[size=medium]
2. ADVANCED THEORY[/size]
2.1 Amplifier output stage topologies: SET vs Push-Pull vs OTL
Often, amplifier components are arranged in certain particular patterns – these patterns are known as
topologies.
This section covers the common tube amplifier topologies.
SET
SET stands for `Single Ended Triode'. SET designs are always Class A and usually cathode biased. [Refer to section 2.2 for more information on tube biasing]
Triodes typically output low current into a wide voltage swing. This is not especially useful – speakers and headphones need a high current into a narrower voltage swing. The solution is to use an output transformer to convert the former into the latter. The secondary of the output transformer is connected to amplifier output.
Being connected directly to triode anode output, there is usually a large extraneous DC current flowing through the output transformer. Hence output transformers for SET designs are usually beefy (and extremely expensive) affairs with gapped cores that are able to handle this DC offset without saturating. One may infer (correctly) that output transformer quality is of paramount importance in SET design.
Given that the triodes are biased into Class A, there is disparity in the amount of gain between the positive and negative half-cycles. Hence most of the distortion that occurs is even order (primarily 2nd). Given that chosen triode operating parameters are usually happy compromises, SET designs in general have poor measured distortion figures.
Despite their poor linearity, SET designs are extremely popular. The likely cause of this is the relatively benign 2nd order distortion - many feel that it is that which gives SET amplifiers a ‘magical’ sound.
Push-Pull
Push-Pull (PP) designs can be Class A, AB or B. PP topologies always involve pairs of tubes, with more pairs of tubes paralleled for higher power output if necessary.
In PP amplifiers, pairs of tubes are driven in opposite phase to each other -
In Class B amplifiers, one half of the tube pair handles the positive portion of the cycle (input signal) while the other half of the tube pair handles the negative portion of the cycle.
In Class AB amplifiers, one half of the tube pair handles the positive cycle plus a little of the negative cycle. The other tube handles the negative cycle plus a little of the positive cycle.
In Class A amplifiers, both tubes handle the entire signal cycle – each tube handles both positive and negative completely.
Each side of a tube pair is connected to opposite sides of a centre-tapped output transformer – in effect one ‘pushes’ the signal through the transformer, while the other ‘pulls’ it through the transformer. The signal is in effect ‘re-stitched’ within the output transformer.
Class B PP amplifiers are usually avoided in audio. Why? When output signal ‘re-stitching’ occurs, a particular form of distortion known as crossover distortion often occurs with Class B amplifiers. Crossover distortion sounds extremely bad and is extremely obvious.
Class B PP
Class AB PP amplifiers are the most common variation of PP design. When ‘re-stitching’ the output signal, crossover distortion is avoided due to the ability to ‘overlap’ each portion of the output signal handled by each individual tube.
Class AB PP
Class A PP amplifiers ‘re-stich’ the 2 separate but complete ‘push’ and ‘pull’ waveforms in the output transformer. As can be inferred, Class A PP designs are also the most inefficient. Many feel that they give the best sound though.
Class A PP
PP amplifiers have some advantages over SET amplifiers. As mentioned earlier, two signals are summed in the output transformer. If some care is taken to match the tubes or bias them appropriately, this means that DC offset current can be more or less cancelled out during output signal summation. The output transformer hence only has to handle signal current. We would then be able to use a smaller (and much cheaper) output transformer.
This also means that PP amplifiers can get along with cheaper power supplies – any anomalies in the signal due to the power supply (e.g. power supply hum) are also cancelled out in the output transformer.
All this means that PP amplifiers typically have better measured distortion levels than SET amplifiers.
However, PP amplifiers have more higher order odd distortion than SET amplifiers. Why? As mentioned earlier, even order distortion occurs when there is uneven gain on either the positive or negative portion of the signal. Given that one half of the tube pair amplifies an inverted signal, and assuming that both tubes in the tube pair are reasonably matched, even order distortion cancels. On the other hand, odd order distortion sums.
OTL
OTL stands for `Output Transformer-Less'. Despite a recent surge in popularity, OTL topologies have typically been regarded as ‘fringe’.
While properly designed output transformers perform superbly, OTL designs attempt to avoid the inherent non-linearity of output transformers (no matter how little) by avoiding them completely.
Tubes designed for audio use typically have high output impedances and are not suited for driving low impedance loads like headphones or speakers directly (i.e. without an output transformer). OTL designs use special tubes to get around this problem. However, another problem replaces the one just solved - these tubes (examples include the 6AS7/6080 and PL509) are typically not designed for audio, and so may have questionable linearity.
Not having output transformers, OTL designs usually have abysmal efficiency.
