Output transformers are one of the most important factors in the sound of a tube guitar amp, but they don't get discussed as often as other components for a few reasons: output transformers, or OPT's, are fairly expensive, they're inconvenient to swap out, and there aren't many transformer options available for most amps.
How They Work
The output transformer couples the signal from the output tube or tubes to the speaker. The tubes feed one coil, which is magnetically coupled to another coil through literally a bunch of iron, and the speaker is connected to that second coil. Most of the signal makes it through the transformer without becoming too distorted, and the rest becomes heat.
Why is a transformer needed at all? It's the same reason a car's engine isn't connected directly to its wheels: the engine turns optimally within one speed range, and the tires at another. Think of the output transformer as an amp's transmission – a one-speed, fluid-coupled transmission. The tubes are like an engine that's most content going a few thousand RPM, and the speaker is like the wheels turning at only a few RPM. The important thing to remember is that the load on the transformer's output – i.e. the speaker – is “felt” by its input, or primary.
To illustrate this, think of a fixie, or fixed-gear bike – one where the rear hub can't coast or freewheel. If the bike is moving, so are its pedals. When you climb up a hill, it's harder to pedal. If you're going down a steep hill, the pedals might be turning as fast as you can keep up with. The load on the bike's wheels (gravity) is reflected back to the driving force (your legs). A tube amp's output transformer works the very same way: it reflects the speaker load across its secondary back to the tubes via the transformer's ratio. At the risk of belaboring the bicycle analogy, just as a multigear bike lets you choose how hard you have to push the pedals to go up a hill, output tubes can function into a range of loads, affecting power output as well as distortion magnitude and character.
You'll often see a specification say that an output transformer has something like an 8000 ohm primary and an 8 ohm secondary. It would be more accurate to say that that transformer will present an 8000-ohm load to the output tube(s) if there's an 8-ohm load connected to its secondary. In other words - and this is a more useful spec – that transformer has a 1000:1 impedance ratio.
So what happens if the speaker load is 4 ohms instead of 8? That OPT with its impedance ratio of 1000:1 will now present those output tubes with a reflected load of 4,000 ohms. They may or may not like it. In any case, they'll operate differently, deliver a different power level, and a different sound. Or they could melt and cost someone a lot of money.
In the same way, removing two output tubes from an amp that has four of them doubles the impedance reflected to the remaining tubes. Of course, this can be compensated for by halving the speaker load, e.g. from 16 to 8 ohms, or from 8 to 4.
It's important to realize an 8-ohm speaker doesn't present a pure 8-ohm load to the output transformer the way an 8-ohm resistor would. A speaker's impedance rating is its average resistance to an AC signal. It'll be 8 ohms at one or two frequencies, and more or less than that at all others. So that changing load, which is being affected by which notes are being played and how that speaker reacts to them, is always being reflected back to the amp's output tubes.
This touches on why it's generally a bad idea to operate a tube amp with the speakers disconnected: that's an infinitely light load being reflected back to the tubes. They don't like that. For that reason, the output jacks on many amps actually short the OPT's secondary when the speakers are unplugged, offering a bit of short-term protection from tube damage. With tubes, a heavy load can be better than none at all.
I mentioned that the output transformer is a little like a fluid-coupled transmission. That's because output load variations like high or low frequencies or fast transients are not all reflected purely back to the output tubes. The output transformer can be seen as a non-phase-linear bandpass filter, and its inherent non-linearity can compromise the effectiveness of negative feedback loops that include it.
The output transformer in a tube guitar amp is an iron-core transformer. It's essentially two coils wound around the same block of iron. That iron is sliced up into thin layers, or laminations, so that currents don't form in the iron that would interfere with its ability to transfer signals efficiently from one coil to the other.
The coils are wrapped one on top of the other onto the same form or bobbin, which is then slipped over the iron core's center shaft. In practice, part of the primary is wound onto the form, then some or all of the secondary, then more primary, and so on. This is called interleaving. In high-end audio transformers for tube hi-fi amps there can be a lot of interleaving, which helps the magnetic coupling of the two coils, reduces losses, and improves frequency response.
From a purely technical standpoint, the output transformers in a lot of great sounding vintage amps were actually a weak link. Transformer iron is expensive, so some amps' transformers were undersized and would saturate at high signal levels. They also had a minimum of interleaving, which also helped keep the costs down. Frequency response is not critical for electric guitar: the low E is about 80 hertz, and typical guitar speakers were only functional up to about 5kHz tops, so why bother with a high-quality output transformer?
