Listening pleasure...
Listening pleasure is affected by the the equipment we own and how its performance integrates with the
room it is in. Regardless of age we all want to transport out minds away from the day to day drudgery of life.
I find the best "transport of the soul" occurs at a live performance without any electronic equipment and it is not all so easy to re-create the atmosphere of the concert venue in ones home. 
Our pleasure with replay of recorded music depends on our listening room's character, the recording methods used, whether the artistes were in form on the day, our moods, choice of music, the source of the music, cd, vinyl, tape, radio etc,
speakers chosen for the job, amplifiers, cables and other ancillary items.
Of all these listed things, the only things I may provide to enhance the experience of music are the amplifiers and perhaps the speakers, and some sensible advice about the rest.

Class A amplifier basics...
My experience leads me to believe that the most enjoyable amplifier sound quality results with the use of vacuum tubes with plenty of class A operation.
And so what is class A?
There are a fair number of classes of operation of the active devices such as tubes, mosfets and transistors.
The classes of operation which concern us in audio are A, AB and B. These classes are for "linear circuits"
which do not employ digital switching or pulse width modulation.

For those non technical minds, class A operation of amplifier devices has been used since the vacuum tube was invented in about 1903. At that time it was found that the world's first active amplifier device, the triode, could have the current flow
between anode and cathode either increased or decreased by application of an electrostatic field from a third "grid"
electrode in a vacuum tube.
The first amplifier was a class A single ended triode amplifier. Its first use was to convert small amounts of audio
or radio frequency power in an input circuit to larger amounts of power in output circuits with a good linear
relationship of voltage amplitude change between input and output.

The single class A device such as a triode is set up with a steady dc idle current flow from anode the cathode and equal to
about 1/2 way between the maximum and minimum possible flows for the triode.  The grid which has no current flow
input or output has an applied voltage which controls the electron current flow between anode and cathode.
Generally, triodes or other tubes do not directly couple to the speaker and cannot cope with the magnitude of speaker currents so a transformer is placed between tube anode and speaker to magnetically relate the tube current and voltage changes to thoseoccuring at the the speaker. The transformation of high voltage x low current changes within the tube to
low voltage high current in the speaker is a manifestation of magnetic phenomona and has no direct link to class A action in the
tubes.
 
In class A operation the idle current flow is varied by a small input voltage signal so that the idle current then increases and reduces either side of the idle reference value. It does so, in a manner which exploits the most linear region of operation of the device, and so there is no sharp switching on and off of current as in the case of class B amplifiers.
The current in the vaccum tube is called the anode current, and consists of trillions of electrons flowing from a heated cathode of special emissive coating to an anode with a positive voltage which attracts the electrons which are negatively charged.
A positive change in grid voltage causes more current to flow, a negative voltage change in grid voltage causes
less current. If the grid voltage becomes negative enough, the anode current can be turned right off, and if the grid is made
positive enough the anode current reaches a maximum determined by the the cathode emission cabability and anode voltage
and its distance from the cathode. The grid is a mesh of wire placed between the anode and cathode. When the grid voltage
is raised to a positive voltage, ie, above zerovolts, the grid attracts electrons and the grid has grid current.
In most triodes the grid is at a negative voltage called the grid bas voltage with respect to the cathode voltage.
The change of grid voltage in most circuits is limited to going from its idle negative voltage to 0V or to
twice its idle negative voltage.

Conventions about current flow were established before anyone knew what actually current was, ie, movement of electrons.
The difference in electric charge or potential values was known, and could be positive or negative, and it was assumed current had one direction, from positive to negative, so we say anode current flows to the cathode. But in fact the
electrons flow from cathode to anode. The old conventions are still used, and the context should be noted and this will keep you on your mental toes when considering discussion about amplifiers and how they work.

Class A action in a single triode or other types of tubes such as beam tetrodes or pentodes is like one man using a bush saw to cut a log. His motion of the saw is slightly different in each direction, and if we were to draw a graph of his motion with time
we may see that the motion is not a pure sine wave or saw tooth wave either, and that there is an amount of mainly
even numbered harmonics above the frequency of the saw cutting strokes.
And so it is with a triode, the main distortion artifacts are even numbered, 2H, 4H and so on.

Class A action can be achieved with two tubes each acting as the single tube but with complementary action.

Imagine our bushman who is cutting the log is joined by a friend  who takes the other end of the bush saw and together
they push and pull the saw alternately, so while one is pushing, the other is pulling. If the motions of the saw could be
graphed we would see a more equal motion each side of the centre point, and the distortions in each man's contributions
to the saw motion are largely cancelled out, and so it is with two tubes working in a push pull circuit, where each one is set up like a single ended class A tube but which each has a signal production which is 180 degrees out of phase.
The transfomer allows the out of phase signals to be combined to produce a single phased signal for a speaker
and for the power of each tube to become combined into twice the power of one tube.
The two men will cut a log faster than one man alone.

