Throughout this thread I intend to explain the ins and outs of as many tube preamps as I can. This is also going to a place to ask any questions about preamps, including tone stacks and phase inverters. I'm hoping some tube veterans will chime in as they see fit.

I'll post some preamp designs, and give explanations on how they work, and the tonal responses of the preamp. How to stabilize a preamp that just seems to wanna hiss or buzz. Let's look at some designs. This is also a place to submit your own preamp ideas, say why you like it, and ask questions on them.

Note: This is not a complete thread yet, I'm adding as I go.

1st post: Some basic knowledge of building a tube circuit. Links!
2nd post: Clean preamps.
3rd post: A little bit of distortion
4th post: Metal distortion
5th post: Buffers, Where to use them? What to use?
6th post: Pentodes, Solid state gain stages
7th post: Feedback, what exactly does it do?
8th post: Stabilizing the preamp. How to tune the circuit.
9th post: Tone stacks
10th post: Phase inverters
33rd post: The Mu Amp and SRPP: What in the heck are they anyway?

The Amp building resource thread
GB&C Starter Pack
How vacuum tubes work
Fun with tubes (lots of good stuff)
Tube Amp FAQ
Types of distortion
Tube VS SS Distortion
Elliott Sound Products
Valve Wizard
Pentode Press: Gain calculator (they have other calculators as well)
MosFET Follies article
MosFET Boosts
JFET Boosts
Frank's Electron Tube Data Sheets
Duncan Amps - TDSL Tube Search
Lead Dress in Tube Amps
Tube Amp Debugging
Duncan Amps - Technical Pages
HV Regulator (Good 300v supply for preamps and smaller amps)

Capacitors: a guide to types and habitats
Capacitor types and their uses
Capacitor characteristics
Secret Life of Pots
Bias Supply networks
Potentiometers, the beginner's guide to pots
AWG Wire Table
Tailoring Pots (pdf)

Safety Tips
Last edited by blandguitar at Aug 4, 2011,
Clean preamps

Okay, let's look through this, it's got a supply voltage of 275, not too common in an amp, but not uncommon either. This design is simple, use capacitors to block DC, Direct current, from the following gain stage, this eliminates any voltage offset.

How to design a gain stage: We'll need a datasheet for the tube we want to work with. I typically use Tung-Sol datasheets, but all the brands will have similar datasheets. http://tubedata.tigahost.com/tubedata/sheets/127/1/12AU7.pdf

First, we pick our supply voltage, then we pick an anode load resistor. Using ohms law, E=IR, we can do a lot with tubes. I chose 275v since it's a nice large voltage, which is great for headroom. Pay attention to the maximum voltage on the datasheet, it's 300v for the 12AU7 tube, so we're good. Next, I chose a 10k anode load resistor, that means that for every ampere of current through the resistor, the voltage will drop by 10 thousand, to put it into perspective though, for every milliampere, mA, or 1/1000 ampere, the plate will have 10 volts dropped across it.

To find the maximum amount of current through the resistor, we use ohms law, E=IR, 275=I*10,000 I= .027A or 27mA. So we look at the data sheet, find the graph that plots plate voltage versus plate current, with the control grid's voltage graphed as well. It's on the 4th page of that datasheet I linked. Then, draw a line from the 275v plate voltage to 27mA plate current. This is our anode load line.

Next, we pick a voltage for our control grid. I chose -6v with respect to cathode, this should give decent headroom, without distorting the signal much. If you look at the curves of the grid voltage, you'll notice how they aren't symmetrical, and vary slightly with the grid voltage. This means that no matter what, we will have some distortion, but fortunately, it's largely 2nd harmonic distortion, which we like in small amounts, even for clean.

But, how do we make the grid naturally rest at -6v? If you look at the graph, you'll notice that there is the plate current scale, we do just a little bit of guesswork to find where we think the -6v curve would be. Then we find where the -6v curve intersects with our anode load line. Then look directly to our left, and see that it's roughly 8mA on the plate.

That brings us to the cathode resistor. This is what we'll use to "lift" the cathode by 6 volts, afterall, we don't need the grid to be negative with respect to ground (0 volts DC), we just need it to be more negative than the cathode. If we have 8mA running through the tube if its grid is at -6v with respect to the cathode, we can easily find out the value of the resistor needed. E=IR, 6=.008*Rk therefore Rk=6/.008=750 ohms. The closest standard value you can find will suffice, this isn't hifi afterall. Another trick you could use is set-up a 1k trimmer pot, and play with it until you find the tone you like.

I just duplicated the resistors for the second triode, it will give the same results.

But wait, what's the capacitor for? The Capacitor from the cathode to ground is a cathode bypass capacitor, it literally bypasses the resistor for AC since capacitors only allow a change in voltage (AC like a guitar signal) to flow. This allows us to keep the DC bias of -6v while effectively boosting the signal to allow more gain, this also slightly distorts the signal, but for one stage, you wouldn't notice anything other than the increased gain. If you use a smaller capacitor, you won't get a full A.F. spectrum boost, you can create a treble boost by limiting the capacitor to 1nF-100nF with small values boosting less bass, and can create midrange boosts (midrange+treble) as well as treble.

If you look at the datasheet, on the first page, there is maximum circuit resistance, this allows us to set the input impedance to almost any value we would like, since tubes have a larger output impedance than solid state, it helps to have a large input impedance. Typically a 1M is used, but lower can be used as well, this resistor also has a purpose of why it must be there, as electrons flow, the grid can get hot and emit electrons, the 1M resistor allows these electrons to flow to ground. It also keeps the grid at 0v bias, unless using a negative grid bias power supply. (We'll get into negative grids without a cathode bias later, this is called fixed bias.)

