Some tube amps have more high frequency roll off than ss amps. SET amps have harmonic signature not found in p-p amps. Those are common audible differences people talk about when describing "tube sound". The high frequency roll off can be simulated with ss amp by the use of equalizer. Differences in Amp Sound: What’s the Truth? | AudioXpress
And vice-versa, of course. Ditto solid state.
But as SY said earlier, there are so many contradictory descriptions of tube sound that it boils down to nothing.
Unless you ask most people these days. They are likely to say "warm and mellow" or some such. At least the marketing folks will say that. Marketing has a big influence on perceptions.
From link said:
This seems like a pretty reasonable conclusion. I think amplifier choices are rooted in forming a preference for how and when the system distorts. Speaker interaction plays a large role in this and room interaction can either make it more or less detectable due to comb filtering, decay, and room modes. I think the most effective money I've spent in this hobby has been on speakers and acoustic treatments. I love building amps though. 🙂
Mmmmm ...
Firstly:
Glad to make your acquaintance, Mr Cheever (Dr Cheever by this time?)! Your 79 page thesis of 1989 has been in one of my files for more than a decade. Thanks; excellent reading! To be recommended.
Secondly:
You are wrong, SY! 🙂 See, especially these days there are so many mediocre designs of both topologies on the market that it all depends what has been listened to! ( 😉 Sorry, SY - couldn't resist!)
To topic:
There should be no audible difference between properly designed examples of either topology (says he with no immodesty, simply indicating that basics are so often not being respected). But yes, I can agree where designs are moderately in order, it is mostly audible 2nd and 3rd order harmonics present in tube amplifiers (and nothing else) that make them sound 'pleasant' (over-real?). I recall reading that investigation decades ago indicated that a few % of second harmonics added 'fullness' to the music - it is after all everything an octave higher; not unmelodious and not much imagination required to agree that that will make a lot of music sound fuller/warmer - a description often used. Also it was apparently found (can others confirm?) that a small degree of 3rd harmonic products made strings and brass in particular sound 'crisper' - again no stridency (er - mostly). (As Douglas Self said: Perhaps one needs to add a 'niceness' control to ss amplifiers.)
But often over-feedbacked ss amplifiers are harmonic multipliers, and that is where listener fatigue comes in [a separate subject on its own. (Sorry, Mr Putzey is wrong there.)]
Decent output transformers have distortion lower than the tubes feeding them, and they are not bandwidth-challenged; a 20 Hz - 50 kHz spec is not difficult to get. Only in some tube amplifiers .... 😱 Also I must respectfully disagree with over-compensation necessarily bringing phase shift/frequency compensation into the mix in tube amplifiers. In many ss amplifiers 'C-dominant' starts cutting well into the audio region - not to speak of integrated circuits.
Unfortunately the classic tube amplifiers of yesteryear were mostly limited to below 20W, which might make honest comparison difficult.
PS: I have not checked, but much of this might be in Mr Cheever's thesis. I would repeat: It is well worth studying.
Edit: PS 2: I would particularly accentuate Mr Cheevers' middle paragraph in his post #4.
Firstly:
Glad to make your acquaintance, Mr Cheever (Dr Cheever by this time?)! Your 79 page thesis of 1989 has been in one of my files for more than a decade. Thanks; excellent reading! To be recommended.
Secondly:
You are wrong, SY! 🙂 See, especially these days there are so many mediocre designs of both topologies on the market that it all depends what has been listened to! ( 😉 Sorry, SY - couldn't resist!)
To topic:
There should be no audible difference between properly designed examples of either topology (says he with no immodesty, simply indicating that basics are so often not being respected). But yes, I can agree where designs are moderately in order, it is mostly audible 2nd and 3rd order harmonics present in tube amplifiers (and nothing else) that make them sound 'pleasant' (over-real?). I recall reading that investigation decades ago indicated that a few % of second harmonics added 'fullness' to the music - it is after all everything an octave higher; not unmelodious and not much imagination required to agree that that will make a lot of music sound fuller/warmer - a description often used. Also it was apparently found (can others confirm?) that a small degree of 3rd harmonic products made strings and brass in particular sound 'crisper' - again no stridency (er - mostly). (As Douglas Self said: Perhaps one needs to add a 'niceness' control to ss amplifiers.)