Some people believe that old transformers have some sort of special “mojo”, owing to the type of construction and materials used, like the paper or the formula of their varnish. It would be interesting to conduct a double-blind listening test with all other parameters held constant to see whether these things actually make a detectable difference. Although they might be present in an amp that sounds good, the reasons for that sound probably lie elsewhere. Insulating materials in a transformer don't behave like the dielectrics in capacitors, which are electrostatic devices. What matters most in a tube guitar amp output transformer is that its materials can stand up to several thousand volts and possible high temperatures without melting, burning, or becoming perforated.
So if the guitar amp output transformer is a lo-fi piece of iron, how can it have so much impact on the amp's sound? We've discussed how output tube efficiency and distortion characteritics will depend on the speaker load that's reflected upstream via the transformer. Low-efficiency coupling means less volume and more heat, so amplifier designers typically want those tubes to be seeing a load that allows them to deliver the most clean power. But we guitarists like a little dirt, preferably from several places at once. The way to find the sweet spot where the output tubes are distorting in a way we like is by being willing to sacrifice a little bit of efficiency. This is a worthwhile trade-off, as many tube amps don't sound their best until they're impractically loud anyway. In general, a lighter than optimum load favors 3rd harmonic distortion, whereas a heavier load favors 2nd harmonic distortion.
To a limited extent, it's possible to experiment with tweaking the load on the output tubes by changing the speaker load and/or the amp's output impedance selector. But don't do this indiscriminately, as it can permanently damage your amp's costliest components. Plan your experiment and calculate the new load on the output tubes, checking it against the dissipation limits on the data sheet.
A Helpful Tip
There's a way to roughly measure the impedance ratio of an unknown output transformer. Two important facts make this possible. First, the impedance ratio is the square of the voltage ratio. This means a transformer with a measured voltage ratio of 30 to 1 has an impedance ratio of 30 x 30, or 900:1. Second, a transformer's primary and secondary are interchangeable. This means you can apply an AC voltage to the secondary, and read a higher voltage on the primary, which makes for more accurate readings. Connect a signal generator capable of driving a low impedance (like a tuner app on your phone fed through a small signal transformer), and connect it to the mystery transformer's secondary. Measure that AC voltage, then measure the voltage across the entire unloaded primary. Divide the two voltages and square the result to get an approximation of the impedance ratio. As we did before, multiply the chosen speaker impedance by that impedance ratio to get the primary impedance, or load presented to the output tube or tubes.
Wrapping it Up
Like pickups and speakers, a tube amp's output transformer is an electromagnetic device that plays a crucial (and often overlooked) role in forming one's overall tone. Keeping in mind how the transformer reflects (“transforms”) speaker impedance back to the output tube or tubes can be the basis for some informative experimentation. It might even save those precious output tubes from a spectacular demise.
Let's take a look at the pros and cons of the different techniques that have been used to build guitar amplifier circuits, and make an attempt to clear up some terminology issues that have crept in.
As the oldest and simplest circuit construction technique, point-to-point wiring evolved from the original form of “breadboarding”, a method for building one-off circuits that involved twisting or soldering component leads onto nails driven into a piece of wood -- literally a cutting board for bread. The nails could be placed to match the needs of the circuit. If a breadboarded design was put into production, the same layout could be transferred to the inside of a chassis, with the nails replaced by short terminal strips where needed.
The key to identifying point-to-point wiring is in its name. If components are wired together directly, e.g. a resistor soldered between an input jack and a tube socket with no wires or boards in between, that's point-to-point wiring. Other methods discussed below are sometimes referred to as point-to-point because hand wiring is involved, but there are some very important differences.
True point-to-point wiring minimizes the number of connections between components, so it's potentially more reliable than methods requiring more connections or distance. Also, point-to-point wiring can have a Z-axis, so the circuit can retain some of its inherent three-dimensional characteristics. This is important because it allows sensitive signals to take the shortest possible routes and avoid interference (hum, noise, oscillation) through unwanted inductive coupling.
Point-to-point wiring can also be physically robust. The circuit can be resistant to damage if the chassis is twisted or flexed. There is no board to crack or to interfere with heat flow around components.