Whenever driving something, the source should have an output impedance well under the input impedance of what you are driving. Tubes are high-ish impedance, the output impedance of an average 12AX7 stage is about 40k ohms. The output of a 12AU7 is lower, but still not nearly as low as solid state. When driving a low impedance with a high source impedance, you can lose a good deal of signal. This is why we use a higher input impedance than source.

Next, we have the resistor that is attached to the grid, 10k to upwards of 1M is used for this. We can tailor the frequency response with this resistor, the input capacitance is relatively easy to figure out.
Where Cin is the input capacitance, Cga is the grid to anode capacitance, A is the gain of the stage, and Cgk is the grid to cathode capacitance.
The grid to anode capacitance is amplified by the gain of the stage since the anode voltage is changing much more rapidly than the cathode.

The gain of the first stage is 12, so Cin=1.5*12*10^-12+ 1.6*10^-12
=19.6pF, which we can round to 20.0pF.
Next we pick a frequency to attenuate at
We can also turn it around and find the frequency from resistance.
f=1/(2*pi*R*Cin) Pi~3.14 and R is the grid stopper's resistance value plus the output resistance of the source.
f=1/(6.28*20*10^-12*[22k+Zout]) The resistance of a passive guitar pickup can vary a lot, put to simplify things, let's call it 100k.
With this, even a large impedance signal like a piezo pickup could have no noticeable treble loss, a piezo pickup can have an output impedance of several Megohms.
f=3.6khz, which is which will attenuate the upper harmonics of the pickup.

The next reason we have for that grid stopper, the resistor in series with the signal, attached to the grid, is to prevent blocking distortion. Blocking distortion happens when the changing voltage is too much for the capacitor to discharge properly. Unless the following stage is being overdriven, this isn't of much concern; however, most modern amps distort more than older designs did, and you should always include grid stoppers on all of your stages. Small values may not prevent blocking distortion, and large values will attenuate treble just as the last equation proved. So a value of 10k-100k will suffice. Notice that even with a volume potentiometer, we still use a grid stopper, the proceeding grid must always have series resistance to prevent blocking distortion.
Common K Equations pdf.pdf
Clean Preamp Analysis pdf.pdf
Last edited by blandguitar at Jun 16, 2011,
Distorted tones

Fortunately, all the same rules apply as with a clean preamp. The difference is in the biasing. (More on this in tuning the preamp)

I'm going to do two designs for this, a trebly classic rock tone, and a bluesy mid boosted tone.

Here's the blues preamp:

If you notice, there is a voltage drop between the two tubes. This adds filtering to the power supply, PS. By using a dropping resistor, the ripples in power supply are made smaller, combined with a filtering capacitor, you can increase and tighten bass response as well as stabilize the PS. The easiest way to calculate the resistor needed is to add the current of the stages after the resistor, then use E=IR to solve for the resistor. There is a 35v drop between the two tubes. Then using E=IR you can calculate the current through each resistor.
For the first stage.
1=I*3300 I=.0004mA
The diode stages are a little more tricky, and can't be reverse engineered with E=IR, instead, I used 2 .6v silicon diodes to create the necessary voltage drop. Using a diode for a voltage drop is a reliable way to get a fixed cathode voltage. It pretty much has 0 resistance to AC, so the capacitor isn't there to boost the gain. The capacitor is there to minimize any switching noises the diodes would create.

From the datasheet of a 12AX7 you can detmermine the current through the diodes by drawing a load line from 125v to 1.25mA, and finding the resting current at the point 1.2v, this is roughly .3mA or .0003A
So together, the stages use .7mA of current. The two gain stages are out of phase with one another since using a tube to amplify signal inverts the signal, this keeps the current through the dropping resistor relatively constant. So 35=.0007*R R=35/.0007=50k, 47k or 51k will work just fine. Using a 22uF capacitor will give a hearty bass response as well as ripple filtering. The beginning stages are the lowest voltage as they are the stages with the small signal, therefore needed less headroom, as well as being more susceptible to power supply ripple. The more you filter the power supply, the less ripple there will be, but the voltage will be lower the more it's filtered as well.

The 12AU7's are slightly warm biased, this should help create the bluesy tone. The large-ish grid stoppers should reduce treble response slightly, combined with the partially bypassed cathode resistors should create a mid-boosted tone. This will not only help cut through the mix, but with a light distortion, I personally find preferable. To decrease the treble response further, you can reduce the grid-leak resistor to 470k, this will shift the response down an octave.

The capacitors on the potentiometers are there to retain the treble when the pot is turned down, the inter capacitances of the pot as well as the first order filter of the DC blocking cap and potentiometer will drain treble, the pF value caps can help retain a tonal balance throughout the entire sweep of the pot.

This is pretty simple, combine a good bit of gain, with a cold bias. A cold bias is a bias that is towards the 0v grid point on the graph. A warm bias sits more naturally towards cut-off as you get a more negative point on the grid. If you look at the datasheet, you'll notice that the more negative the voltages get, the more bunched up they become, the less current will flow. This creates a softer distortion, and largely 2nd order harmonics. Grid current limiting occurs when a grid becomes positive with respect to cathode, and sharply clips giving off strong 2nd and 3rd harmonics.

