The Sound of Vacuum Tubes, Part ITHE 12AX7/AU7/AY7 SERIES OF DUAL-TRIODE PREAMPS 12/01/2004 7:00 AM Eastern
This the season to snuggle up to a warm ampli-fire with a good book (or this magazine). Amplification is the topic this time — specifically, the sound of vacuum tubes. I must have been a good boy this year because thermionic specialists Groove Tubes (GT) and Electro-Harmonix (EHX) filled my stocking with goodies. Preamp tubes will be evaluated this month; we'll look at power tubes next time. Researching and writing about this subject is like trying to control a volcano, so expect to see continued reports throughout the year and expanded online versions at www.tangible-technology.com and www.mixonline.com.
A triode vacuum tube and a single transistor are the most basic forms of amplification. The initial difference is obvious: Size! Tubes are “big” on many levels, but in the most literal sense, real estate dictates population density. This form of sonic discipline, an enforced minimalism, is generally responsible for the color we associate with tubes, especially guitar amps. Neither color nor the lack of sameness is exclusive to vacuum tubes. (Think API and Neve modules.) It's quite possible to design a tube circuit that is more linear and less like the expected signature by using — for example — a combination of less-aggressive — sounding tubes, more gain stages and some negative feedback.
Miniaturization (in the form of the transistor and followed by the integrated circuit) ultimately increased the number of gain stages that can be squeezed into ever-smaller spaces. The resulting performance “improvements” have reduced distortion in the window between nominal levels all the way up to clipping (guitar levels). When done right, linear amplification — the proverbial wire with gain or “neutral” sound — has its place in every facet of the recording — especially the monitoring process — so long as we can still play with our crayons.
Because more than 24 tubes were evaluated, to avoid potential mix-ups, shrink tubing was applied as a label. Each tube was cycled several times through a series of tests. Both the hardware and the test parameters were fine-tuned along the way to make the results as meaningful as possible. (I've included an extensive spec comparison chart online.)
The first step was measuring each triode section on a TV-7/U military-type tube tester; the resulting number is only significant relative to the minimum reading as stated in the operator's manual. Some applications require identical performance from each partner in a dual-triode, such as when driving a pair of power tubes. For a screaming-hot preamp, the gain becomes a priority, hence the use of the 12AX7, aka the 7025 and ECC83. High gain comes at a price, an increased potential for noise and microphonics, but there are alternatives.
The preamp emulator is a versatile test circuit with three modes: high gain, low gain and variable treble boost; the latter was not used for these tests. For the 12AX7, the difference between high- and low-gain options was about 6 dB and less for the other tube types.
Before attempting to measure distortion, it was necessary to determine the point just before the onset of asymmetrical hard clipping, which, at about +35 dBu, generated 10-percent distortion using the GT-ECC83 (selected for the V-1 guitar amp position) as reference. The level was then dropped until the high-gain and low-gain distortion readings saddled 1 percent. This turned out to be 18 dB and 12 dB down from +35 dBu — for the respective gain options — for the same GT-ECC83. Consider this window the nominal level region. Note: This tube is modeled after the European cousin to the 12AX7; hence, the name.
From low gain to high gain — a mere 6dB change — the distortion increased two to four times, with one exception. For all tubes save one, the distortion was primarily second-order, an octave up from the 1kHz test frequency. Here's your big clue, folks: In music, octaves add richness to an arrangement. In an audio circuit, harmonic enrichment makes things sound bigger. When the Groove Tubes 12AX7M was scrutinized for the reason it behaved differently than all of the others, it was found to have about as much third-order harmonic distortion as second-order.
In the test circuit, adding a cathode bypass capacitor increases gain (and distortion) by eliminating a form of negative feedback. This simple circuit variation is the cause of the distortion and not the 6dB level change. Without the bypass capacitor, the distortion at +35 dBu is cut in half.
