Why Tubes Sound Better

Tube amplifiers sound better because of the euphonic distortions they add to the music, as well as plenty of other reasons I'll cover below. These are subtle effects most audible to musicians and very dedicated music lovers; casual listeners (people who "listen" with their eyes open while doing something else) usually won't notice, but sometimes the difference is so obvious that people's wives will comment that "wow, that sounds much better" when people use tubes at home.

Tube amplifiers measure poorly in the lab specifically because of these added distortions, but these distortions are often a part of what make them sound better.

Even today with an all-digital infrastructure from recording studio to home, professional studio microphones for decades and decades and decades have used tube pre-preamplifiers inside the microphones themselves. Today their outputs are fed to tube preamplifiers before being digitised for recording, mixing and distribution. We use tubes simply because they make the music we create sound better: smoother, warmer and cleaner.

Ditto for guitar amplifiers used in creating music. The ways that tubes distort when pushed to the edge are much more musical than the artificial sounds that come from transistor amplifiers when overdriven. Some transistor guitar amplifiers attempt to mimic tube distortion, but that's a different article.

Of course these are all very broad generalizations, and this is just as much due to circuit designs used with tubes or transistors as the devices themselves, but what are the distortions and other reasons tube amplifiers sound better?


Order Harmonic Distortion

Tube amplifiers have much more distortion than solid-state amplifiers, but most of it is second-order, which is quite musical. That's why it's called "harmonic" distortion.

Second-harmonic distortion is exactly the same note, an octave above. Ditto for higher-order even harmonics; they are also the same note more octaves above.



Harmonic Distortion Components at 2 W. Rohde & Schwarz Audioanalyzer UPV


This is a logarithmic scale; in this typical case 99% of the distortion power is second-harmonic; the second harmonics are about 20 dB higher than any other harmonic.


Progressive Distortion

Not only is tube amplifier distortion harmonious, it increases as things get louder - exactly as they do in a musical performance. As instruments play louder, or as you hit a percussion instrument or piano key more strongly, they generate more harmonic content. As notes decay, the percentage of harmonic content drops again.

Tube amplifiers mimic this. A good tube amplifier like the Woo Audio Amplifier increases its distortion directly with output level across three decades of voltage, or a million-to-one power range:


THD LVL LO4 37 Ohms

Tube Woo Amplifier THD versus level at 1 KHz, 37.5 Ω load.


By comparison, here's how a typical solid-state amplifier, in this case a Crown Amplifier D-75, lowers its distortion with level, and then suddenly clips like crazy (the nearly vertical line on the right):



 THD versus power. Crown Transistor Amplifier


Note that the Woo graph is in terms of voltage output, and the Crown plot is in terms of power. In fact, the Woo plot covers a power range of over 6 million to one, while the Crown plot only covers a power range of 50,000 to one. With this progressive, "dynamic" distortion, tubes add sharp attacks while retaining long, floating sustains for every musical note.

Just like our ears, musical instruments and just about everything else natural, tube amplifiers have the least distortion at the lowest levels. This is why a tube amplifier can sound great played softly, while with transistor amplifiers people are usually needing to turn it up to have it sound best. Honestly, I don't bother using my dbx 3BX dynamic expander with a tube power amp, since it adds too much dynamic impact.


Optimum Sound at Optimum Levels

Tube power amplifiers sound their best at the volumes at which you actually want to enjoy them. Just like digital systems, solid state amplifiers measure and sound their worst at low levels, and have their best performance at close to their maximum output levels where no one ever actually plays them. For normal use with normal music at normal levels, most of us enjoy our music at about 1mW ~ 1W long-term RMS, or about 0.01W ~ 10W peak. For most applications, a 30 WPC amplifier is about right.

What's sad is that the few consumer magazines that try to publish lab results usually only plot performance down to 100mW, when in fact the most relevant power range at which we enjoy most amplifiers is from 1mW to 1W. What happens below 100mW is extremely important; that's right where most of our music lives! Sadly even if you pay $150,000 for a pair of overpriced solid-state amplifiers, you'll see its reviewer said many nice things about it, but he still said "the more I cranked them, the better they sounded". So for $150,000 they don't sound best at the levels I want to enjoy them? Follow the money; I don't take ads from manufacturers. Don't let me stop you if you want a 100 WPC amplifier, but you don't need it unless you like to crank it, have a big room or inefficient speakers, or enjoy very wide dynamic range classical music at concert-hall volume.


