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

[DonH50]

Can you and Ethan get a room? One with a ring? Actually, not sure Ethan did anything to start this one...

Technical note: IMD is actually larger in magnitude than HD if you work out the math, so not only is it more objectionable because it is non-harmonic in frequency, but it is also louder than harmonic distortion.

Transformer hysteresis is another source of tube amp distortion but I am not sure how relevant/significant it is in actual designs. Nor am I sure how much transistor thermal modulation contributes to audible distortion. I can attest transistor thermal modulation is a huge pain when designing data converters (ADCs and DACs), especially more conventional designs, and it adds ugly settling tails to amplifiers. Good thermal design can mitigate the effects for LF signals but not for radar systems (since radars were mentioned earlier figured I could get away with it again).

 

[thedudeabides]

esldude gave you the short answer, and these give a much more complete explanation of how fidelity is defined, and what amounts typically matter enough to care about:

Audiophoolery
AES Audio Myths video
AES Damn Lies video

--Ethan

Thank you. I'll scan the material but it's likely stuff I've heard / read before.

The "short" answer is probably sufficient but in the end after all is said and done, I will continue to use my ears (as I have for the past 35 years or so) as the final arbiter of value and sound quality.

 

[Phelonious Ponk]

That's for sure. I would place the best of the modelling guitar amps at about 60% of the abilities of the amp they are modelling. They just don't have the guts!



Some people here have been talking about the magic of tubes. Of course there is no magic. But the idea that we have had everything sorted out for the last 40 years is bogus too. The problem here has to do with understanding how we hear things. We know the mechanics of the ear. Not so much about how the ear is able to convert sounds to neural signals. Even less about how the brain processes those signals on arrival!

But we can say with certainty that the more we know about how the ear/brain system works, the better we can get things to sound. In effect, using engineering to solve the problem. For the most part, right now the engineering has been used to make the equipment look good on paper. The problem is that the paper does not jive with the ear/brain system's perceptual rules. So we can hear things that are hard to measure (trace amounts of odd ordered content) and there are other things the ear just does not seem to care about at all, but are easy to measure, like the phase of a sine wave.

IOW: the 'magic', the mystery is still quite literally in our heads. We're not going to be able to undo millions of years of evolution. Anyone here have any idea why the ear is so sensitive at upper midrange frequencies for example? This is far above human vocal tones... This is an example of what I mean by mystery (if you want the answer I can answer it BTW).

But if you *understand* that the ear is so sensitive in that area, and that it uses higher ordered harmonics to tell how loud a sound is, wouldn't you want to avoid making distortion in that region??

I'd bet the farm I could dial in my tone on the modeler, dial in a Deluxe with fresh tubes and biasing, and neither one of you guys could consistently tell me which was which. Two real vintage Deluxes can easily sound farther apart. and that comes from someone who has owned 3 of them. And guts? Guts is definitely not the problem, if there is one. I could make you void your bowels with this amp. I think I might be talking to a couple of guys who fooled with a modeling amp or two in the store, concluded what they expected to conclude and moved on. They are like anything else. You have to learn how to use them.

"Don't have the guts..." Now that's funny.

Tim

 

[Atmasphere]

What you miss is that the magnitude of distortion matters much more than its specific makeup. Nobody will hear 0.01 percent distortion whether it's mostly odd or mostly even, and even budget amplifiers often achieve distortion that low. Further, the notion that odd-order harmonics are more objectionable is simply false. It depends on the context, and how far away (in frequency) the added harmonics are compared to the fundamental due to the masking effect. A clarinet produces mostly odd overtones, so adding a 2nd harmonic will change its timbre more noticeably than adding a 3rd harmonic which is already present anyway. Further, even or odd THD is much less important than IM distortion, which is vastly more objectionable than any THD. Since THD and IMD are always present together, IMD is the most irritating type of distortion.

--Ethan

In this you are wrong, operating out of ideas over 40 years old. What you don't seem to understand is one of the more basic human hearing perceptual rules: how we detect sound pressure. We can hear odd ordered harmonic content that is so low as to be unmeasurable. This is because the ear/brain system uses these harmonics to sort out how loud a sound is. So the ear is very sensitive in this area. How much, we are still finding out as research into human physiology progresses.