Note the diodes in this design as well, they provide an easy bias for the triode. You can use different types of diodes, LEDs will typically have a larger forward voltage drop and get easily achieve a 1-4v bias. The cap across is a 100nF again, the capacitors on the cathode to ground only need to be rated roughly 1.5 times the bias voltage. Although, it's easy to get 12-25v capacitors, any in this voltage range will suffice, and be cheap. You can reduce the capacitor to 47nF on the diodes as well. 100nF is convenient and for the voltage range will most likely be the same cost as a lower value capacitor.

The colder bias in this design will give a more rough, classic rock type of overdrive compared to the warmly biased blues preamp above. The coupling capacitors will reduce the amount of bass in the preamp. Notice that, even though we are reducing the bass in the preamp, that the first stage is fully bypassed, you always want a full audio frequency spectrum boost in the first stage of your preamp, this makes the signal less susceptible to interference. To lessen the harshness of the distortion you could add another diode in series with either of the stages, the 2nd having more influence since the signal is boosted again before it. Or, use a capacitor from anode to power supply+ which would be 175 for the first two stages, or 200 for the 2nd two stages. One more method to calm the harmonics is to hook up a capacitor from anode to cathode, this causes negative feedback reducing the signal size of the harmonics, with more feedback with increasing frequency. The final capacitor is made large to allow all of the signals to pass with relative ease, so that it could be used as a pedal, if using this as an amp's preamp, the final capacitor can be made smaller.

One last thing that is introduced more heavily in this design than the others is the different anode load line resistors. These resistors allow you to not only tailor the gain of the stage, but also the way the anode voltage is modulated from the grid voltage. I'll go into detail in the tuning of the preamp section.
Distorted Preamp Analysis.pdf
Last edited by blandguitar at Jan 8, 2011,
Heavy Distortion

This is what made me wanna start this thread, I had a new idea as far as I knew, but no way to share it.

(The attached file has a slightly updated schematic and as a well a pdf for portability/printability.)

Quote by RG Keen
A grid driven more positive than the cathode changes suddenly from an almost infinite impedance to a few K. In the case of a 12AX7, it's very roughly 5K. In the case of a power tube, maybe more or less. If the driving impedance is less than 1/10 (roughly) of the positive biased grid impedance, it hardly notices the change. The output impedance of a MOSFET follower is well under 1 ohm in most cases. A cathode follower is perhaps 100 ohms to 1K. The MOSFET is a bigger horse.

This got me thinking, what about a distortion, very coldly biased, and using a source follower to drive the proceeding grid positive, this is the schematic that followed. I haven't built this yet, so I have no idea how it sounds. I'll experiment once I buy some more MosFETs.

The hard clipping caused by grid current limiting is explained above, the rapid impedance change from 1Meg ohms to a few thousand ohms loads down the circuit before it, and even a normal cathode follower doesn't truly have the power to drive the grid positive, the extremely low output impedance of a source follower can not only drive the grid positive, but can do it well. I took the HT voltage and multiplied by 2/3 to find the rough voltage for the first tube. The divided that voltage by .002, this allows 2mA to flow through the MosFET so it doesn't overheat. Then I took the load line from the B+, the voltage source, and divided by 100k. Then drew the load line the same way as the clean stage above. I then looked at the voltage directly below where the control grid=0v intersects the load line. This gives me the rough voltage the rest of the anodes, plates, will rest at, I then divided that voltage by .002 to find the resistance for the other source followers. I gave adjustments that can be made to accommodate different voltages.

The capacitors from anode to cathode help tame the harmonics of the distortion and can help smooth out any harsh tones. The low output impedance from the source follower means that we have to use larger capacitors to get the same frequency response. These might get brought down to 100nF or even 47nF, that will need to be experimented with. The low source resistance may or may not prevent blocking distortion, So I added a 2k2 interstage resistor, in addition, voltage followers have a tendency to oscillate, and using a resistor can help prevent it.
MetalMadness-TubeyGoodness pdf.pdf
Metal Preamp Analysis.pdf
Last edited by blandguitar at Jan 8, 2011,
Using and understanding buffers.

A buffer is a unity gain, or near unity gain stage, they are used to isolate the previous stage from the following stage.

A buffer combines a large input impedance with a low output impedance. A buffer is also called a voltage-follower, the output voltages is the same, or "follows" the input voltage. All non-op amp followers have a gain in the .9# range, or within 10% of the initial voltage. Designing, you can change the gain, input impedance, and output impedance to fit your needs.

A buffer also offers another advantage, a large current boost, which combined with a low output impedance makes a buffer ideal for driving a loaded down stage, such as a tone stack, power tubes, or any low impedance load.

Designing a standard buffer

Every discrete (non-IC) active device has to follow the laws of mother nature. For most of our purposes, we use 3 terminal devices. There is the control terminal; which is the gate in FETs, the base in bjt (bipolar junction transistor), and the grid in vacuum tubes. There is a low impedance current path; the source in FETs, the emitter in bjt's, and the cathode in tubes. And lastly, there is a high impedance current path; in FETs, it is the drain, in bjt's, the collector, and in tubes, the plate/anode. Both plate and anode are commonly used, and it should be recognized that they are the same thing.

To put it simply, the low impedance current path "follows" the control voltage. This is slight compression, therefore the gain at the low impedance current path is less than 1, however, there are things that can be done to improve the performance. Higher load resistances improve the linearity. If using a resistor to load the low impedance path, higher resistances allow less average current to flow, and lower resistances allow more average current to flow.

