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 by means of audio - visual electronic media.
Today's emphasis is on the visual from TV screens and audio is the "poor cousin" in
our AV systems. But even if people are not conscious of their audio sytem performance
they will benefit if that experience has hi-fidelity, just as a high resolution picture on a big
home theatre screen allows us to immerse ourselves in the action, plot, and drama of the
movie we may watch.

I find the best "transport of the soul" with music occurs at a live performance without
any electronic equipment and with acoustical instruments which evolved over the last
many centuries. It is not so easy to re-create the atmosphere of a good concert venue
in anyone's home. However, the recording studio may convey to us a more intimate
experience with the musicians than can be had at a concert. 

Our pleasure with replay of recorded music depends on our listening room's character,
the recording methods used, whether the artists 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. Everyone has slightly different
hearing abilities, and psycho-acoustic abilities. Some will want sub-titles to be displayed
during all operas and most pop music to understand the words of songs but others will
not need any sub-titles. Some will hum along easily while others are tone deaf but still
enjoy music.

The only things I may provide to enhance the experience of music are the amplifiers and
perhaps the speakers if you can afford the price of what I make of what older pre-loved
equipment I can reform with re-engineering to perform better than when originally
purchased. I may have some sensible advice about everything else. Much music has
been produced with a great deal of digital processing; it may never have existed as sound
in the air of a venue or studio and has not passed through a microphone, and has only been
monitored as sound in a composer's headphones or tiny speakers at a PC or bookshelf
speakers, all with restricted abilities for providing hi-fi sound.

It is no wonder that we may be dissapointed when we try to expand what the composer
heard on his headphones and with full range speakers when we try to fill a room with
sound. We still wish for the best experience when we expand a recording to fill a room.
And we wish that the ambience of a cathedral or large venue be reproduced in a room
so it is imperative that the room exhibit low reverberation and few resonances so that the
recorded ambience information in music not be marred by a room's qualities. We wish
to hear what our speakers have to give and not what our room may add or subtract.

The listening room is most important. It is also the the most expensive item
to replace and one cannot shop for new listening room at a hi-fi shop. But we can mainly
only make do with what we have bought or rent, so we should take steps to reduce
reverberberations and resonances to a minimum which means floors should have thick
carpets or rugs, walls may have absorbant tapestries or absorbant panels, ceilings may also
have lattices and there should be enough soft furniture to absorb and damp reflected
sound waves. Widows should have curtained glass. But all this depends on what a wife
may tolerate, so gluing empty used egg cartons all over a ceiling may not be acceptable
even if it is an excellent acoustic idea. Large bags of chopped plastic or styrene foam in
"bean bag" seating make cheap excellent acoustic absorbers. There should be no
perception of any echo, and when you clap your hands there should be no sense of
any return signal. I have been in many rooms with appalling acoustics with hard floors
and walls with little of anything to absorb reflected sound and the use of good sound
gear is utterly in vain. One fellow has a very high output moving magnet cartridge, a
preamp with far too much voltage gain, a sensitive Quad405 power amp and 1970s
Electrovoice speakers with very high sensitivity. His room had bare timber floors and
plasterboard walls. He could not use his system. Even at volume set at 1, the sound
was too loud, jangly, muddled, and damn awful. Another had a similar room and with
all SET tube amps and horn loaded speakers and the effect was the same - LOW-fi.

Loudspeakers are the next most important thing after getting the room right.
There is no use spending thousands on an exotic CD player and amplifiers if the speakers
cannot produce a flat response with low THD and IMD. Speakers differ widely in colour
and clarity, and what we want is clarity without colour. It should be easy and non critical
to find the best position to give the best stereo imaging and balance between channels.
I do not like sub-woofers which should never ever be needed. I prefer "full range" floor
standing speakers. There should be 3 ranges of frequencies covered by 3 different drivers
within each speaker so that bass extends from 25Hz to 250Hz, midrange from 200Hz to
3kHz, and treble from 3kHz to 20kHz. This usually means the speakers need to have
large bass drivers and large enclosures to reproduce tones with any plausible hi-fi. And
so often it is a wife who will object to such things cluttering a loungeroom. I can only
suggest that whatever large speakers with good bass be tried before buying and a wife
be persuaded to see reason by genteel trial and error of her favourite recordings. I once
shared a house with a lady of 44 who had excellent hearing. But she preferred mostly
country and western music and she hated my speakers which stood 1.6 Metres high
and weighed 300Kg. To her they were two elephants in the room. But she had a couple
of CDs by Garth Brookes, the Texan crooner. On her awful 1972 Pioneer system Garth
just sounded like any other squawky country faker. I asked to bring Garth to my system.
Within 5 minutes she was able to hear whatever it is in Garth's voice which makes women
wobble at the knees. Ah Yes!! Good Stuff!!. Size didn't matter, the larger the better.
Then I introduced her to Bach and classical, and music became a whole new experience
and it led to a close relationship which lasted for 10 months of bliss - music should be an
aphrodisiac, along with a log fire, good food, and a glass of red.
Unfortunately, many audiophiles miss the benefits of a better social life music should bring.
We are becomming dominated by the damn PC and locking ourselves away and only
allowing cyber-space communication with others.  

With a well prepared room and good speakers, there is a lot to be gained by the best
selection of CD player, TT, tuner, amplifier, and perhaps cables and other parts used
within amps such as tubes and capacitors. I do not have any recomendations and you
just have to try before you buy.

The sound quality is ALWAYS going to be the sum of the recording quality over which
we have no control. We only have control over our room, speakers and other source equipment
choice.