Optimum Power Output

Tube power amplifiers are usually rated for the power you actually need and will use, like 8 to 80 watts per channel. Solid state amplifiers tend to be rated up to stupid and dangerous power levels as prices increase. No one needs or uses 300 WPC except public address systems; 300W will melt any single speaker. Sadly not only do you wind up paying for and having to lift these solid state beasts, they sound even worse down at the rational levels at which you will actually enjoy them.


Soft Clipping

Tube amplifiers overload gradually. Add more input and they distort more, but there is no precise level above or below which they suddenly start to clip. Contrast this to solid state, where there is a very definite clipping point. Here is a McIntosh MC240 at clipping with a random collection of weak old tubes:



McIntosh MC240 with old tubes clipping around at 25 watts per channel.


These waveforms are highly asymmetrical because each old tube was quite different from the others, but the key here is that even though they are clipping, there are no sharp edges to the waveform. (This old amplifier and its weak old tubes still sounded fantastic at normal levels.) When a solid state amplifier clips, there is a sharp edge where it looks like someone simply took a pair of scissors to the tips of the waveforms. The sharp edges of solid state amplifier's waveforms at clipping give rise to insane levels of very high order ultrasonic harmonics, which are what blow out tweeters.



In my younger days I used to think people who worried about capacitors were crazy, but in 1980 engineer Walt Jung along with Dick Marsh published the best articles I've seen to date about picking capacitors  where they explain exactly how and why some capacitors are much better than others. Making a long story short, while electrolytic capacitors are superb for power supplies, their high dissipation factors and mysterious recovered voltage effects make them poor for coupling high-quality audio signals.

Most solid-state designs use electrolytic coupling capacitors because they are much smaller and much less expensive than good (film) capacitors in the values needed, typically 10 ~ 220 µF at 15V or less. Electrolytics like this sell for pennies apiece and are the size of pencil erasers, while film capacitors in these values sell for about ten dollars each and are about the size of a baseball — and there may be dozens in even the simplest amplifier. Solid-state equipment usually uses polarized electrolytic capacitors to decouple DC, and that's bad. Not only are electrolytic capacitors far from optimum for coupling audio, they need a polarizing voltage which is usually absent in these designs. Most of these designs hope that no one notices, since the DC voltages aren't enough to blow up the capacitors when the DC voltages are backwards from what the capacitor needs, but it's still suboptimal.

In contrast, because of the much higher voltages and lower capacitance values needed, tube amplifiers almost always use much better film (or 50 years ago, paper) capacitors for audio coupling. Tube amplifiers usually use coupling capacitors rated between only about 0.022 ~ 0.47 µF, but usually rated at 600 volts! Film (usually polyester or polypropylene and less commonly polycarbonate or Teflon) capacitors abound in these ranges.

Whether designers want to or not, the voltages and impedances involved usually sentence solid state designs to electrolytic capacitors, while tube amplifiers usually can use film capacitors. As Jung and Marsh observed directly, for all they know, a big reason people may prefer tube amplifiers is precisely because they usually use much better capacitors for the audio signal path.


Less Negative Feedback

Tube amplifiers have less free gain to throw around, and as Mitch from Manley Labs loves to point out, tubes have more linear transfer functions than transistors, so tube amplifiers usually need and use much less negative feedback. While feedback itself isn't really bad, what is bad is sloppy basic amplifier linearity that needs a lot of negative feedback to test well. It's easy to identify an amplifier that works with a lot of feedback to cover up its basic flaws: it has a distortion curve that rises with frequency and has foolishly low output source impedances (high damping factors), as most Crown amplifiers do:



Crown Amplifier D-75 THD versus frequency at 1 W


High Output Impedances

Tube power amplifiers have higher output source impedances, also known as lower damping factors. This is because of their lower negative feedback and their output transformers. While solid state amplifiers tend to have near-zero output source impedances for damping factors higher than the wire used to connect to the speakers, a typical tube amplifier has an output source impedance of about an Ohm, or a damping factor of only about 8. Not that this is particularly good or bad, but this lower damping factor does change the frequency and transient response of the system when connected to a real loudspeaker. It will tend to lower the low-frequency damping, which can have the effect of blooming the bass, which some may prefer.