IMD is quite audible, but Don here seems to have the better grasp:

Technical note: IMD is actually larger in magnitude than HD if you work out the math, so not only is it more objectionable because it is non-harmonic in frequency, but it is also louder than harmonic distortion.

 

[DonH50]

Thanks. Curious: How low is the odd harmonic content we can hear? "Unmeasurable" is awfully durn small considering we can measure below -120 dBFS. I do not know, just trying to understand how well we really hear. Haven't had a physiology class since college, and psychoacoustics is not something I know a lot about.

 

[Atmasphere]

If the amp has 0.001% THD the odd ordered and higher orders will be audible. They will manifest as brightness and harshness.

 

[Atmasphere]

I'd bet the farm I could dial in my tone on the modeler, dial in a Deluxe with fresh tubes and biasing, and neither one of you guys could consistently tell me which was which. Two real vintage Deluxes can easily sound farther apart. and that comes from someone who has owned 3 of them. And guts? Guts is definitely not the problem, if there is one. I could make you void your bowels with this amp. I think I might be talking to a couple of guys who fooled with a modeling amp or two in the store, concluded what they expected to conclude and moved on. They are like anything else. You have to learn how to use them.

"Don't have the guts..." Now that's funny.

Tim

That might be an expensive farm. We noted this phenomena when recording a band several years ago- the lead had a modelling amplifier. Boy was he happy when we swapped in the real thing in place of the modeling. The recording sounded better too. So I don't buy your response, not by a long shot.

 

[esldude]

If the amp has 0.001% THD the odd ordered and higher orders will be audible. They will manifest as brightness and harshness.

Got any references for this?

 

[DonH50]

If the amp has 0.001% THD the odd ordered and higher orders will be audible. They will manifest as brightness and harshness.

Thanks, pretty incredible (-100 dBV). That is with musical source or test tones? I have in the past conducted experiments and I could not hear THD that low in my system; I fell between 1% and 0.1% using headphones (Stax and my old Sennheisers). Hard to believe speakers can deliver that low distortion.

 

[Mark (Basspig) Weiss]

Mark,

Did you change anything other than the tube rectifiers? Can you quantify the "immense drop in source impedance"?

Other changes were made at different times, but the biggest change in performance came with the switch over to SS rectifiers. The supply stiffness lowered the effective output impedance. One way to illustrate that is to measure the output voltage while toggling the load from 8? to 4?. With the vacuum tube rectifier, output voltage dropped by almost half. With SS, the drop was about 15%. The amp was acting more like a voltage source than a current source after the SS mod. That single change also dropped THD an order of magnitude. All other changes (the recapping I did) had only minuscule effect. Each change was done, documented in terms of performance effects and then the next change was done, allowing me to see which modifications had what kind of impact on performance.

 

[Mark (Basspig) Weiss]

Mark,
first, thats an interesting statement and I wonder if you can recall where you obtained that information, or if its just through your extensive experience in audio repair.

Also, your post shows that making electrical differences to his amp made audible differences, so a change was noted, and in this case, he felt these were improvements. No magic there, just electronics. And I presume he has preferences for what he listens to.

About the perception of harmonics? Both from testing done at an AES convention two years ago, and from my own testing.

If you have a sound editor like SoundForge, you can use the synthesis generator to produce a fundamental and up to three harmonics at the same time. The ratio of harmonics can be set in absolute dB, making it a good tool for demonstrating at what point people can hear the presense of additional harmonics.

Even order harmonics are hardest to detect. In fact, up to 3% THD may go unnoticed on simple waveforms. But if the harmonic is not an even-order multiple of the fundamental, harmonics can be detected at low as 1% THD. These numbers may be skewed by speaker systems that sound especially bright or emphasize the harmonic frequencies more than the fundamental, but on a flat system it's pretty hard to hear even order harmonics until they approach -25dB of the fundamental amplitude. The presence of odd harmonics stand out far more than even harmonics. Even at -40dB under the fundamental, a 3rd harmonic will be easily heard. As they go up in order, or further separated from the fundamental, the masking effect becomes weaker, and the harmonics become more audible. You can test these concepts yourself if you have some basic sound editing software.