Some history about Class A. 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 wire mesh
"control grid" placed between an anode and cathode in a vacuum tube. The amount of
electrical power applied to the grid may be extremely small but the amount of electrical power
developed by the anode and cathode may be extremely large.
Power = voltage x current. If a grid circuit has extremely small current change the "power gain"
of a vacuum tube is enormous even if the tube is a 300B and generates only 8 watts of audio
power. 

The first audio amplifier was a class A single ended triode amplifier. Its first use was to convert
small amounts of audio frequency power from a microphone to a larger amount of power
in output circuit to permit transmission of the signal to a long distance telephone cable, or to
power headphones or a loudspeaker. The triode allowed this to be done with a fairly linear
relationship of voltage amplitude change between input and output. The triode was like a slide
projector which could convert the little picture on movie film into a brilliant big picture on a screen
while preserving all the picture information fairly accurately.

The single class A device such as a triode is set up with a steady dc "idle" current flow from
anode to the cathode. The idle current is then able to be varied so that the current change varies
between being nearly cut off or nearly doubled during each wave cycle.

Sine Wave cycles are described also as "sine curves ", and to understand what a sine wave might
be I suggest if you try studying at http://en.wikipedia.org/wiki/Sine_curve
If after 2 days you are still struggling to understand what a sine wave might be, you may have major
difficulties with just about everything else mentioned at this website.
The wikipedia site is a bit inconsiderate though because they try to use mathematical formula to
immediately define the sine curve, but nobody has a clue about the math they quote unless they
have spent years at a university. But sine wave motion of many things in your environment is fairly
common. When a bell rings, the metal flexes back and forth in a spring motion to generate a sine wave.
Its lowest frquency is the "fundemental" frequency but other harmonic frequencies are also present.
A guitar string plucked at 1/2 way along its length swings side to side to make a sine wave which is the
string's lowest possible fundemental frequency. If the string is plucked near the bridge, there are many
other harmonics generated up to perhaps the 25th harmonic, so many sine waves are present. The
oscilloscope will show the wave form generated in one string, or multiple strings or multiple
instruments or singers and the wave form does not look like a nice simple sine wave. This is because
many sine waves of different changeing amplitudes, frequencies, and rates of phase change exist.
The sum of all such waves in music can be displayed on an oscilloscope and they always move
positively and negatively either side of a zero volt axis.

Mr Ohm set all this out in about 1850, and I suggest you spend another sunday studying Ohm's Law.

Output transformers.
Most triodes or other tubes are not directly coupled to the speaker and cannot produce the
magnitude of speaker current. The tube is capable of producing far more voltage signal than the
speaker requires so a transformer is placed between tube anode and speaker to magnetically transform
the tube current and voltage changes to those occuring at the the speaker. A typical type 6550 tube
might generate 10 Watts of output power. At its anode there may be 200Vrms and 0.05Arms
applied to a an output transformer primary. The load experienced by the tube = V/I = 200 / 0.05
= 4,000 ohms and power in the load = V x I = 200 x 0.05 = 10 Watts. Let us suppose the OPT has a
voltage ratio of 28.8 : 1 so that the secondary winding produces 7.7Vrms and 1.29Arms at a 6 ohm
speaker connected, and there is 10 Watts delivered to the speaker. The "impedance ratio" of this OPT
is voltage ratio squared = 666:1 Thus a speaker of 6 ohms is made to appear to a tube as 4,000 ohms.
 
The transformation of high voltage x low current changes in the tube to low voltage x high current in
the speaker is a manifestation of electro magnetic phenomena. I'm sorry to have to inform you that
your sunday domestic bliss may now be utterly ruined by the exasperating and infuriating study of
electromagnetic behaviours.

The transformer is necessary to transform the power from vacuum tubes to suit speaker loads.
The tubes are like a high revving car engine which needs to use a LOW gear ratio in a gear box
to drive up a steep hill with 6 fat policemen in the car. The transformer is an electronic gear box
and basically both transformer and gear box work on the principle of the lever where a wide swing
of the hands with low amount of force is transformed to a smaller movement with much higher force.
When we use a spade to dig up the garden, we are using leverage to move earth.

In class A operation the change in idle current flow is varied around the centre point of the most
linear region of operation of the device and there is no sharp switching on and off of current as in
the case of class AB amplifiers. The current in the vacuum tube is called the anode current, and
consists of trillions of electrons flowing from a heated metal cathode with a special emissive
coating to an anode with a positive voltage which attracts the electrons which are negatively charged.
The grid placed between anode and cathode has an idle steady voltage applied, Vdc, controls the
flow of current. The steady voltage which controls idle current anode to cathode is called the
*grid bias voltage,* and often the idle current between anode and cathode is called the
*plate bias current,* or anode bias current. A positive change in grid voltage causes more anode current
to flow, a negative voltage change in grid voltage causes less anode 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 the anode voltage swing capability.

The grid is usually a helical coil of special fine wire supported on metal rods and placed within 
the anode which is sheet metal box. The cathode is a metal tube which emits electrons when heated.
it is placed inside the helical grid and it is heated to about 900C. The distance between grid and
cathode is much smaller than the distance between grid and anode. This difference in distance
allows the grid voltage to have maximal effect on anode current without too much interference
by the effect of anode voltage on the
current. ( more 4 U2 study....  :-)

When the grid voltage is raised to a positive voltage, ie, above zero-volts, the grid attracts electrons
and the grid generates grid current. When this grid current begins to flow the linearity ceases and
a limit of operation has been reached.

The region of tube operation between tube current cut off where Vg is made very negative and where
grid current flows is the useful power producing region for hi-fi amplifiers.