Zero DC Offset

Solid-state amplifiers almost always have some minor amount of residual direct-current (DC) voltage present at the output. Even the oldest solid-state amplifiers with single supplies and big electrolytic output coupling capacitors will have some slight leakage — but that's the least of your problems with ancient solid state amplifiers. DC offset is usually negligible and less than 10 mV, but every little bit of DC offset does displace the woofer cone slightly, and can lead to added speaker distortion. Lab tests for noise and distortion and everything else ignore the DC offset, but your speakers won't. Tube amps are transformer coupled, and by definition (and my measurements) have zero DC offset.


Safer for Speakers

When a transistor amplifier has a shorted transistor or other problem, your speaker will blow unless their cables are fused. When a tube amp has a problem, the output transformers saves the speaker from shorted output tubes.


Completely Isolated Outputs

The best tube amps, like McIntosh, have separate feedback windings, so the output taps are completely isolated from everything else. This lets us series or parallel the outputs to our heart's content for mono or increased power outputs, or driving just about any imaginable load impedance. For instance, the McIntosh MC240, MC275 and MC225 in mono can be wired happily to drive any of 2Ω, 4Ω, 8Ω, 16Ω or 32Ω loads as their optimum impedance. These amplifiers also have higher impedance outputs (typically 150Ω, 600Ω in stereo and many more in mono) for driving very long lines at high powers.

This also means that there is no worry about interference or ground loops, no matter how you wire the outputs. Contrast this to solid state amplifiers, whose outputs are almost never isolated from anything, and may never be connected in series or in parallel. The mono options on stereo solid-state amplifiers are bridge connections, which tend to lower performance.



Again not particularly good or bad, but tube amplifiers usually have some form of microphonics, meaning that any tapping of the amplifier will result in audible output. Put on headphones and tap your amp, and you'll probably hear it. In practice this usually doesn't mean much, but it can mean if you have an amplifier close to your speakers and it's microphonic that you'll actually add what amounts to a little bit of extra echo or reverberation as the signal flows from speaker to amplifier (after a delay of about a millisecond per foot of separation) and then comes back out the speaker to repeat the cycle. Microphonics can lead to a warmer, fatter sound from this extra reverberation. Playing LPs in the same room as your speakers also can lead to this effect; record players are highly microphonic.


So why use transistors for audio?

This is exactly what we asked in 1960 when transistors had already taken over in aerospace and military applications. It turns out the advantages of transistors back then weren't relevant to audio. Aerospace needed low power dissipation, tiny miniature size and resistance to vibration, but none of this applied to audio amplifiers for the home. Back then transistors were also wildly expensive, far more than tubes. You could pay a few hundred dollars for a transistor in 1960 dollars, while tubes sold at every corner drug store for a dollar or two.


So why did tubes go away?

Just as the iPhone took over the world in much less than ten years (the iPhone was only introduced in 2007), transistors became cheap and plentiful, and by 1970 tube amplifiers were as unwanted as an old cellphone is today. Tubes were old and hot and easy to break, and transistors were the future and the space program. Just like today; who want an old dumb-phone versus a new iPhone? Transistors were with-it, and tubes were boring. Transistors never need replacement; they just work. Vacuum tubes, like car tires, have to be replaced every few years. When people were able to get amplifiers that never needed service, they were as popular as tires that never needed replacing.

Back in 1970, speakers and audio sources were terrible. The best audio source available in the home was a live FM broadcast — but it was all mono until 1961! LPs were stereo but scratchy, and prerecorded stereo reel-to-reel tapes were also popular, but at 7-1/2 IPS quarter-track with no noise reduction, they had as much hiss as records had clicks and pops. There was no way to bring the real sound of the studio or stage home, unless you hauled a professional 35mm mag recorder, weighing about a thousand pounds and reaching from floor to ceiling, home, and had access to original master reels. You could bring an Ampex 350 home (only about the size of a washing machine), but you'd still need the master tapes — and there was no noise reduction yet to quiet the hiss.

