Tube Guitar Amp Overdrive
This webpage is available in .PDF form: Tube Guitar Amp Overdrive.pdf
While distortion is to be avoided at all costs in audio amplifiers, it is the most important aspect of guitar amplifier design. The primary reason tube guitar amplifiers are favored over solid state amps is their overdrive tone. Very few people can tell the difference between a high quality tube and solid state guitar amp in a blind test when playing clean. But push the amp into overdrive distortion and the difference can be stark. Tube overdrive tends to exhibit softer, asymmetric clipping, time-variance and nonlinear distortion that is almost impossible to mimic using solid state devices. Electrical engineers will tell you that solid state amplifiers can faithfully mimic tube clean and overdrive tone but I say why bother with something that is trying to mimic the real thing. So I admit up front I have a tube guitar amplifier bias.
NOTE: On this page I do not use "conventional current flow." Conventional flow assumes current flows from positive to negative because Benjamin Franklin guessed wrong on the direction of electron flow. On this page I describe the actual flow of electrons from negative to positive and from cathode to plate because with tubes it's important to understand the actual movement of electrons through the circuit.
Table of Contents
Quick Review of How Tube Amplifiers Work
This short review may help you understand the following discussion on tube overdrive. In a vacuum tube negatively charged "space charge" electrons flow from the hot cathode, through the control grid to the positive charged plate. It's the high voltage DC on the plate that pulls space charge electrons from the cathode, through the control grid and onto the plate. The guitar AC (alternating current) signal voltage charges the tube's control grid which acts as the control valve of the tube. During non-overdrive conditions the control grid stays at a negative voltage compared to the cathode.
The no signal voltage difference between the grid and cathode is the bias voltage. The bias voltage sets the tube's operating point. You must understand this concept to understand tube amplifier overdrive. Anything other than the guitar signal that changes the voltage difference between the grid and cathode shifts the bias voltage and operating point.
Like electrical charges repel one another so when the control grid has a strong negative signal voltage (negative voltage = excess electrons) they repel the negatively charged space charge electrons in the tube and slow their flow through the grid to the plate. Negatively charged electrons don't like to flow through a negatively charged grid. Opposite electrical charges attract one another so the positive plate pulls the negative charged space charge electrons to it.
When the grid has a positive signal voltage (positive voltage = scarcity of electrons) its blocking power is reduced and space charge electrons flow through the grid to the plate at a higher rate.
Standard Vacuum Tube Preamp Circuit
Guitar signal voltage enters the tube on the left at the Grid and exits at the Plate. This tube is a 'triode' which means it has three electrodes, the cathode, grid and plate. Space charge electrons flow from the cathode through the grid to the plate.
The signal voltage on the grid is amplified because a small voltage change on the grid creates a large electron flow change through the grid which leads to a large voltage change on the plate. The voltage changes on the plate are the amplified guitar signal. See How Tubes Work for more information.
An audio signal is distorted when there is a change in signal wave shape from input to output. Since tube amps' transfer curves are nonlinear they will always cause some distortion even when played "clean." But this low level distortion adds texture and warmth and fattens up the tone. Perfectly amplified electric guitar with no distortion can sound cold and sterile. That's why solid state audio amps make electric guitar sound so bad.
Guitar AC Audio Signal Voltage
Fender Stratocaster connected directly to an oscilloscope. Each 'wave' on the oscilloscope is caused by one string vibration. The left side of the graph shows a high open E string pluck followed by a pluck of the low open E. Voltage is on the left scale and time runs along the bottom. The top half of the signal is positive voltage and the bottom half is negative. The guitar's signal voltage and current alternate between positive and negative.
If an amplifier's transfer curve is a straight line (linear) you will have no distortion. The output signal will be an exact, but larger version of the input. Tube amplifiers' transfer curves are not straight lines, they are non-linear so tube amps exhibit nonlinear distortion even when not being overdriven.