Although even numbered  harmonics are cancelled by push pull action there is still distortion artifacts produced either by the two men on the saw or by the vacuum tubes. They are mainly odd numbered harmonics, 3H, 5H, 7H etc.

With class A PP tubed circuits the odd order distortion products are usually about 1/4 of the even numbered
harmonics produced in a single tube circuit, or where paralleled single tubes are used, which is like having both
men on the same side of the saw to cut the log.

Class B.
Class B action in devices is usually confined to a pair of devices working in push pull.
There is a very awkward analogy to the two men cutting the log. Imagine one man only applies force to the his end of the saw
between the half way point  then towards himself and back to the centre. Then he lets go of the saw. Immediately
the other man does the same thing, and the first man takes over for 1/2 a cycle of the cutting wave.
This is class B action. Now if you asked the men to cut a log like this you'd have a union strike on your hands.
They would hate to have to stop-start their sawing like this. The graph of the wave motion of the saw would show
serious "glitches" at the half way point of the saw as each man tried vainly to time the gripping of the saw and the letting go.
The wave form of the sawing would have "switching artifacts", or crossover distortions, usually consisting of many high numbered  and odd numbered harmonics.
The class B PP amp has both devices with no anode current flow at idle, ie, with no signal. Current only flows in one of the
devices when the signal either moves positive or negative, and at the zero crossing point one device switches off
and the other switches on.

But with electrionic devices, there are no unions in control of working conditions and the tubes can switch on and off
at a rate up to hundreds of megahertz. Nevertheless there are considerable artifacts produced at the crossover region
of operation and a class B PP circuit with tubes has a much higher number of odd numbered harmonics.
So the devices don't mind switching on and off but they make a mess of doing it like the guys on the saw.
Tubes and transistors both suffer the same problem.

In a class B tube amp the two PP tubes are connected to the load via an output transformer in the same way as a class A circuit
except that the idle current has been reduced to zero. To reduce the crossover distortion the anode current at idle is increased to an intermediate value which is between no current and class A current. So each tube can then work
in class A for more than 1/2 of the part of the wave cycle, but becomes switched off for the extreme wave form changes.
This is class AB action, and is analgous to the two guys on the log working together on the saw for
say 1/2 the total swing distance but each man relaxes for the last 1/2 of the saw travel alternately.
Many PA amps with high power were constructed with a low idle current in each tube so that the gross effects of pure
class B switching distortion were avoided but only a small amount of class A was produced. The efficiency of low bias current class AB is high and if the distortion is reduced with NFB then the sound outcome is fine for PA.

Class AB.
This intermediate way of working is called class AB action, and it is used in 99% of PP linear amplifiers today.
The distortion is very nearly as low as the pure class A PP amp for the first few watts of power but above
the first few watts the distortion radidly increases to much more than the class A PP amp.
But with PP triode and ultralinear amplifiers the portion of class A power available before crossover distortion occurs
is usually enough to cover all the needs of most listeners and the class AB high power region of operation
is reserved for drum beats and cannon shots in Beethoven's 1812 overture. The use of class AB allows for
less idle power, ie, dc voltage x dc current, in the tubes and for greater efficiency and long tube life.
You can't beat class A amplification for maximum fidelity.
However, high powered pure class A amps are rarely required, since most people only use a few watts of audio power in their day to day listening. Therefore, class AB becomes the most sensible way to allow us to hear extremely fine sound, and yet enjoy the luxury of having an enormous power reserve to cater for the transients in highly dynamic music. In class AB, at least the first 20% of the AB power power maximum is pure class A which is determined by the idle current in the devices, and then above this threshold level each tube in a push pull amp output stage takes turns to produce +ve and -ve extremes of voltage swings by itself.  All my class AB1 push pull amps use a large % of class A in their operation.
This results in substantially linear (distortion free ) amplifiers even without any applied negative feedback to correct distortion or reduce noise and enough class A power to cover all the music except very short lived transient drum beats.
Triodes are most inherently linear whilst operating in class A, and while transfering across to AB without generating too many evil sounding harmonic artefacts because their switching off is a gradual process not a sudden one as with beam or
pentode tubes.
A typical 50watt PP amp using a pair of 6550 will be set up to make 54 watts AB into 4 ohms, with about 20 watts of class A
and then 35 watts of nearly all class A into 8 ohms, so that all that we listen to will be handled by the class A action,
regardless of the load value.

More info about A and AB operation is elswhere on this site, and of course much is available from old books on tube audio written before 1960 and by my father's generation.

Nearly all small signal audio amplifiers use cascaded stages of devices set up as I have described above, sometimes they work as a single unit alone, single ended, or in pairs within a push-pull circuit.
This is true of nearly all tube or solid state small signal circuits, such as phono amps, microphone amps, line level preamps and the input stages of power amps.
The class A signal circuits always run warm and are inefficient, but for small signal and/or low power use the inefficiency is of no concern  and the benefits are low distortion, and wonderful music.