Too low of currents reduce the effectiveness of a buffer, and therefore the resistance must be calculated based on average current. A constant current source to load a buffer offers 2 advantages, the constant current allows it to drive heavy loads, and the high AC (signal) resistance makes it extremely linear, which is good for hi-fi, but not necessary for guitar amps.

Know the Voltage Offset of the Devices

Most buffers consist of one of two topologies. Either using the low impedance terminal of a device as an output with a resistor to a power supply rail and the high impedance terminal connected to a power supply rail of opposite polarity, OR using localized negative feedback or manipulating the resistors in the supply rails-device path. This all sounds very confusing, so a couple examples will clear it up.

These designs do not use negative feedback from the high impedance terminal to the control terminal to stabilize the buffer. To understand how these buffers work, some digression is necessary.

Let's look at the triode tube, when the grid voltage is positive with respect to the cathode, a massive amount of current flows, so to get anything usable out of the triode, the grid needs to be negative with respect to the cathode. However, why create a negative supply voltage for all of the grids when a resistor can be stuck between the cathode and ground. The current flows through the tube as well as the cathode resistor, the tube will "self-bias" by the current flow through the tube creating the voltage offset. This same action works for all depletion mode devices, including JFETs and some MosFETs. This self-bias voltage is the voltage offset necessary for proper operation of a gain stage or buffer. The low impedance terminal will be the control terminal voltage PLUS the voltage offset. This can be 0-4v roughly. (This hold true for NChan FETs and tubes, for PChan FETs, the source will be attached to the positive power supply rail through a resistance and the source will be more negative than the gate.)

Enhancement mode devices require the control terminal to be at a more positive voltage (for NPN bjts and N channel Enhancement Mode MosFETs, negative for PNP and P Channel), and the low impedance terminal will be the control terminal MINUS the voltage offset, in reality it's addition still, the offset is merely positive or negative. If the voltage of an enhancement mode device from control to low impedance terminal
is less than the required voltage hardly any current flows, certainly too little to be of much use to us, so bias, or a static set of conditions are assumed, this is done by means of a bias voltage or Vb Vref Vr Va or pretty much any V_ combination imaginable.

Down to numbers. In small signal tubes, the bias voltage is typically 1-4v in hopes that the grid never goes positive. In depletion mode FETs, the gate can even exceed the source, however, a source resistor is typically used to limit the amount of current flowing through the device, which is another reason why this biasing arrangement can be considered self-biasing. Enhancement devices. In bipolar junction transistors, the offset is from the base-emitter diode voltage drop, depending on the size of the device and whether the device is Germanium or Silicon, and is .3v to ~2v+, power transistors may have a larger voltage drop than typical. MosFETs that are enhancement mode typically have a 2-5 volt gate-source voltage drop. In enhancement mode MosFETs, the offset voltage is called the Threshold Voltage.

When designing with larger voltages (25v+) the voltage offset is less critical, however, when designing with 9v-12v systems, the voltage offsets must be accounted and designed for. Most devices are close enough to their "typical" offset voltage that the circuits can be designed with some wiggle room, or an adjustable bias voltage to compensate. There are also multiple ways to test devices for characteristics.

JFET Matcher

For enhancement mode devices, one can set the control terminal voltage to 1/2 the supply voltage, short the high impedance terminal to the power supply rail, and attach the low impedance terminal to ground through a resistor. Then measuring the gate-source voltage or base-emitter voltage directly is a simple task.

Geofex: MosFET Follies
AMZ Basic Buffers

One Last Note On Buffers:
With a voltage follower (buffer) the output should idle at the middle of the power supply voltage to give the maximum signal swing possible. This is why understanding the voltage offset of a device is important, and especially important in low voltage designs. The control terminal should idle at 1/2 supply voltage plus any offset be it a positive or negative offset.
Last edited by blandguitar at Jun 17, 2011,
Using small signal pentodes and solid state gain stages in the preamp.

For every "amplifying device" except bjt's, the device typically requires almost no current from the source. Infact, the change in current flow due to a voltage change on the control terminal is given in siemens, the base unit for transconductance. The conductance of a device is the reciprocal of it's resistance, or it's ability to allow current to flow. Transconductance is the "transfer" of conductance, the control voltage causes a change in current in the device. So for a device with a gm (transconductance) of 1mho or 1S (mho is the backwards spelling of ohm, the unit of resistance, and will be seen on older datasheets such as tubes) For a linear device, the gm is constant, however, no semiconductor nor tube is linear, therefore the transconductance changes with dynamic conditions.

For tubes, the gm can be anywhere from 1-10mS (millisiemens or .001Siemens)
Some common preamp tube gm's (12A_7s)
12AT7: 4-5.5mS mu: 60
12AU7: 2-3mS mu: 20
12AX7: 1-1.6mS mu: 100
12AY7: 1.75mS mu: 44

A pentodes gain and gm is highly variable based on the relationship of the screen voltage to the anode voltage.

Pentodes have a much larger gain than triodes and typically need resistors in the cathode circuit to help counterbalance the massive gain seen at the anode. A completely bypassed cathode resistor of a pentode gives a gain of approximately A=gm*Ra where gm is the transconductance of the pentode and Ra is the anode load resistor. After a glance at a pentode datasheet, one notices two types of Ia-Va graphs, one for pentode characteristics and one for triode characteristics. The pentode characteristics rise sharply after Va is increased from zero, but then flattens out, this is good because for almost any plate voltage, the characteristics of the pentode remain relatively unchanged.