Conventions about current flow were established before anyone knew what actually current was,
ie, movement of electrons into one end of a peice of wire and out the other end, and propelled
by the applied voltage across the wire, and some field effects. Field effects and exactly why current
flows is beyond my capability to explain here right now, and I suggest you study more books or
websites about it.

The difference in electric charges or voltage potential values was known, and could be positive
or negative, and it was assumed current had one direction, from positive to negative. So many
years ago it was agreed anode current flows to the cathode from anode, because an anode is
always has a much more positive voltage than 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.
We may never teach a poodle to walk backwards, but humans like myself have become use to
silly conventions when we think of current flow + to - while really it is electrons going the other way.

Sawing logs using class A action.
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, saw tooth, or square wave. But we would find there is a fundemental
sine wave frequency of saw strokes plus 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 men each
working like one man, but with complementary action.

Imagine our bushman who is cutting the log is joined by a friend who works the other end of the
bush saw 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, because they can produce twice the power of one man. But two
men might operate the same end of the saw and also each produce the same power. When
you have two tubes in parallel the amplifier action is single ended, but called "parallel single
ended" or PSE.

Although even numbered harmonics are cancelled by push pull action there are 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.

Distortion.
It should be remembered that with class A vacuum tubes the THD rises from 0% at 0.0V
at the output to a maximum of about 5% at maximum output with a single ended triode.
A single ended tetrode or pentode may produce 13% THD when serious THD suddenly
becomes visible on an oscilloscope monitoring the THD.
 
This is at "clipping", when very little further increase of voltage amplitude swing is possible. 
At 1/10 of the "power at clipping", the THD might be approximately 1.3% and 3.3% with
SE triodes or beam tetrodes. With a pair of the same tubes in a push-pull class A circuit with
each tube experiencing the same loading of a the SE tube load, the THD might be 2% for
both triodes and beam tetrodes, and at 1/20 of the maximum PP clipping power which equals
twice the power of one tube in SE, the THD will usually be less than 0.5%. From an early
time in tube technology development the PP circuit was considered much superior to SE.
But most people could not afford PP amps and didn't need them because most needed
to only listen to a radio or play a 78 record. So before 1950 nearly everyone heard
reproduced sound with lots of distortions. But when the PP amp is constructed to
work in class AB, the THD at clipping may well equal the higher amount measured in
SE amps and the PP amp does not produce any better sound than the SE amp.

I have quoted THD figures for tube operation with no NFB applied and for low levels
of power because most listening is done where amplifier power averages less than 1/10
of the manufacturers claim for maximum power.
 

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 grabs his end of the saw
and begins to continue the saw's motion and does the same thing as the man on the other
side had done. So each man handles the saw *alone* for only 1/2 of each sawing wave. 

This is class B action. Now if you asked the men to cut a log like this you'd have a union strike
and maybe they'd become very agitated. 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 audio signal present. Device 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, and the
THD % at low levels where the amps are used can be distressingly high.
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, when asked to operate in class B.

Class AB.
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 are 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 is produced before switching and class B operation begins.
 
The efficiency of operation is poorest in pure class A tube amps and can only reach a maximum
of 40% in most class A amps. A class A amp consumes high power constantly and it does not vary
with signal production. If the power supply provides a constant 25W of power to a class A amp
circuit, you get 10W of audio power.

In a pure class B tube amp the input power at idle = 0.0W, and the power supply need only provide
input power with signals present, and a typical PP amp may have maximujm input power of
60W at full output from which 67% is audio power, or 40Watts.

In a 25W capable class AB tube amp the input power at idle is less than the class A amp, say 12W.
There may be a maximum of 6W of pure class A produced with the rest of the power in class AB.
The power supply has to supply 12W at idle but this rises as audio power increases to say 25W when
the power supply must produce 50W. Thus efficiency is low at low levels but increases to about
50% at maximum audio output. Most tubed Push Pull amps are configured this way and THD
is often only 3% at maximum output which canbe reduced to 0.6% with global NFB.
But up to the limit of pure class A the class AB amp may have almost no more THD than if
the amp were biased with higher idle current to allow only class A operation. Because most people
will require only 1 watt or less at each stereo speaker for average levels the class AB tube amp
can offer excellent sound. Ideally, at least 10Watts of class A is needed from any amp using
modern insensitive speakers and there should be another 20W available for headroom.

The accountants employed at amplifier companies have chosen to use as little idle current
as possible in class AB amps. This means the amp runs cooler so parts can be crammed together,
and construction costs are lower and company profits higher while reducing the quality of hi-fi
in your listening room.

The class AB tube amp with generous idle currents will probably survive against the onslaught
of digital aliens.

You can't beat well constructed class A amplification for maximum fidelity.

With today's average loudspeaker sensitivity one needs a minimum of 30 Watts per channel
to provide the ability to cover most transients without clipping or amp overload, or any speaker
overloading. 95% of most listening is done using no more than 2 watts average.
If we had a pure class A amp able to make only 10 watts we might find distortion artifacts
during loud music passages to be alarming and unacceptable even though the sound levels
below 10Watts was enchanting.

