Why Tube Amps Sound Different (and better) Than SS Amps

[Atmasphere]

Well actually tubes color the sound to seem as if they have more details or more real music. Those differences you hear, are generated by the tube equipment. The input signal actually didn't have it to start with. The tube equipment added it in.

This is entirely 100% bogus. You can't *add* detail. Equipment can loose it, once gone it cannot be recovered.

Aside: Is your OTL circuit patented (i.e. is there a schematic I could see?) Curious...

There is a thread on DIYAudio.com called 'What tubes for a tube amp'. I placed a schematic somewhere on that thread- probably about a year ago or more.

Miles, you don't need a study to know that it's not just an opinion that tubes color the sound, or to know that the color comes from the tube equipment itself. All you need are some fairly banal measurements of the source signal and the output of the tube gear. Or if you don't believe in simple FR and distortion measurements, you can do a null and hear the difference for yourself. This is not to say that it's bad or good, just to say that it is. And I can't point you to peer-reviewed and verified studies on the subject, because no one would put in that kind of effort to confirm the obvious.

This is also not to say that there isn't some as of yet unmeasured and immeasurable ingredient in tubes that raises their fidelity beyond current measurements and reveals detail from the source material that would be obscured by any other technology with the same distortion, compression and noise. Sure...it could happen.

Tim

Something that should be pointed out here is that tubes are often in single-ended configurations, which can lead to certain distortions- which in turn impart certain colorations because the ear interprets distortion as tonality. If you run transistors in similar configurations you can get many of the same colorations. It has a lot to do with topology. Its harder to build single-ended transistor circuits as the distortion is a lot higher. But a notable example of what I am talking about is an amplifier made by Sunn for guitars, back in the early 70s. It was mostly single-ended, and so had a fairly rich sound, much richer than most transistor amps would seem capable of. The amp was single-ended right up the output stage, which was push-pull. So it made a lot of the 2nd harmonic.

You can build tube circuit single ended a lot more successfully because they are so much lower in distortion. So in a preamp it can be single ended (and zero feedback), yet the distortion can be pretty hard to measure! I do believe that many designers go for these types of distortion on purpose though.

Note tubes also have thermal modulation and distortion effects. Feedback solves a host of ills, for either technology. Personally I think the transformer dominates tube amplifier characteristics and discussions of electron flow in semiconductors vs. vacuum and plates/grids/etc. is somewhat moot. The added harmonics from tube amps often leads to a "fuller" sound people enjoy, and I am not immune to that, but it is not as accurate.

I have never heard Atma-Sphere OTL amps so cannot comment on them.

Don is spot on here. The Atma-Sphere amps, FWIW, are fully balanced and differential, so they don't have that 2nd harmonic so often associated with tubes. In fact the even orders are canceled not just at the output of the amp, but at each stage. At one time we were the only ones doing this, now there are plenty of others. I'll just put it this way- if the circuit is fully differential it will not have that 'ever-loven tube sound'. It will be lacking in coloration. This is how transistors do it BTW.

The transformer argument only applies to power amps. And SS amps can also exhibit FR variations (etc.), just typically less than transformer-coupled tube amps.

IME the best tube and solid-state preamps sound very similar. Tube preamps tend to have a little more noise but much greater headroom, something I find especially good for handling the glitches from some DACs and ticks/pops from most all phono cartridge/records. That said, tube preamps with very low distortion seem to often garner the dreaded "sounds like SS" label, as I mentioned in an earlier post talking about the differential tube preamp I designed and reviews of some of the commercial tube preamps.

The distortion characteristics of tube and SS devices, and preamps, are well-known and can be readily measured. Not by me now as I don't have access to good audio-range test equipment at my current workplace, but I have done it in the past. My old SP3a-1 had <0.001% THD at modest (~1 Vrms) output, at my measurement floor and certainly as good as SS. 2HD tends to dominate the output of tube preamps, 3HD (and related IMD products) for SS. And of course many (most?) tube preamps still have much higher output impedance than SS preamps. Even with cathode-follower outputs my ARC was pretty load-sensitive; it did not like the ~10k load of several SS amps nearly as well as the ~100k load of my tube amps and higher distortion resulted with low loads. I did try using a FET buffer for a while but my final system had amps high enough in impedance it wasn't needed.

IIRC, in the audio range tubes tend to have less flicker and popcorn noise than most SS devices, though I am not certain of that (long, long time since I tried to compare noise characteristics). Using a microwave transistor at audio is usually a very bad idea.