As the screen voltage is increased, the headroom of the pentode is increased, and the input sensitivity decreased, the output is less sensitive to grid voltage. This allows for a larger clean input signal for a given output voltage swing. As the screen voltage approaches and surpasses the anode voltage, the screen draws more current. Often times, the screen grid is supplied from a voltage divider or series resistance from the anode supply, by using a capacitor from screen to ground, the screen voltage can be relatively stabilized rather than varying the voltage with the screen current. Allowing the screen voltage to "sag" under load will give more compression and a loss of some of the tightness of the sound. Some compression may be good however, and combining filtering and series resistance can lead to interesting results.

These are taken from the Phillips EF86 small signal pentode datasheets.

The first picture is an example of the anode characteristics of the EF86 pentode with the screen grid set at 140 volts. To understand the importance of the screen grids voltage, some explanation of the pentode is necessary. A triode consists of an anode, a control grid, and a cathode. The electrons travel from the cathode to the pentode, however, the negative charge on the grid slows down the flow of electrons so that the voltage of the grid controls the flow (current) of the valve/tube.

A pentode has two additional grids, the first, suppressor grid is usually at cathode potential (voltage) or slightly above cathode potential. The other additional grid is the screen grid, which is typically at or under the plate voltage, but positive with respect to the cathode. Unlike triodes which come from the abscissa of the anode current-voltage graph, pentodes sharply bend upwards and flatten out the the right. Choosing the proper load line and bias current is the same as with triodes, the transfer characteristics are merely different.

The screen grid draws current, however, and this is usually 1/3-1/4 the anode current as long as the screen voltage isn't too large with respect to the anode, for small signal pentodes, this current is almost always negligible except when finding idle current. Some data sheets include grid 2, g2, currents in addition to anode characteristics. Because the grid is typically at a more negative voltage than the cathode, the grid draws almost no current, however the cathode is essentially in series with the screen grid and anode. The screen grid and anode are essentially in parallel, to sum this up, the cathode current is equal to anode current in addition to the screen grid.

The characteristics of the anode current based on screen voltage and grid voltage. Under the assumption that the screen voltage stays constant, one can determine anode current from the grid voltage. For instance, at g2=60v, g1=0v, the anode current is Ia~3mA; at g1= -2.3v Ia~0mA. Just the same as we estimated current based on grid voltages between those given in the datasheet to find an idle grid voltage for triodes, we can estimate an entire set of characteristics for a given screen voltage, if g2=80v, the characteristics will be roughly in between the characteristics for g2=60v and g2=100v.

These characteristics are for a plate supply voltage of 250v, as the plate supply voltage increases, the current does as well for each curve; and as the supply voltage decreases, the current decreases. This is of nearly no concern, but should be made aware of, a good idea would be to measure these characteristics at Va=50v, Va=250v, and Va=VmaxIdle which for an EF86 is 300 volts at idle. Note, an EF86 can idle at a max of 300v, but the instantaneous maximum voltage is 550v. Tube datasheets generally give the maximum idle voltage, whereas solid state components give the maximum voltages/currents period.

Since we can estimate plate current based on screen voltage and grid voltage, if we set the screen voltage to 60v, and vary the grid voltage, we'll find something similar to the above image. Note that the characteristics still curve or "knee" towards Va=0v, however, if I recall properly, as the screen grid voltage is decreased this knee becomes sharper.

The screen grid affects how much clean headroom a stage has, larger screen voltages allow larger input swings, and smaller screen voltages make the pentode overdriven much easier. In fact, by making the screen voltage variable, one could control the gain of the stage simply by adjusting the g2 voltage.

The Determination of Quiescent Voltages and Currents in Pentode Amplifiers

Mu-stage Philosophy Info on Constant Current Sinks (CCS), geared towards pentodes.

Geofex: Designing a MosFET Boost
JFET Amplifier Stage

Using Solid State Devices to Actively Load Gain Stages.
Active Loads and Signal Current Control
The Under appreciated Hybrid Mu-Follower
Last edited by blandguitar at Aug 4, 2011,
Tuning and stabilizing the circuit.

The anode load line:
By manipulating this value, we can rotate the line to get just about any curves we want. We can limit, or open up, the gain. Different load lines will intersect the grid bias curves at separate points. This also allows us to tailor the total harmonic distortion generated, even a clean tube stage will have more distortion than a clean solid state gain stage. The output is modulated, changed, slightly in any case.

An update to the anode load line, interesting things happen to the waveform when put through non-linear amplification, where the curves are changing most are where the most interesting things happen. For a triode, too low or two high limits gain, however, with a pentode the load line changes where the signal intersects the "knee" of the ec=0 curve, for guitar it is advantageous to have the anode load line intersect at or below the knee. Naturally, there is no set rule, just a suggestion, play around with different loadings.

Bias: In short, biasing is setting the tube stage to react how we would like. A more negative grid allows less current through the valve/tube and therefore has a larger anode voltage. (Because a higher grid voltage means a lower anode voltage, and vice versa, the signal is inverted, or flipped upside down.) By changing the cathode resistor, we change the voltage that appears at the cathode, which allows us to pick the way the signal is changed. As you move towards a move negative grid voltage, the anode characteristics are more bunched up, this creates a much softer, smoother distortion. The effect is a largely 2nd order distortion, which means that other than the fundamental frequency, the 2nd harmonic (an octave up) is strongly present in the signal. This bunching of the signal causes a natural compression and sustain.