Hybrid amplifiers
Hybrid amplifiers attempt to combine the best of both worlds - the prized tube `sound' with the extremely low output impedance and high current capability of solid-state. Often, a tube stage is placed before a final solid-state output stage.
There are countless variations of this topology, the only thing common to them all being that they involve both tubes and solid-state. Hence, they will not be covered in this guide.
2.2 Tube biasing?
Tubes need to be biased in order to function properly as low-distortion amplification stages.
But what is tube biasing? I'll quote a commonly used analogy - imagine
a car engine.
We want to tune the engine so that the revolution speed (RPMs) at idle is neither
too fast nor
too slow. In other words - while idling - we do
not want it to be consuming more fuel than it should (overly high RPM), nor do we want it teetering on the brink of stalling (too low RPM).
Now apply the same logic, and you are "biasing" a tube.
Biasing a tube involves setting the 'default' DC voltage (i.e. when no signal voltage is applied) for the control grid. Remember that a tube is a device that is a `nominally on' device - that is, electrons will always attempt to flow from cathode to anode. The grid (and the voltage applied to it, known as the bias voltage) is the only thing standing in the way of that.
Bias your grid to too positive a voltage, and too many electrons will flow from cathode to anode (consuming more fuel that it should). Bias your grid to too negative a voltage, and too few electrons will flow from cathode to anode (teetering on the brink of stalling). [Remember - different charges attract, like charges repel - electrons are negatively charged particles]
To sum up, the amount of DC bias voltage you apply to the control grid determines several things about that tube amplification stage: power output capability, distortion figures, maximum headroom, gain, efficiency and most importantly, the class of operation of the amplification stage (A, AB, B etc).
There are 2 "normal" methods
(and 1 proprietary method) of biasing - cathode biasing and fixed biasing. The
proprietary method is PrimaLuna's "Adaptive AutoBias", and will be elaborated upon in a bit.
Confusingly,
fixed biasing is the biasing that is actually
user-adjustable! In fixed biasing, a trimming potentiometer (i.e. trimpot) is adjusted by the user to determine DC voltage to the grid. As the operating parameter of tubes change as they age, increasingly frequent adjustments will need to be made to the bias adjustment. One would also need to take into account manufacturing variances between tubes of the same make and type, and bias each one individually.
Cathode biasing involves the placement of resistors between cathode-ground and grid-ground. While not needed any user maintenance, it effectively means that there is now a big capacitor in your signal path - not necessarily the best compromise for perfect sound.
Be sure to check with your amplifier manufacturer which type of biasing your tube amplifier uses.
PrimaLuna's "Adaptive AutoBias" is a proprietary "on-the-fly" autobiasing. Essentially, it operates like fixed biasing, but with multiple sensors that monitor audio signal voltages, tube voltages, etc and feed a microchip controller this information. The microchip then figures out the "best" bias for the moment for the power demand and applies it.
While this is certainly a novel idea, and possibly one of the bigger (biggest?) leaps in tube circuit design in recent times I have not heard it personally - and thus cannot comment on the
audible (as opposed to theoretical) improvements it brings.
2.3 Regarding tube-based rectifiers...
the GZ32 tube rectifier, this example made by Mullard
Tube rectifiers are inefficient, expensive, have high output resistance, have large voltage drops across them, need separate heater circuitry, and in the case of some examples of recent make - are notoriously unreliable.
Taking the above into consideration, why are tube rectifiers still used?
Tube rectifiers have one clear advantage over silicon, and that advantage is so convincing as to win many converts - rise time. Rise time refers to the time needed for output voltage from the rectifier to rise from zero to operating levels when fully loaded. For tube rectifiers this is a gentle curve that extends over 5s or so. This greatly reduces inrush current generated by various capacitors down the line and potentially saves many a regulator.
For silicon rectifiers, the voltage rise line is a hard, almost vertical line (that has led to many a blown regulator for the author of this guide).
Many die-hard proponents of tube rectification claim that tube rectifiers `switch' more cleanly than their silicon counterparts, leading to less resonance within the power supply circuit. Mr. Morgan Jones, author of "Valve Amplifiers, 3rd Edition " debunks this, stating "the author's [M. Jones] experience is that both types of rectifier produce switching spikes and it is the smoothing/snubbing arrangements that are important."
Finally, there is another undeniable advantage - albeit a small one - that must be accorded to tube rectifiers: they certainly are much prettier to look at when compared to their silicon cousins.
2.4 Tube datasheets
Tube datasheets are useful things. Tube datasheets were the technical notes released by tube manufacturers. Every tube type has its own datasheet.