But often over-feedbacked ss amplifiers are harmonic multipliers, and that is where listener fatigue comes in [a separate subject on its own. (Sorry, Mr Putzey is wrong there.)]
Decent output transformers have distortion lower than the tubes feeding them, and they are not bandwidth-challenged; a 20 Hz - 50 kHz spec is not difficult to get. Only in some tube amplifiers .... 😱 Also I must respectfully disagree with over-compensation necessarily bringing phase shift/frequency compensation into the mix in tube amplifiers. In many ss amplifiers 'C-dominant' starts cutting well into the audio region - not to speak of integrated circuits.
Unfortunately the classic tube amplifiers of yesteryear were mostly limited to below 20W, which might make honest comparison difficult.
PS: I have not checked, but much of this might be in Mr Cheever's thesis. I would repeat: It is well worth studying.
Edit: PS 2: I would particularly accentuate Mr Cheevers' middle paragraph in his post #4.
Last edited:
I should say at the outset that I have nothing to contribute to this discussion in terms of measurements. However, the discussion does interest me for a few reasons. I've been modifying, designing and building hi-fi as a hobby for a good 40 years, during half of which time I was a busy freelance professional musician playing classical, jazz and rock/soul/blues. I was conservatoire trained, and have a good ear and a very good musical memory.
So my goal throughout this period was quite simple - to create a sound system that brought me closest to the actual sound of the instruments I heard every night when I was playing. Most frequently these were vocals, drumkit, piano, and brass and wind instruments, plus all the orchestral instruments when I was doing classical work. These acoustic sounds are imprinted and engraved in my head. I can recognise their timbres just as easily as we can, with familiarity, recognise the distinctive voices of people we know.
So my progress through years of design was to remove, through a process of rejection and replacement:
1. Anything that veiled this acoustic sound, i.e. masked details and reduced transparency
2. Anything that distorted the timbre of acoustic instruments away from the characteristic timbres I was hearing every night I played.
This process led me to singe ended stages and DHTs in all stages because they sounded to me more transparent than an equivalent indirectly heated triode (similar mu and characteristics). It led me to eliminating as many capacitors as possible from the signal path (coupling, cathode bypasses) and using only polypropylenes in the PSU. It led me to coupling stages with high quality transformers. And it led me to filament bias for all stages, so in my amp I ended up with just 26 into 4P1L into PSE 4P1L. This all produced the closest approximation to the acoustic timbre and tone of the real instruments I played with. Timbre and tone were the single most important differentiators between what was developed and what was rejected.
So as I said at the outset, I can't produce measurements, but I can produce results in terms of actual working amplifiers. As we all know too well, we can't send the sounds of these over the internet, and I'm not asking anyone to take my word for anything. What I would offer in terms of opinions is that:
1. You can't design anything without mathematics and engineering theory, but you can develop designs by listening to them
2. I reject all the common adjectives about "tube sound" such as warm, mellow etc etc. Tubes should have no added sound. But this doesn't mean that tubes don't matter, because inferior tubes can colour sound and reduce transparency and faithfulness of timbre.
3. Most of the work in designing for clarity and timbre is in the circuit itself.
My approach has been to reduce audible distortions, using a common reference of familiar acoustic sounds. It's not the same as reducing measured distortion but it can produce useful results as long as A B comparisons are done as rigorously as possible. Not double blind (useful but impractical) but always respecting the nul hypothesis that changes can be inaudible or barely audible, that they should be repeated as much as possible to see if they consistently produce the same results, and that change can be positive or negative and should not be hyped up for any personal reasons.
The two processes aren't mutually exclusive, and for most of us I'd venture to suggest they are complimentary. In theory they should actually produce the same results.
So my goal throughout this period was quite simple - to create a sound system that brought me closest to the actual sound of the instruments I heard every night when I was playing. Most frequently these were vocals, drumkit, piano, and brass and wind instruments, plus all the orchestral instruments when I was doing classical work. These acoustic sounds are imprinted and engraved in my head. I can recognise their timbres just as easily as we can, with familiarity, recognise the distinctive voices of people we know.
So my progress through years of design was to remove, through a process of rejection and replacement:
1. Anything that veiled this acoustic sound, i.e. masked details and reduced transparency
2. Anything that distorted the timbre of acoustic instruments away from the characteristic timbres I was hearing every night I played.