 

[Mark (Basspig) Weiss]

Yes, it is push-pull. At the time of comparison, the amp also had solid state rectification (beefed up over stock). In the meantime I have gone to tube rectification, but my observations are quite opposite of yours with tube vs. solid state rectification. They can be read in my review here:

http://www.whatsbestforum.com/showt...trol-MB-external-power-supplies-for-tube-amps




In my case with tube rectification the noise floor has dropped, resulting in much greater resolution (I still hear nothing in the background, which is a slightly different topic). Dynamics are more effortless than before. Bass is powerful and even more controlled. Everything sounds cleaner (you can read it all in detail in my review).

While lowering the level of electronic noise, the external power supply with tube rectification apparently also has stabilized the HT (B+ rail) supply, as advertised. Yet as has been pointed out in a discussion, this is only possible because a) the new external power supplies are truly massive (they would never fit in my amps) and b) my amps are low power output -- I guess the scenario simply would not work with an amp the type of which you modified for your customer.

How massive the power supplies are you can see from images here (there is a rack where you see four of them below a BorderPatrol amp; the last picture on the page shows the inside of a power supply):

http://parttimeaudiophile.com/2012/10/24/rmaf12-borderpatrol-take-me-away-to-the-living-voice/

With tube rectification as regularly implemented I might indeed have all the problems that usually are solved with solid-state rectification.



Yes, but those peak levels are rarely reached. And for normal listening, 95 dB is really loud (not for you I guess, from posts of yours I read elsewhere on this forum). After two minutes of the final brass chorale of Bruckner's Fifth Symphony at 95-97 dB my ears feel the pressure, even though there is hardly any perceived distortion in playback. Chamber music typically does not go beyond 85-90 dB (mostly the lower value), but if you put the microphone closer than the usual listening positions in the hall you may exceed those levels at that position handily of course. It also depends which chamber music. If it includes percussion it may be a different story.

I can see a massive tube rectifier supply being pretty stiff, especially if it has a large capacitance storage. One can parallel multiple rectifiers, too. But in comparison to a single 5U4 vs. SS diodes, the stiffness is quite improved with the SS. Granted, if you use ordinary silicon diodes, you'll experience the 120Hz spikes as the diodes switch on significantly past zero crossing of the power input. The use of fast-recovery or Schottky diodes greatly offsets this problem.

I made my orchestra measurements from the 4th row center, where my microphone array was hoisted 18' above the audience. Further away, in the balcony, the bass levels are even higher due to room gain. This was a relatively modest orchestra setup, for a Beethoven piano concerto. In contrast, something like the Cleveland Symphonic Winds, performing a suite of Holst marches would be quite a bit louder. Certainly achieveable with 105dB-efficient loudspeakers though.

 

[Phelonious Ponk]

That might be an expensive farm. We noted this phenomena when recording a band several years ago- the lead had a modelling amplifier. Boy was he happy when we swapped in the real thing in place of the modeling. The recording sounded better too. So I don't buy your response, not by a long shot.

Several years is a lifetime in that technology. I have a second generation POD. It's probably 60% of the real thing. FWIW, I don't buy your response by a long shot either. I was figuring we just had a slightly different experience until you said "don't have the guts." Then I knew we were talking about completely different products.

How did you record that modeling amp? Which amp was it? What kind of mic did you use?

Tim

 

[Al M.]

Even 5% is still a lot. The sum of all the harmonics must be more than -32dB below fundamentals in order to fall below the masking effect of human perception. That happens around 1-2% for most people. I, myself, am sensitive to much smaller amounts of distortion. Your amp sounds like a parallel push-pull arrangment, not an SET, so it's distortion (especially even order) is probably much lower than 5%. Probably more like .5% or better. In such a case, and assuming solid state (not tube) rectifier, and a stiff, low impedance power supply, the amp can sound a lot like solid state.

Ok, I read on Wiki about push-pull vs. single-ended, and now I understand why the distortion of push-pull is much lower than the nominal distortion of a single tube (or transistor), at least if it's true what the Wiki article says:

In analog push–pull power amplifiers the two output devices operate in antiphase (i.e. 180° apart). The two antiphase outputs are connected to the load in a way that causes the signal outputs to be added, but distortion components due to non-linearity in the output devices to be subtracted from each other; if the non-linearity of both output devices is similar, distortion is much reduced.

http://en.wikipedia.org/wiki/Push-pull_output

That's interesting, and would explain why I didn't have the idea that my push-pull tube amp distorted significantly more than the SS Spectral (neither as nasty sounds nor as euphonic coloration). I suppose the distortion from my speakers is greater. As I understand it, usually speakers are the most distorting component in an audio chain, with THD somewhere between 0.5 - 2 % at higher levels if I'm not mistaken.