A tube transfer curve is a plot of control grid voltage vs. plate current. Since the transfer curves are not straight lines or linear, nonlinear distortion will be created during amplification even at low volume.
Overdrive distortion is a type of distortion that occurs when the AC input signal voltage swing is too large and the output signal is clipped causing the addition of harmonic and intermodulation distortion. Vacuum tube clipping can occur due to cutoff and/or grid clipping. When an AC signal's lower, negative lobe is too large it is clipped or flattened by cutoff. Cutoff occurs when the input signal's negative lobe on the grid goes so negative that all electron flow through the tube is shut off. Even if the signal on the grid goes more negative the output signal cannot respond because electron flow is already shut off so the output signal is capped and flattened. This change in wave shape from input to output is overdrive distortion.
Tube amplifiers don't actually clip due to true saturation (unrestricted electron flow from cathode to plate) because as the control grid approaches a positive voltage the signal on the grid is clipped well before the point of true saturation. For tube amps the term "saturation" really means the onset of grid clipping.
Understanding grid current is important to understanding tube overdrive distortion. Electrons flowing from the control grid upstream can cause bias voltage shift, cutoff clipping, grid clipping and blocking distortion, typically in that order.
During normal operation a tube flows negatively charged electrons from its cathode, through the control grid, to the positively charged plate. Since opposite electrical charges attract, the negative charged space charge electrons are attracted to the positive plate. The tube's bias voltage places the control grid at a negative voltage compared to the cathode voltage. Since like electrical charges repel one another a negatively charged grid has excess electrons which repel the negatively charged space charge electrons in the tube. The more negative electrons on the grid the more blocking power the grid exerts on the space charge electron flow.
When a large input signal voltage hits the control grid the positive lobe of the signal pushes the grid toward a positive voltage. As the grid becomes more positively charged the more negative charged space charge electrons will be attracted to it in the same way space charge electrons are attracted to the positive plate. Remember, a positive voltage is a scarcity of electrons so a positive guitar signal on the grid pulls electrons off the grid. If the pull is too hard it will actually start pulling space charge electrons in the tube onto the grid and into the input circuit. This is an important concept to understand so I'll say it again. When a guitar audio signal is too large for a tube circuit the excess positive voltage on the grid will pull space charge electrons in the tube onto the control grid. This is how grid current begins to flow.
As space charge electrons are captured by the grid, grid current will begin to flow upstream toward the grid leak resistor and coupling cap. This is known as grid conduction or grid current. Grid current is direct current (DC) and will charge an upstream coupling cap with negative voltage and cause a voltage drop across a grid stopper and grid leak resistor.
Grid current electrons want to flow to ground but the grid leak resistor restricts their flow so they build up on the grid and inside the coupling capacitor. This "build up" of negative electrons is a negative voltage bias shift.
Typical second preamp gain stage. During overdrive electrons are captured by the tube's grid and flow through the 220K grid stopper resistor to charge the coupling cap and through the grid leak resistor to ground. The grid stopper and grid leak resistance restrict the flow of electrons causing a buildup of electrons on the grid. This buildup of electrons on the grid repel the space charge electrons that are trying to flow from the cathode to plate.
Grid current will push the grid more negative which shifts the grid's bias voltage toward cutoff. In other words as the electrons flow from the grid to the coupling cap and through the grid leak resistor a voltage drop forms across the grid leak resistor which shifts the grid bias voltage negative toward cutoff. As the bias voltage changes, the tube's operating point changes. This dynamic grid voltage bias shift is very nonlinear so it generates nonlinear distortion along with harmonic and intermodulation distortion.
Asymmetric Soft Clipping
The sine wave input signal is shown in green and the clipped output signal is shown in blue. The bottom, negative lobe of the output signal has been rounded off and flattened due to cutoff clipping. Since the lower lobe has been distorted and the upper lobe has not this is an example of asymmetric clipping. The rounded shape of the lower lobe clipping makes it soft clipping. Soft asymmetric clipping is the most common form of tube overdrive distortion and what separates it from solid state transistor distortion. Photo by Boris Dugonich.