This process led me to singe ended stages and DHTs in all stages because they sounded to me more transparent than an equivalent indirectly heated triode (similar mu and characteristics). It led me to eliminating as many capacitors as possible from the signal path (coupling, cathode bypasses) and using only polypropylenes in the PSU. It led me to coupling stages with high quality transformers. And it led me to filament bias for all stages, so in my amp I ended up with just 26 into 4P1L into PSE 4P1L. This all produced the closest approximation to the acoustic timbre and tone of the real instruments I played with. Timbre and tone were the single most important differentiators between what was developed and what was rejected.
So as I said at the outset, I can't produce measurements, but I can produce results in terms of actual working amplifiers. As we all know too well, we can't send the sounds of these over the internet, and I'm not asking anyone to take my word for anything. What I would offer in terms of opinions is that:
1. You can't design anything without mathematics and engineering theory, but you can develop designs by listening to them
2. I reject all the common adjectives about "tube sound" such as warm, mellow etc etc. Tubes should have no added sound. But this doesn't mean that tubes don't matter, because inferior tubes can colour sound and reduce transparency and faithfulness of timbre.
3. Most of the work in designing for clarity and timbre is in the circuit itself.
My approach has been to reduce audible distortions, using a common reference of familiar acoustic sounds. It's not the same as reducing measured distortion but it can produce useful results as long as A B comparisons are done as rigorously as possible. Not double blind (useful but impractical) but always respecting the nul hypothesis that changes can be inaudible or barely audible, that they should be repeated as much as possible to see if they consistently produce the same results, and that change can be positive or negative and should not be hyped up for any personal reasons.
The two processes aren't mutually exclusive, and for most of us I'd venture to suggest they are complimentary. In theory they should actually produce the same results.
Last edited:
And indeed, when amplifiers are designed to merely make a small signal larger (i.e., low distortion and noise, flat frequency response in the audio band, low output impedance, swift and uneventful overload recovery), they will sound alike in ears-only testing. For any desired harmonic or frequency response coloration, again one can design a tube or solid state device to produce that effect. It may or may not be more convenient to choose one or the other, but that's engineering.
The difficult effect to replicate is microphonics, but even that could be readily done today with inexpensive transducers and some DSP, if that were the goal.
We discount the "coolness" factor of tubes and the effect on the user's mind at our peril.
Its difficult (but not impossible) to design a one or two stage transistor amp with only local feedback. This, in my experience, is the recipe to achieve what andyjevans has just described - transparency and Timbre.
The more we stray away from this formula the more we lose those essential qualities - and its very difficult to achieve in a multistage SS amp - but relatively easy in a tube amp. it is only really the field of DIY which affords us the luxury of building these exotic beasts as it is still very common for commercial tube amps to be a of a Williamson type design.
As for the lack of importance of OT to the mix, my experience is that things happening out at 50khz are very audible, I prefer a gentle role of starting around 30khz with no peaking. It maybe that ultra high bandwidth SS amps have something to offer, but most people I know prefer some degree of bandwidth limitation and that is somewhat automatic with OT.
Just my thoughts.
Shoog
The more we stray away from this formula the more we lose those essential qualities - and its very difficult to achieve in a multistage SS amp - but relatively easy in a tube amp. it is only really the field of DIY which affords us the luxury of building these exotic beasts as it is still very common for commercial tube amps to be a of a Williamson type design.
As for the lack of importance of OT to the mix, my experience is that things happening out at 50khz are very audible, I prefer a gentle role of starting around 30khz with no peaking. It maybe that ultra high bandwidth SS amps have something to offer, but most people I know prefer some degree of bandwidth limitation and that is somewhat automatic with OT.
Just my thoughts.
Shoog
Last edited:
I did hear an astonishingly good SIT amp with one of those obsolete Sony devices - as good as a SET with tubes. Just one stage. As Nelson Pass says of his new SIT amp:
"Do simple SIT amplifiers sound the same as single-ended Triodes (SETs)? That would be a matter of opinion. Without an output transformer with its significant limitations, the SIT has a bandwidth and distortion edge. On the other hand, some people like the sound of their transformers.
Me, I like to listen to the sound of different amplifiers. I don't see a practical point in actually duplicating a SET amplifier. The SIT gives me an opportunity to listen to the best that a single gain device can do. Surrounded only by a few well-chosen passive parts, this transistor speaks with a single voice.