Some solid history and class B....
Since the advent of solid state circuitry in mainstream audio engineering by 1960, the operation of output stages has drifted to nearly class B operation where two devices are crudely switched on and off to amplify the positive and negative parts of a given wave cycle. The solid state devices only have a few milliamps of idle current each and thus produce a negligible amount of class A power.
The voltage sent from the active terminal on your amp, ( the red terminal ) to your speakers varies by going +ve or -ve in voltage while the ground terminal of your amp, ( the black terminal ) is kept to a zero signal reference level known as 0V.

The crude class B operation of solid state devices is largely tamed by a simple technique of corrective
circuitry known "negative feedback".
Class B operation is much more efficient than having hot running class A output devices which consume considerable power even when not producing any music, since the devices in class B have almost no idle current flow.
Efficiency itself is renowned for sounding not so good in the domestic scene, but nevertheless nearly all amps
are virtually class B, efficient, and cleverly biased for low crossover distortion and set up with plenty of NFB.
To make 40 watts of audio power in class A, you need to have about 90 watts of input power power at least dissipated in the output devices, regardless of how the output devices are hooked up. In a 40 watt solid state class B amp the input power
may only be 2 watts. The cost of production of a 300 watt class B amp is about the same as a 50 watt class A amp with solid state. The first solid state class B amps with a tiny idle current were truly awful amps with a bad sonic signature. Good music was sacrificed for efficiency.

However, due to device manufacture and circuit topology improvements over 45 years the solid state amps have become a lot better than they were, and at least they have impressively low distortion measurements.

But now trends are toward pulse width modulation and digital signals and I don't know where we are heading with
fidelity. I think tubes will stay as long as production can continue where it does, ie, in mainly emerging or stagnant economies like those of former "iron cutain countries" such as Russia.
Digital and PWM amps will replace the existing class B in at least in the mass market budget amps and home theatre amps.
The 100 watt rated PWM amps I have seen and heard are remarkably light and small
and have 95% efficiency so large heatsinks are not needed, and they sound as well as many "conventional"
class B solid state amps. 

Yes I did it too....
I have made a couple of fine solid state amps with a high amount of class A, and extremely low distortion measurements, but I still think the tubes have the edge if judged in subjective A/B listening comparisons.

Just measurements?....
In the past total harmonic distortion ( THD ) measurements were very important to people trying to decide which
amplifier to purchase because it was assumed that what measured best also sounded best.
But the THD measurements do not tell the whole story about why so many people I say to me that I am not wasting time building tube amps. I have done numerous listening tests with tube amps where the THD was only just below 0.1% on a tube amp, and only barely below audibility, and yet the tubes yield a more natural and realistic audio experience with life, warmth, body, bloom, dynamics, detail and conveyed emotion which seem better preserved than a solid state amp which may measure with THD at 0.003%.
THD is a starting point for measurement and once we know how much it is with a fixed sine wave we can calculate the much more dreadful intermodulation distortion harmonic products, IMD, produced by the presences of more than one signal frequency as a result of the non-linearity expressed in terms of a THD figure.
Basically the higher THD, the higher IMD which is like a background hashy noise in the music,
something grating or edgy about the music, like background ripping sound rising and falling in level with the music.
It seems to be subjectively worst in solid state amps so they seem to need all the negative feedback that can be applied, but in tube amps the IMD which measures much higher than in an SS amp the subjective effect sems to be is much less, and especially in the case of single ended amps with a single output tube operating in class A.
I can only suggest that the tube amp's poor measurements  just do not matter because the spectral nature of its IMD distortion is less painful or noticeable to ears than the lesser quantity of spectrally more complex IMD produced by a solid state amplifier.
The "objectevists" are welcome to believe that I am incorrect about this since they say there are virtually no THD/IMD
that can be measured easily with well made solid state amps. I leave them with their version of their gospel,
but too many listerners tell me they prefer the tubes and say the the tubes make things sound more accurrate, and less clinical, dried out, music-less, cold, harsh etc.

At a recent live concert of a pianist playing a Yamaha grand piano in a small venue, we were all impressed by the Liszt being played, and the power of the instrument. To reproduce levels like that piano did is no mean feat and will take some serious engineering; do not expect a lone 300B for each channel with low power ability speakers to be able to do it
unless the speakers are horn speakers which are perhaps 10 times the efficiency of normal dynamic speakers.
So measurements do matter; one may need twenty 300B in an amp to do justice with a grand piano played with vigour
to reproduce the levels heard when standing nearby the grand.
Of course we would never stay long beside the grand and would prefer to sit about 10 metres away where the levels are easier to bear. Anywhere further away might be nice if there was an orchestra plaing as well. And those levels don't need so many 300Bs.

