The Sound of Vacuum Tubes, Part 2TESTING MULTIPLE POWER TUBES WITH AMPLIFIERS 1/01/2005 7:00 AM Eastern
After submitting last month's column on preamp tubes, I set a record by making it Web-ready within the week. While this alone was worth a pat on the back, one particular value-added link tempered my feeling of accomplishment — the original “Tubes vs. Transistors” article written by Russ Hamm way back in 1972. The paper — presented at the 43rd AES convention — is longer than just about any magazine article today. Hamm included circuits, graphs, waveforms and tables, all successfully correlating the data that best represents the perceived differences between tubes and transistors.
My favorite line appears just five paragraphs in. One engineer who admits that there might be some marginal difference in the sound says, “You just have to get used to the nice, clean sound of transistors. What you've been listening to on tubes is a lot of distortion.” All this back in the day when analog tape was the only capture medium. In a word, prescient!
Each new technological improvement — transistors and op amps, ICs and then digital — heightens our awareness of the newly possible and what's been lost. The magical character of vintage gear, when it actually exists, is a complimentary distortion akin to the resonance of the “right” piece of wood on any acoustic instrument. Hard to imagine that some of today's “precious relics” were almost discarded as tired and inconsequential. I once saw Pultec equalizers on sale at a Canal Street surplus shop for $95.
I often thought that had the Record Plant been able to stay open just a few more years, it might have successfully transitioned into the New York branch of the Rock and Roll Hall of Fame's Sonic Museum division. As employees, we often wished for a few new pieces of outboard gear, but every control room had a few essential pieces — Pultecs and LA-2 limiters — and everything in our outboard room is to die for now, from guitar amps to compressors, vintage mics and EQs.
WHAT IS A POWER TUBE?
We mostly rely on the 12A?7 series of preamp tubes for voltage gain. All are dual-triodes (two independent three-element devices in one glass package). Power tubes are generally pentodes (a five-element device that can also be wired as a triode). And while they are called upon to amplify voltage, driving a “load” such as a loudspeaker requires the ability to maintain current without fatigue, which translates into power, the ability to do work.
I tested more than two dozen power tubes using three very different amplifiers, one of which was a Fender Pro Reverb. Of the listening tests, two tubes yielded consistent (and sometimes even astounding) surprises: the EL-84 (in a GT adapter that makes it work in a 6L6 octal socket) and the KT-66. The EL-84 pair (safely protected by the adapter) allowed the amps to achieve a smooth overdrive at a much lower level. The KT-66 pair delivered a more solid bass tone with more headroom (and much louder) than whatever 6L6 tubes had been installed — adding up to three very different sounds right off the bat. Both Groove Tubes and Electro-Harmonix versions of the EL-84 were tested.
SPECS, CLASS AND BIAS
Groove Tubes provided one amplifier that was particularly well-suited to the task of testing such a wide range of output tubes. Its Soul-o-Single, a Class-A “super-Champ,” sports an oversized output transformer and cathode bias resistor capable of supporting everything from the meek EL84/6BQ5 to the monster truck KT88/6550 — all of the tubes in between, including the 6V6, 6L6, 6CA7/EL-34 and KT-66. Electro-Harmonix (EHX), Sovtek (Sov), Svetlana (Sve), Groove Tubes (GT) and new old stock (NOS) tubes were also tested.
The EL-84/6BQ5 is rated for 5.7 watts @ 10-percent distortion in Class-A mode, with 250 volts at the plate. A 6L6 is rated at 6.5W for the same plate voltage — hardly what you'd expect, the difference being that the 6L6 plate's design maximum is 350V, extending the power capabilities to 11.8W. The KT-66 is essentially an industrial-strength version of a 6L6, delivering about five times the power — 25W in Class-A — with up to 500V at the plate. (According to a GE data book, circa 1973, the KT-66 cross-references to a 7581A.)
Note: The Soul-o-Single amp, used for both sonic auditions and the bench test, delivers 400V to the output transformer (399V at the plate), which is normally more than a 6V6 wants to see, but the self-biasing circuit safely optimizes the current.
Guitar amps squeeze many more watts out of tubes than what has been detailed here so far. The higher-powered amps are push-pull: a two-tube minimum, or multiple thereof, in some bias variation of Class-A/B mode. High plate voltages and well-matched output transformer impedance (on the plate side) can combine to yield from three to six times more power than Class-A because each tube is “processing” a little more than half the wave.
