Why Tubes?

Some thoughts about vacuum, silicone, electron speed and negative feedback and some ... magic.

Nothing is perfect in the real world. Some thing are simply better than others. Are tubes better than transistors and IC? "What? Vacuum tubes?" - "There's no use anymore." - good if posted by someone who never had been listen to audio equipment build with tubes.

Seems in 2003 vacuum tubes or electron valves are still best choice and heavily used not only for audio amplifiers, but also irreplaceable in microwave ovens /each have at least one build in/, medical X-ray's, for the time being are best for picture tubes and professional hi-power/high frequency applications where tubes are unsurpassed yet. Tubes are not very good for computers and digital applications /still first electron computers was build on tubes, and probably was consuming power for the small neighborhood/; can be used, but are not as good as IC's for cell phones, miniplayers, kids toys, automation, robotics and some other areas of consumer and mini-micro professional electronics...

Why tubes anyway - they have average no more than several thousand hours between failures, they need additional energy for the heater /filaments/. Tubes waste a lot of energy as heat, need also inconvenient several hundred volts supply. Audio tubes require heavy and ugly transformers, some tubes look like ../beep/.. and are expensive. Still, tubes for almost one century are the best choice for consumer, professional audio equipment designers and musicians. The expressions "tubes sound natural", "great valve tone", "soft and mild sound" are small part of opinion of many musicians and guitar players. Why's that and comparing to what? Here's one possible explanation for the growing and glowing tube amplification charisma.

We know that the music and sound have attack and decay, tone, pitch /dynamic-harmonics-frequency/... Speed of attack/decay is covered completely by any electron tube "front of climbing" - no "sound attack" in hearing range can be faster than a vacuum tube reaction. Even with sudden input overload tube produce soft clipping instead of harsh unpleasant clip of overloaded transistor.

Well known is that every audio amplifier introduces certain amount of harmonics not present in the original musical picture - and that's we call distortion. Some folks may say that tubes have lower distortion comparing to transistors and that is true, some say tubes also have wider bandpass - true - drawback here is the output transformer /output transformer may never surpass the bandpass of directly connected speaker to tube OTL amp or even some MOSFET's/. Others may say that tubes have lower noises - not always, wider dynamic range - 100% true. And all that combined with lower audible total harmonic distortion /THD /. And the human factor magically influencing all that... All answers may be hidden in the physics of vacuum tubes - it's structure and operation.

Physically tubes are completely different than transistors. The "substance" inside them is "vacuum". All possible air has been evacuated /from a 1 000 000 000 air molecules all have been pumped out except 1/ and space inside the tube is coming closer to perfect vacuum environment. There are still a lot of air molecules /at least 100 000 000 000 air molecules per cubic inch/ left inside this "vacuum", but if achieved - that's good result. The "electron stream" has very few obstacles on the way from the cathode to the anode. The better vacuum means better tube. So in receiving tubes grid act as a "static" voltage stopper and regulator of the "electron stream". Grid, mounted between the cathode and the anode of the tube an class A will not draw current or current is so small that is negligible/if the vacuum is good/. This feature /no current drawn from the grid/ is also reducing distortions quite a bit. In Class A2 grid draws current and this changes the input voltage - as a result we have increased distortions and somewhat reduced dynamic performance. And still vacuum tube amplifier in class A2 is audibly better than transistor amp in class A. "Sweet Tube Sound". Let's see where the difference might be:

The speed of "electron stream" within properly heated vacuum tube environment depend on plate voltage. It is estimated that the average electron travels in that vacuum environment with speed between 5940 km./sec. at 100 Volts applied on the plate - to closely 60 000 km./sec. at 10 000 Volts - the higher the voltage the higher the speed. The distance plate/anode/ - cathode within the tube are rare over 1-1,5 centimeter in length /anode diameter - 3-4 cm/. So that electrons are coming to the plate/anode very fast - almost instantly aprox. 0.000000000166 sec or 0.000166 usec /9 zeros after the decimal point in seconds/ at slowest speed. At higher voltages this speed becomes almost 1/5 from the light speed - different law may apply above that speed - the electron becomes heavier, because of the speed /see Einstein/, and in audio we almost never use 10KV plate supply.