Enough is enough....
I aim to ensure the music will be heard as a beautiful experience, and to be as close to the natural acoustic event as possible, so the equipment needs to measure very well, but not any better than it needs to.
This is certainly true of class A tube amps, because they are substantially linear before negative feedback is applied, so not much feedback is needed compared to solid state.  Class A amplifiers in general have THD that is proportional to output speaker voltage, so that although an SET amp with no applied loop or global negative feedback might make 5% THD at 10 watts, at 1/10 of a watt needed to reproduce a female voice with sensitive speakers the THD will be 0.5% or less and not noticeable and she will sound quite acceptable to all.
On the other hand solid state would be non-listenable without NFB, and needs as much NFB as can be applied, since the harmonic spectra produced by the devices is more objectionable than those produced by tubes.
I myself like to apply some loop or global NFB  with SET or push-pull class A amplifiers at least to ensure the
output resistance of the amp is below 0.5 ohms so it is a small fraction of the speaker impedance, so the response
flatness is not seriously compromised. This means ensuring the damping factor,  RL/Ro =  at least 8 or more.
99% of Speakers are designed to operate with amplifiers with Rout < 1 ohm.

Thus simple tube circuit topology can be employed with a minimum of devices, but which doesn't spoil music because it is too simple. I am sure that those who have the ability to read schematics will see that all my schematics show the "simple but not too simple" theme.

Negative Feedback, multi grid tubes....
In amps with multi grid output tubes without any applied loop or global feedback, the noise of tubes and power supply noise is reduced to a low level using excellent power supplies, and since class A operation is used there is substantial linearity and low noise and wide bandwidth. However unlike pure class A triode amps mentioned above some global NFB usually must be applied reduce output impedance to a low level of around 0.5 ohms or less, since without it a pure beam tetrode or pentode connected amp has an output resistance way to high and perhaps 25+ ohms at the output terminals. Even ultralinear connected output stages have Rout = say 5 ohms which is still way too high so up to 15dB of global NFB must be used, or some combination of two loops of NFB used for a total of up to 20 dB. Such amounts of NFB will reduce Rout to 0.5ohms or less and reduce THD from say 2% at clipping without NFB to 0.2%, with NFB. At a couple of watts many 50 watt class AB amps will thus make less than 0.05% THD and the sound is excellent for normal listening levels which are less than 2 watts average..
My recent designs of SE35 amps have distortion to match that of a good PP design,
since they exploit complementary distortion voltage cancellation between amp stages.

Triodes and NFB....
SE or PP Triode amps using triode connected beam/pentode tubes, KT88/EL34 etc and real triodes such as 2A3, 300B, 211 and 845 need only about 12 dB maximum of global feedback, if any at all, depending on the output transformer impedance ratio and load value to be used.
At the time I am drafting this report, march 2006, I am currently building a pair of 45 watt SE mono amps each using a pair of KR845 tubes in parallel from KR Audio in Prague. These large transmitting triodes may not need any global NFB applied because the plate resistance of these tubes combined  = 1k and is much lower than the load resistance = 6k so without any global NFB the damping factor would be tolerable. Usually with a slightly elevated Rout of say 1.5ohms,
the bass performance often improves since the speaker impedance in the low bass region is often well above the nominal speaker value so a dB or two of increased bass is heard, and because many small sized modern speakers have poor bass
for full range music a little bass boost helps!

Triodes have NFB inside them already....
Triodes or beam & pentode tubes which are triode connected have internal electrostatic negative feedback voltage acting between the anode and the electron stream and grid voltage.
In a triode, when a +ve voltage is applied to the grid, the anode voltage swings -ve with increased load current.
The +ve grid change encourages an increase in electron flow from the cathode but the -ve moving anode voltage is getting less +ve, and so electrons are less inclined to move towards the anode.

So two voltages are together working intimately to control the electron flow, and the anode voltage action opposes the action of the grid voltage. So if any distortion voltage appears at the anode which is not present at the input grid, it to is applied to the electron stream but in a way which opposes its own production, since a -ve distortion voltage would tend to cause less electrons to flow and hence tend to stop the -ve going distortion voltage appearing.

The feedback effect is at maximum when the load value is a highest number of ohms, ie, when no current change occurs
in the triode. So hence I like to use constant current source loads in signal preamp triode circuits.

The triode is said to be a voltage device, and it is true true because there will be changes to the voltages of electrodes
even when there is no current change to the idle current flow between anode and cathode.
The voltage gain in triodes is determined by the relative distances between the cathode grid and anode.
These distances determine the effects of voltage field intensities and the effect on the electron stream.

In power circuits, the best sonic outcomes with triodes is where the load values are highest possible but there is then low efficiency, so a balance has to be struck between loading and maximum power output capability.
My pages on load matching loads to power tubes emphasize the need for careful loading of tubes.

The feedback in triodes is due to the electrostatic voltage effects and in itself the current/voltage relationships is not perfectly linear beacuse the anode current variation with anode voltage is proportional to the square root of a number cubed.
Professor Child describes NFB in triodes better than I can in Terman's 1937 book, Radio Engineering which is
something all the anti NFB fetishists all ought to read twice.

