You better add diaphram distortion to the mystery, you still don't have it.
dave
true. one character told me that he can't stand the sound of ceramic drivers. on one hand it would seem implausible but then maybe ceramic acts in some sort of a weird piezo way ? you never know.
you probably wouldn't go wrong with beryllium + brick wall linear phase crossover though.
right. so where are the amps that do this like they mean it ?
I don't understand you. If a pre-amp designer has achieved his desired performance without nfb then why bother to do it? Most would not care to reduce their distortion to 0.01 if they had already achieved 0.1 % open loop.
Consider that the majority of commercial SS companies are run by marketing bods who believe that bragging about 0.0001 % is the only route to sales success, whereas the majority of tube companies are smaller ventures often run by owner-designers, and they really don't care that much about the thd numbers. (One larger company using tubes, Nagra, currently quote "less than 0.03 %" for their current commercial PL-L linestage).
As for commercial power amps that "do this like they mean it" I can not help, as I do not follow all of the commercial releases. However Aston / Langford-Smith published an updated Williamson design (Radiotron A515) that produced 0.025 % thd at 10W, and that was in the 1950s.
hm that amp is 40 years old, and it doesn't look all that complex. shouldn't we be able to do even better today ?
where are the 0.001% THD tube amps ?
Here's an old one that pulls it off (Linky). It just takes stable feedback.
Of course, a Citation II can be improved upon in several ways without increasing feedback.
However: There is no reason for a designer to get too carried away with THD numbers. First, numbers don't sell tube amps anymore, and don't fully characterize an amp's distortion performance. Second, at a certain point, it's better to focus on improving overload handling and stability with weird loads, with better transformers and good tuning. Just because an amp can perform well into a resistive load, doesn't mean it will perform well into a loudspeaker, so it is smart to prepare for all possibilities. If you are already at the limit of audibility (~0.01% THD, primarily order < 3), it's likely best to move on to other design parameters.
in SS amps distortions are reduced by several orders of magnitude using NFB.
what would it take to reduce distortion of a tube amp to say 0.001% THD ?
Tube amps can be designed exactly the same as SS amps. No, you don't have P type tubes, so a current mirror load isn't possible, but you can still get silly gain and DC coupling by cascading diffamps and whatnot. (Hint: a voltage divider keeps voltages steady between successive plates and grids. A negative bias supply helps.)
Tektronix did it for two decades of tube scopes, creating a tubological masterpiece of DC to 85MHz bandwidth in their last all-tube boatanchor.
It's perfectly possible to construct circuits with as much gain as SS (all with pentodes: diff amps, VAS, follower output, etc.), the low level stages posessing enough bandwidth that traditional dominant-pole compensation will keep the circuit stable. The only difference is, the OPT isn't DC coupled (except for Dave Berning), so you also need a dominant zero to keep it from motorboating at LF.
The ultimate product of such an amp will sound very "SS", with no more tube distortion (rich 2nd harmonic, some 3rd, nothing else).
Likewise, an SS amp can be built along the same principles as a tube amp, generating similar numbers. The difference is, there are no SS triodes, and only MOSFETs come closest in terms of fundamental transfer curve (quadratic, not 3/2 power); BJTs are always exponential.
With no feedback, it is therefore the case that SS devices have higher distortion than tubes (mind that's not counting triodes, which have internal feedback). However, they also have higher gain, so there is more NFB available for a desired amount of gain, and therefore lower distortion possible under that condition.
Tim
The ultimate product of such an amp will sound very "SS", with no more tube distortion (rich 2nd harmonic, some 3rd, nothing else).
so you're saying tube amps are made to sound tube ? that nobody bothers to reduce distortion because then they wouldn't sound tube any more ?
In speakers there is mechanical distortion due to non linearity of motors and cone breakup, compression, resolution limits due to sensitivity limits by mass inertia, diffraction, geometrical alterations of the wave launch, plus signature of the diaphragm material's multitude of small scale resonances, contributing an aftertaste mainly during their motion fade out phase. What you call plastic, papery, etc. So to resume the speaker issue grossly.
While large amounts, relatively to electronics, of low order harmonic distortion from speakers can be demonstrably difficult to pick out in controlled subjective testing, unavoidably the logical question arises: How do we listen to amps differences through them at all?
The magnitude of total harmonic distortion is one thing, and its profile is another. But for the same, say preferable profile and its origin, when lowering it by an order, what we really do is we augment the linearity of an amp by an order.
More information comes out. That is all. And you can hear it through your speaker. It may have its own set of stronger limitations, but you just lose through it a portion only from what it is fed already. So you retain a bigger piece if the pie is kept as whole as possible until transformed in the acoustic domain.