Bias towards the 0v grid to cathode voltage (with the cathode resistor.) Once the grid reaches 0 volts, the current flows rapidly through the tube, and drops a large resistance very fast. Coming back down, the voltage is a little more gentle. The rising signal gives a hard clipping of mainly 2nd and some 3rd order harmonics in the signal. The 3rd harmonic is 2 octaves up the fundamental, 4th is 3 octaves above. The harmonics produced go into 4th and 5th order as well as higher, while the higher the harmonic, the less noticeable it is. The voltage of grid to cathode as it becomes negative than the cathode again is softer than when it becomes more positive than cathode. A high source resistance will make the clipping more drastic, and a lower source resistance can often help soften the distortion.

Note: No matter which form of clipping you try to use, too extreme of a bias will make the signal clip, and possibly very heavily.

With cathode biasing, the resistor helps to keep the grid at a fixed voltage from the cathode, as the grid is more positive, more current flows through the resistor, so the cathode raises in voltage, forcing the grid to be more negative. And as less current flows through the grid, the voltage at the cathode drops slightly, forcing the grid to be more positive with respect to the cathode.

Cathode bypass capacitors: A capacitor across the cathode resistor can maximize gain for the stage. Basically, the capacitor blocks DC, so it allows you to keep the voltage offset necessary for amplification. But the capacitor allows AC to flow, this helps alleviate the effects I stated in biasing by keeping the cathode-grid voltage relatively constant and reducing the resistance "seen" for higher frequencies.
If you're interested in seeing the equation for the capacitive reactance, please look at the GBC starter pack.

If you look at that graph, you can see that a smaller capacitance allows higher frequencies to flow, and larger capacitances give a full spectrum boost. By choosing the bypass capacitor, we have another way to tailor frequency response of the preamp. Another thing that you can do is to put a pot in series with the capacitor to set up as a variable boost, the resistance inline with the capacitor will make the AC signals "see" a larger resistance, which will look like a smaller capacitor. If you have a convenient value for you cathode resistor, you can connect lug 3 to the cathode, and lug 1 to ground, then attach the wiper to the + end of the capacitor.
Last edited by blandguitar at Jun 18, 2011,
Tone stacks


All first order filters can be broken down into f=1/(2*pi*R*C) That gives us the -3dB roll-off frequency.

So we can break down each of these tone stacks into their relative roll-off frequencies.

(I'm going to type the math up for this soon.)

Jack Orman's Presence Control:
AMZ Presence Control A very usable control, the impedances could be scaled to work with tubes as well, though, a buffer would minimize losses as well as allow the controls to remain unchanged.

Pentode Press links for tone stacks:

Their info is a much better analysis and more complete for designing a tonestack than the information available in this thread, I would highly recommend reading the following links.

Some info on James (Passive Baxandall) tone circuits:
Analysis of the James Tone Stack
The James Tone Stack: Creating your own design

Hiwatt CP 103 Tone Stack:
Analysis of the Hiwatt CP103 Tonestack
Hiwatt CP 103 Tonestack: Creating your own design

Gretsch Chet Atkins Tone Stack:
Circuit design of the Gretsch Chet Atkins tone stack

Fender Bassman/ Marshall Plexi:
Fender Bassman 5F6-A Vs. Marshall Model 1987
Low and Midrange Frequency Response
High-Frequency Response
Creating your own design
Last edited by blandguitar at Aug 30, 2010,
Phase inverters

Fun With Tubes Phase Inverters
Phase Splitters

I attached a document containing MosFET Phase Inverters.

A MosFET is an enhancement mode device, you can buy depletion mode, but they're much harder to find. Enhancement mode means that you have to adjust the bias to get them to conduct/amplify. The zener diode allows a relatively steady bias of the MosFET, the 1M resistor in series is to prevent too much current from flowing through the diode, overheating it. With the LTPs (long tailed pairs,) one is done to bias with respect to ground. The other simplifies things in a sense and uses any negative voltages, these can be accomplished with center tapped transformers. As an example a 250v-0-250v transformer. You can use this to bias the power tubes and the MosFETs if the negative voltage needed for the MosFETs is around -20v or a more negative voltage. I would say that exceeding the voltage necessary to penetrate the glass layer on the diode should be the maximum used. The resistor in the tail end will raise the voltage slightly, so it should be safe as long as you don't penetrate this.

The diode from the source to drain should be equal, actually slightly less, to the maximum rating of the device. Many MosFETs have this included internally, if they don't, add it.

The 20k trims are to adjust any imbalance, reducing distortion in the power amp from mismatched signals. These can be excluded if you'd prefer to not have them. The capacitor is to affect the driving ability of the MosFETs into higher frequencies to reduce oscillation, I haven't calculated a value yet.

That's it so far, I've still gotta add in tube PIs of my own creation- Stay tuned!

Phase Splitters Not on designing, but shows the majority of VT phase inverters, good overview.
HV MosFET PI pdf.pdf
Last edited by blandguitar at Mar 1, 2011,
Good idea, can't wait to see updates : ) .
Quote by MH400
a girl on the interwebz?

You have 2 options.

1. Tits.
2. GTFO.

Ibanez UV777P
Ibanez RGD2127FX
Ibanez RG3120TW
Ibanez RGD7321
Ibanez RG6003FM
Ibanez SA160
Jackson Slatxmg3-7
Baron Custom Amps K88
Rivera Knucklehead TRE
Fryette Sig: X
Randall RM4 /w Modded modules
Mesa 4x12
Bogner 4x12
Peavey 4x12(K85s)
Quote by Steve46
thanks alot ice condition!! your the breast!