As grid current shifts the bias voltage towards cutoff it makes it more likely that there will be more negative lobe clipping than positive lobe clipping. This asymmetric clipping adds sweet sounding even order harmonics to the output signal.
To summarize, an overdriven input signal's positive charge on the control grid will lead to it being attractive to the space charge electrons flowing through the tube. As the space charge electrons accumulate on the grid and begin to flow, the grid voltage shifts negative toward cutoff--this is a grid bias voltage shift. This shift makes it more likely that the negative signal lobe will get clipped at cutoff. When one signal lobe is clipped more than the other the clipping is asymmetric.
Grid clipping occurs when grid current flows into the input circuit and loads down the positive lobe of the input signal. Grid clipping distorts the input signal. When grid current begins to flow the grid current caps the grid's maximum positive voltage which clips the positive lobe of the input signal on the grid. A larger positive signal voltage simply causes more grid current to flow and does not add to the flow of electrons to the tube's plate so the output positive lobe is clipped.
The tube will then faithfully amplify the clipped input signal and send it out the plate. In other words when grid current begins to flow it greatly reduces the tube's input impedance and the positive half of the input signal is loaded down to the point of clipping.
A tube amp circuit with global negative feedback will typically grid clip harder, with sharper waveform corners, than the same amp without feedback. The transition from clean to distorted also tightens up with negative feedback. The Marshall JTM45 is an almost exact copy of the Fender 5F6-A Bassman but the Marshall uses almost three times more global negative feedback. This is the main reason the overdrive tone of the JTM45 seems more "aggressive" than the 5F6-A.
If grid current flows long enough it will charge an input coupling capacitor. The coupling cap acts as a reservoir to store the electrons flowing from the grid. This reservoir of electrons will slow the voltage change back to normal bias (bias recovery). A certain amount of bias shift can lead to a pleasing overdrive dynamic but too much bias shift can be a problem.
Here's an online Power Tube Grid Bias Excursion Calculator.
Excessive grid current can cause a sufficiently negative bias voltage shift to cause blocking distortion. Blocking distortion can cause a complete dropout of the output signal voltage when the positive signal lobe is grid clipped while simultaneously the negative signal lobe is clipped at cutoff caused by bias shift (see Randall Aiken's explanation of blocking distortion for more info). Unlike most other forms of overdrive distortion, blocking distortion always sounds bad. "Farting out" is a popular term used to describe severe blocking distortion.
Because grid current causes a change in the bias voltage, grid current alters the flow of electrons from the cathode which can charge or discharge a cathode bypass capacitor so a bypass cap can affect bias shift depth and recovery time. The larger the bypass cap value, the longer the cap can support a bias shift. Using a very small or no bypass cap will shorten bias shift recovery time. This is another reason modern high gain amps use very small cathode bypass caps such as the .68uF found in many Marshall preamps.
Controlling Grid Current and its Effects
Heavy grid current and the blocking distortion it causes can be controlled by increasing the coupling cap charge time or decreasing the cap's discharge time. The cap's charge time can be slowed by placing a grid stopper resistor between the grid and coupling cap but a grid stopper will make grid clipping harder because the grid stopper will accelerate grid bias negative voltage shifts. The cap's discharge time can be quickened by reducing the value of the grid leak resistor but this attenuates the input signal. Power tube grid leak resistors are usually a much smaller resistance value than preamp grid leak resistors (220K vs. 1M) because of the greater amount of grid current they must sink.
Since low frequencies carry more energy they will charge a coupling cap more efficiently. Reducing the size of the coupling cap will reduce the bass content of the signal and slow the cap's charge time. When the input signal hits a small coupling cap bass frequencies are reduced downstream so the tube grid gets a filtered signal with less bass frequencies. When grid current begins to flow upstream to charge the cap it has less bass frequency energy to charge the cap which slows its charge time. This is one reason why high gain amps often use very small .0022uF coupling caps. The other reason small coupling caps are used is to reduce bias shift recovery time which also helps prevent blocking distortion. As stated earlier using small cathode bypass caps can also reduce bias shift recovery time. The Fender 5E3 tweed Deluxe uses huge .1uF coupling caps, no preamp grid stoppers and large 25uF preamp bypass caps which allow grid current to run wild for a perfect storm for "farting out" with extreme blocking distortion.