In vacuum tubes below 1% of the electrons will "bump" into something on their way... Nothing in the vacuum slows them down and form of the signal input-output remains the same.../some passive elements - capacitors, coils - in the "external" wiring could resist and change the form of the signal, but we have to live with that, no matter tube or solid state is employed. Special care should be taken for all passive elements around tubes - especially reactive ones - transformers, chokes and capacitors /for Hi-Fi amps you better forget about electrolitics/.

Transistors and IC and MOSFETs are made from different "substance" - silicon or metal-oxide. The electron travels differently within this - like a football player - looking for "empty" spaces to go through and bumping in the other team players - and that is a lot more "solid state" obstacles on the way to the "anode - collector". Electron stream "speed" in Solid State Devices /semiconductors/ is not over than few kilometers/second. Sand /silicon/ made transistors are slowest in a row - just around 10-15 meters/second per volt. Distances in transistors and IC are getting smaller and smaller - just some several microns, but even with that small distance /and reduced power/ the "electron stream" will be slower than a vacuum tube.

Most transistors are current regulated devices - the base also draw current. That changes the input voltage - which increasing distortions significantly. And all that - the slow speed and current on the base gives relatively low dynamic and higher distortions - silicon transistor as stated is much slower than tubes at the same power. There is first major difference between vacuum tube and the transistor: the dynamic/distortion combined is better in tubes. So if you like transistors go with MOSFET. MOSFET's are not perfect but operate differently /"similar" to vacuum tube operation/ - they are faster and better than silicone and also does not produce as much clicking noises as the sand made power transistors does.

High levels transient distortions are often associated with class B amplifiers - but "sand" made amp may produce it even in class A. Also solid states may become overloaded easily - sudden overload will clip badly producing annoying sounds that means high level of nonlinear distortion. Here usually is employed negative feedback /NF/ to help reduce this effect.

All solid state made amplifiers have some kind of deep negative feedback, needed to reduce the distortions inheritant for those devices. The negative feedback help to reduce the THD on paper at the same harming the dynamic range and adding upper level harmonics - very unpleasant coloration of the sound. And as the French say - sound makes the music. Transistor or IC amps are "dead sand" without feedback of some kind - the distortion levels may surpass 15 % easily at low levels, and some 50 % at max power.

Transistors and IC cannot live without feedback - in conditions, where no NF is used, operation of the transistor amplifier may come to the point where the transistor may blow up. Designing power transistor amp with no NF we should use some very low power specifications - transistors to be secured 1000 %. For example if we construct transistor amplifier having 5 Watt undistorted /1-2% THD/ output without feedback the calculated output should be at least 50 Watt. That's no guarantee how such system will perform and what we should expect as sound quality.

Due to large currents and low voltages all transistor and IC amps also need extremely large electrolytic capacitors - known to be worse choice for audio, no matter if servicing the power supply /smoothing/ or some blocking functions. Electrolitics are producing noises and non-linearity in the whole system - and that's big drawback for transistor amps and MOSFET. Less smoothing capacity will increase transients, because of the large currents drawn from the power supply. Using paper in oil capacitors here /to reduce noises and nonlinearity/ is unwise - the size and price will be unbearable, better buy few tube amplifiers instead.

High voltages and low currents used in tube amps may easily be smooth or block with not very big paper in oil capacitors, known to have good "speed", much less noises and very good linearity. Power supplies silicon diodes may also spoil the music if used in tube amplifier, so building good tube amp better use vacuum rectifiers only corresponding chokes and paper capacitors.

Any musical instrument have a "attack", "sustain" and decay time. "Attack" could be very fast - as loud punch of the piano key. Fast "attack" can be simulated with squire wave test input and scope the output on every amplifier - tube or solid state. With the dual trace scope you can monitor simultaneously input and output at different levels. In good amp the output should not look much different than input for frequencies even below or above hearing range at all input levels. Tube amplifiers are known to have a very good dynamic range. In best amps - the output should follow the input so fast that no curving will appear on the scope. The feedback sometimes tend to improve this test and fool the eye, so best way to establish the quality of the reproduction is to listen to the equipment.