The screen in multigrid tubes interrupts the voltage field from the anode from affecting the electron stream so beam and pentode tubes have higher gain than triodes, but much worse spectral distortion, hence external loops of NFB must be connected.
The jury is still out after 60 years and has not given its verdict on whether beam or pentode tubes can perform as well as triodes with appropriate NFB applied.

NFB and sales....
There is no need for me to copy the absurd efforts of many 1950 makers to win sales by achieving vanishingly small THD figures by applying 30 dB or more of feedback, and no reason to copy designs with "unity gain" output stages such as the McIntosh range, or some EAR designs, which use over 40 dB of NFB, such as EAR509.
In some cases this much feedback combined with rather non linear tubes such as PL509 working in very close to class B conditions is counter productive and can make the  sound subjectively worse than amps with much less maximum output power but with more class A and better tube choice and much less NFB.

Circuit development....
I have slowly developed circuits which I know sound better than all previous trials of  derivatives of the main amplifier topologies invented by 1960, such as Peter Walker's Quad II with CFB windings in the output transformer, D.T.N Williamson's landmark design of 1947, Mullard's 520 design, and the McIntosh, and finally various versions of single ended amps where I continue to explore to extract the finest musical experience possible.

These designs all have their pros and cons, and since the early 1990s a complete renaissance has occurred with the advent of the Single Ended amp with transmitting tubes, multiple tubes in parallel, or just a lone 2A3 or 300B.

I hope I have promoted further developments of circuits and topologies which work better than the amps made in 1955 when the mean minded accountants employed by audio gear manufacturers were alive and well and minimized the quality of what was sold to the public.

My favoured PP amplifier uses about 12.5% to 20% of CFB windings on the OPT with KT88/6550/KT90, but uses an SET input stage and pair of EL84 in triode as the balanced longtail pair driver.
 
Amplifier manufacturers would never have done this in 1955, since accountants ruled the costs of production and stifled
many a good design from ever being produced. Most mainstream manufacture resulted in amps being made down to a price, not up to a quality. The EL84 is a small nine pin "mini" power pentode which makes a truly excellent sounding triode driver tube to drive output tubes when connected to work as a triode, since it is almost exactly equivalent to 5 halves of a 6CG7 or 6SN7 all wired in parallel. The production of lowest common denominator quality is still finely controlled by accountants, and much imaginative design is still being stifled, and only ever finds itself expressed in some expensive hi-end brands, or in the efforts of those dedicated to to perfection in sound quality, but who do not mass produce or have any damned accountants on the payroll. I can add, subtract, multiply and divide, and I don't employ a bean counter to be able to compete with the lowest
quality amplifiers.

Iron cored wound components...
My attitudes have led me to wind all my own output transformers and chokes and power transformers because the sonic quality available is better than can be generally be purchased off the shelf, and the manner in which off the shelf  iron cored wound components are now prepared leaves a lot to be desired unless one pays many times the price of the cheapest items.
I have never been able to get anyone to wind my designs of output transformers etc for a reasonable price and for delivery when I want them.
For high powered designs like my 300 watt amps there is nobody who prepares any ready made wound parts and all must be custom designed and wound, and it did not take me long to build myself a lathe and practice the skills of winding transformers. In earlier times there were many people involved in the transformer winding trade but like many trades in Australia transformer winders of any capability are hard to find because of the imports of transformers made in countries where labour costs are 1/100 of what they are in my country. All the tradesmen and women of earlier times when demand was high have
mostly grown old and retired or have passed away. The quality of asian made transformers leaves a lot to be desired.
But the raw materials I source have never been better and the sound with my components is second to none and the measurements are better than the text books would predict. Part of my earnings made when I sell an amp comes from the work of winding transformers. I do not have to pay someone else a truckload of cash to do it.

Future amplifiers...
In 2004 I was able to design a new batch of amplifiers and I invested a small fortune to employ a local sheet metal working company, CanFab, to make a range of machine pressed and fully welded chassis with 2 mm thick steel plate so that the new range will all have chassis quality equal to the best anywhere in the world.
Thus the new range of amplifiers will depart from the older models and have a more uniform common appearance, although still be able to purchased with a wide selection of tube types.
The types of chassis I hace are shown on my page 'future amplifiers'.
All power amps will only be supplied as mono blocks, because a stereo amp weighing at least 35 Kg is a hazardous weight which could cause a back injury. Mono block construction with all transformers properly potted is also part of a "no compromise" attitude.  The amplifier market is littered with cheap models which I could never compete with, so I really only wish to compete with the best available brands. I hope people to see the value I offer, as well as hear it.

Preamps....
All preamp circuitry is 100 % class A operation, and since the dynamic range of tubes is so enormous, the operation at low output voltages assures negligible distortion production and sound you would die for, without using enormous quantities negative feedback error correction techniques.
Triodes are predominantly used, in a variety of topologies to suit the purpose.
Whilst some would say you cannot beat class A integrated silicon chip opamps, I have yet to hear anything as emotionally engaging, and as warmly inviting as a well made tube preamp.
The only time I have included something solid state as a voltage amplifier device in a signal path is in the case of  my latest moving coil cartridge phono amp designs where the first device at the input is a 2SK369 j-fet which only has to produce a few millivolts of output. Its a suitably quiet enough device to use with such small input signals, quieter than all triodes known
and distortion is negligible at such low levels.