Besides having to be carefully planned in three dimensions, point-to-point wiring has a few things working against it. For one, it must be performed entirely by hand, so it is time consuming and very prone to human error. Soldering (and desoldering) components that are above one another is tricky work, sometimes leading to accidental component damage. And because it's so labor intensive, commercial point-to-point wiring has traditionally been performed by low-paid workers unfamiliar with the products they're assembling. This is true in almost all commercial electronics assembly, but it has greater impact on something as error-prone as point-to-point wiring. In a production environment, speed has often been valued over thoroughness.
It should be added that point-to-point wiring looks like a dog's breakfast to some, especially if the layout respects circuit flow and empirically minimizes signal interference. This has nothing to do with the amplifier's quality, usefulness, or reliability, but some people are uncomfortable unless the components in their amplifier are all in parallel rows, with the wires all bent at 90 degree angles and bundled together. While it's true that orthogonal paths minimize signal coupling, this approach is often taken to extremes for aesthetic reasons at the expense of circuit performance.
It's been said that a true point-to-point amplifier is harder to repair than one with a flatter layout. Whether this is a relevant really comes down to how much of the amp's life is going to be spent being taken apart and worked on, versus turned on and played.
Point-to-point wiring is found in pre-1948 Fender amps, smaller early Supro and similar amps, a few modern high-end amps, and very little else in the guitar amp world.
Turret / Terminal Board
Like point-to-point, turret board board construction also began life as a prototyping technique, but one that employed a generic circuit board that could be reused for other projects. Turret boards (or terminal boards) usually consist of two parallel rows of evenly-spaced terminals swaged into a strip of bakelite, phenolic, or fiberglass. Rarely, a circuit-specific turret board is made, giving the builder more control over circuit flow and component layout. Either way, component leads and connecting wires are soldered onto the terminals. Wire connections are then made to chassis-mounted components, like switches, controls, jacks, and tube sockets.
Turret board construction has some advantages. A prototyped circuit can be transferred directly to a finished one-off project, or even put into production as is. Strong, reliable connections are made by wrapping the component leads around the terminals and soldering them. The boards used tend to be thicker than printed circuit boards, and as such they're relatively less likely to crack.
In a production environment, turret board construction offeres advantages in consistency, as assemblers can copy a sample board and spot errors more easily than with point-to-point wiring. But it's still time-consuming work, and the same assembly quality vs. cost drawbacks of point-to-point still apply.
Perhaps more than any other method, generic turret board construction imposes a two-dimensional form on the circuit. This might not seem like a disadvantage (after all, we're used to seeing amplifier circuits shown this way on schematics) but a two-dimensional layout doesn't represent actual signal or current flow. Parallel component placement also means that most circuit elements will likely be farther apart or closer together than in an optimized layout. What's more, the additional wires and solder connections between circuit elements in turret board construction inherently compromise reliability: with more connections, there's simply more to go wrong.
Turret board or terminal board construction is found in early Vox, Marshall, and Hiwatt amps, as well as other (chiefly British) brands.
Employed in Fender and Gibson amplifiers for decades, eyelet board construction is similar to turret / terminal board construction, but provides some key advantages.
Consisting of a sheet of insulating material (traditionally paper-based) with an arrangement of tin-plated rivets installed in a pattern specific to the circuit being built, and eyelet board can be seen as a compromise between point-to-point and printed circuit board construction. Because an eyelet board is made for a particluar circuit, the components can be arranged in patterns that aid circuit flow and reduce unwanted inductive coupling. I say “can” because most eyelet boards tend to stick to fairly linear component arrangements, making it easier for time-pressured assembly workers to avoid obvious errors. Linear layouts also help minimize board area, further reducing cost.
One big advantage eyelet boards have over turret boards is that an eyelet board makes it possible to reduce the number of connections, improving reliability. You're more likely to find four adjacent components sharing a single eyelet than a single turret.
For early manufacturers, eyelet boards provided another key advantage: they were dirt cheap -- both to build and to populate -- yet they provided consistent results.
The main drawbacks of eyelet board construction stem from how it was (and is) typically implemented in vintage mass-produced amps and their countless replicas. As mentioned above, the circuit layouts originally governed by cost-cutting concerns are rarely ideal, so hum and noise levels are typically higher than necessary. Early examples of the paper-based boards absorbed moisture – not a desirable feature in high voltage equipment. The resulting changes to the damaged boards' physical and electrical characteristics could cause some amps to oscillate, especially if they'd been relying on stray capacitance from the nearby grounded chassis surface for stability.