Loudspeakers made in 1960 were usually far more sensitive than today with very light
cone weight and light voice coils such as Tannoy dual concentrics and Altec drivers, 10
Watts seemed to be enough. But since the 1980s the bass content and drum content of
music has increased enormously and it became cheaper to make speakers which had less
sensitivity. This assisted marketting claims about the "higher power handling cabability"
of the newer speakers. The advent of solid state meant there was plenty of cheap power
to drive the new speakers. Many sounded worse than what they replaced. But many of
the newer speakers which are less sensitive had a much flatter and truthful response, ie,
they had far less distortions than the old sensitive lightweight speakers of 1960.
So we DO NEED 4 times the average power which amplifiers had in 1960. In 1960, most
10W amps used a pair of 6V6 or EL84 tubes in class AB and many gave 15W max and sound
was very good with the right speakers. Today, the cost of purchasing a 50W class AB amp
with a pair of 6550 is no higher in terms of weekly wages than what the 10W amp cost in 1960.
We now live in times where everything has become cheaper in real terms. I recall spending $8
on an LP in 1978 and in today's money it would be $100. But a CD may only cost $20 and
have more music time. In a 50W class AB amp we can produce the first 15 watts in pure class
A which is 4 times the pure class A in the 1960 class AB amp battling to make 15 watts total.
There are people who cling to their 15 watt amps, for example their stereo Leak 20 or some
other thing with low power tubes but the possible fidelity is just not there all the time.
We also live in times where Chinese manufacturing costs are a small fraction of the cost
in the US, Europe or Australia. Some of this translates to cheaper prices in your local hi-fi store
and much cheaper on-line prices.  

There is more info about A and AB operation is elswhere on this web-site and of course much is
available from old books on tube audio written before 1960 by my father's generation. The wealthier
friends of my father had amplifiers with a pair of 6L6 or 807 which could give 40 watts AB. Some
had Quad-II amps with a pair of KT66 for 22 watts of class AB. But they were only 5% of anyone
who claimed to be interested in hi-fi. The advent of the KT88 in 1957 and 6550 at around that time ended
any worries about having enough power with low distortion. 

Nearly all small signal audio preamplifiers use cascaded stages of tubes in simple class single
ended mode. Despite whatever I said about lonesome solitary fellows sawing logs and the high
distortions of single ended amps, such negative statements about SE amps applies only to SE power amps
with low power capability, say 10W or less, and where tubes used are in beam tetrode or pentode mode
without any loops of local NFB in the output stage.

In preamplifiers using SE triodes such as 6CG7, 6SN7, the THD at 10Vrms output can be between 0.1%
and 0.4% at 10Vrms. But for most power amps only 1Vrms input is needed to make 50 Watts for clipping,
and average preamp rarely rise above 0.1Vrms where preamp THD is often less than 0.01%, or about 1/5
of whatever the power amp may produce. And this would be true where the preamp had no loop NFB
anywhere while the power amp had a lot.

I have developed SE amplifiers using passive techniques of natural THD cancelation between output
and driver tubes thus giving less THD than many PP amps. For example, a single EL34 in pure pentode
mode will give 9 watts of pure class A but THD may be 13%. But with my local CFB circuit and the right
driver tube the THD is never any worse than a good PP amp of the same power. The SE35 with 4 x EL34
is an example where THD is very low but the amount of NFB is also low. Its perfomance gives fabulous fidelity.

Therefore I have never thought it necessary to build preamps with push pull or "balanced" circuitry.
When signal voltages rarely ever rise above 1Vrms, the huge dynamic ability of the SE circuit topolgy
offers very low THD. A typical tube preamp might be able to produce 60Vrms maximum output
at 3% THD. But the THD is about proportional to output voltage so at 1Vrms the THD = 0.05%.
At 0.1Vrms the preamp THD may be only 0.005% !

A Burr Brown opamp with 30 transistors and j-fets within could be used to get 
THD = 0.001% when 60dB of NFB is used, ie, a heck of a lot of NFB, but you won't hear better music.
Below 0.1% there is virtually no sonic improvement and some argue SS is always cold and lifeless.
I could als use a single j-fet such as 2SK369 to replace a triode but the j-fet will produce about
1% THD per volt of output. This is very poor linearity compared to say a 6CG7. But the j-fet
at 0.05Vrms output will make only 0.05% and with some local current FB the THD can be 0.01%
and lower than any following stage amp stage. The j-fet has 1/10 of the input noise of most triodes
so it is an ideal device for the input to a phono amp. The j-fet THD has a similar spectral content to an
SE triode with mainly even numbered harmonics, ie, mostly 2H, and at low level the 2H just does not matter.

So my circuits use simple topology with far less complication and device numbers used by many
other makers. They succeed at proving how clever they are, but there isn't any more music.

Class A signal circuits always run warm and are extremely inefficient because a typical 6CG7 triode may have
Ea x Ia = 150V x 0.005Amps at idle giving Pda = 0.75 Watts, Output power would be a tiny fraction of the idle
power and heater power in filaments is 2 Watts per triode section of the 6CG7. But for small signal preamps
and for low level input stages the inefficiency is of utterly 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 two or multiple output transistors has become based around class AB operation with very
low bias currents which amount so 25mA dc per transistor. It is virtually class B operation.
If there are two power tranistors connected as a complementary pair and if the rail supplies are
+/- 50Vdc, the maximum idle power liberated as idle heat is 100Vdc x 0.025A = 2.5 Watts and
hence the heatsink of the amp barely gets warm if the amp is not producing any audio power.
This helps reliability because for SS amps the reliability declines 10% for every 10C rise in
temperature above standard room temperature of say 22C. The maximum class A power with
collector current = 25mA for an 8 ohm speaker is 0.01Watts, and its not very clean because the
transistors are operating in their turn-on turn-off non linear region. The two transistors will have
different "polarity" with one being NPN and the other being PNP and usually the NPN handles the
positive 1/2 of the wave cycles while the PNP handles the negative 1/2 of the wave cycle.
The NPN and PNP transistors may have have very different characteristics and the mismatch
can be similar to the difference between an EL34 and 6L6.