To be sure that you are listening music /not what transistor audio equipment designers suggest for music/ we need probably all vacuum triodes made amp with no negative feedback. Triode amplifiers does not require negative feedback /NF/ of any kind because the THD with carefully designed SE triode in class A - may not go over 2 - 3 % at max power/dominantly second harmonic/. All you need if you have such amp is very good speakers, but for that some other time...Some authors in their efforts to explain why vacuum triodes have such low distortions will tell you that a triode already has "internal or structural feedback" . Keep in mind that transistor was made to replace the vacuum triode and was constructed with different physics but almost same math description...

Push-pull vacuum tube amplifier /if properly build/ may have similar results with a bit different picture of harmonics /mainly 3-rd harmonic/. Negative feedback which is option for push-pull tubes is a must for transistor amplifiers - 15 % or more THD is something usual for silicone without feedback. Solid state transistor or IC made amps will also follow the input /slower than tube amp -curving may appear in the output/ and will employ deep NF to reduce the distortions which harming the dynamic even more. Because of the feedback the dynamic range may drop at least 8-10 dB, changing at the same time phasing of the signal and adding some upper level harmonics, making the reaction of the system even slower. And music is gone - who like to drive a car with hand break on.

Total "all stages" deep feedback may be very good for transistor amplifiers, ghetto blasters, automatics and robotics, but not for Hi-Fi Audio systems. At some degree one stage 4-5 dB NF will not ruin the music picture if only employed around output tube /usually power pentodes - they have twice higher than triode distortions/ and one stage NF may improve THD figure significantly, but two or more stages feedback makes the system behave like a pound of soft dough - you pull one end but unpredictable getting thinner on the other end or in the middle and getting sticky. NF may create also unpredictable intermodulation and transient distortions, which is worse than simple reduced dynamic range. Music reproduced from such "all stages" deep NF amp sound machined and mechanical no matter tube or transistor /transistor is worse thou/.

Symphonic orchestra /let's say 90 dB dynamic range - or less if badly recorded/ is good example for how proper reproduction should behave. Playing pianissimo and immediate "attack" - instant fortissimo or the whole orchestra is going crescendo /very loud/ - that often happens in the simphony. All good amplifiers will cover this dynamic range at their rated power, but only the best amps and speakers will perform it without noticeable distortion at this instant higher level. Here's the strength of properly made vacuum tube amps - they will follow the music with no severe dynamic/distortion combined at overload, or added "clicking noises" and "metal or sand" coloration as transistor amp does. That's why the true amplifier should be made with VACUUM TUBES without "all stages" feedback and if possible no feedback at all. But for that you need very good output transformer and proper schematic design.

Why tubes, anyway? Some folks may like red wine, some vodka... Some may like the concert sound of the opera hall, others of country music on a portable radio. Nothing wrong with it. Anyone should judge for himself. Myself...I like tube made amps and red wine. Somebody else may love transistor sound and white wine. And why not? But I hope that tube amplifier fans are growing number and more and more people are joining the club...

About some magic: Is our expectations from the tube amplifier is improving it's qualities? Seems magic.

- it is well known that the observation and expectations of some process of chemistry or physics have influence on the results of the process. Is that... magic.

- I'm not sure if expectations and observation of the performance of the tube build amplifier gives improvement in the performance... or that goes to all amps... It is unmeasurable at that point , but it is worth trying.

If you have experience or know more about this "magic" subject drop me a line...

P.S. There have been questions I've received about use of magnetic fields in vacuum tubes. Magnetic fields may influence the work of vacuum tube - may introduce hum /through the heater for example/, buzz or have other side effects. Magnetic field may also be useful. It is long used in vacuum tubes since the development of the magnetron. Magnetic field may only change direction of the electrons attracting for example the "electron stream", but cannot slow or accelerate the electrons - it may just change the direction or maximum block it. To use a "magnetic grid" instead of normal one we need much more input energy /power/ than usual, so that seems useless for audio equipment. Magnetic field may be used to build /imaginable at that point/ "beam power triode" if used properly with some permanent magnets - but constructing tubes is another field of research, not subject of this article...

Some "new" tube designs, I've only heard about may use a some sort of magnetic field for extra regulation or "improvement"...but I haven't seen one yet.





All rights reserved. © 2003