Even if you were home for the baton drop of a live broadcast of your favorite music, or had your 35mm mag recorder spooled up, speakers were horrible back then — no one could hear the subtle things that made tubes better! While today we take uncolored frequency response and good transient response for granted, even the very best all-time classic speakers like the EV Patrician, Altec 19, Altec A7 and Acoustic Research were horribly colored and had ridiculously bad transient response by modern standards. Not that those speakers can't sound great, but none of them have the transparency of modern speakers — and almost no one actually had those top-of-the-line classic speakers back in the day. I recall a review of I think it was an AR-3a that mentioned how it had so much transient ringing that your couldn't see where the tone stopped and the ringing started on the oscilloscope photo!

Solid state gear was so dry-sounding in the last days of tubes that it helped compensate for the warm, muffled, woolly and ringy speakers of the time. There were no dome tweeters, just 2" paper cones! Sure, some people used electrostatic tweeter panels, but with big 15" bass-reflex cabinets on the bottom. Speakers were so cloudy and obscure-sounding that anything that help cut-through sounded better, unless of course you were one of the lucky few with Quad ESLs. Add never needing to change tubes, and within less than ten years people had hauled all their glorious tube gear to the dumpster.


So why tubes today?

Simple: because they sound better, and today our recordings and speakers are good enough that we finally can appreciate the subtly magical things they do for our music. Today everything else in the chain is so transparent that one added layer of "tube sound" is just about perfect.


AES Research

As an engineer with a BSEE, patents and an Audio Engineering Society (AES) member for about 40 years, I know that if you have two amplifiers with the same gain, same measured frequency response, same measured output impedance and low enough measured distortion and noise, that they will sound the same so long as they are operated within their linear range. Regardless of amplifier topology, tube, transistor or a lot of hamsters with little cups and strings, they will sound the same if you match the basic measured characteristics and the noise and distortion are low enough. I participated in an AES research project at one of our conventions in Los Angeles around 1989 to see if there was any audible difference between a top-quality VTL tube amplifier and a cheap Lafayette integrated transistor amplifier.

Each amplifier had its gain and frequency response matched with soldered RC networks, and I presume a build-out resistor network added to the Lafayette's output to match the VTL. The test was double blind. We were played the same selection twice, and all we had to do was guess was if it was the same amplifier each time, or it it was different. We weren't trying to hear if one was better or worse; all we were trying to see was if there was any difference at all. Being the AES, these guys were hard-core and actually did haul an Ampex 350 with tube electronics, and were playing 15 IPS half-track tapes that were recorded entirely with tube electronics, and these were played through speakers in a hotel room. All this, and I couldn't hear any difference at all. So which is it: do tubes sound better, or is it all the same? Both!

If you match everything, they will sound the same, but in real life, things are never matched. In reality, tube amplifiers tend to have enough distortion that does make them sound different. Frequency responses vary, and output impedances certainly vary wildly from one kind of amplifier to the next which has a big impact on the sound. Likewise, I can't hear anything in a motel room. The sound bounces all over, and you're hearing the room more than the music. When I listen over speakers, even big speakers, I listen close so I can hear the music and not the room.

Also, with all the many layers of tube sound in what was playing, I don't know that adding or subtracting just one more would be audible. Play what people really play, which today is pretty much direct digital from the studio, and one stage of tubes or not will be more audible. I bet the tube fog was so thick you could cut it with a knife in that test, so there's no way we could have heard anything, especially in a motel room listening mid-field, even if the amplifiers weren't matched at all. Also we humans can't make real comparisons if there is silence for more than a second. Strange but true, human auditory perception is so fickle and mysterious that once you've got a second of silence between selections, half the time our brains start imagining differences that weren't there. The best way to hear subtle differences is with electrostatic headphones to hear just the music and not the room, and then switch selections with an instantaneous switch while the music is playing.