Tubes are not bad....
They have been with us now for over 100 years, since the first triode was made in 1903.
The power tubes have have been getting better for audio use ever since and world production is said to be increasing at up to about 10% per annum despite the opposition from usually much cheaper solid state.
The best small signal tubes are NOS made 30 years ago, but many fine samples of  Russian production tubes sound OK.
Computer control of the manufacturing, selection, and quality control is common nowadays.
Some lesser known types have begun to be re-produced after a lull of 35 years.

Tubes wear out. Power tubes usually last about 5,000 hours and small signal tubes up to 10,000 hrs.
If a power amp is used 2 hrs a day for 1 year, that is 730 hrs, so you can get maybe 7 years from a pair of tubes.
One may get an occasional early random failure of a tube. Plug a new one in. Replacements are a part of life.
If you break a plate or glass while washing dishes is it a major drama?
The reliability of well made vacuum tube gear is much improved since the 1950s because of the quality
of much more reliable surrounding components such as capacitors.
If one has to spend $250 after 6 years to re-tube an amp, just consider the fixes so often needed for solid state amplifiers.
I get one amp a week coming here for a fix, many have been "fixed" before, and it nearly always means difficult
and tedious diagnosis, careful desoldering of fused solid state devices and replacement with care so as not to damage
fragile printed circuit boards.
So please do not insist that solid state is always more reliable and that it lasts forever.
I try to set up the output tubes in my amplifiers so they are never working at more than 70% of the maximum design ratings.
This  tends to extend the life of the tubes.

You can't beat wide bandwidth....
A good amp will have wide bandwidth before the application of negative feedback.
The bandwidth before NFB is applied should be at least 15 Hz to 50 kHz  at full rated power to the nominal load resistance, with less than 30 degrees of phase shift between 20 Hz and 20 kHz.  This means the amp has a small delay within the amp before negative feedback is connected, and complete stability is unconditional, when feedback is connected.
Without any global NFB 99% of all amplifiers, including solid state, will display a roll off at HF whose phase characteristics must be tailored to reduce phase at extreme HF, before NFB is applied. Some reduction of open loop gain beginning above
30kHz is not only desirable, but essential to stabilise an amp when used with capacitive loads, or no load at all. In good tube amps the open loop gain ( gain without global NFB ) at 20Hz and 20kHz is only 1dB below the 1 kHz level and thus when NFB is applied, response is flattened, phase shift reduced, stability is unconditional, and the output impedance is no lower at
20kHz than at 1kHz.

First we must get the best operating points for the active devices, and then perhaps there is some merit in Auricaps, or in some exotic rare NOS Siemans triodes.

In may 2006 I tried an audition of Auricaps in one channel and Wima polypropylene caps in the other channel and heard
no sound change when using a switched mono signal between channels. Nor did the other guy at the AB test.
At first he said he liked the selection I had made to the channel with Auricaps. I let him choose which he preferred
without knowing which caps and channel I had selected. Then later later he selected the "better" channel with Wimas, twice, and adamantly, and I had to convey the sad news that he was not picking which capacitors were Auricaps any more often than
random chance would predict. I am sure that if I'd said I I'd made a change between caps and actually changed nothing that he
may have said he heard a change. I didn't go that far in my trickery.
So why did the Wimas sound the same as the Auricaps? I reckon its because the same guy made both channels :-)

However, I am very happy to fit whatever "special" parts someone wants me to fit, just as long as they are prepared to pay a the extra costs. Everyone must feel happy with component choice made acording to their belief.

Tube choice, and NOS.
The tube choice does make a difference, ( and there is much chatter amoung audiophiles about which brand of 6CG7 sounds the best after the NOS Siemans :-)
I am wary sometimes of NOS because one never really knows the real history of tubes one buys across the Internet .
Honesty amoung tube traders is by no means universal, and nobody would know if the NOS small signal tubes they buy
have already been used for 1,000 hours.  Power tubes will show slightly dulled edges in the gettering within the tube
after 1,000hrs, and a used power tube is usually easily spotted. This insinuates that some folks will sell their expensive
tubes as new after 1,000 hrs if the market price for them is high, as would be the case for NOS GEC KT66.
Say they paid $100 each for the NOS originally, and get $95 on resale after 1,000 hrs. Then they buy another set of
NOS. But were they also conned about NOS from the dude who sells? Who knows.

I get too many good reports about the Russian made power tubes and I will not use NOS power tubes unless
someone places them on my bench and then I refund them the price I have allowed for the russian tubes in a new amp.
I cannot be fairer than that, and it suits me because I don't  have to gurantee tubes for 90 days that I did not supply!

