Amplification, a bit more precisely (Patreon)

I realized that I used some vague language in describing exactly how guitar (and bass, and radio, etc.) amps work, and I considered adding this to pt. 3 of the series, but it would take too long, and god knows I don't need to make Dan sit and listen to me blather any more.  So, for those of you who are interested, here's a further-reaching explanation:

The main preamp tube in virtually all modern guitar amps is a 12AX7 (called an ECC83 in Europe); there are other types, but nearly all are dual triodes, meaning they each have a pair of 3-part common gain stages in them.  Those 3 parts, you may remember, are:

• the cathode, which emits a stream of electrons as it's heated by a filament

• the anode or "plate," which is being hit with high-voltage direct current (AC vs. DC is very important here)

• and, in between them, the grid, which regulates flow between the two

The grid is biased negative, i.e. set up to idle at a negative charge.  For the sake of simplicity, and because these are common values, we're going to say that the grid is biased -1.5v, and the anode is receiving +150v DC.

When your guitar string vibrates over the pickup, it generates a voltage wave with both positive and negative peaks.  That's what "alternating current" means.  Now, this caught me up for a stupidly long time: "-1v" does not mean "1 volt less than zero," which makes no sense.  It means "1 negatively-charged volt."  It's still a volt, equal in magnitude to +1v.  A guitar puts out a very small voltage wave; Fender used +/-0.03 volts as their test voltage for amplifiers, so that'll do. 

(Note: when I write "+/-0.03 volts," it means "the signal wave has a positive peak of +0.03v and a negative peak of -0.03.")

That AC voltage wave heads to the grid of the first tube, sitting there idling at -1.5v, and as the wave hits the grid, it creates a second, bigger AC voltage wave using the +150v DC as its base.  How?  Two ways:

• Guitar string goes up (let's say; it could be down, too, depending on your guitar, pickups, etc.) and sends a positive voltage to the tube's grid.  The grid goes more positive, meaning it starts to attract the beam of electrons that's firing off the cathode; this beam meets the +150v DC from the anode and briefly drives it more negative, creating a voltage change of, say, -1.5v.

• Guitar string goes down (or whatever) and sends negative voltage to the grid.  This drives the grid more negative (remember, it's negative to begin with), so it no longer accepts the electron stream from the cathode, and it more strongly attracts the +DC from the anode.  That creates a voltage spike in the opposite direction – say, roughly, +1.5v.

See what's just happened here?  We're using DC for the way it interacts with the tube's grid, but DC doesn't fluctuate up and down.  We've taken the DC as a base and modulated it with the AC guitar signal to create a new AC wave of +/- 1.5 volts.  With an input voltage of +/- 0.03v from the guitar, that's 50 times the size of the original signal.  Not bad.  

(A lot of vintage amps without master volumes, like ’60s Fenders, would use that kind of gain stage by itself as "input gain" and then send the signal on to reverb, tremolo, tone controls, whatever.  But if you "cascade" 2 or 3 or 5 gain stages into each other, as Randall Smith of Mesa/Boogie discovered, you've got a high-gain amplifier.)

The +DC we've just modulated to get our new amplified signal, however, is only useful as raw material for creating that larger AC swing; it has to be amputated before it gets to the next triode stage and blows something up.  So, right after the gain stage, there's a small capacitor called a coupling cap that will slice off the +150v DC and allow the +/-1.5v AC to pass through.  

(Incidentally, the part of the amp that takes electricity from a wall socket and turns it into the +DC power supplies for each tube stage is called a rectifier, and can be either a tube itself or an arrangement of silicon diodes.  The Mesa Dual Rectifier is so called because, on the original version, you could switch between various rectifiers.  That change will have a huge impact on how the amp feels to play: with silicon diodes, it'll seem bold, punchy, and immediate; with a 5U4 tube rectifier, it'll be "saggy" and "squishy," because the tube takes milliseconds longer to get up to the necessary voltage; and with a 5AR4 tube, also known as a GZ34, it'll be somewhere between those extremes.)

... "But wait," the keen-eyed among you may be saying, "a positive signal from the guitar created a negative voltage spike?  And a negative one created a positive spike?"  Yes!  That's why the common triode gain stage is called an "inverting" stage: it reverses the polarity of the signal every time that signal passes through another gain stage.  That's not just a curiosity; it's crucial to the way amplification works.  

Since the amplified guitar signal has two "lobes," i.e. its negative and positive peaks, both lobes need the chance to act as positive signal, which is the lobe that can create clipping (distortion, compression, etc.) by driving that -1.5v grid bias nearly up to 0v, neutral, and thus forcing the tube to turn itself off to save its life.  If the common gain stage didn't invert, you'd end up with a guitar signal that was half mega-amplified and half inaudible.  Plus, this way, the positive and negative lobes of the signal are asymmetrically clipped, i.e. each amplified in its own way, and asymmetrical clipping creates harmonics that human ears tend to like.

Except for the necessity of this inversion, however, positive volts and negative volts are, in themselves, relatively unimportant until you get to the power amp, and here's the class A vs. class AB thing I mentioned.  In class A, where the power tubes are on 100% of the time and amplify 100% of the signal, those tubes have to be biased hot enough (i.e. have to be running at high enough output power) that they can cover the full voltage swing of the signal.  That swing might be, say, 200 volts or more in each direction by the time it hits the power amp.  It's got to be hot enough that a -200v input doesn't drive the grid so negative that it just turns off, but still cool enough that a +200v input doesn't drive it into complete unmusical overload.  Delicate balance, and it'll cost you $100 for rebiasing every time you need to change a tube.  This all means a class A power amp will tend to be noisy (because it has to amplify everything), unstable (because it's right up near its max output all the time), and hard on tubes (because they're always running hot).  It will give you more power-tube distortion, though, and will arguably make the power tubes a bigger part of the sound, if that's what you want.

Class AB uses a tube called the phase inverter to produce 2 versions of the signal, one in phase and phase-flipped (i.e., version 2 will have a negative peak wherever version one has a positive peak, and vice versa).  It then the "correct" version of the signal to half the power tubes and the phase-flipped version to the other half of tubes.  That means these tubes (let's say we've got 2, for convenience's sake) have a much easier time: each tube only needs to conduct half the signal before it gets a chance to rest, each can be biased much colder (run at lower percentage of its capacity) because it only has to fit half the waveform into its operating range, and, consequently, tubes will tend to last a lot longer in class AB.  

The main possible drawback of class AB is in amplifiers that have simple, small preamps and rely on power-amp saturation for their tone; since AB is biased colder, it's much harder to drive into clipping.  But if the amp's tone is mostly shaped by the preamp, then class AB will do a much better job producing a faithfully amplified version of that preamp signal.  It can also generate more power than class A with the same number of tubes, for complicated electromagnetic reasons (involving 2 spools of wire wrapped around an iron cinderblock called an output transformer) that I don't entirely understand yet.

MSJ