The crude class B operation of solid state devices is largely tamed by a simple technique of corrective
circuitry known "negative feedback". This circuit technique has been used since about 1925 to make
devices operate more linearly than they would without NFB.

Negative Feedback.

NFB cannot ever reduce THD to zero. Luckily, we don't need to ever have 0.0000000000% THD.

But the reduction of THD, IMD, and phase errors and amplitude errors or any other amplifier artifacts
can be much reduced by NFB.

THD with NFB = THD without NFB / ( 1 + [A x ß] )

where the NFB is a typical case of "series voltage negative feedback",
( and not any other kind of NFB as classified in text books  ),
1 is a constant for all equations,
A is the "open loop gain" of the amp, ie, the output voltage divided by the voltage between the two input "ports."
ß is greek 'beta' which stands for the "fraction of output fed back to input",

Let us consider a typical example of a solid state amp :-

In this example, A = 10,000, and ß = 0.05.

THD without NFB = 8%, at 1dB below clipping.

THD with NFB = 8 / ( 1 + [ 10,000 x 0.05 ] ) = 0.016%

The amp is set up so the internal voltage gain is typically = 10,000. This means that without NFB
you only need 0.002Vrms applied between TWO input terminals to make 20Vrms appear at the output
with perhaps a 6 ohm load giving 67 Watts of audio power.
Without NFB, only one input is driven by an input signal with the other taken to 0V.

From the output terminal there are two resistors in series to 0V which act as a "resistance divider" so
that when say +20 Vrms is at the output, +1Vrms appears at the junction of the two resistors.
The +1Vrms is connected to input 2 of the amp. Input 2 is an "inverting input" and a positive going
voltage applied causes a negative going output signal.
The +1Vrms voltage is called the "negative feedback" signal, because it is a fraction of the output
signal and it contains a fraction of the distortion appearing at the output terminal. In this case the
fraction is called ß and = 1/20 = 0.05. Suppose we measured 0.016% THD at the output.
It means distortion signal = 0.004Vrms. ( 4mVrms ). The amount of this THD signal fed
back via the resistance divider to the input 2 port = 0.0002Vrms. ( 0.2mVrms ).

The arrangement of the amp with TWO inputs means that if the same input voltage is applied to both
inputs simultaneously, there is no output signal produced, and only the DIFFERENCE between the
voltages applied to each input is amplified by the input devices and consequent amp stages.
The amp is said to have "differential gain" of say 10,000.

In the example so far, there is +20Vrms at the output with a 6 ohm load, and +1Vrms
of NFB signal applied to input 2. The input applied to input 1 must be slightly greater than the signal
at input 2. The signal between input 1 and input 2 must be 2mVrms in this example.
So the signal to input 1 will be 1.002Vrms. When probing around in an amplifier with a voltmeter
it s impossible to accurately measure the difference between input 1 and input 2, let alone
measure the distortion voltage of 0.2mV easily.

But between the two inputs there is +0.0002V of distortion signal, and this appears at input 2,
an "inverting input."

The distortion signal of +0.2 mV is amplified x 10,000 to become -2.000Vrms at the output.

How can this be when we measured distortion of only +4 mV at the output?

The answer is simple. Without the "resistance divider and the applied loop of NFB", the amp would
have had +20Vrms signal with +2.004Vrms of distortion signal which is 8% THD. As the fraction of
the output voltage fed back is increased above zero the distortion is instantaneously amplfified to cancel
itself so that in this case -2.000Vrms of invisible signal is generated to subtract from the +2.004 Vrms
of distortion which would appear were there no NFB loop present.
The residue remaining after you subtract 2.000V from 2.004V = 0.004V which gives 0.016% THD.

Class AB operation with miniscule idle current is much more efficient than having class A with hot
running devices which consume considerable power even when not producing any music. But
efficiency itself is NOT renowned for making music sound any better, but nevertheless nearly all SS amps
are virtually class B, fairly 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 idle 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 AB amp is about the same as a 50 watt class A amp with solid state.
However, the heat liberated by devices in a class B amp quickly rise once audio power rises above
a few watts.

An amp with rails of +/- 45V may be able to generate 28.3Vrms into an 8 ohm load
which is 100 Watts of audio power. I have often measured this. The above graph shows how
much heat is generated in the complementary NPN and PNP devices on each side of the PP
circuit. There is over 30W of heat generated at only 10W of audio power so class AB SS amps
need to have well rated heatsinks especially if 4 ohm loads or lower are to be used. Sadly many amps
will smoke themselves to death all too easily with their inadequate heatsink size and too few
output transistors. At least 2 x NPN and 2 x PNP TO3 or TO3P package devices should be used for
any amp expected to produce 100 Watts. Heatsinks should be as large as possible. A similar graph to
the above could be drawn for a 100W tube amp but heat is mostly radiated away from anodes
through glass. At least 4 x KT88 or 6550 should be used for 100W. Tubes also overheat if they
have speaker loads which are too low, and smoke can be generated unless there is a protection
circuit in place to turn off the amp if one or more tubes conducts too much current for too long.  

However, due to device manufacture and circuit topology improvements over 50 years the solid
state amps have become a lot better than they were, and at least they have impressively low
distortion measurements with Halcro boasting of 0.0001% THD at 200Watts at all frequencies
below 20kHz. This is impressive, and so is their unaffordable price. But their heatsinks are well
made and the switch mode power supply is located away from audio circuits at the bottom of the
H shaped tower.

