise… adjusting the simulation's sine generator to have VSINE • nominal gain be within a few volts of clipping, and the “other behavior” of the circuit shows up… lots (lots) of “odd harmonics” as soft-clipping occurs, along with a notable increase of even harmonics too.
But since you (we, and often just I) am in the nuts-and-bolts digital simulation-to-prove-ap-point game, it is REALLY easy to “change out” the “digital tube model” for a semiconductor model, a 3 terminal device, and drop it in.
Notably… of course, it is 'biased wrong'. So, the initial results are pathologically silly. If with some determination, corrective 'bias' is applied digitally to allow the RA (now a RC technically!) to be conducting at about the ½ B+ point with no input signal, and an appropriately scaled sine excitation is applied to achieve a VC A/C of about ¼ to ½ B+ (RMS), well lp, and behold.
Taken from signal to Fourier frequency plot, one sees … a big fundamental, and a bunch of harmonics. Depending on whether one's digital simulation machinations included a properly sized RBASE signal voltage-to-current converter or not (or of an appropriate size, or not), one can see all kinds of different harmonic spectra.
To whit, some configurations sim-out with substantially lower ALL-harmonic energy; others with higher harmonic energy than the first valve sim. Some, with injudicioiusly chosen sine generator excitation voltage (or inappropriate RBASE) show a LOT of … what turns out to be … clipping harmonics. High in odd harmonics.
Oy, that must be it!!! Ye of hard-won lore excitedly chant.
Mmmm, hmmm… Yah, sure … clipping sounds as pleasant the scratching of nails on a chalkboard. Who knew, right? Well… I could have tol' you that, y'know. If you go back and figure the right 'biasing' and 'voltage → current' resistances, suddenly the same digital sim circuit shows no clipping, remarkable (by eye) linearity, and evne when a carefully sclaed Fourier( VIN - kVOUT ) difference-spectrum (which removes the fundamental quite well) is applied, the harmonics are below –30 dB relative to the fundamental. And not particularly 'odd-enhanced'.
Sigh…
⋅-⋅-⋅ Just saying, ⋅-⋅-⋅
⋅-=≡ GoatGuy ✓ ≡=-⋅
IO ≈ k1VKAVG1k2³⁄₂
Wherek1 & k2 are higly dependent on valve design
This in and of itself does not deliver “even harmonics”, contrary to what the post mentioning it opined. Oh, to be sure, one can (analytically) source a sine wave of some voltage, drop in a couple of well-chosen k1 and k2 values, a suitable B+ supply, a suitable RA anode resistor, and then capture its (RA) voltage response to the sine wave. And use the fabulous Fourier Transform to convert that from time-domain to frequency domain … to plot out the “harmonics”.
And sure enough, there is a spectrum that has a strong fundamental, a 2nd harmonic, a 3rd, a 4th, 5th, 6th … n-th. Voilá!
Yet, if you are doing that sim, try changing the sine VRMS to ¹⁄₁₀ its first value. Now what to the harmonics look like in the frequency plot?
Ah… quite a bit different.
Notably, much less compared to the fundamental.
Likewise… adjusting the simulation's sine generator to have VSINE • nominal gain be within a few volts of clipping, and the “other behavior” of the circuit shows up… lots (lots) of “odd harmonics” as soft-clipping occurs, along with a notable increase of even harmonics too.
But since you (we, and often just I) am in the nuts-and-bolts digital simulation-to-prove-ap-point game, it is REALLY easy to “change out” the “digital tube model” for a semiconductor model, a 3 terminal device, and drop it in.
Notably… of course, it is 'biased wrong'. So, the initial results are pathologically silly. If with some determination, corrective 'bias' is applied digitally to allow the RA (now a RC technically!) to be conducting at about the ½ B+ point with no input signal, and an appropriately scaled sine excitation is applied to achieve a VC A/C of about ¼ to ½ B+ (RMS), well lp, and behold.
Taken from signal to Fourier frequency plot, one sees … a big fundamental, and a bunch of harmonics. Depending on whether one's digital simulation machinations included a properly sized RBASE signal voltage-to-current converter or not (or of an appropriate size, or not), one can see all kinds of different harmonic spectra.
To whit, some configurations sim-out with substantially lower ALL-harmonic energy; others with higher harmonic energy than the first valve sim. Some, with injudicioiusly chosen sine generator excitation voltage (or inappropriate RBASE) show a LOT of … what turns out to be … clipping harmonics. High in odd harmonics.
Oy, that must be it!!! Ye of hard-won lore excitedly chant.
Mmmm, hmmm… Yah, sure … clipping sounds as pleasant the scratching of nails on a chalkboard. Who knew, right? Well… I could have tol' you that, y'know. If you go back and figure the right 'biasing' and 'voltage → current' resistances, suddenly the same digital sim circuit shows no clipping, remarkable (by eye) linearity, and evne when a carefully sclaed Fourier( VIN - kVOUT ) difference-spectrum (which removes the fundamental quite well) is applied, the harmonics are below –30 dB relative to the fundamental. And not particularly 'odd-enhanced'.