Of course no system is perfect, and I suspect most of the problem measuring differences lies in the difficulty of recreating real-world test scenarios (test loads, signals). It is a lot of effort, and after you do all that a bunch of people will jump in and say it does not apply to their system. They'd be technically right but by and large I think the differences are overstated. My perception of the music changes much more with my mood, time of day, weather etc. than by rotating through several components in my system. That said, load (and other) interaction among components is one of the areas I feel most promising for explaining differences, and one of the very hardest to measure and correlate.

There was a discussion of electron flow earlier in this thread.

Again Don has made good points here. Out preamps are fully balanced and differential too, and using a patented Circlotron output that is direct coupled. They have no trouble driving 600 ohms (in fact our MP-1 preamp can drive 32 ohm headphones quite nicely). You need that if you want to drive very long interconnects without HF roll-offs. But more to the point, the better we (as an industry) make a tube amp (or preamp), the better we do it with transistors (and assuming in the case of amps they are both driving a speaker that they are both happy with) the more they will sound more like each other rather than differently, because they are sounding more like music.

 

[Phelonious Ponk]

No.

I didn't figure it would, and that was my only point to Myles: As far as I can tell, there isn't much debate over the position that tube amps and preamps typically (that's a pretty important word I didn't use before) change the input signal more than solid state. There are probably exceptions. I'm not qualified to enter into the topology discussion, so I won't.

Is the difference between input signal and output signal too simple a way to gauge the transparency of an electronic component? I think it might actually be the only way. How it got altered in there I'll leave to others.

Tim

 

[microstrip]

Taken from the Spectral site - I have posted this link before in the thread. My bold. http://www.spectralaudio.com/DMA400/DMA-400%20With%20Framed%20Pix.htm It refers to exactly the same effect studied in the 90's and named "distortion thermique".

The Quest for Superior Amplification Devices

In developing new and advanced amplifier topologies, we are constantly researching available semiconductor transistors for premium devices which will be superior in Spectral high-speed analog applications, including a new generation of SMT transistors. Many of these surface mount semiconductors we could use for new designs have quicker responses and possess excellent amplification potential, yet their inclusion into carefully designed Spectral circuits would not live up to our expectations. Many were found to damage resolution and clarity. We believed that perhaps thermal stress and charge settling might be responsible since electronic evolution has favored smaller semiconductors whose junctions might respond to transient heating and produce thermally active errors. To explore this possibility we developed new sophisticated sampling tests at Spectral using simulated music waveforms to screen promising new semiconductor devices. A typical evaluation would create isolated transient events at semiconductor junctions so that heating would be similar to that from amplifying music and correction responses expected from loudspeaker loads. Such events require responses that demand brief energy bursts, which produce quick temperature surges. Then junction voltages, which should be constant might change to memorialize the event and to initiate error responses or thermal tails. Since most integrated circuit op-amp amplifiers have similar difficulties, the test methodology was made practical from working with these consumer devices. Performance limits became quantified to hearing acuity as well as human perceptual ability. These new test methodologies have now consistently verified the Spectral philosophy of inherently fast - low stress amplification to achieve ultimate resolution and accuracy. When instantaneous waveform accuracy is achieved at parts-per-million, the listening experience can become most detailed, transparent and involving. Our search for superior amplification devices led us to recent discrete semiconductors intended for premium television and computer displays. Their hefty construction and advanced manufacturing processes create a substantial and very fast device that is free of thermal tail and memorialization of error issues. These new devices can operate over a wide range of voltages and currents while being capable of excellent gain linearity, quickness and ability to amplify very high frequencies. The DMA-400 achieves its unmatched clarity and resolution from a combination of these breakthrough performances.

 

[microstrip]

(...) End result, for example, i highly respect Nelson Pass, and he advocates a transfer function that is not perfectly linear, he advocates a bit of even order (mostly second order) harmonic distortion. No magic as seen from where I perch. (...)

Tom,

You are still and again referring to the old memories of the old days when everything was simple ...

From Stereophile : Fortunately, the Pass XA60.5's distortion is predominantly the subjectively innocuous third-harmonic in nature at both moderate (fig.8) and high (fig.9) powers,
Fig. 9 Pass Labs XA60.5, spectrum of 50Hz sinewave, DC–1kHz, at 120W into 4 ohms (linear frequency scale).