Trends are now toward pulse width modulation and digital signals and some digital sounds
better than other digital and I now think much digital is equal or better than most generic brand
solid state class AB which will die out like dinosaurs within 10 years simply because digital PWM
is cheaper to make for the same sounding result. 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 AB solid state amps.  I think tubes will stay
as long as tube production can continue where it does in former "socialist countries" such as
Russia, and in China which is officially socialist, but with Chinese capitalism added because as
Deng said, "To grow rich is glorious."

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 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
presence 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 rising and falling with signal levels and in time with the rythym in the music. IMD seems
to be subjectively worse in solid state amps so they seem to need all the high amount of NFB
which 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 tube or multiple parallel output tubes operating only in pure class A.

I can only suggest that the tube amp's poorer measurements just do not matter because the
spectral nature of its IMD distortion is less painful or noticeable to ears than the smaller quantity
of spectrally more complex IMD produced by a solid state amplifier. The "objectevists" are
welcome to disagree with me because they say there is virtually no THD/IMD that can be
easily measured 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 accurate, and less clinical, dried out, music-less, cold, harsh etc.

At a live concert I was impressed by a pianist playing a Yamaha grand piano in a concert hall.
The audience was allowed to come on stage afterward and some could play well. To reproduce
levels heard right in front of the grand would take some serious engineering; do not expect a lone 300B
for each channel with low sensitivity and low power ability to be able to do it unless the speakers
are the best 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 reproduce the highest
levels heard when standing near the grand. Of course most of us could 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 then
those levels don't need so many 300Bs.

Distortion, noise, Negative Feedback, triodes, multi grid tubes....

The problem of hum noise generated by power supplies can usually be easily reduced by
adequate and simple passive methods of RC or LC filtering of DC rail voltages. Hum noise
from filament heaters or in directly heated cathodes can only be overcome by using DC
supplies to all heaters or directly heated cathodes prone to hum from AC current flow.

There is some possibility of hum generation by means of low level magnetic induction of Vac
in signal and 0V rail paths from the presence of mains transformers near the audio amp tubes and OPT.

Such noise may be minimised by use of power supplies on remote chassis placed a metre away
from the audio amp.

My pages on the SE55 monobloc power amps each with a pair of 845 output tubes shows what
is involved with implementing best practice for SET. Noise is extremely low and with little
reliance on NFB.

With beam tetrodes and pentodes,
the distortion and Rout is all much higher than with pure triode amps. My website pages explain
how I achieve low levels of distortion and low output resistance while using multigrid tubes.

I like to achieve THD less than 0.4% at 1dB blow clipping for all amplifiers and I like to see
Rout = 1/10 of the loudspeaker load. Often these aims are bettered especially with PP amps.
However my SE35 amps have distortion to match that of a good PP design, since they exploit
complementary distortion voltage cancellation between amp stages.

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 +going voltage is applied to the grid, there is a -going anode voltage with increased load current.

The +V grid change produces a more positive electrostatic field which increases electron flow from
the cathode to anode. The -V anode voltage produces a more negative electrostatic field through out the
triode and even between cathode and grid, so the anode voltage reduces the electron flow to the anode.

The electron flow in a triode is controlled by TWO voltage field effects, one from the grid voltage
and the other from the anode. .

So two voltages are together working intimately and instantaneously and at all frequencies to control
the electron flow and the anode voltage action opposes the action of the grid voltage.

The voltage field effects in a real triode such as a 300B are effectively the same to a KT88 or other
beam tetrode or pentode which has its screen grid connected to its anode so that it works like a triode.
In the case of the "triode strapped" KT88, it is the screen grid which propagates the voltage field to
alter the electron stream to itself and to the anode.

If any distortion voltage appears at the triode anode which is not present at the input grid this voltage
is applied to the electron stream in a manner which opposes its own production. If there was a -going
distortion voltage at the anode it indicates increasing current in the RL. This -going voltage lessens the
anode current and opposes the increasing current. In effect, the anode voltage field effect is the reason
why triodes have such low dynamic anode resistance.

In the case of a KT88 with screen connected to anode, the effect is the same as a real triode. But the
fact there is a screen allows any fraction of anode voltage field effect to be applied by the screen to the
electron flow. This can be done using taps on an OPT winding as it is done with an "Ultralinear" OPT.

One may ask more questions about this and you need to consider a KT88 more closely. The screen
grid is a second control grid using a helical wire coil mounted on support rods but placed between the
main control grid and anode. Screen grids were designed to allow a fairly unimpeded anode current flow
of electrons from cathode to anode, and yet the screen blocks roughly 95% of the the electrostatic field
effect due to anode voltage changes. The screen needs to be at a high positive dc voltage to maintain the
velocity of electrons on their way to whatever is positive. About 10% of tube current from the cathode
strike the screen wires and are absorbed and form the screen supply current. The other 90% of electrons
miss hitting the screen but continue on to the next positively charged element, the anode, where they are
absorbed and forn the anode current. The anode voltage can change over a wide range but anode current
does not change much. The dynamic anode resistance of a KT88 beam tetrode may be 30,000 ohms when
there is a fixed screen voltage. If the screen is tied to the anode to make a triode the KT88 anode resistance
becomes about 1,000 ohms.

If the screen has 50% of the anode signal voltage taken from a tap on an OPT the tube is intermediary
between a beam tetrode and triode, and the name 'Ultralinear', or UL has been given to this form of use of a
multigrid power tube. With "50% UL taps" the dynamic anode resistance could be about 3,000 ohms
which means this simple application of local NFB gives a very worthwhile reduction of the high beam
tetrode anode resistance. The tube gain between control grid and anode will be roughly twice that of the
triode connection depending on load. The beam tetrode capability of producing up to 45% efficiency in class A
is largely maintained and 40% efficiency is typical with UL. The 50% UL connection reduces THD spectra
to being very similar to triode connection and with a huge reduction of odd order products such as 3H and 5H.
UL operation of beam terodes and pentodes originated soon after WW2 and became the the most popular way
to connect multigrid power tubes to give approximately twice the power of triode connection yet with triode
characteristics of low Ra and low THD. A huge number of amps use the UL connection.