Now if you stress your speakers, don't they sound less resolute among other possible additive nuisances? OK you just raised their distortion. You worsened their linearity. Info passing ability got lower. If you could cut your speaker's THD in half at all volume levels they can play, you could hear more info. Speakers and acoustics ARE the Eastern Front, but they will unmistakeably let you listen that you changed a cartridge on your TT or that you got yourself a new CDP. How? Because they twist what they can pass to a point but they don't create it.
Now you may add 2nd harmonic for testing through a processor and try to pick it up. Well you may be amazed, it can even go up to 10% before you get your act together. But this is an additive. Its not a filter. Try to listen for things that used to be there before than to pick up new. Bingo. It masks stuff more than it sticks out itself.
THD and noise are resolution limits, no matter how low they are, or what kind of mechanics are preferable to create them and at what magnitudes they can be intrusive, cut them lower, and your info goes up.
Say you prefer 1% SE no loop feedback triode amp than some 0.001% BJT high feedback amp. Cut both THD in half due to using a more linear triode, a new cool mesh plate model with expensive metal & chemistry, superior vacuum and alignment, and better, new generation, intrinsically more linear output BJTs. DON'T touch anything else. Both amps gonna sound superior, passing more info through the same struggling speakers. Because 0.5% or 0.0005% are still a 6dB more linear dose of your preferred poison.
Now, twisting the nature of an amp's circuit or upping NFB to get its THD lower and then comparing your sound before, is not apples to apples. Its a new amp that you may prefer its linearity profile at some THD magnitude threshold. Or not.
So a quick answer to your above question for well thought out and competent sounding tube amps with some distortion and why it was left at some standard, may be: ''Because they knew better''.🙂
While large amounts, relatively to electronics, of low order harmonic distortion from speakers can be demonstrably difficult to pick out in controlled subjective testing, unavoidably the logical question arises: How do we listen to amps differences through them at all?
The magnitude of total harmonic distortion is one thing, and its profile is another. But for the same, say preferable profile and its origin, when lowering it by an order, what we really do is we augment the linearity of an amp by an order.
More information comes out. That is all. And you can hear it through your speaker. It may have its own set of stronger limitations, but you just lose through it a portion only from what it is fed already. So you retain a bigger piece if the pie is kept as whole as possible until transformed in the acoustic domain.
Now if you stress your speakers, don't they sound less resolute among other possible additive nuisances? OK you just raised their distortion. You worsened their linearity. Info passing ability got lower. If you could cut your speaker's THD in half at all volume levels they can play, you could hear more info. Speakers and acoustics ARE the Eastern Front, but they will unmistakeably let you listen that you changed a cartridge on your TT or that you got yourself a new CDP. How? Because they twist what they can pass to a point but they don't create it.
Now you may add 2nd harmonic for testing through a processor and try to pick it up. Well you may be amazed, it can even go up to 10% before you get your act together. But this is an additive. Its not a filter. Try to listen for things that used to be there before than to pick up new. Bingo. It masks stuff more than it sticks out itself.
THD and noise are resolution limits, no matter how low they are, or what kind of mechanics are preferable to create them and at what magnitudes they can be intrusive, cut them lower, and your info goes up.
Say you prefer 1% SE no loop feedback triode amp than some 0.001% BJT high feedback amp. Cut both THD in half due to using a more linear triode, a new cool mesh plate model with expensive metal & chemistry, superior vacuum and alignment, and better, new generation, intrinsically more linear output BJTs. DON'T touch anything else. Both amps gonna sound superior, passing more info through the same struggling speakers. Because 0.5% or 0.0005% are still a 6dB more linear dose of your preferred poison.
Now, twisting the nature of an amp's circuit or upping NFB to get its THD lower and then comparing your sound before, is not apples to apples. Its a new amp that you may prefer its linearity profile at some THD magnitude threshold. Or not.
So a quick answer to your above question for well thought out and competent sounding tube amps with some distortion and why it was left at some standard, may be: ''Because they knew better''.🙂
if tubes are more linear than SS then from where does the high THD come from ?
"High" THD relative to what? If you're comparing an OTL solid state design with just essssssssssssss-loads of NFB to running a low-u triode SET open loop, then, yes, the triode SET is going to have a higher THD figure. Running a solid state amp open loop would show even higher THD figures, and probably sound positively hideous. The only use for an open loop SS audio amp is the audio strip of communications xcvrs, where fidelity is neither necessary, desirable, nor legal.
is it the low gain ? the output transformer ? unwillingness to design a proper circuit because it would be too expensive to implement ? an actual craving to have an amp with tons of THD ? something else ?