Amplifier designers refer to datasheets in order to choose an appropriate tube, determine operating points, design optimizations, component selection and the like. For end-users like us, the main use of tube datasheets is in researching tube substitutions.
Below are a couple of sites that have comprehensive databases of tube datasheets. Some have pdf scans of the actual datasheets, others have compiled the relevant data into html pages, and yet others have done the dirty work for you and listed out the tube substitutions for each tube.
http://tdsl.duncanamps.com/tubesearch.php
http://www.duncanamps.com/tdslpe/
http://hereford.ampr.org/cgi-bin/tube?index=l
http://www.qsLnet/wa7zcz/area2/page58.htm1
http://www.arrakis.es/-igapop/referenc.htm
2.5 Tube testers
Often while buying tubes, you will often come across readings supplied by 'tube testers'. What are tube testers? In the distant past, tubes were sold 'as-is', with no returns or refunds. Tube testers were used then to prove on-the-spot tube operability.
There are 2 major types of tube testers - emission testers ('dynamic/plate/cathode conductance' testers) and mutual conductance ('GM') testers. The latter utilises a superior method of testing a tube than the former, though neither type is useful at all if the tester is not recently calibrated.
Emission testers work by simply measuring `emission', which is the ability of the cathode or filament to emit electrons. Results are typically displayed as mA (milli-amperes). However, the cathode's ability to emit electrons is at best secondary to the importance of the grid's ability to modulate this same electron flow. Sometimes electron flow may originate mostly from a single `hot spot' on the cathode/filament. Emission testers do not take this phenomenon (which to some extent negates the modulating capability of the grid) into account. In addition, emission testers usually utilise a low current when testing tubes - in audio applications, current draw is almost always higher than in an emission tester. Hence at best, emission testers are useful only as a quick way to separate dud tubes from working ones.
On the other hand, mutual conductance (or GM) testers work by analysing the tube's ability to amplify an AC signal (typically 1kHz). AC plate currents are then read off and may be averaged for some time to give a GM value. Results are typically in µmho (micro-mhos). Usually such testers incorporate scales which indicate where your tube stands in a REPLACE-FAIR-GOOD scale. Some better GM testers test give results over a range of frequencies while other even better ones utilise the rated plate voltages (as opposed to a compromise lower voltage) when obtaining results.
Excellent GM testers were made by Hickok and B&K. eBay is the usual source for tube testers.
the author's B&K 747 GM tester
Remember that the ultimate test for any tube is still the actual amplifier you're putting the tube into. Many tube testers will NOT tell you anything about microphonics, real life span or how a tube sounds.
Tube testers are useful only for testing for faulty (and hence dangerous) tubes, matching tube sections and giving rough estimates of remaining lifespan.
2.6 The proper care and feeding of tubes
• The greatest chance for tube failure is at turn-on. Minimize on/off cycles for your tubes to prolong their lives. For example, if you will be listening 3 times within 4 hours, instead of switching the amp on/off 3 times, leave it on for the entire four hours! Better still, ask your amp manufacturer whether he can implement a ‘heater standby voltage’ mode.
• Avoid storing your tubes in drawers. The brief shock whenever you close the drawer is not good for tubes at all – a grid wire may break, or a oxide cathode might crack, or a anode might fracture, or a mica plate might crack, or etc...
• Original tube boxes were rarely, if ever, made of acid-free paper. To slow their inevitable degradation, store them in an airtight box with dehumidifier sachets in a dark and cool place.
• NEVER EVER tap your tubes with anything to test for microphonics while they are running. At higher temperatures (like that found in a running tube), everything is more fragile.
• For tubes with metal bases that are rusty: light rust can be wiped off with a cloth soaked in lemon juice or vinegar before being cleaned with a dry cloth and coated with some dielectric oil. For heavy rusting, gently sand off affected areas, then follow the steps for combating light rusting
• A rattling sound from your tube base is bad? Wrong! The sound is usually due to some dislodged glue, and is usually harmless.
• For tubes with cracked bases, you may attempt to secure the base by wrapping it with some electrical tape. This is a temporary band-aid type measure at most though.
• Clean off any external deposits on the glass envelopes. When running, these deposits heat up and cool down at a different rate from the glass, promoting uneven expansion and stressing the glass at that point. Your glass envelope will crack in time if you ignore them.
• Tube markings and base markings are soluble in most solvents. Be careful around them.
• You can clean tarnished tube pins with some fine wire brushes, isopropyl alcohol and cloth. Remember to clean away all the dirt you dislodge though!
• A tube that can rotate slightly (roughly 1 mm or so) within its base is usually perfectly alright. Be careful with it though.