Even without listening, the distortion and circuit variations clearly demonstrate that the “big magic” most often attributed to tubes is primarily circuit-related — but, of course, the tubes help! This is not simply due to tubes providing 100 to 1,000 times more distortion than an op amp, but from tubes' broad window of “harmonic embellishment” from nominal to just before clipping. An op amp at the threshold of visible symmetrical clipping generates about 0.5-percent distortion, comprising mostly odd-order harmonics.
Lest you become fearful that all tube circuits have this much distortion, note that with the exception of guitar preamps, most hi-fi/preamp designs employ more finesse along with a bit of negative feedback to lower distortion. For example, total harmonic distortion plus noise (THD+N) for the D.W. Fearn mic preamp is 0.2 percent, and this is after passing through two transformers, two gain stages and an impedance converter — that's two or three tubes' worth, none of which are of the 12AX7 variety. The 12AY7 has 3 to 5 dB less gain, depending on the circuit, and about one-third the distortion even in the high-gain mode.
I ran two tests for noise (measured and by ear) for spectral (hiss and “rocks”) and microphonics. All low-noise metal-film resistors were used in the test preamp as a way of minimizing the color and noise variables. From a guitar-o-phile's perspective, the carbon variety may be preferred, and I am told that modern carbon types perform better over time than their 40-year-old predecessors. The “spectral” noises, when present, were generally constant. Very few of the tubes had major noise issues, partly because several had been pre-selected. Noise measurements and audio samples are provided online.
Microphonic describes a mechanical sensitivity that can be tested by tapping on the tube and listening. Ideally, the response to such stimulus should be a muted “thunk.” A resonant “ding” is a sure sign that the tube will “sing along” if located in the same space as the speaker. Microphonic tubes have a greater potential to feedback and squeal.
There are many issues and opinions about the sound of tubes, not the least of which are the design variations of supposedly identical types such as the 12AX7/7025 and ECC83 family. While the critical stuff is hidden to the eye, the plate structures are the most obviously visible part of any tube. Of my samples, Mullard, RCA and Telefunken look similar, but never identical, as are what I believe to be earlier Sylvania versions. Later Sylvania/ECG/Phillips types look like elongated versions of GE's style. There are black plates and gray plates, smooth plates and corrugated plates, long plates and short plates, and some Euro variations that look like other tube types such as the 6DJ8 and 6GH8. The point of this detour is that these are all models for today's versions and it is obvious that everyone is experimenting.
If you'd like to know more about the history of these tubes and how modern tubes are selected, the large-format Tube Amp Book (from Groove Tubes) is a wonderful resource. It might surprise you to learn that GT reveals the sources for all of its tubes. On the opposite coast, look closely at the tubes that have come through the Sovtek/EHX pipeline (via New Sensor Corp.) and you'll see their heritage, as well. All pay homage to their forebears. I believe you can still get good tubes by paying extra for the pre-tested versions. Buying untested, off-the-shelf tubes is a crapshoot.
Recording, audiophile and guitar enthusiasts may appear to have dissimilar needs and preferences, but it's not as if tubes were made specifically for guitarists, then or now. The 12AX7 was a consumer-grade tube then, and while the 7025 was its low-noise “select” version, in both cases, it was designed to minimize the tube count by providing more gain than any other tube. As such, it tends to be wild and unruly, so if your guitar amp (V-1 or V-2 position), mic preamp or Chinese tube mic is too noisy or microphonic, try a selected low-noise, low-microphonics version.
Conversely, replacing the 12AX7 with a 12AY7 will chill things quite a bit. With the 12AX7/AU7/AY7 Series, you are free to experiment without fear of damage — even if it's not a perfect circuit match. With known good tubes, performance is then in the hands of the designer. I certainly learned how circuit variations affect distortion. Next month, we'll examine power tubes.
Eddie thanks Aspen Pitmann and Steve Good at Groove Tubes and Rick Stevenson at Electro-Harmonix for supplying the tubes, as well as D.W. Fearn for his assistance.
Click here for more specs, broken down by manufacturer and tube type.
For an even deeper look at tube technology, visit Eddie Ciletti's Website at www.tangible-technology.com.