 

[Phelonious Ponk]

Tom,

You are still and again referring to the old memories of the old days when everything was simple ...

From Stereophile : Fortunately, the Pass XA60.5's distortion is predominantly the subjectively innocuous third-harmonic in nature at both moderate (fig.8) and high (fig.9) powers,
Fig. 9 Pass Labs XA60.5, spectrum of 50Hz sinewave, DC–1kHz, at 120W into 4 ohms (linear frequency scale).

And I think you long for a world that is so complex that it is impossible to conclude anything that could contradict what you want to believe in. In the meantime, while we wait for worlds to come to being, the debate is robust and the music is beautiful. Enjoy.

Tim

 

[Gregadd]

Greg, you listened to a DHT heated tube amp recently and kept it to yourself?

Not recently trying to get power, linearity,and low distortion makes the cost prohibitive for most audiophiles.
Hint I traveled cross country to heart it.

 

[MylesBAstor]

Miles, you don't need a study to know that it's not just an opinion that tubes color the sound, or to know that the color comes from the tube equipment itself. All you need are some fairly banal measurements of the source signal and the output of the tube gear. Or if you don't believe in simple FR and distortion measurements, you can do a null and hear the difference for yourself. This is not to say that it's bad or good, just to say that it is. And I can't point you to peer-reviewed and verified studies on the subject, because no one would put in that kind of effort to confirm the obvious.


This is also not to say that there isn't some as of yet unmeasured and immeasurable ingredient in tubes that raises their fidelity beyond current measurements and reveals detail from the source material that would be obscured by any other technology with the same distortion, compression and noise. Sure...it could happen.

Tim

Can't have it both ways Tim. You can't condemn audiophiles and manufacturers for not providing enough facts/measurements and then simply dismiss something as being obvious. Not the way science is done. If every hypotheses was right, we'd have cured every disease by now.

And the measurements are no where as simple as you would suggest. Nor is as gear without it's own set of issues that even SS designers own up to. Read some of the lengths that respected designers like Johnson and Linkwitz go to eg. Settling times and tone bursts, to evaluate the performance of audio equipment and speakers.

Truth be told, the performance of the best SS and tubes are more similar than dissimilar. Yet each maintains it's own signature. Why? Because in addition to trying to accurately reproduce the sound of the recording, designers are also trying to emulate the best parts of the other technology. Do you think that even someone like Dan D'Agostino doesn't respect the sound of tube gear?

 

[MylesBAstor]

Not recently trying to get power, linearity,and low distortion makes the cost prohibitive for most audiophiles.
Hint I traveled cross country to heart it.

So you were holding out on us!

 

[esldude]

Tim perhaps you can supply us with the reference (s) to the peer reviewed scientific paper (s). Otherwise it's just speculation and conjecture that has no more validity than anything else posted. Hell ESL couldn't even explain why OTL amps (and I might add other topologies-not to mention the original recording) didn't fit into his *theory*. That ambience isn't created by tubes; it's there on the master tape not to mention the hard drive. No it's lost during the production and recovery process.

At least Ralph provided some scientific studies.

Didn't realize I needed to explain every known topology to comment on the subject of those in this thread. Or to comment upon those I had experimented with and whose results had nothing to do with an OTL. I have only heard 3 OTL's and two of them were DIY units. My comment was depending on the load they sounded anywhere from clean and quick to smoky and colored.

As DonH50 later comments we know it is possible to add things to the original signal in the form of harmonic distortion and intermodulation distortion. Things that weren't in the original signal. If those reach a high enough level, and in your general xfmr coupled tube amp they can, then what you hear will be altered. I haven't shown those exact effects create space and added apparent detail. But that is what it sounded like to me. As that is at least a common description of tube amps superiority it fits they might be heard that way due to what they add to the signal. As to whether or not they add to the signal more than good SS amps, well that is pretty much an open and shut case. They do, it is regularly measured.

Are there some other tube topologies and even this topology done to higher refinement that don't do that? There might be or are, but that isn't really the topic. Those are far and few in between. You further have the history of ARC gear that as it improved and developed specs closer to SS equipment, a complaint was it sounded more like SS equipment.

And while one could talk about pre-amps etc., I haven't found them to be so noticeable as tube amps. They also are often quite low in distortion if used with the right connecting equipment. Others have made the claim they too give you that tube sound like an amp. That is their idea to explain, not mine. It hasn't been my experience nor my claim.