All of the above, and none of the above. Design a VT amp with ridiculously low THD figures? Sure, I can do that. Add more gain stages, perhaps include a VAS using something like a 12BY7A high gain pent, and pour on the NFB: glass op-amp. However, would you actually want to listen to that thing? I highly doubt it.
THD is not the be-all and end-all of sonic excellence. If it were, we could just shut the whole forum down, go to the Big Box store, and grab any number of solid state amps. Audio nirvana, closest you can reasonably get to the ideal of "wire with gain". But that's not audio nirvana by a long shot. That's why we're here.
With the designs I did, high levels of NFB are neither necessary, nor desirable. Load on more than 12db(v) of gNFB, and you get a "solid statey" sound that just isn't very good. 20db(v) of gNFB was positively horrible. I'm sure that the THD would have measured much better, but the sound was significantly worse.
Not all distortions are created equal. You can hear that as well as see it on the o'scope screen. If you o'scope a pure sine wave, you won't notice any disturbance until you add 10% of h2, and probably not until you've added more like 20%. With higher and higher order harmonics, it takes less to cause a noticeable disturbance. If that 1.0% THD is all h2, you won't notice. 0.1% THD, if it's all higher order harmonics, will be all but unlistenable. Also, h2 and h3 are musically correlated. Higher order harmonics are dissonant. You have to pay attention to what you're getting as well as how much.
Of course, THD makes for a nice bragging point for the marketing dept. This is why Norman Crowhurst's suggestion to assign weights to the harmonic order was rejected, both then and now.
so you're saying tube amps are made to sound tube ? that nobody bothers to reduce distortion because then they wouldn't sound tube any more ?
More likely he is saying that tube amps are made to perform well, and that distortion has not been reduced because it already sounds fine.
on the other hand many people can't tolerate the sound of plastic cones even though the amount of distortion caused by the plastic cone is probably minor.
That is not THD...
dave
Your question is backwards. It should read "what permits low THD in SS amps?" The answer is feedback.What causes high THD in tube amps ?
Bear in mind that audio is a hobby. There is no objective "BEST". There is whatever floats your own personal boat.
Some hobbyists aim for high fidelity, which usually equates to low distortion measurements in amplifiers, complex multi-way speakers with fancy crossovers which compensate for speaker deficiencies, 24-bit 96 kHz sources, etc.
Others go for musical excitement, usually preferring amplifiers with some 2nd & 3rd harmonic enhancement like SETs and transformer-coupled triodes, speakers with less refined measurements but a more lively sound, vinyl records, etc. They want the sense of having the performance in the room with them, and in some degree it is being created in playback.
Regardless, listen to as wide a range of systems as you can. You'll come to know what you like and what annoys you. You'll realize there are trade-offs as in any engineering discipline. There are very dynamic systems often using tubes and horn speakers which grab your ears and make you jump. There are very accurate systems with feedback amps and planar speakers which seem to disappear. More in one area of performance usually means compromises in other areas. Listen to live acoustic music to calibrate your ears.
What is very seldom spoken of is the source recording. Like sausage, you really DON'T want to know what your music has had done to it before it went onto that CD or disk. I have seen enough recording sessions to know how badly most recordings are made. Engineers and producers have those hundreds of knobs at their disposal, and it's hard to resist using them to "sweeten" the product.
So enjoy the music. Try to forget the gear and let a performance carry you off. If you're listening to your wiring or your amp or whatever, either it's broken and distracting you or you're entirely missing the point. OTOH, if you'd rather listen to a capacitor than a symphony, I guess that's your prerogative.
Some hobbyists aim for high fidelity, which usually equates to low distortion measurements in amplifiers, complex multi-way speakers with fancy crossovers which compensate for speaker deficiencies, 24-bit 96 kHz sources, etc.
Others go for musical excitement, usually preferring amplifiers with some 2nd & 3rd harmonic enhancement like SETs and transformer-coupled triodes, speakers with less refined measurements but a more lively sound, vinyl records, etc. They want the sense of having the performance in the room with them, and in some degree it is being created in playback.
Regardless, listen to as wide a range of systems as you can. You'll come to know what you like and what annoys you. You'll realize there are trade-offs as in any engineering discipline. There are very dynamic systems often using tubes and horn speakers which grab your ears and make you jump. There are very accurate systems with feedback amps and planar speakers which seem to disappear. More in one area of performance usually means compromises in other areas. Listen to live acoustic music to calibrate your ears.