NOS tubes which have had no use since manufacture may have been sitting on a shelf for 45 years and micro stresses over that
period could have fatigued the glass or allowed some gas entry. When fired up after such a long sleep some rapidly fail.
If I have provided a NOS signal triode, typically 6CG7 which I do know is genuine NOS, I always provide a replacement freely on an exchange basis if there is any early failure.
It is not something I have to do very often.

Careful tube testing!!!!!
The answer to ensuring people get the best possible performance from tubes I supply is to test the tubes well before sale.
I have  a couple of test rigs where I set up the tubes in a single ended amplifier with adjustable settings for
bias, idle current, anode voltage supply, screen voltage supply and load. I then measure the gains of the tube with and without a load or with different loads and using algebra I work out the Ra, Gm, and µ for the tube under test.
A Cathode Ray Oscilliscope is used to inspect the wave forms and a distortion meter is used to measure thd.
With small signal tubes after measuring voltage gains with different loads for Ra, Gm and µ, the grid is grounded
and the anode noise is amplified 1,600 times by a low noise amplifier which is band pass limited from
near dc to 20kHz. The noise displayed on the CRO screen, and fed to a power amp and speaker so I can hear it.
Needless to say the power supply to the tube under test is very quiet and the heater supply is DC.
The noise produced by the noise amp is negligible compared to the noise from any tube under test.

Where say a 12AU7 is being tested the gain from grid to anode may be say 14. With the following signal pre-amp,
any noise at the input grid is amplified 14 x 1,600 times, ie, x 22,400.
If there is 2uV of grid input noise with dc to 20kHz in the 12AU7, then the signal measured at the preamp output
= 22,400 x 2uV = 44.8 mV, easily measured and seen on the CRO. The noise should be a constant amplitude
without large hum levels and without the meter needle jerking about as levels bounce due to gas or poisoned cathodes,
or from cathodes where electron emission is failing or become sporadic.

So if I see 45mV of noise at the noise amp output with 12AU7, I know the equivalent input noise is 2uV, a quite good figure
for any small signal tube. Often when I make this test of a 12AU7 that I have removed from an amp or that someone has given me, the noise level measured can be 5 times the 45mV level and the tube is not in great shape. The tube is given a slight
tap with a pencil. The resulting noise signal should cease quickly, and sound like a dull thud in a speaker used to monitor
the noise with my ears. Where the pencil excites a large noise, and with a following long ring tone like a bell, the tube is then
deemed to be microphonic. Some are so bad that whistling to the tube will cause whistle tone to display a sine wave on the CRO and be easily heard in the monitoring speaker. Some tubes are so microphonic that one may get acoustic
feedback which will typically be a loud audible tone between of 500Hz to 5kHz when the gain is turned up even without music.
Many microphonic tubes are sensitive to specific frequencies at which the internal electrode structure vibrate,
rather like a tuning fork.
Microphonic tubes in phono preamps are most prone to acoustic feedback, and even where no feedback could occur but where tubes are somewhat microphonic then the sound could be coloured by the tube being modulated by the
acoustic vibrations from speakers. Thus a tube which which tended to vibrate at middle C would sound OK
if the key of the music was in C but not if in B flat.
Therefore phono preamps are best placed away from speakers. The common placement of between two speakers
on an equipment stand is not always the best.

I have seen NOS tubes fail such investigative testing, and mil-spec tubes that were anything but non microphonic or quiet.
But I have also seen 12AU7 and 6CG7 which sometimes measure only 1uV of input noise and with low microphony
compared to the rest. Usually these there the best sounding, and are dead quiet in any amp.

Tube testers.
I do not own a tube tester as was used in the old days and which was very convenient to tell a tech if there was life in a tube.
Complex radio mixer tubes would be hard to measure the way I do.
I test radio tubes in the radio by measuring gains or replacement with a known good tube. I do not have to test TV sets
where tubes were used only very briefly, and now there are no transmissions to suit the old TV sets. Thank goodness.
With so many varieties of tubes by 1960, the tester made a lot of sense, but it cannot tell us all about the tube that a fussy audiophile should know if he is to hear noise free music from a phono stage without effects of microphony affecting the sound quality.

Damper rings and tube addons.
Damper rings and other tube attachments have a very small effect on a tube that is microphonic.
Some metal enclosures of small signal triodes is beneficial by damping tube motion screening the tube from hums but it otherwise causes the tube to run hotter, and hot tubes mean shorter life. But with many twin triodes the idle power level is insignificant and such tubes as 12AX7 used in a phono stage the screening cans prevent the 12AX7 being hummy without the cans. I would never use rippled metal cylinders that are not grounded.

The proper place for tube that fail my test is in the bin, or returned from whence they came with a demand for a refund.