In any case, I have learned something, thanks Mark.

 

[Orb]

About the perception of harmonics? Both from testing done at an AES convention two years ago, and from my own testing.

If you have a sound editor like SoundForge, you can use the synthesis generator to produce a fundamental and up to three harmonics at the same time. The ratio of harmonics can be set in absolute dB, making it a good tool for demonstrating at what point people can hear the presense of additional harmonics.

Even order harmonics are hardest to detect. In fact, up to 3% THD may go unnoticed on simple waveforms. But if the harmonic is not an even-order multiple of the fundamental, harmonics can be detected at low as 1% THD. These numbers may be skewed by speaker systems that sound especially bright or emphasize the harmonic frequencies more than the fundamental, but on a flat system it's pretty hard to hear even order harmonics until they approach -25dB of the fundamental amplitude. The presence of odd harmonics stand out far more than even harmonics. Even at -40dB under the fundamental, a 3rd harmonic will be easily heard. As they go up in order, or further separated from the fundamental, the masking effect becomes weaker, and the harmonics become more audible. You can test these concepts yourself if you have some basic sound editing software.

Mark you touched upon something that actually makes much of this specific debate a bit academic for both subjective and objective posters; the only study I know that did dbt focusing on THD with music only managed to have thd down to 4% accurately audibly perceived (and by trained listeners), think they started the training in double digits and worked way down.

Cheers
Orb

 

[microstrip]

(...) Some people here have been talking about the magic of tubes. Of course there is no magic. But the idea that we have had everything sorted out for the last 40 years is bogus too. The problem here has to do with understanding how we hear things. We know the mechanics of the ear. Not so much about how the ear is able to convert sounds to neural signals. Even less about how the brain processes those signals on arrival! (...)

Ralph,

Happy to read some more commonsense in this tread. I am amazed that people can thing that magic of tubes is something related to wizards and magic potions. Magic in audiophile jargon just means what you described - something we perceive but can not fully explain with our current instrumental and perceptual audio knowledge.

We also use it to describe a system when everything is in tune and the system sounds excellent and very enjoyable, due to a great synergy between room, system and recording - then we refer we are getting the "magic" of it. This word also reflects the sometimes hazardous experimental way of getting it, as this point of equilibrium depends on many aspects and just changing one can spoil it - spoiling all the "magic".

As I have experience with Atma-sphere electronics - I still own an MP1 mk3 and from time to time an MA2 - I would like to add that your electronics sounds the opposite of the typical full and slow sound most people associate with tubes. When matched with adequate sources, cables and speakers it sounds very dynamic, harmonically rich but on the lean in the positive sense - this means much "faster" than most solid state designs. And yes, your OTLs need adequate sources and speakers to show their incredible absence of sound artifacts.

 

[microstrip]

Other changes were made at different times, but the biggest change in performance came with the switch over to SS rectifiers. The supply stiffness lowered the effective output impedance. One way to illustrate that is to measure the output voltage while toggling the load from 8? to 4?. With the vacuum tube rectifier, output voltage dropped by almost half. With SS, the drop was about 15%. The amp was acting more like a voltage source than a current source after the SS mod. That single change also dropped THD an order of magnitude. All other changes (the recapping I did) had only minuscule effect. Each change was done, documented in terms of performance effects and then the next change was done, allowing me to see which modifications had what kind of impact on performance.

Mark,

Thanks. I am astonished by this result - it suggests that the engineers who designed the power supply did a poor job and the power supply was not well matched to the amplifier - or the tube rectifiers you were using had low cathode emission. Replacing tube rectifiers with diodes increases the voltage of the B+power supply under load, and as sometimes the drivers and input take their supplies from the same source this effect must be considered - typical tube rectifiers have an equivalent dynamic resistance around 50 - 100 ohms. Most probably it was a low feedback design, and the increase in distortion due to power supply modulation could not be compensated.

According to your measurements we can think that the amplifier was optimized for an higher power supply voltage than what it really used!

 

[Al M.]