The best bosom in all of UG.
This is a great idea. This is probably what you were talking about in your PM. I was kinda tired when I read it.

This is definitely a good place for people to graduate from the brief descriptions in my thread.
This is the thread.

I figured if anyone was starting to design tube stuff, and had any questions, it could go here, so it's for experiments as much as anything else.
Last edited by blandguitar at Aug 17, 2010,
I'm curious as to how that mosFET/tube distortion hybrid will sound ... Any idea on that?

You should deffffinitelyyyy post sound clips once you've got it.
Quote by Steve46
thanks alot ice condition!! your the breast!

The best bosom in all of UG.
I'm gonna set up a tube testing board once I'm finished with my piezo preamp, I probably won't have the MosFETs until after christmas though. I'm not sure how much it'll distort, or it's tone. I've just never seen it done before.
Quote by blandguitar
I'm gonna set up a tube testing board once I'm finished with my piezo preamp, I probably won't have the MosFETs until after christmas though. I'm not sure how much it'll distort, or it's tone. I've just never seen it done before.

What Mosfets do you plan on using for this?
It doesn't matter, or shouldn't, even a big power MosFET should do the job fine. Anything with a source-drain voltage breakdown a good bit higher than the HT should do well. Really anything rated for the entire HT voltage should work well enough, the anodes will rest at about 2/3s the HV. I'm gonna try different ones for different voltage levels. Running the B+ at about 165 is a convenient value since an iso transformer will do.
Quote by blandguitar
It doesn't matter, or shouldn't, even a big power MosFET should do the job fine. Anything with a source-drain voltage breakdown a good bit higher than the HT should do well. Really anything rated for the entire HT voltage should work well enough, the anodes will rest at about 2/3s the HV. I'm gonna try different ones for different voltage levels. Running the B+ at about 165 is a convenient value since an iso transformer will do.

Alright, I wasn't sure if you were gonna get any specific ones. Don't get NTE parts, as they are usually 2 to 3 times the price of other similar parts.
I just went with decent Vdss values, low output resistance, and low capacitance MosFETs, low for a MosFET anyway. I went through the mouser catalog to find them since I'll have enough cash to do a big order come winter.
The gain can be calculated by the equation A= [-mu*Ra]/[Ra+ra]

22uF is a fully bypassed cathode resistor. A stands for the amplification factor, or gain. I don't remember the mu of a 12au7 offhand.
Ra= plate/anode resistor.
ra=internal resistance at the anode.
so does anyone have a layout or schem for a late 60's or early 70's marshall plexi style preamp? I love that tone! Trying to get the Alex Lifeson tone (Finding My Way, La Villa Strangiato).
Just found something, is this all that's in the preamp? Its a Plexi preamp, but it looks too simple.

EDIT: It's confusing me, I don't see values or pots or anything that should be in a preamp. Can anyone explain it?

And found the tone stack, again no details.
Last edited by plexi123 at Feb 7, 2011,
That's just the input stage. As for the scheme, googling typically does very well by me, don't forget to look at images as well as links.

That's an analysis of the FMV (Fender Marshall Vox) tone stack. Check out the book Guitar Preamps for Guitar and Bass by Merlin Blencowe, it'll teach you a ton about how to design stages.
Last edited by blandguitar at Feb 7, 2011,
Quote by blandguitar
That's just the input stage. As for the scheme, googling typically does very well by me, don't forget to look at images as well as links.

That's an analysis of the FMV (Fender Marshall Vox) tone stack. Check out the book Guitar Preamps for Guitar and Bass by Merlin Blencowe, it'll teach you a ton about how to design stages.

Yeah, this stuff is new to me. I appreciate all the help I'm getting on this! I've been trying google for a while now, but can't find what I want. As of right now, I can't afford to buy a book and parts (waiting on money for parts until my Boss GT-3 sells), I'm just trying to get the information ahead of time so when I get the money, I have the information and can order the parts quicker.
The money on Merlin's book is money well spent, and if you wanna do a homebrew design, I'd say buy it before parts. alibris.com also offers used books for cheap. Fun with tubes, in addition to the links I posted above are good sources of information.
Hi. With a dual/twin Triode, what's independent? the cathode is shared or independent? Can we set two stages of pre-amp with a single twin triode? I'm asking because I'm looking at a Marshall amp schematic (JCM800 2204) that does this I think
We can see the output plate of the V2a tube going to the grid entry of the V2b(same tube) does this amplifies the same as two tubes?

In the V2b (in the schematic) the out plate doesn't go to the next tube it goes straight to the powersupply, does the "sound" exits throught the catode?...I'm a bit lost after this schem.
The triodes are independent, the heaters are not. If you are only using 1 triode, ground the cathode of the unused triode.

The V2b is used as a cathode follower, a buffer. It lowers the output impedance and isolates the previous gain stages from the tone stack, allowing less noise and lower losses from the tone stack.

Any vacuum tube or semiconductor set up as a gain stage, the current flows through both the plate and cathode resistor, or in the case of FETs, the drain and source.

The physical construction of these devices create a negative feedback to the input, and whatever the input does, the cathode/source follows it. Now, we can then compare the resistor values of the plate/drain to the cathode/source, the same current through them means that the voltages are proportional to the resistors. If the resistor at the plate is twice as big as the cathode resistor, the voltage gain will be 2.