It doesn't sound quite like anything else."
Rigorous A B comparison isn't rigorous unless it's double blind. It's not easy to set up DBT but once you get over the learning curve, it's not so impractical for comparing amps.
Sighted test doesn't produce meaningful results in this context. We can't control bias.
Sure we can. Many amps have a bias control and a bias meter or test point. Very easy to control. 😛
I quite understand that bias exists - I'm a psychologist and I'm used to testing in other contexts, so I freely admit it is an issue, even when you try to allow for it. But are such results completely meaningless? Well, I don't think so, but confidence in results is nowhere near what would be needed for significant statistics. But does it need to be? In making a choice between A and B we may be satisfied with 60% or 70% reliability. That's roughly what you'd expect from personality tests, and they would not be so widely used if they were not useful despite their flawed reliability. I'm not making any strong claims here - just that sighted A B comparisons can be "useful" if they are better than random choice.
Last edited:
You mean people only do things which are useful? You must know a different set of people from me.
Andy I'm definitely with you on the timbre issue. That what I listen for, and want it to sound like real instruments. If that involves something other than perfect THD or whatever, so be it. I want fidelity to the sound of acoustic instruments and spaces. Usually that involves fidelity to the original signal, but there are fudge factors there (what's audible, and how).
In general I find DHT SET to a good job at bringing me the timbre and texture of instruments, but I have run a few SS amps that do as well, or even better. Gary Pimm's Tabor is one example, some of the Keneda and Hiraga class-A amps are another.
Typical Solid State adds an edgy coloration that sounds fake to me. Some P-P valve amps add an overly mellow glow that also sounds fake. Fun, but fake. 🙂
In general I find DHT SET to a good job at bringing me the timbre and texture of instruments, but I have run a few SS amps that do as well, or even better. Gary Pimm's Tabor is one example, some of the Keneda and Hiraga class-A amps are another.
Typical Solid State adds an edgy coloration that sounds fake to me. Some P-P valve amps add an overly mellow glow that also sounds fake. Fun, but fake. 🙂
The only "tube sound" I'm aware of is that they sound something like "PAFF!", especially if you drop one on a concrete floor. Transistors, on the other hand, sound like "Clink!", especially the TO-3 types.
It's all audiophoolery. Tubes (and BJTs and FETs and chips) don't have any "sound". Circuits, speeks, program material and how the mix was done -- these have "sound". Any time you see any such claims, better get a tighter hold on your wallet because someone's trying to put one over to justify an inferior, distorting design.
We saw this with the "MOSFET sound" that was the result of trying to use the same SEPP topologies originated for BJTs applied to MOSFETs. Complimentary NPN/PNP types are a good deal more complimentary than are N-Channel and P-Channel pairs. The MOSFET "complimentaries" have a wicked x-over in Class AB. This distortion was marketed as something cool: i.e. "The MOSFET sound". YUCK!
It's easier to design a good sounding hollow state amp, but you can also screw it up by selecting bad loadlines, inappropriate tubes, and sloppy open loop designs in general, then try to sweep your open loop mistakes under the carpet with gobs of NFB.
It's a bit of a challenge when it comes to solid state, but I've dunnit: made very good sounding transistor amps, and there is no reason for SS to be sounding as bad as it does.
And a tick from me too. Once you have got a system to work right, on at least one occasion, you have no trouble from then on distinguishing conventional hifi quality, no matter how expensive or "perfectly measuring", from something that creates subjective fidelity.
A key factor is that the majority of audio systems make a mess of the low level detail when you attempt to produce near realistic volume levels. Normal THD measurements don't pick this up - but the human hearing system has no trouble picking up what's going on, and the result is instantly dismissed as being "fake".
Yes, if you are searching for objective data.
The problem with subjective comparison is that the listeners can perceive a difference even when there is no. For example, speaker cable comparison, the listeners noticed wider sound stage, more details and ...etc when they were told which cable is being used (expensive one obviously). They all agreed with the results and went home happy. What they weren't told was that the person doing the cable switching never physically switched the cables but told the listeners that he switched it. All along, the listeners were subjected to same cheap lamp cord speaker cable.
The outcome is subjected to whatever the external influence the listeners are faced with. You can change the test result easily. What usefulness does that provide in finding out certain qualities of the product?
- Status
- Not open for further replies.