As I have experience with Atma-sphere electronics - I still own an MP1 mk3 and from time to time an MA2 - I would like to add that your electronics sounds the opposite of the typical full and slow sound most people associate with tubes. When matched with adequate sources, cables and speakers it sounds very dynamic, harmonically rich but on the lean in the positive sense - this means much "faster" than most solid state designs. And yes, your OTLs need adequate sources and speakers to show their incredible absence of sound artifacts.

Yes, tube amps can sound incredibly "fast" and clean -- mine sounded just as fast and clean as the commonly acknowledged very "fast" sounding Spectrals in the comparison in my system. But indeed, speaker matching can be critical for tube amps, and the benign load of mine works perfectly with my amps (while the Spectral of course can drive almost anything).

 

[mep]

At times I have regretted starting this thread because we seem to rehash the same things we have always hashed out and yet sometimes I'm glad because we at least are talking and some people are maybe learning something new.

Francisco-I found my copy of the VTL book which is tattered and torn because some time ago one of my dogs chomped on a corner of it. Believe it or not, I'm missing pages 51-53. The beginning of the book opens up with a series of quotes from some famous high-end people which mirrors much of what has been said on this thread. I would post those quotes, but I don't want to get into any copyright issues. I also came upon an old issue of TAS where MF challenged David Clark to a double blind test for amplifiers. The test involved a VTL 500 watt tube amp, a Crown amp, and a Threshold amp. MF scored a perfect 5 out of 5 with his ability to identify each amp correctly. David Manley brought a Studer C-37 and nothing but master tapes and dubs of master tapes were used for the test. David Clark tried to rig up the box that was used to plug all of the amps into so that it favored the Crown amp. David Manley was a little too smart and caught what Clark was up to and snipped out the added components that Clark had installed which was going to help the Crown with the speaker load and was going to reduce the output of the Manley amp to make it lower than the other two.

 

[Gregadd]

the problem as i see it is the perennial misunderstanding of measurments vs ear/brain interface.

there is nothing in an audio signal or combination of them that we can not measure. If your background is not a lifetime of a broad exposure to different electronics, then you may disagree but what exactly are you able to base a counter argument on?

Well, you say, I hear differences. That does not mean there is any problem with measurments. Measurments dont hear through your unique ear and brain system.

To me, the big gap here is of course understanding, that for example, there are many more measurements than THD or twin tone IMD, but this industry fails to list them. That does not mean that audio electronics is not fully understood at the electronics level.

What is not fully understood, is what happens at the ear/brain level, and of course, that is the 900 pound gorilla in the room.

It seems hard for folks to make this distinction for some reason.

I started a thread asking who can hear changes in wiggles in air better than can be measured by instruments...no arguements ever came. audio is changes in wiggles in air resulting in chages from wiggles in electronic signals...it is not black magic at the measurment level.

this thread started out with an idea which the OP i think said he might not be totally happy with the way the thread went, but, folks can not for some reason understand the utter simplicity of simply changing the harmomic spray, thus the changes in the way the air is wiggled, and that is why each amp made sounds different, and if two so called identical amps are made, and you hear no differences, measurments will reveal the differences.

audio is a very narrow band in the scheme of electronics, and we have gone way deeper than that through the years...we can move individual atoms and place them to spell out the name IBM, and lots more.

I had a circuit that i could adjust for predominately even or odd distortions, and the even at extremes sounded more the same all across the band, and the odd at extremes made all the music sound more the same. in between i could adjust for any effect i wanted, mellow to hyper detailed. all with just one special tube and circuit. it is really all harmonic spray,which we can measure, but in stereophile you only see a few base frequences looked at, at a few input levels, and this does not take into account phase but other measurments can.

harmonic spray ladies and gentlemen, is what electronic (also transducers but they do other stuff, ie they really are the primary tone controls) circuits themselves do.

the electronic harmonic spray is the finger print AND it is measurable, all of it, as long as it is measured in the SYSTEM it is used in!

Everything man treasures he perceived first! Then he has to understand it completely.Aft
error that he designs a metric. Following that is the design of a tool or machine to perform. No machine exist without mans understanding. Moreover the use of those machines/tools is as much art as science. I ask you what is greater- the creator or the creation?
One thing is sure we will make no progress as long as we assume none is to be made.
My earlier facetious statement notwithstanding. When our result is perfect we will knowo we have measured everything.