I did find a good SS to not be heard when inserted into the signal chain or removed. The same couldn't be said of the xfmr coupled tube amp I tried it with nor an SET I have used in the same fashion. As both were similar to the generally available examples it is a jump to conclude that covers all of the genre. But not a very large one considered how the gear measures, and that they have a signature sound. The only controversial aspect would be my opinion the good SS amps are either transparent or very close while others though valued for their sound are corrupting the input signal to achieve it.

 

[Atmasphere]

As DonH50 later comments we know it is possible to add things to the original signal in the form of harmonic distortion and intermodulation distortion. Things that weren't in the original signal. If those reach a high enough level, and in your general xfmr coupled tube amp they can, then what you hear will be altered. I haven't shown those exact effects create space and added apparent detail. But that is what it sounded like to me. As that is at least a common description of tube amps superiority it fits they might be heard that way due to what they add to the signal. As to whether or not they add to the signal more than good SS amps, well that is pretty much an open and shut case. They do, it is regularly measured.

Are there some other tube topologies and even this topology that don't do that? There might be or are, but that isn't really the topic. Those are far and few in between. You further have the history of ARC gear that as it improved and developed specs closer to SS equipment, a complaint was it sounded more like SS equipment.

This might have something to do with negative feedback, which is known to add brightness to any circuit.

I am very used to customers conflating detail and brightness. They are not the same, although it is possible to seem to be more detailed because something is bright. But you can have a relaxed presentation that is detailed as well. IMO/IME if you have detail and a relaxed presentation at the same time, now you are getting somewhere as the detail does not arise out of brightness. And it may be because there is less distortion, which by itself can mask lower level signals (detail).

 

[microstrip]

And I think you long for a world that is so complex that it is impossible to conclude anything that could contradict what you want to believe in. (...)

Tim

Tim,

Although I can not understand you sentence, I am suspecting that this is related to our definition of "magic" ... Something so complex that simple interpretations all fail and succeed simultaneously.

But yes, the interest comes from the new aspects arriving to the debate.

 

[Phelonious Ponk]

Can't have it both ways Tim. You can't condemn audiophiles and manufacturers for not providing enough facts/measurements and then simply dismiss something as being obvious. Not the way science is done. If every hypotheses was right, we'd have cured every disease by now.

And the measurements are no where as simple as you would suggest. Nor is as gear without it's own set of issues that even SS designers own up to. Read some of the lengths that respected designers like Johnson and Linkwitz go to eg. Settling times and tone bursts, to evaluate the performance of audio equipment and speakers.

Truth be told, the performance of the best SS and tubes are more similar than dissimilar. Yet each maintains it's own signature. Why? Because in addition to trying to accurately reproduce the sound of the recording, designers are also trying to emulate the best parts of the other technology. Do you think that even someone like Dan D'Agostino doesn't respect the sound of tube gear?

I'm not suggesting that all the measurements are simple, just that the ones needed to reveal the sources of the audible signature of the overwhelming majority of tube amps are. Are there exceptions? Probably. I thought I already said that.

 

[MylesBAstor]

From a well known designer of both equipment and transformers who prefers to stay anonymous:

The ringing will always be a factor on any transformer not properly matched or terminated. But the better the winding structure and the better the understanding of the operating conditions will provide a very wide band transformer with negligible ring at a very ultrasonic frequency. A good output transformer can offer very fast rise times and low in band distortion. (my note: of course this depends on the proper interweaving of primaries and secondaries.)

 

[Phelonious Ponk]

Didn't realize I needed to explain every known topology to comment on the subject of those in this thread. Or to comment upon those I had experimented with and whose results had nothing to do with an OTL. I have only heard 3 OTL's and two of them were DIY units. My comment was depending on the load they sounded anywhere from clean and quick to smoky and colored.

As DonH50 later comments we know it is possible to add things to the original signal in the form of harmonic distortion and intermodulation distortion. Things that weren't in the original signal. If those reach a high enough level, and in your general xfmr coupled tube amp they can, then what you hear will be altered. I haven't shown those exact effects create space and added apparent detail. But that is what it sounded like to me. As that is at least a common description of tube amps superiority it fits they might be heard that way due to what they add to the signal. As to whether or not they add to the signal more than good SS amps, well that is pretty much an open and shut case. They do, it is regularly measured.

Are there some other tube topologies and even this topology done to higher refinement that don't do that? There might be or are, but that isn't really the topic. Those are far and few in between. You further have the history of ARC gear that as it improved and developed specs closer to SS equipment, a complaint was it sounded more like SS equipment.