Printed Circuit Board (PCB)
PCB technology has been ubiquitous in electronic equipment production since the 1960's, mainly due to its low cost and high consistency. In its simplest form, a copper-coated board of phenolic or fiberglass is printed with an image of a pattern of conductors or “traces”, then soaked in acid to remove the remaining unwanted copper. Holes for components and their leads are drilled, usually by a CNC machine performing a programmed series of steps. Various other stages were added to circuit board production over time, such as plating and screen printing.
Components were installed and soldered onto early PCB's by hand. Later, processes were developed to perform all the solder connections at once by dipping the loaded board into a bath of molten solder or passing it over a wave of molten solder (wave soldering). Automated soldering techniques are sometimes thought to be inferior to hand techniques due to the difficulties of performing real time quality monitoring, but they brought new levels of consistency to electronic equipment production, as did machine placement of components.
Layouts for printed circuit boards were originally made by manually placing strips of tape and donut pads on a sheet of mylar. This allowed for some very organic – and even artistic – layouts. As PCB design software emerged, layouts became more linear, and traces more angular. While this approach facilitates the densities required for digital circuitry as well as automated component placement, it can be detrimental to high voltage, high impedance analog circuits like those in tube guitar amps.
Most printed circuit boards in guitar amplifiers are single-sided, meaning there are copper traces on one side only, and components on the other. Most amp circuits are simple enough that the density advantage of a second copper layer isn't needed. But double-sided PCB's with plated holes do a much better job of holding components in place. The plated holes act like miniature eyelets, increasing both the electrical contact area and the connection's total board-to-copper adhesion strength.
PCB creation produces more chemical waste than any of the other construction methods described above. Even with the much-needed banning of toxic lead from electronics production, PCB manufacturing remains a subtractive process with a negative environmental impact.
A trend that contributed to the bad reputation PCB's have earned in tube amplifier circles was a bold attempt on the part of manufacturers to further reduce production costs by mounting everything they could on PCB's, including tubes, heat-producing resistors, controls, switches and jacks. Anyone who fixes amplifiers can tell you that these design shortcuts are prime causes of failure, and can make for nightmare repair jobs. Even today, most mass-produced amps have their front panel controls mounted on printed circuit boards, as hand wiring is deemed too expensive by their manufacturers or their customers.
Although you may never see one in a tube guitar amplifier, modern multilayer circuit boards with surface-mounted components on both sides actually resemble point-to-point construction in that they strive to minimize connection lengths and retain a three-dimensional circuit architecture. And as with point-to-point layouts, designing multilayer PCB's is not a trivial task.
Combinations of the above construction techniques are used in both commercial and home-built amps, and more combinations are possible. For example, many British amp brands have for years employed single-sided printed circuit boards that closely resemble the earlier turret or terminal boards they replaced. One the one hand, this allows manufacturers to sell their products less expensively (and/or more profitably), but on the other hand it retains the layout restrictions of linear turret boards and the strength limitations of single-sided PCB's.
Most older amps using boards of some kind employed a little point-to-point wiring here and there. An example would be the resistors on Fender input jacks or Vox cathode follower tube sockets. Passive components are also sometimes soldered directly onto tone controls. But while manufacturers usually did this for convenience, individual builders can integrate sections of point-to-point wiring where these would most benefit the circuit itself.
Builders who choose to make their own turret or eyelet boards have an excellent opportunity to craft their circuit layout to optimize component placement and flow. And desiging a true point-to-point version of a chosen circuit is an even more rewarding challenge.
Wrapping it Up
From the abundance of good amplifiers old and new, it's clearly possible to make a decent product using any of the above circuit construction techniques. That being said, some of those techniques are clearly more oriented toward saving money in a large-scale production environment where every penny is counted. Anyone planning to build an amplifier should consider examining their design goals and options before settling on an approach simply because it's popular or was used in the factory that produced a revered amp decades ago. The myriad design and component options available to today's builders would likely thrill the commercial amplifier designers of old. If you want to honor their greatest achievements, make the design decisions they might have made if their hands weren't tied by the cost cutters down the hall. A little extra thought and planning can make all the difference between an imitation and a masterpiece.