The Mind's Eye

All musical perception is purely intangible. We can't put a finger on a musical image and point someone else to what we're seeing as we can on a painting, piece of sculpture, a musical score, a book or a photograph. Because musical images are created entirely in our imaginations, what we think we are going to hear is often what we hear. This is why otherwise reasonable people think they hear huge differences in foolish (but high-profit) items like cables or power cords. Even though there is no real difference, they hear very real differences that just aren't there. The differences are very real in that listener's vivid imagination, but no where else. This is why we use double blind tests where neither the subject nor the presenters know what's being heard when we try to do scientific research, like the AES research above.

Music is all about using our imaginations. This is a very good thing and why music is such a powerful art form. This is why a good HiFi can recreate the original listening experience. Unlike a TV or movie, close your eyes, and you can be seeing and feeling the same things that you do in the concert hall. I close mine and see the performers, see them moving around, breathing, moving valves and keys, turning pages, and then I see the music itself. You have to concentrate, and if you listen carefully and keep your eyes closed, you'll see the music, too. If you think a nice, warm glowing tube amplifier is going to sound smooth, liquid and warm, it will! Our imaginations are very susceptible to suggestion; that's the whole point of music.



Anyone picky enough to be reading this article probably also worries about how to enjoy music after a limited nuclear exchange. Presuming you have a way to power your amplifiers, which again you people probably do, true tube amplifiers are far less susceptible to the EMP (electromagnetic pulse) effects of nuclear weapons. While most solid state gear that's connected to the power grid or antennas will probably be toasted in microseconds, your all-tube gear will probably be OK. Air-testing of nuclear weapons ended before transistors took over.



If everything transistorized tries to claim it's as good as tubes, why not just go with tubes? It's funny how some people will pay $50,000 for a frou-frou solid-state amplifier just because it claims to sound "almost as good as tubes," instead of simply stepping up to a real tube amplifier. Geesh, be prepared to buy a few tubes every few years, and you too can bask in glorious sound for a lot less money than you'd pay for overpriced solid state designs that don't sound as good. Now that I use tube amplifiers for my power amplifiers, they makes all my solid state amplifiers that I used to love sound awful by comparison. Yes, I always thought those tube people were crazy, but as soon as I actually opened my ears and listened, the real world effects are marvelous.



For monitoring accuracy, of course use solid state, but when you want it to sound great for enjoyment, it's tubes all the way. Use solid state monitor amplifiers when you're producing music so you can hear exactly what you're laying down, but when you want to kick back and have it sound as good as possible when you're all done, tubes are it. When a transistor amplifier alters the sound, it almost always makes it worse. When a tube amplifier modifies the sound, it usually makes the music sound better.

Crummier tube amplifiers will have more of the distortions that make tube amplifiers sound like tube amplifiers. If you really want to hear the "tube sound," get a TubeCube 7 (3 WPC, $180) and you'll hear how smooth, liquid and warm tubes really sound — but it only puts out enough power for desktop or background use. For a much higher quality tube amplifier that has enough power for many home Hi-Fi uses so long as you're reasonable with playback levels, the Elekit TU-8200 (8 WPC, $699 in kit form) is superb. It self-biases so you never need to match tubes or tweak it.

For the ultimate, get a classic McIntosh MC225 (25 WPC), MC240 (40 WPC) or MC275 (75 WPC), which are the best-designed tube amplifiers ever made. They excel for their stable designs (no bias adjustments or matched tubes ever needed) and have extremely low distortion due to their unique design. They have enough power for anything, and are unflappable for their ability to deliver seemingly limitless low bass response. These are all 50 years old today and you'll pay at least a couple of thousand dollars used, and when you get yours, you'll know why people pay such ridiculous prices. They really are that good. Of course the McIntosh, when operating to its original specifications, has such little distortion that it sounds less "tubey" than weaker amplifiers. If you're playing a McIntosh that hasn't been serviced in a decade, then it's probably out of spec or needing new tubes, in which case it will have more distortion and a more "tubey" sound. Here's where the art comes in: just how much euphonic distortion do you want?

For most people with reasonable budgets, go for the new Elekit TU-8200. If you like it loud and have unlimited funds, or like to crank the bass without biamplification, get a used McIntosh MC240. The new version of the MC275 is probably pretty good for the rich and unadventurous, but it's a different design than the classics and I have not tested it.

Mr. & Mrs. Ken Rockwell, Ryan and Katie.