And while one could talk about pre-amps etc., I haven't found them to be so noticeable as tube amps. They also are often quite low in distortion if used with the right connecting equipment. Others have made the claim they too give you that tube sound like an amp. That is their idea to explain, not mine. It hasn't been my experience nor my claim.

I did find a good SS to not be heard when inserted into the signal chain or removed. The same couldn't be said of the xfmr coupled tube amp I tried it with nor an SET I have used in the same fashion. As both were similar to the generally available examples it is a jump to conclude that covers all of the genre. But not a very large one considered how the gear measures, and that they have a signature sound. The only controversial aspect would be my opinion the good SS amps are either transparent or very close while others though valued for their sound are corrupting the input signal to achieve it.

I don't know much, but I know this: When you push tubes to the point where their distortion and compression are no longer an effect we can debate, but obvious - in electric guitar amps - preamp tubes and power tubes distort and compress very differently. That hasn't kept manufacturers from attempting to sell hybrid amps that claim to get the classic tube sound from preamp tubes into SS power amps, but they've never succeeded. They sound different. Pushed hard, preamp tubes are much grainier, much more saturated. And they compress faster and more severely. I wouldn't think you would want any of that going on, no matter how subtle, in a hifi amp. Power amp tubes? The guitarist in me completely understands why just a little of that sound could be very attractive. One of the best guitar sounds in the world is the "clean tone" of a Fender Twin or Super Reverb. That's loads of headroom, very low gain (and distortion) from the pre and a bunch of big 6L6 tubes cruising along, never breaking sweat. It is warm, lush, fat and crisp all at once. The distortion doesn't sound like distortion at all. It just sounds like a great guitar tone. Yeah, I'm a tube guy.

Tim

 

[microstrip]

For those really interested in transient thermal distortion and wanting to go through almost 40 pages of text in french, graphs and oscilloscopes screens filled with measurements the 1984 articles published by l'Audiophile, although still under copyright, can be found in the net.

The titles are
La distorsion thermique - elle existe, je l'ai rencontree (Héphaïtos)
La distorsion thermique- tube contre transistor (Héphaïtos)

 

[Al M.]

For those really interested in transient thermal distortion and wanting to go through almost 40 pages of text in french, graphs and oscilloscopes screens filled with measurements the 1984 articles published by l'Audiophile, although still under copyright, can be found in the net.

The titles are
La distorsion thermique - elle existe, je l'ai rencontree (Héphaïtos)
La distorsion thermique- tube contre transistor (Héphaïtos)

My French isn't good enough to decide from the abstract which one has less thermal distortion, tube or transistor. Can you please help?
Thanks.

 

[Ethan Winer]

1. a good amplifier should not have a 'sound' --it should be a straight wire with gain. No particular signature, neutral ... That's a few of the major points off the top of my head.

Exactly Mark. The differences between tubes and solid state are well known, understood fully, and simple to quantify using the standard metrics of fidelity. In every case competent SS has higher fidelity than tubes, even if some people prefer the sound of tubes (which is fine). So why are there still arguments? Tim said it very well:

Because the measured and well-understood conclusions are not something that some tube lovers are willing to accept.

--Ethan

 

[MylesBAstor]

Exactly Mark. The differences between tubes and solid state are well known, understood fully, and simple to quantify using the standard metrics of fidelity. In every case competent SS has higher fidelity than tubes, even if some people prefer the sound of tubes (which is fine). So why are there still arguments? Tim said it very well:




--Ethan

Well I guess we can throw away two or three decades or more of recordings made with tube electronics since they're garbage.

But we all know you only listen to two CDs since your system is so revealing that those are the only two recordings you can stand to listen to. So that's a moot point.

 

[thedudeabides]

I'm almost afraid to ask Mr. Winer but ask I must.

What are your parameters for determining "higher fidelity" in amplifiers or other audio hardware?

And what do the words "competent" and "standard metrics" mean?

GG

 

[microstrip]

My French isn't good enough to decide from the abstract which one has less thermal distortion, tube or transistor. Can you please help?
Thanks.

Al M.

Tubes are shown to have much less thermal distortion than transistors in general. However, once the proper semiconductor is selected and the operation point properly established there is not a winner anymore - both can have excellent results! The article describes the tube/transistor war, but the author is a transistor "partisant": as he says once you find a problem, you must find a solution, not just stop because of it.