People sometimes give me a box of 50 tubes.
I always welcome such electronic orphans. They usually come with many in the old 1958 cartons and with an assurrance they are "all really good".
Unfortunately, 70% of such tubes are usually the old tubes which were removed from a TV or radio
for one of many reasons. Old techs hoarded their pulls from old gear; maybe they hoped some miracle treatment
would rejuvenate the failed tube which could be sold again as new.
The miracle treatment they witnessed was the advent of solid state, which led them to the eye specialist for glasses to see what they were doing, and some heavy re-learning, and a lot more work to deal with "queer circuits" used in the early solid state gear.
So I usually cannot afford to pay well for such aquisitions of old stock tubes because I have to take the
the time to test each one if it actually is useful for audio.
Maybe there were about 7,000 varieties of vacuum tube produced in 1960 but I could easily
build fabulous amps if I was limited to no more than about 50 type numbers.

Add on bits and pieces....
Many of these plug-ins and add-ons such as cables and power conditioners do absolutely nothing except make a listener feel good about what he has been able to achieve by purchasing something for his or her system.
Whilst any expensive cable or tweak is not going to do any harm to a system, it is important to get the room, then the speakers, and then the amplifiers and source up to a high standard first, in that order before worrying about cables and other minor tweaks. Music at home is an armchair experience, and sometimes the thought of re-plastering the walls and ceiling with an extra layer of  gyprock, and adding acoustic panels, buying a nice thick large floor rug to reduce reflections and reverbrations, or moving to a better house is painful compared to visiting a downtown store and shopping for new cables.
But once a total listening environment is established with some effort, music can be reproduced with great satisfaction, and with some fondness and respect for the hardware, especially if the amps are tubed.
Probably the greatest contribution to vinyl replay is to have clean records. They become dirty with gunk adhering to the grooves and a proper record cleaner with vacuum brush used with proper cleaning fluid is the only way to do wonders for
vinyl which then may play better than any other medium, despite the advent of the CD.

Audio news groups....
I could go on, but how and what I think can be witnessed in the occasional postings I make to the discussion groups such as rec.audio.tubes, where there is some gentlemanly discussion sometimes and more rarely at aus.hi-fi, where the discussions are often full of acrimony and bovine manure.  Since the end of 2004, diy discussions on Usernet publicly accessible
news groups has declined a lot mainly due to the acidic behaviour of a few who flail around madly with abuse for anyone
who makes the slightest error when discussing audio craft. The flames are fanned when those who have made a technically wrong statement try to save their skin and not look like fools. Then there are those who let fly with the most
ill-informed and ill mannered and grandiose lies. Without moderators, the public discussion groups become sewers
and many have left forums such as rec.audio.tubes to chat peacefully on a myriad of little groups where moderators
control the behaviour. Unfortunately, many of those in the lesser know groups remain at the beginner level for years
and the folks in them are allergic to their pet theories being challenged, even when I have done it politely, so I don't have a presence on such little cliques of audio interested people without lot to say which would keep me interested.

In preparing this 4th Edition of this website I have hoped to answer most thinking men's questions if they simply read
the site thus saving me time repeating myself on news groups where something typed in today is then never ever read again,
and we never know how long the archives will be maintained.
The usernet groups are available without password access by simply bringing up the list of groups from your server by clicking 'subscribe' in your file menu window of your inbox, and entering the above NGs titles under 'search for group', and clicking to subscribe to the group.
I have posted a couple of book-fulls of info to rec.audio tubes and aus.hi-fi over the 5 years after 2000 and a search of the google groups archives will indicate would bring up a fair amount of info if a search is done under my name and
any tube audio subject.

I am indebted to many others I have met via the Internet, since their questions and answers have led me to a deeper understanding of what I am doing, and prodded me towards incremental improvements in the topology details of my amplifiers.

I was very lucky to know a fair amount BC, ( before computers ) so I was able to discern the difference between truth and nonsense when I joined to the Net in 2,000. I'd done my apprenticeship without the madness of the Net interfering
or causing a waste of time.

I probably owe my deepest respect to the authors who wrote the
Radiotron Designer's Handbook, 4th edition, 1955, about 1,600 pages.
But within 15 years after 1955, we'd been to the moon and opamps and logic integrated circuits were the new thing.

I probably have 20 books on electronics based on the tubed past and from based on recent modern times, and the detail found within these books surpasses nearly everything that may be found on the Internet.

As a globe full of restless people we are drifting towards a homegenous future where all the electronics will be digital and nobody will be able to build anything for themself anymore, and circuit sizes will not allow any repair or alterations or understanding.

And maybe there won't be much new music which will satisfy us greatly. Every time I hear electronic music
generated by someone almost totally untrained at any music school and using a mouse to make strange sound
on a PC I am never spiritually rewarded; usually the new music grates, is irritating, says nothing except
"whatever, whatever, whatevert" musically, and is boringly repetitive because the clowns who compose such rubbish are very limited people who don't understand that each line of music must take us on a new vista. I could say the same about much modern art, sculpture, architecture, town planning, but don't let me waste your time.

I doubt a PC could invent any music remotely as satisfying as creations by Mozart or Beethoven,
and I won't care that if I live to 90 that I may be the last person to still like analog and tubes, and
a spin of a big black disc.