Let us consider a typical KT88 ( or almost identical 6550 ) a little further, with Ea = +400Vdc, and
screen supply = +400Vdc and control grid biased at say -55Vdc to give Ia at about 50mA. The dynamic anode
resistance, Ra with a fixed screen supply will measure about 32,000 ohms. The grid gm will be approximately
0.0055 Amps per volt, or 5.5mA/V. For all tubes, amplification factor, µ, = gm x Ra.
Thus the amplification factor of the beam tetrode = 0.0055 x 32,000 = 176. The Ra and gm I use is not the same
most data sheets show but those data sheets quote gm and Ra for Ea = 250V and Ia = 140mA and nobody in their
right mind would ever set up a 6550 or KT88 in such conditions for class A operation. The conditions of Ea = 400V
and Ia = 55mA is far more common and typical of many amps made since 1957 when KT88 were first produced.
gm and Ra vary considerably depending on Ia; the relationship is not a very linear entity. The µ for a tube may
easily be measured using a very high value of inductance, say a 100H to bring DC current to the anode but allow
the anode voltage to change at audio frequencies because the reactanceof the inductance is so high at the AF.
100H at 1,000Hz has XL = 628,000 ohms and this load value is so high that the voltage gain will become
close to the calculated µ value. A CCS anode DC current supply would be better than a choke, and the measured
µ would be a more reliable way to determine µ than by calculation. BTW, gm is found by shunting the anode to
0V and preventing any anode voltage change and then applying a small grid voltage change, say 1Vrms, and then
measuring anode current change. Ra is found by having the anode connected to B+ via the CCS and connecting a
1,000 ohm resistance to a 10Vrms signal voltage source ( via a 10uF dc blocking cap). The signal voltage across
the 1,000 ohms gives the AC current flow to anode. The anode to 0V signal voltage is measured and Ra = anode
signal voltage / anode signal current.

The same tests on the tetrode may be done to discover the Ra and gm and µ when the screen grid is used to control
Ia with an applied input signal rather than the control grid. For this test we keep the input grid biased at a fixed
-55Vdc and find Ra and gm as described above. We should find gm of the screen for a small voltage of say 10Vrms
will be 0.83mA/V, and that Ra will be close to the same value as the beam tetrode operation. It would be found that
the screen µ will be about = gm x Ra = 0.00083 x 32,000 = 26.5.

Eqivalent model for UL connection of 6550 or KT88.

Suppose there is a class A anode load of 6,200 ohms.
Gain from screen drive = µ x RL / RL + Ra = 26.5 x 6,200 / ( 6,200 + 32,000 ) = 4.3.
Gain in beam tetrode mode with fixed screen voltage
= µ x RL / RL + Ra = 176 x 6200 / ( 6,200 + 32,000 ) = 28.56.

I hope the above schematic model of an SE UL connected 6550 or KT88 sheds light on how to calculate
UL gain and Ra for any position of the UL tap along the anode primary winding of the OPT.

The above could also be used to establish gain of the tube where screen is at a fixed B+ voltage and
there is cathode feedback windings.

Some further derviations will give the calculations for Ra and µ and Gm of g1 for CFB
use and with or without UL taps.

But for most people, it is enough to know CFB windings are more effective than UL taps
to reduce Ra and THD while keeping voltage gain high enough to avoid a grid drive signal
more than 70Vrms.

Any tube is said to be a voltage device, and it is true true because there can be changes to the
anode voltages 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.
Triodes are said to be volatge sorces because their Ra is usually lower than the RL they power.
Pentodes and tetrodes are said to be current sources because their Ra can be much greater than the RL
they power.

My pages on load matching loads to power tubes emphasize the need for careful loading of tubes.

I have slowly developed circuits which I know sound better than most basic PP amps
devised before 1960 such as Peter Walker's Quad II with CFB windings in the output
transformer, D.T.N Williamson's triode amp of 1947, Mullard's 520 design,
and the McIntosh. Thereare few SE amps which will sound as well as the SE35.
I continue to explore possibilities to achieve the finest musical experience.

In earlier times there were many people working 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 1/30 of the local costs here. All the local
tradesmen and women who were employed to satisfy local demand from WW2 to 1990 when
demand was high have retired or passed away. The quality of asian made transformers leaves
a lot to be desired.

Presently I have a large stock of power and output transformers and chokes which I may use in
any future projects. Most is NOS and quite acceptable.

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 1.6 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 because tubes have such a high dynamic range,
the operation at low output voltages assures negligible distortion production and excellent sound.
There is no need to use large amounts of negative feedback error correction techniques, and yet
the SNR is excellent and and N&D remains very low, even with a phono amp; see my preamp pages.
Triodes are predominantly used although I have used triode strapped frame grid pentodes and j-fets
in MC phono amps. These sound so good in comparison with other expensive hi-end preamps that
I find it difficult to know how to improve my techniques. While 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.
I have only ever included a solid state device as a voltage amplifier in a signal path in my 10-tube Preamp
and Rocket phono amps. These have a high Gm j-fet at the input, 2SK369, which only has to produce a
few millivolts of output. It is much quieter than all triodes known and distortion is negligible at such
low levels. The idea of using cascode circuit topology with j-fet & triode for a phono amp was first
implemented by Allen Wright in his Musical Fidelity Four Valve Preamp of 1988 which received rave
reviews. See http://www.vacuumstate.com

Tubes 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. Most tube production for audio use is for guitar amps resulting in large
quantity production which enables production of no-guitar amp tubes cheaply and profitably.