Now. say we only want the signal from the cathode, this allows us to set the idle current to whatever we want. After all, having a plate resistor just gets in the way, we use the cathode resistor, in any gain stage, to set the idle current, and in this case, removing it does us no harm. The idle current, which is the idle plate voltage divided by the cathode resistor of the follower.

I use [2/3*HT]/i=Rk

In which, HT is the voltage supplying the plate of the 2 stages, and i is the idle current that you want. Rk is the cathode resistor value that you need.

After the tonestack, you have the phase inverter, which makes 1 signal, 2, which are 180 degrees out of phase with each other, to feed the power amplifier.
Tnx Just to be sure. In the same schematics.
if the input is in high It will be amplified by V1a and V1b, and if it's in low if will only be V1a?
Mu-Amp and SRPP: What in the heck are they anyway?

In short, the Mu-Amp, the SRPP, and the Constant Current Source loaded stages can be summed up like this:

CCS: Uses an active device to maintain a constant current, the load line on the graph is approximately horizontal at the desired current. Because the stage uses a high impedance to obtain high gain, it has a high output impedance and should be buffered or signal loss will occur. Output is taken from the bottom device.
SRPP: High gain stage using a CCS to maximize gain, signal taken from the low impedance terminal of the top device. High output impedance, harsh overdrive.
Mu-Amp: Large gain, less than CCS or SRPP, but still larger than a normal resistor loaded stage. Pleasant overdrive characteristics, used in many guitar stompboxes with FETs.

The CCS: The CCS loaded stage is simple, the AC impedance of a current source is theoretically infinite, therefore the load line is a horizontal line at the current you bias at. The voltage bias of the bottom tube is E=Rk*I, where I is the current set by the current source. Improvements can be made with devices such as MosFETs or pentodes that have a larger transconductance, lastly, using a LM317 (remember to stay within maximum voltages here) in current sink mode works wonderfully, better than any single discrete device. The main advantages to using the LM317 are that it's small, just a 3-pin SIL package and a resistor, and very much cheap, and lastly, it's dependable, the LM317 is consistent, and makes for a fantastic CCS load. If interested in using the LM317, look at a relevant datasheet, the resistor side goes to the anode.

SRPP: The SRPP is the shunt regulated push pull, the top triode acts like a cathode follower, but since it is directly taking the grid voltage from it's own cathode resistor, it loses the low output impedance of the cathode follower. A good place to use this is to drive a constant load, such as headphones, a reverb tank, or any other known load. The best way to learn about designing these stages is simply to outsource to those that had done much more comprehensive research than I. The TubeCAD link is pretty good as are the Valve Wizard's thoughts on these topics and can be found on his site.
TubeCAD: SRPP Deconstructed

Mu-Amp: The mu-amp is called thus because the gain is said to approach the mu (voltage gain factor) of the tube; in practical use, it is actually less than the mu, but still larger than a resistor loaded stage. The mu amp is actually a cathode follower combined with an actively loaded voltage gain stage. It has the impressively low output impedance of a cathode follower, with additional gain compared to a resistor loaded stage. This makes it near ideal to drive a tone stack (some say that tonestacks should be unbuffered). The top tube uses the output of the bottom tube to drive the cathode follower via the Cg capacitor. The load seen at the bottom triode is essentially a constant current source, the self-bias of the upper triode forms a CCS as long as the upper triode doesn't run into either power supply voltage (B+/HT or ground). The upper triode is bootstrapped by the lower triode, and in turn, the upper triode forms a larger load on the bottom tube. Rg2 sets the DC bias of the follower (remember that the output of a voltage follower is essentially constant in reference to the input voltage), for a voltage follower, we want the output to rest in the middle of the power supply for maximum signal swing. Therefore we divide the available power supply voltage between the two valves. The remaining half of the power supply is where we want the bottom tube's anode where idle at. It is somewhat strange that the cathode follower uses it's own cathode load to supply the input to the grid, but that's how it works, and it works very well.

Why Use These?
Why would one use 2 devices to amplify the signal once rather than using 2 stages? There are several advantages; fewer amplifications stages, this means less coupling capacitors and other "accessories" to the stages, drive ability, the Mu-Amp is a great buffer/amplifier in one, and the SRPP can be tailored to any specific load needed and can make for a good driver, PSRR, power supply rejection ratio, by using an active load on the amplifying stage helps greatly reduce the power supply noise from coupling into the audio signal. Even though the input stages are typically the most filtered from power supply noise, the signal at the input is weakest and therefore most susceptible to noise, using an actively loaded stage helps boost the signal more than a single stage normally would and reduces the amount of power supply noise injected into the signal. Lastly, each other these stages will react differently than a typical resistor loaded stage, and they will distort differently, meaning a different tone.


With all of these circuits, solid state can be subbed in the upper triode's position, and in many cases is even better. Depletion mode HV MosFETs can be found at suppliers such as Mouser, if staying under 30 volts, JFETs may even be used. Enhanced mode devices, such as enhancement mode MosFETs and bjts could also be used, but not nearly as readily due to circuit modifications needed.

Active Loads and Signal Current Control
The Under appreciated Hybrid Mu-Follower
Last edited by blandguitar at Nov 9, 2011,
hrmmm... wonder if there's any coincidence between this new addition to your tut and a recent conversation about tube driven reverbs and the SRPP :p
Epiphone Les Paul Plus Top
Jet City JCA5212RC (SLO Modded)
Ibanez WD7 Wah
Mad Professor Sweet Honey Overdrive
TC Electronic Flashback Triple Delay
TC Electronic Trinity Reverb