The interaction of grid current and coupling capacitor charging is a form of time-variance nonlinear distortion, meaning the distortion varies over time as the coupling cap charges and discharges. This time-variance can be controlled by altering the value of the coupling cap, grid stopper and grid leak resistors. A little grid current and the resulting drop in bias voltage toward cutoff is a dynamic source of sweet sounding asymmetric soft cutoff clipping. The dynamic bias shift caused by grid current and coupling cap interaction at the right level can be magical and is the hallmark of guitar tube amplification and its sweet overdrive tone.
Asymmetric Clipping, Duty Cycle Modulation and Even Harmonics
If a tube is biased perfectly between grid clipping and cutoff it is possible to get equal (symmetric) clipping where both the positive and negative lobe of the guitar signal are equally cut off and flattened. But even when biased perfectly bias shift will cause asymmetric clipping so asymmetric clipping is much more common in tube amplifiers than symmetric clipping.
Symmetric Soft and Hard Clipping
Since both the upper and lower signal lobes are clipped the same amount this is symmetric clipping. The tube's output signal upper, positive signal lobe has been clipped or flattened by grid clipping. The transistor's output signal upper, positive signal lobe has been clipped by saturation. The tube and transistor's lower, negative lobes have been clipped by cutoff. Note the sharp cutoff of hard clipping. Solid state amplifiers tend to clip harder than tube amplifiers which can sound more aggressive, raspy and less musical. (original image credit: Wikimedia Commons)
The Marshall 1987 amplifier's preamp first gain stages are intentionally biased off center to induce early asymmetric clipping. The Lead channel uses a 2.7K cathode resistor to shift the bias toward cutoff and bring on early negative lobe cutoff clipping. The negative lobe clipping will occur earlier than the positive lobe (which gets clipped much later by grid clipping).
Marshall 1987 first preamp stages are biased warm and cool with 820 and 2.7K ohm cathode resistors. Marshall also employs a very large 320uF and very small .68uF cathode bypass caps, and .022uF and .0022uF coupling caps to differentiate the overdrive tones of the two channels. The Lead channel uses a cathode resistor of 2.7K to shift the bias toward cutoff to cause asymmetric clipping. It also uses a very small .68uF cathode bypass cap and .0022uF coupling cap to reduce bias shift recovery time and prevent blocking distortion.
With asymmetric clipping one signal lobe carries the undistorted musical content while the other lobe carries the distortion. Asymmetric distortion also adds even order harmonics and makes the distortion sound more warm, musical and distinct from typical solid state distortion.
The Marshall 1987's normal channel biases the first gain stage off center to the warm side using a small 820 ohm cathode resistor to get early positive lobe grid clipping. All things being equal grid clipping tends to be harder than cutoff clipping. Biasing the two 1987 channels differently offers up different clean and overdrive tones on each channel.
Clipping will generate odd and even harmonics but asymmetric clipping adds mostly more pleasant sounding even order harmonics. Asymmetric distortion is often described as sounding "creamier" with less fizz than symmetric distortion.
It is common in tube guitar amplifiers that one gain stage will clip one signal lobe, then in a following gain stage the undistorted lobe will get clipped making the distortion more symmetric.
An AC guitar signal voltage causes electrons to alternate their direction through a circuit. During the positive half of the signal swing the electrons will move one direction and during the negative half they reverse course and move the other direction. With a symmetric AC signal the electrons will move the same distance back and forth through the circuit so there is no real movement of electrons on average, in other words there is no direct current (DC) component to the signal.