Sigh…
⋅-⋅-⋅ Just saying, ⋅-⋅-⋅
⋅-=≡ GoatGuy ✓ ≡=-⋅
All parts of the audio chain (including the listening room) divert from the original recording as a result of distortion. We can go to great lengths to diminish the individual distortions but it's practical to eliminate the most offending source to end up with a reproduction system that some call high end. That includes room threatment as well, or use of very high quality head phones which are not free of distortion either.
Funnily people arrive at the interchange of components to compensate for individual flaws. Ofthen you wander in snake oil territory with very expensive interconnects and alluded magical properties of sources and amplifiers. Personally I am content with some degree of imperfection, when most offending distortions are dealt with. It's a misinformation tubes have "sounds" as it's the circuit design that leaves some imperfections, only to a certain amount just tolerable.
The next step could be to focus on distortion of sonic output. There are (but less known) ways of precompensation to counteract driver induced distortion.
I think it mainly has to do with very good marketing.
There are so many other ways to make anything else sound and behave exactly the same like a tube. Control theory 101.
(well, and marketing 101)
Last edited:
Alarm bells go off for me when I see the word sonic, its frequently a sign of phoolish nonsense, and the plural is _always_ a sign of such 
I suppose it when people use the word as a generic term for "something sounds wrong but I don't understand why nor want to analyse the performance using measuring equipment that would confirm/deny suspicions" There are many metrics for "sound quality": distortion, reverberation, flatness of response, noise level, phase errors, channel matching, etc - all can be measured and teased apart.
Don't get me started! Too late.
I suppose it when people use the word as a generic term for "something sounds wrong but I don't understand why nor want to analyse the performance using measuring equipment that would confirm/deny suspicions" There are many metrics for "sound quality": distortion, reverberation, flatness of response, noise level, phase errors, channel matching, etc - all can be measured and teased apart.
Don't get me started! Too late.
And still I enjoy my tube radio giving very FR limited, noisy and distorted FM-mono
If I exchange it for my plastic portable transistor radio, it gets shut off within the hour. That's to say there is distortion and distortion...
Being a guitarist, I think I can say with confidence that it's more than marketing. Read Merlin Blencowe's excellent book on designing vacuum tube preamps for guitar amps. You shape the tonal response of the amp using all sorts of tweaks like where to use a smaller than normal cathode bypass cap, where to allow overload of a stage vs. where to put a stage with max headroom, tweaking tone control circuits, warm bias vs. cold bias, etc. For a 'clean' tone, there's this sweet spot you can get where clean tones are somehow 'saturated', with some dynamic compression of the attack and a nice fattening of the tone from (probably) added harmonics (2nd? 2nd and 3rd?). If you've ever played through a 1960s Fender Deluxe Reverb or a Gibson GA-50, you know what I mean. Once you're used to that, playing through a solid-state guitar amp or a high-powered 'clean' tube amp (like Fender Twin Reverb) will leave you thinking they have a 'hard' tone.
A tube amp can have a 'hard' tone too. It's not the tubes, it's how you use them. Competent guitar amp designers (yes, some do exist) can get a wide palette of colorations built into an amp's 'clean' sound. Of course it has more to do with the speaker cab than anything else. But the amplifier part can lend a flavor to the tone.
Anyway, this topic is about hi-fi amps, so I'm sorry for the digression. I gave up on my beloved PP 2A3 amp a few years ago, in favor of a pair of active, DSP-corrected, self-powered speakers. However, I do enjoy putting an autoformer volume control and tube line-stage/phase splitter in front of those active speakers. In my opinion, the biggest improvement came from the autoformer volume control. I'm not saying this is 'good' or 'the best'. It's just what I'm enjoying these days.
Sorry for the blabbing...
--
Last edited:
@rongon - "autoformer volume control". Never heard of such a thing until your post. Read the explanation at (Autoformer Volume Control AF-28PC/PB – DIY Hifi Supply). Wondering why more of these dont sit at the input of modern tube amp designs; seems like they'd go hand in hand with transformer interstage coupling for an "everything 'former" amp design!
That seems a pretty sensible approach.
A tube is a basic amplifying device like a transistor or FET or MOSFET or SIT.
Except for some SITs, the triode is the most linear device devised by man so far.
But the design and execution are paramont. More so than the device used. And one cannot forget that any amplifier has to be considered as a system with the speakers it will be used with.
Far too many tube amps are “tubey”, with the kinds of distortions that were common in the old says before we had decent caps and could use silicon for support purposes.
Some amps are designed to exploit this, gobs of 2nd order that give a syrupy sweet sound. And lots of (mainly) inexpensive chinese amps that have circuits that have long been superseded. Many of them the same circuit (some different values). An embarrasement AFAIC. Gives tubes a bad name. Many of the latter could be considered as good pre-made kits, amps that are just screaming out to be tweaked or have wholescale changes made.
But there are also superb tube amplifiers out there. Many of them diy.
dave
What is the original?
And all amplfiers have their own problems — which will change depending on the speaker — and in the end it is down to what compromises the listener can most easily live without and those they can’t.
dave
I like to flip that around. The original recording has is information. We want the chain — from there to our ears — to loose as little information as possible.
dave
- Status
- Not open for further replies.