Tubes wear out. Life expectancy of power tubes is about 5,000 hours and for small signal tubes about
10,000 hours. Life expectancy is only what we may expect, but there must be a median age where
just as many perish below the age as above it. Some tubes will fail after 1 hour of use, while others
last 60 years. So one should remember all electronic devices have a "random failure rate". When the first
electronic computers were made they used perhaps 10,000 small signal tubes such as 12AX7 and
the random failure rate for 12AX7 is about 1 failure per 10,000 tubes in each hour so a technician
was needed to replace faulty tubes perhaps after each day of use, rather too often. If you had a total
of ten 12AX7 in your hi-fi gear expect 1 tube failure over 1,000 hours. This would certainly be true
in a guitar amp where tubes are subject from damaging vibrations from a big speaker only inches away
from the tubes which have their life rattled out of them. In hi-fi amps the vibration is minimal, and random
failures are far less commom below their median age. Luckily, a 12AX7 costs less than a dinner at McDonalds.
So we should keep a few spares. I have aquired maybe 100 in various condition between NOS and
damned awful, ie, they still work, but have become noisy and microphonic. I've thrown out those which
have failing emission after 20years+ in some amp.

If a power amp is used 2 hours each day for 1 year, that is 730 hrs, so you can get maybe 7 years
from a pair of output tubes. One may get an occasional early random failure of a tube. But one should
try to find the cause of the failure, and a technician may be needed. In many old amps, failing coupling
capacitors or faulty speakers damaged the tubes. Replacements are a part of life. If you break a plate
or glass while washing dishes, is it a major drama?
The potential 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. Alas, many tubes fail now because owners
don't undertsand how to adjust bias correctly and the amp manufacturer has made the amp so badly that biasing
is difficult, dangerous, and can only be done by a courageous technician.

However, if one has to spend $250 after 6 years to re-tube an amp, just consider the repairs needed
for solid state amplifiers. I get one amp a week coming here for a fix, many have been "fixed" before,
and repair 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. Its not unusual for
tube amps to have bandwidth of 10Hz to 50kHz before any NFB is applied, and at the maximum
rated output power. To give this, a wide bandwidth OPT and low µ triode input stages are required.
Many older amplifiers made in the 1950s and 60s failed the optimal requirements because OPTs had
too little interleaving and pentode or high µ triodes were used. I have serviced or re-engineered
many samples of such amplifiers such as Leak 2020 based on the Mullard 520 topology. I like
all my amps to have bandwidth > 20kHz before NFB is applied. This means that after NFB has
been applied there is no possible use of the amp which will result in oscillations of spurious frequencies.
This means the amp is unconditionally stable with NFB, and usually will give 5Hz to 65kHz of bandwidth at
up to 1/2 of maximum power. Rout will be constant for all F between 20hz and 20kHz.

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, but is subject to subjective opinions and prejudices.
There may be much chatter amoung audiophiles about which brand of 6SN7 or 6CG7 sounds the best.
But NOS Siemans are remarkably good, IMHO. I am sometimes wary of any NOS tubes 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 1,000 hours, and a used power tube is usually easily spotted. This insinuates that some folks will
sell their expensive tubes as new after 1,000 hours if the market price for them is high, as would be the
case for NOS GEC KT66 or KT88. Say they paid $100 each for the NOS originally, and get $80 on
resale after 1,000 hours. Then they buy another set of NOS. But were they also conned about NOS
tube seller before they got them? Nobody knows.

I get too many good reports about Russian 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 50 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.

Testing tubes.
Before I complete constriction of any new or re-engineered amp I will have spent days testing
and optimising the function of everything so all tubes I select will be run for days and turned on
and off perhaps 50 times. The amp itself is therefore the "tube tester". I do 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. An Oscilloscope 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 in the 12AU7, then the signal measured at the preamp output = 44.8 mV, easily
measured and seen on the oscilloscope. The noise should be a constant average amplitude over the
bandwidth 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
which is 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. If the slight impact by 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 at the tube
will cause whistle tone to display a sine wave on the oscilloscope 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 music was composed in B flat. Therefore phono preamps are best placed well 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 military-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 are the best
sounding, and are very quiet in any amp.

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. Most tube testers which have been donated to me had so many faults
from testing faulty tubes or using wrong switch settings that I just stripped them for useful parts and
discarded what was left. I test radio tubes in a 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 old tubes and I always welcome such electronic orphans.
They usually come with many in the old original cartons and with an assurance 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 or many reasons. Old techs hoarded their "pulls" from old gear; maybe they hoped some miracle
treatment would rejuvenate the failed tube which could be used again in a repair job or sold again as new.
No such miracle treatment was evolved and what old techs did see was the advent of solid state
which led them to the eye specialist for glasses to see what they were doing, and much learning to adapt to
technology change. 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 is actually is useful. Maybe there were about 7,000
varieties of vacuum tube produced by 1960 but I could easily build fabulous amps if I was limited
to no more than about 50 type numbers.

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 learn so much before I began using a PC in 2000. I was able to tell the difference
between truth and nonsense when I joined Internet discission groups 2,000.
I'd done my apprenticeship without the madness of the Internet 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. This is certainly the case with a PC motherboard with
thousands of parts on it. They run for years until they are obsolete, when we have no use for them.

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,
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.