But an asymmetric AC signal voltage, where one lobe is larger than the other, will cause the movement of electrons more in one direction than the other which is a DC current. As an extreme example, if you completely clip off the positive lobe of an AC signal it becomes a pulsing DC signal--electrons move in a pulse in one direction. So asymmetric signal voltages have a DC component that will charge capacitors and cause voltage drops over resistors which can cause grid bias voltage shift.
The term "duty cycle" can be used to describe asymmetric signal voltage. A symmetric signal has a balanced 50% positive and 50% negative voltage so we say it has a duty cycle of 50%. An asymmetric signal will have an imbalance of positive and negative signal voltage so its duty cycle has moved away from 50% so we can say the duty cycle has been "modulated."
Screen Current's Effect On Pentode and Beam Tetrode Power Tube Overdrive
Screen grid current, much like control grid current, causes nonlinear distortion. Overdriving a tube that has a screen grid results in screen grid conduction and screen voltage drop which leads to lower tube gain. In pentodes and beam tetrodes screen current has more affect on overdrive tone than control grid current because screen voltage drop not only shifts the tube's operating point along the transfer curve but it actually changes the shape and slope of the tube's transfer curve. This dynamic transfer curve morphing adds tons of nonlinear, harmonic and intermodulation distortion. This is the main reason most people prefer the sound of power tube distortion over preamp (triode) distortion.
As a tube's plate voltage swings low with the guitar signal its electron pulling power drops. A pentode or tetrode's screen grid is located between the control grid and plate. The screen grid is held at a high constant positive voltage to help pull electrons from the cathode even when the plate voltage swings low. The extra pull of the screen grid greatly increases the tube's gain.
When a pentode or beam tetrode is overdriven the plate voltage goes low enough that the positive high voltage screen grid attracts negatively charged electrons. The lower the plate voltage the more attractive the screen becomes. As electrons accumulate on the screen grid its voltage drops (excess electrons = negative voltage) which decreases its ability to boost gain. This overdrive gain shift distortion is the primary difference between triode and pentode overdrive tone (triodes have no screen grid).
Beam tetrodes have their control and screen grids wound so the grid and screen winding are in precise alignment. This alignment shields the screen grid from the electron beam which reduces screen current by over half compared to a pentode. With less overdrive screen current the beam tetrode's screen voltage remains more consistent and contributes less distortion to its overdrive tone. True pentodes do not have their grid and screen aligned so the screen picks up many more electrons which causes more screen current and a larger screen grid voltage drop. True pentodes such as the EL34 flow more screen current and therefore offer up more gain shifting distortion than beam tetrode power tubes.
Pentode Power Tube
Actual Pentode Grid Structure
Innermost control grid is wrapped around the cathode, next is the screen grid and outermost is suppressor grid. Note how the control grid and screen grid do not line up with one another. This exposes the screen grid to many more electron impacts and the resulting increase in screen current.
Beam Tetrode Power Tube
A beam tetrode's control grid and screen grid are precisely aligned to shield the screen from electrons to reduce electron impact and screen current.
Since true pentodes flow much more screen current than beam tetrodes an amp designed for pentodes like the EL34 needs to limit screen current to prevent excessive current from melting the screen which usually shorts the tube and blows the screen resistor. A beam tetrode amp can get away with a small low watt screen resistor (some Fender amps have no screen resistor at all) but a true pentode amp needs protection. A typical pentode screen resistor is 1K and rated for 3 to 5 watts. I believe all modern tube amps should have a screen resistor of 470 to 1.5k and 5 watts to handle extreme, long term overdrive. Since a screen resistor will restrict the flow of electrons off the screen and cause a buildup of electrons, a screen resistor will increase screen voltage drop and gain shift distortion. Increasing the resistance value of the screen resistor will increase the overdrive screen voltage drop and therefore increase the drop in gain caused by the lower voltage screen. Adding screen resistance also slows the screen voltage recovery time which tends to "soften" the overdrive tone.
