By Bryan Geyer
I returned from Korea in early 1954, and soon became absorbed in the new “hi fi” craze. Of course, that meant embracing vacuum tubes. Transistors were very new—not ready for prime time—so tubes were the sole option. Tube failures were common, and their heat would often cook some of the adjacent components, but I reasoned that those faults could be my gain if I learned basic radio/TV repair, so I built (from kits) a tube tester, a signal oscillator, a simple oscilloscope, and I bought a new Simpson model 269 multimeter. That marked the start of my 32 year career in the electronics industry.
Do recognize that vacuum tubes are not high precision parts. Why not? Well, to start, all of the tube makers present their products only by listing generic “average” or “typical” performance data. They don’t provide (or control) any of the specific operating characteristics*, so vacuum tubes of a given type can vary widely. Further, all tubes exhibit a constant, gradual, and persistent life cycle drift; plate current fades, grid bias shifts. So a tube’s functional day-to-day performance is never precisely the same. This natural cyclic drift starts when the tube initially enters service, and ends when the tube dies from cathode depletion failure—barring all of the many other modes of premature demise that might intervene (e.g.: open filaments, vacuum leaks, gassing, microphonics, atypical distortion, excessive hum/noise, and damage from external mechanical shock). As a consequence, vacuum tube circuits are not the best means to assure stable circuit performance; there’s simply no optimally constant operating phase. Regardless, for some 70 years tubes were all that we had. Circuit design got stale toward the end of that era because the chassis space available for tube sockets limited potential complexity; also because tubes are so woefully inefficient. But creative innovation blossomed when PNP silicon power transistors finally debuted in the mid 1970s, thereby making complementary differential solid state symmetry a plausible alternative.
Personal angst: In 1963 I bought a Fisher FM200B tuner, one of the premier signal-seekers of the day, but its IF stages exhibited incessant drift due to tube aging. I had to perform tedious IF realignment annually. And my 1962 Marantz 8B stereo power amplifier needed semi-annual output stage bias adjustment to hold the measured IM distortion inside 0.5%, plus I had to install four new EL34 output tubes every 2 or 3 years; that’s costly! Indeed, I got so hot to dump vacuum tubes that I finally built my own solid state power amps in 1976, when PNP silicon power transistors finally became affordable. (Refer p.39 of https://www.americanradiohistory.com/Archive-Radio-Electronics/70s/1973/Radio-Electronics-1973-03.pdf.) Hey, I was free at last!
Vacuum tube commerce has collapsed in the ensuing 43 years. All of the domestic, British, Dutch, and German producers are now either defunct (like Tung-Sol, my employer from ’57-’60), or they ceased making tubes decades ago. The entire world market for audio-type tubes is now confined solely to the needs of select audiophiles and guitar buffs, and currently fulfilled only by obscure producers in China, Russia, and Slovakia. All are without previous market recognition. The quality, consistency, and reliability of product from those arcane foreign sources is speculative, and supply will persist only as long as there’s viable demand, so the outlook for affordable access to replacement stock looks dicey. Further, this status prevails at a time when all audio engineers concur that the load-invariant advantage assured by driving the loudspeaker from an ultra-low impedance source is dependent on solid-state design. (The Zout for a solid-state power amp runs < 1/10th of the Zout for a tube amp.) Today’s audio-type vacuum tubes represent the detritus of a dead technology; it’s time to move on.
*Refer any vacuum tube specification sheet. For example, here’s the published data for type 6550, a common power output tube: https://shop.ehx.com/catalog/addimages/6550-tung-sol.pdf. Note that there are no minimum or maximum limits given for any of the various operating characteristics. (This traditional practice is in direct contrast with the semiconductor industry, wherein complete min./max. acceptance criteria is provided for almost every critical parameter on every registered device.)
BG (May 2019)