"type F+" gain section

This circuit starts with the Fender AB763 circuit, as used in many of the Fender amps from the "Silverface" era.  In front of the first triode section, which would be the guitar input on an actual Fender amp, I place the "gain" pot; this allows the gain to be trimmed down, because this section will normally be fed from the input section, which also contains a similar gain stage (i.e., more gain stages than the actual AB763).  Alternatively, this arrangement allows more gain/overdrive to be obtained than an unassisted Fender amp could normally produce.

But for even more gain, I provide two more triode stages with a second tube, accessible with the "pull 2-stage" switch.  The first of these triodes is set up as a cold-clipper, similar to Marshall.  There is a "pull alt voice" switch, which increases the cold-clipper cathode resistor, and also changes the R-C circuitry between the two triode sections.  The gain into these sections is controlled by the second section of the dual-gang "gain" pot; thus, this pot has a multiplying effect on gain when in 2-stage mode, which should give a wide range of control over the distortion tone.

The "pull raw" switch, borrowed from other Fender modification websites, effectively eliminates the tone-stack, producing something closer to the so-called "Tweed" tone.  Gain is higher without the insertion loss of the passive tone-stack; and since the tone-stack fundamentally causes a "mid scoop" no matter what its settings are, the "pull raw" switch leads to more prominent midrange.

"type M" gain section

This circuit is the same as the Marshall 2203 from the "low gain" input onward.  Thus, when the input section "voice" control is set appropriately, the overall signal chain is similar to the 2203 from the "high gain" input.

I use the available 4th triode section for a cathode-follower at the output, to buffer the passive tone-stack and level pot.

One of the circuit elements most responsible for the "Marshall sound", is the first triode section: this triode is set up as a "cold clipper", through having a relatively high cathode resistance (10k), and no bypass capacitor.  This section therefore clips the waveform asymmetrically, which accentuates the even harmonics, and this leads to a tone with a pleasingly bright and intense distortion edge, while still preserving readable dynamics and some of the original resonance of the guitar, similar to a "clean" tone (because half of the peaks are clipped, but the other half preserve their original shape).

In Soldano amps, I've read that a similar cold clipper circuit is employed, but the cathode resistor is even higher, 39k.  In this circuit, I provide a "pull colder" switch which brings the cathode resistor fairly close to the Soldano value: 37k.

Notice, in this schematic and in the one for the "F+" section, I mark a dividing border between circuitry closely based on the named original (there may still be slight differences), and elements I have added which are not part of the original ("BWK-M").

gain sections

Here are my notebook sketches, showing more or less what I want to do with the gain sections.  This doesn't show some of the surrounding bits such as the bypass relays.  My amp will have two "type F+", in positions 1 and 3, with a single "type M" in between (and the bypass switches allow any combination to be enabled, from none to all three in series, but always in the same fixed ordering).

I've drawn dashed lines to separate the portions of the circuits which are essentially direct copies of their namesakes (based upon public-domain schematics), from the supporting portions designed by me ("BWK-M").  (Take care to distinguish these dashed lines, from the ones representing conjoined dual controls such as pots and switches: I should have made them more different.)

The component values in the BWK-M portions, in many cases, are just my initial guesstimates.  These values "should work" to at least produce sound, but no doubt much adjustment will be needed once there is a working prototype.

reverb section -- initial component values

reverb section

Latest schematic for the stereo reverb/effects section:

As you can see, component values have not been fixed yet; this just shows the topology, with the new four-tube circuit.  Given the necessary 7th triode section for inverting the dry signal, I've used the "extra" 8th triode as a cathode-follower to more-effectively drive the reverb transformer.

important note re reverb and effect-loop phase

I haven't started to draw the "official" schematic for the reverb section yet, but I observe that in my previous writings here, as well as in many of the diagrams in my notebook, there is a serious problem: the "dry" signal is out-of-phase with the "wet", at the mix knob.  I had been planning to take the "dry" signal right from the input of the whole circuit section, i.e., non-inverted.  But at the position in the circuit of the "mix" knobs, the "wet" signal is inverted, having passed through one inverting gain stage.  After the mix pots, the signal is inverted again, so the overall reverb/effects path is non-inverting.  The fix is that the "dry" signal must be taken after the first inverting stage.

With my new conception that effect sends and returns need to be line-level and not higher, the effect returns will now share the same input circuits as the reverb tanks, so this phasing problem and its solution apply equally to the external stereo effects loop, as to the internal spring reverbs.

(Note that the "dry" signal for the mix pots, is not the same as either the "bypass" signal which appears at the outputs when the relays are in bypass mode, or the effect-send signal.  Both of these latter are non-inverting signals, as discussed in an earlier post on the reverb section.  Thus, unlike in the case of the effects loops associated with the active EQs, the send signal here will remain always-active.  Also, this effects loop is in-phase.)

Just wanted to get this important correction out there (mainly to myself, so I don't forget).  Schematic coming soon...