The addition of a screen-to-cathode screen capacitor can reduce the chance of oscillation but more importantly add a time-variance factor to screen voltage drop gain shift distortion. The capacitor is charged by screen current and its charge/discharge cycle can alter the character of pentode and beam tetrode distortion with added complexity. The addition of a screen cap and trying different cap values may be worth experimentation.
Grid Stopper & Screen Resistors Mounted to the Power Tube Sockets
470 ohm 5 watt screen resistors (green) added to my Fender 5E3 Deluxe power tube sockets to control overdrive screen current and screen dissipation. The 1.5K 1/2 watt blue resistors are grid stopper resistors which prevent blocking distortion by slowing the upstream coupling cap's charge time. Screen and grid stopper resistors also help prevent oscillation by forming an RC low pass filter with the tube's Miller capacitance. The low pass filter eliminates high frequencies beyond human hearing which helps prevent oscillation.
Output Transformer and Voice Coil Flyback Voltage Spikes
When an amp is heavily overdriven severe clipping can turn the guitar signal into a square wave with very abrupt signal voltage rises and falls. The output transformer and speaker voice coils are inductors whose magnetic fields represent stored energy. The stored energy is used to fight abrupt signal voltage changes. The output transformer and voice coils are what cause the points at the corners of the square wave (see scope capture below). When abrupt voltage changes occur the magnetic fields created by the output transformer and voice coil collapse and generate flyback voltage spikes. These spikes are seen as the "corner points" and "shoulder spikes" in the image below. The spikes also add harmonic and intermodulation distortion.
The shoulder spikes are generated when the signal goes from zero to max and max back to zero voltage. The output transformer and voice coil fight these on and off voltage changes with an opposing voltage spike. These corner and shoulder spikes are part of the aggressive, raspy overdrive tone of the Marshall 18 Watt. Its lack of a global negative feedback loop makes the flyback spikes more prominent than in amps with a negative feedback loop. A negative feedback loop would try to "correct" these spike distortions and smooth them out.
Marshall 18 Watt Heavy Overdrive Output
Square wave with flyback voltage "corner points" and "shoulder spikes" generated by the output transformer and speaker voice coil.
A pentode or beam tetrode circuit with a grid stopper, screen resistor, screen cap and bypassed cathode resistor has a very complex distortion interaction function that can lead to tube overdrive nirvana. All of these components interact with one another during overdrive so tuning all of them to find an overdrive tone and playability sweet spot can be difficult and time consuming.
A pentode interactive distortion circuit is very difficult to simulate using solid state components but I say why bother with a simulation? A tube guitar amplifier's soft, asymmetric clipping, time-varying grid current bias shift and time-varying screen grid gain shift all combine to make tube overdrive tone more musical than typical solid state distortion.
From the AES journal, May 1973 Tubes versus Transistors: Is There An Audible Difference?
Vacuum-tube amplifiers differ from transistor and operational amplifiers because they can be operated in the overload region without adding objectionable distortion. The combination of the slow rising edge and the open harmonic structure of the overload characteristics form an almost ideal sound- recording compressor. Within the 15-20 dB "safe" overload range, the electrical output of the tube amplifier increases by only 2-4 dB, acting like a limiter. However, since the edge is increasing within this range, the subjective loudness remains uncompressed to the ear. This effect causes tube-amplified signals to have a high apparent level, which is not indicated on a volume indicator (VU meter). Tubes sound louder and have a better signal-to-noise ratio because of this extra subjective headroom that transistor amplifiers do not have. Tubes get punch from their naturally brassy overload characteristics. Since the loud signals can be recorded at higher levels, the softer signals are also louder, so they are not lost in tape hiss and they effectively give the tube sound greater clarity. The feeling of more bass response is directly related to the strong second and third harmonic components, which reinforce the "natural" bass with "synthetic" bass . In the context of a limited dynamic range system like the phonograph, recordings made with vacuum-tube pre-amplifiers will have more apparent level and a greater signal to system noise ratio than recordings made with transistors or operational amplifiers (op amps).