What is very seldom spoken of is the source recording. Like sausage, you really DON'T want to know what your music has had done to it before it went onto that CD or disk. I have seen enough recording sessions to know how badly most recordings are made. Engineers and producers have those hundreds of knobs at their disposal, and it's hard to resist using them to "sweeten" the product.
So enjoy the music. Try to forget the gear and let a performance carry you off. If you're listening to your wiring or your amp or whatever, either it's broken and distracting you or you're entirely missing the point. OTOH, if you'd rather listen to a capacitor than a symphony, I guess that's your prerogative.
Some hobbyists aim for high fidelity, which usually equates to low distortion measurements in amplifiers, complex multi-way speakers with fancy crossovers which compensate for speaker deficiencies, 24-bit 96 kHz sources, etc.
Please bear in mind that you are totally wrong. 😉
Hehe! Of course! 🙂
Actually, I probably fall closer to the hi-fi end of the spectrum. I like wideband, low-distortion amps and my friends joke about Henry's DC-to-blue-light amplifiers...
My favorite tube amp is the Citation V (sorry, 6550 fans, I think it outshines its big brother), or the Quad II. Both are low-distortion, high-feedback designs.
I like Bart Locanthi's JBL SS Energizers, which made 0.3% THD @ 40W sans feedback, and much cleaner with NFB. I have a DIY 200W FET amp from 10 years ago which sounds good, and I am enjoying going thru my music collection with my just-finished LM4780 chip amps. Next, I need to break out some of the master tapes of orchestra concerts I recorded for NPR years ago...
Speakers! My all-time favorite I keep coming back to is the Quad ESL 57. Despite its obvious flaws, it just does music closer to life than anything else. 2nd system is a 3-way 12" woof, 4"+5" mids, and Linaeum tweeter homebrew.
I am currently listening to iTunes on the Mac Cube's T-Amp & 2" full-range with DIY sub.
So that gives you a reference to where I'm coming from.
Actually, I probably fall closer to the hi-fi end of the spectrum. I like wideband, low-distortion amps and my friends joke about Henry's DC-to-blue-light amplifiers...
My favorite tube amp is the Citation V (sorry, 6550 fans, I think it outshines its big brother), or the Quad II. Both are low-distortion, high-feedback designs.
I like Bart Locanthi's JBL SS Energizers, which made 0.3% THD @ 40W sans feedback, and much cleaner with NFB. I have a DIY 200W FET amp from 10 years ago which sounds good, and I am enjoying going thru my music collection with my just-finished LM4780 chip amps. Next, I need to break out some of the master tapes of orchestra concerts I recorded for NPR years ago...
Speakers! My all-time favorite I keep coming back to is the Quad ESL 57. Despite its obvious flaws, it just does music closer to life than anything else. 2nd system is a 3-way 12" woof, 4"+5" mids, and Linaeum tweeter homebrew.
I am currently listening to iTunes on the Mac Cube's T-Amp & 2" full-range with DIY sub.
So that gives you a reference to where I'm coming from.
Last time I checked my Pyramid-V prototype it had -70 dB of the 2'nd only harmonic on 40W power. On lower power it went down. That means, on half of an output power it made something like 0.03 percent of THD, right?
What caused that "high THD" in the tube amp? Huh?
Later version, Pyramid -VII had 6P15P pentode LTP driver, and local feedbacks across output tubes. Should it show worse THD? I don't think so.
Did I care about THD?
No.
So, what causes high THD in tube amps? Is it the honest question, or a trolling thread start?
What caused that "high THD" in the tube amp? Huh?
Later version, Pyramid -VII had 6P15P pentode LTP driver, and local feedbacks across output tubes. Should it show worse THD? I don't think so.
Did I care about THD?
No.
So, what causes high THD in tube amps? Is it the honest question, or a trolling thread start?
So, what causes high THD in tube amps? Is it the honest question, or a trolling thread start?
Maybe he came from here: Audiophile Invasion.

So, what causes high THD in tube amps? Is it the honest question, or a trolling thread start?
Possibly a troll, but you have to admit that the general run of tube amps are not especially low distortion. I can think of many very expensive tube amps with truly embarrassing specs. Not so many of the solid state persuasion. For a newcomer, it'd be easy to think tubes are inherently higher THD than solid state, when that's far from the case.
That opening post is over 4 years old...
That's the author's join date 😉 August 3, 2009 is the post date.
- Status
- Not open for further replies.
- Home
- Amplifiers
- Tubes / Valves
- What causes high THD in tube amps ?