GT also supplied the very unique “Dual 75,” a 2-channel (fixed bias), push-pull power amp that can be adjusted for a similar combination of tubes. It was delivered biased for an EL-34 pair in one channel and a 6L6 pair in the other channel. A footswitch can be used to audition either channel or the unit can be run in “stereo.” I schlepped all of this stuff to downtown Minneapolis last summer to demonstrate the sound of tubes to the audio students at the Institute of Production & Recording (www.iprschool.com). It was a blast!
Most of the time, the Soul-o-Single (with a built-in 12-inch speaker) was used for the sonic auditions as its self-biased Class-A amp did not require readjustment for each tube. Its output tranny has three impedance taps at four, eight and 16 ohms. Soul-o's effects send was also used to drive the “dual-75” into a pair of Marshall cabs. In addition, a pair of studio monitors (either Yamaha NS-10s or a pair of Electro-Voice MS-802s) yielded some of the sweetest “clean” guitar sounds ever, with treble so delicate and effortless and a less “honky” midrange than you'd never expect from NS-10s. Placed side-by-side, the warmth generated defied their small size.
BENCH PRESS (THE TESTS)
Starting with two NOS pairs of 6V6 tubes as reference, both were JAN “military-grade” types as manufactured by Tung-Sol in 1952 and ECG/Phillips in 1986 — quite a spread, but all respectable. The initial low-level tests, via Soul-o's 4-ohm taps, were used to determine each tube's sensitivity referenced to the 6V6. (A table is included online at www.tangible-technology.com.) I then optimized Soul-o-Single's gain structure for the lowest THD+Noise: just under 0.3 percent — about 29 dB down from the 10-percent distortion samples. The signal path couldn't be more “direct”: an oscillator into the effects return, comprising one-half a 12AT7 feeding the master volume driving the output tube.
Each tube was then driven to 10-percent THD with the amp's 8-ohm setting driving a 10-ohm load, and then using the 16-ohm setting to drive the same load. This yielded an assortment of snapshots that represent the typical range of Class-A soft overdrive characteristics. The different loads definitely shift the bias so that the 8-ohm tests mostly yielded “fat” waves on the positive swing and a flat-ish line on the negative swing (the EL-84 being the exception). With the amp set to its 16-ohm tap, the upper-wave excursion got phatter, and the flat line on the negative excursion disappeared. The 16-ohm tap driving the 10-ohm load was used to create the 17-percent distortion tests. These tests represent typical asymmetrical clipping if the amp was purposefully (but not excessively) overdriven. Something like this happens naturally to smooth out transients and hot notes.
Keep in mind that a “universal” output transformer is a compromise for most of the tubes. One interesting exception: Both the 6CA7 and the EL34 seemed unfazed by the three output impedance options — there were no significant level changes, whereas all of the other tubes “reacted” to the three load impedance matching options. In all cases, a speaker's impedance on the transformer's secondary is reflected back to the tubes on the primary side. Remember that the given impedance for a speaker is “nominal” — indicating the minimum impedance above the resonance point, but before the inductance rise. This is typically constant over a small bandwidth. At resonance (a low- to low-mid frequency for the woofer), a speaker's impedance is much higher than its specified value.
About the oddest tube of the bunch was the GT-6CA7-GE. Newly manufactured on former GE equipment (in the U.S.), as is the GT-6L6-GE, it had an odd overdrive artifact, a bug that became obvious during the bench tests. Interestingly enough, an NOS GE 6L6 did not do this, but an NOS Sylvania version did, so it's obviously not a new “bug,” but could very well be an amp/tube idiosyncrasy. This is a good excuse to introduce and explain the tests that generated the pictures. (See Figures 1a, 1b and 1c.)
Okay, so we can't do audio samples in print, but the pictures should be worth a few words. Even more than the online table of bench tests, the essence of all the pix is the asymmetrical wave distortion, which generally translates into more even (octave) harmonics than odd, though unlike the triode tests, the odd harmonics are more obvious this time around.
Eddie would like to thank Rick Stevenson at Electro-Harmonix and Aspen Pittman and Steve Good at Groove Tubes for supplying the tubes and amps for this article's research; additional thanks to Doug Fearn, Charlie Fox and Russ Hamm. Get the full skinny online at www.tangible-technology.com and www.mixonline.com.