So yes, tube guitar amps do sound better than solid state.
For further reading check out Guitar Amplifier Overdrive A Visual Tour by Neumann & Irving. It's fairly technical but it's the only book written specifically about guitar amplifier overdrive.
The book includes many graphs to help make the material easier to understand.
A Brief History of High Gain Amps
By Y. Blais
The reason it is the way it is goes back to the "origins of high-gain." We have to go back all the way to the Fender tweed Bassman (5F6A). It combined a bunch of textbook circuits: a pair of input channels (because a 12AY7 has 2 triodes and because multiple inputs were popular at the time), one a bit brighter than the other one (just a treble cap on that volume knob), followed by a second tube that has a DC coupled cathode follower configuration (also used by Vox in the Top Boost circuit) which was used to drive a very lossey tone stack, then a fancy long tail phase inverter with negative feedback driving the phase inverter's second input (because they were trying to make a loud clean amp), into a pair of 5881 power tubes (loud and clean, per 1950's standard).
Then Marshall copied this circuit in the JTM45, putting a 12AX7 in the first slot and connecting it to a closed back 4X12 cab loaded with Celestions. Then the Bright channel was made even bolder in the "Superlead" configuration (JMP years), with its own cathode resistor (2k7) bypassed by a small 0.68uF cap to boost high-mids, coupled through a smaller 2.2nF cap (instead of 22nF). They also put a treble cap on that channel's mixing resistor. That was great, but the amp had to be cranked to ridiculous levels to sound really dirty. They would put treble boosters and the like in front, yet it still wasn't enough.
But modders found that running the Lead channel into the Normal one was the trick. Because of the high-mids boosting and low-cutting of that channel, running it into the Normal channel gives a very mean distortion tone, not unlike boosting with a Treble booster.
Marshall soon caught on and released the 2203/2204 Master Volume (what we commonly refer to as "The JCM800", but it was initially a JMP) which used the cascading trick but through their experimentation decided to make the second triode a "cold clipper": replacing the bypassed 820 ohm cathode resistor with a big unbypassed 10k. If you've ever experimented with a 2203/2204 circuit, you know that keeping the second stage "warm" gives you more overdrive, albeit softer and muddier sounding (good for leads perhaps, but not crunchy enough for heavy riffage).
After that, Marshall experimented with diode clipping and such to get more distortion, but not everyone liked it.
The 2203/2204 circuit works really well with a booster pedal in front (Tubescreamer or Super Overdrive for example), so the next logical step was: let's add a gain stage before it, make the booster built-in using a tube. So what do you get? A first stage that is usually configured for very high gain (typically a 220k plate load) focused mostly on the mids and highs (1uF or less bypass cap, 2.2-4.7nF coupling cap, a series resistors bypassed with a cap before the gain control as a "treble peaker", etc), a second stage that is very similar to the 2203/2204's first stage, and a cold-biased third stage that doesn't add much gain but distorts in a more pleasing way than diodes (Soldano used 39k for the cathode resistor instead of 10k because with that extra gain stage in front, 10k was a bit "too much").
Interestingly, Soldano also added a DC coupled cathode follower to drive an FX loop before the tone stack, which means you get TWO cathode followers, and DC coupled cathode followers have a pleasing compression effect that can smooth things out. That's a big part of the Soldano sound that neither Mesa Boogie (with the Dual Rectifier) or Peavey (with the 5150) replicated.
Others got more gain from the 2203/2204 circuit by configuring the fourth stage (the DC coupled Cathode Follower) into a regular anode driven gain stage (this results in more loss in the tone stack and a different response, but it does give more distortion).
And then you get the Peavey 5150 going all out and combining both the extra "booster" gain stage with the plate driven tone stack for massive amounts of distortion.
So as you can see, all these circuits are just simple ways of getting more gain from a "JCM800" circuit (which itself was just a Plexi, aka a modified tweed Bassman, running the Lead channel into the Normal and adding a Master Volume). There are some nuances though: for example the Dual Rectifier mostly took the Soldano SLO circuit with that 39k third stage clipper, but moved the FX loop to after the tone stack (using an AC coupled cathode follower which doesn't have as much color) AND disconnected the negative feedback circuit while replacing the Presence circuit with a simple treble attenuation control (this explains its massive bass-heavy sound). Other amps replaced the classic tone stack with a Baxandall stack (less interactive, allows treble boost AND cut).
If you want to see a truly different circuit, check out Mesa's Mark series. They look very complicated because of all the switching options, but if you go back to the Mark IIC+ and ignore all the complicated stuff, you mostly get this: a blackface Fender (say a 60's Twin) with the reverb circuit replaced by a gain boost circuit. Look at it, you'll see it: first stage drives a tone stack via the plate, then is followed by a recovery stage (textbook Fender blackface), then you get the 3M3 resistor which is typically used to attenuate the signal to mix it with the reverb, but instead of a reverb circuit you get the "Lead" circuit which is a pair of triodes (aka one 12AX7) configured for high gain. What you get is basically trying to replicate a cranked to hell Fender: lots of distortion AFTER the preamp instead of before it. Also the interesting thing is you get a bit of clean sound mixed in with the extra distortion (what makes it through the 3M3 resistor). The resulting sound however is very mid-heavy, hence the graphic EQ, which is a much more powerful tone shaper than the classic BMT tone stack (especially this one, which runs on much higher voltage than a pedal). This gives you a lot of power, the early tone stack can be used to shape the distortion (crank Treble, cut Bass for a very tight aggressive sound, go more moderate for that "fusion" lead sound), then the graphic EQ lets you do the final shaping. Boosting the lows late will not muddy up the distortion, that's how it can sound so massive while tight (the "Simulclass" power amp where 2 of the tubes run in Triode mode help here too).
I've drawn up schematics and layouts for a dozen high gain amp ideas over the years, but it seems I always go back to the 2203/2204 circuit. I find it more flexible to use boost pedals in front of that circuit because I get more variety depending on which pedal I use, and the classic JCM800 crunch is always there waiting for me, and I don't "need" a clean channel because it cleans up well with the guitar's volume. Also, this saves me adding a fourth preamp tube, and makes it easier to keep hiss and hum down (most 5-stage amps will use DC filaments on at least the 1st tube).
But if I was to build a high gain amp, I'd do it this way: V1a would be a bypassable "booster" stage (like the first stage on a Soldano), V1b would be the usual JCM800 first stage (100k, 2k7/0.68uF, 2.2nF coupling, 470k/470pF treble peaker, 1nF treble cap on gain) and V2a would be the JCM800 cold clipper (10k). V2b would be the warm biased stage (100k, 820R unbypassed) usually driving the Cathode Follower but instead it would drive a MOSFET follower (saves me a 4th preamp tube just for that one gain stage), from there it would be textbook JCM800 (but with a 56k slope resistor in the tone stack and a 330pF treble cap and 47nF bass cap). Marshall themselves used a similar trick in the JCM1 (V2a drives a MOSFET follower driving the tone stack, V2b is a Cathodyne Inverter, that's how they get a JCM800 preamp with just 2 tubes)
I guess what I'm saying is that this typical high gain amp configuration was not developed through thoroughly directed research, but mostly through years of trying to get more and more gain from the classic "Plexi" (but really tweed Bassman) circuit.
By Y. Blais
This webpage is available in .PDF form: Tube Guitar Amp Overdrive.pdf
By Rob Robinette
[ How the 5E3 Deluxe Works ] [ Deluxe Models ] [ DRRI & 68 CDR Mods ] [ Amp Troubleshooting ] [ Tube Amp Forum ] [ My 5E3 Build ] [ Spice Analysis ] [ The Trainwreck Pages ] [ Fender Input Jacks ] [ B9A Prototype Boards ]