----

Ummm... more about this... dry signal taken from existing first stage will not work, because this stage is after the source-switch: the dry signal must not be affected by the source-switch, it must always come from the eq2 output position, i.e., the "natural" input signal for this stage.  It is the contrast between the distorted dry signal, and a reverb signal which is fed from an earlier, cleaner signal location, which the source switch is intended to enable.  If both dry and wet are fed from the earlier point chosen by the source switch, then there's no point to the later stages, they would be feeding nothing.

So... I don't like this, but the initial conclusion would seem to be that *yet another* 12AX7 is needed for the reverb section: a total of four.  Yikes.  I feel a great desire to get this tube count back down again, I was already unhappy with three.

One way to reduce the tube count, would be not to buffer the outputs of the mix pots.  This circuit only has one target (load) to worry about, i.e., the inputs of the power amps.  It's either driving the power amps, or it's bypassed.  Other circuit sections have to be able to handle a number of possible output loads, depending on which bypass relays are energized, and also depending on the position of the reverb source selector (which puts some extra load on whichever output it is monitoring).

My original idea for the power amps had the master-volume pots right at the inputs, then followed by tube stages.  But this will only work well if all the other preceding circuit sections have good, low-impedance tube output drive: if the output of the reverb section is "passive", coming from the mix pots, there will be poor performance directly driving the volume pots, and there may be a noticeable change (mainly, in sound level), switching the reverb between "in" and "out".  The power amps could perhaps be reconfigured to have a tube stage ahead of the volume pots; but if this requires actually adding a tube, then nothing is saved: it only makes sense if the reconfiguration can be done within the current tube-count (I'll look into it).

However, maybe the best way is for me to just, reluctantly, get used to the idea of four tubes for the reverb section alone.  The final stages in the reverb circuit, are closed-loop inverting stages, which give better performance for the resistive mixing networks, because the resulting virtual grounds at the tube inputs, ensure significant separation between the dry and wet signals.  Without the virtual ground, the resistive networks might have enough "leakage" that 100% dry or 100% wet settings are not attainable.

Well, that's my current stream-of-consciousness about all this!  As you can see, it is a work in progress, and significant issues remain in play, yet to be solved to my true satisfaction...

active EQ

Here's the latest schematic for the 3-band active EQ.

As you can see, I have decided to move the effects send/return positions to "inside" the boundary of the tube circuitry.  I.e., the send comes after the first tube stage, and the return comes before the last two stages.

Originally, I had planned to locate the send/return outside of basically the entire circuit, i.e., "right next to" the bypass relay contacts.  This would have meant that the send was always-active, even when the relay was in bypass mode: generally, a preferable situation.  However, due to the higher signal levels of tube circuits, jacks placed in these positions would not have been directly compatible with normal line-level audio gear; an external interface unit would have been required, just like the "Dumbleator" unit which is required to use external effects with Dumble guitar amps.  It's easy to pad down the send signal, but the return signal needs active gain to bring it back up to "tube" levels, and then the issue of inverted phase comes into play.

The clean solution, without adding more active circuitry, is to give up the always-active send, and put the jacks as shown in the diagram above.  The send can be padded as desired, including with an adjustment pot as shown; the return has a fixed, open-loop gain when the plug is inserted (the NFB path to the Bax tone control is broken by the switching jack).  The send signal is thus not always-active; it is only active when the section is selected "in" and the LED is on.  When the section is bypassed, the send is muted, because the input of the EQ circuit is grounded.

Note that there is one fairly significant disadvantage to my current effects loop arrangement: the external effect sees an out-of-phase signal.  As long as the external unit is non-inverting, its output signal is then inverted at the return stage, and everything ends up in-phase.  There should be no audible difference.  However, for certain applications (such as using external test equipment like oscilloscopes), it might be inconvenient or confusing that the loop signal is inverted.

There are two identical copies of this EQ section within my amp, named "eq1" and "eq2": one before the distortion, one after.  This distortion-bracketed-by-double-EQ configuration is a crucial part of my sonic concept; along with the nature of the EQ itself, i.e., active-Bax to give the capability to boost as well as to cut, and shiftable mid frequency to permit production of precisely-controlled formants.

input section

Here are the latest developments concerning the input section.  Also, this begins my first attempt to draw a schematic diagram for the whole thing (piece by piece).  I'm drawing by hand (obviously) in black ink on blue-grid graph paper, then scanning to PDF and processing the images with a combination of "NETPBM" open-source tools, and programs I wrote myself (originally as part of my pierced-foil art project, a different story entirely...).  Final result is a reasonably clear, and small, black & white PNG file.

A few things to note here.  I based the input jack setup on Marshall/Fender, i.e., one jack high-gain, the other with a resistor pad, but I made the pad resistors asymmetrical so that the low-gain input (2) pads down more than the original 1/2 division.  The high-gain input (1) still sees about the same 1M Ohm impedance, with about the same grid-blocking resistor (36k instead of 34k, with the values shown).  This front-end circuit seems to be an accepted "magic" set of values, used in many/most of the big-name (as well as the boutique) tube amps.  So I guess we'll stick with it -- other than making input (2) more useful by giving it the deeper padding.

In other images and text before today, I've been referring to the 6-position switch in the input section, as "HPF" (highpass filter), but now it's called "voice".  That's because it controls more stuff.  Originally, I based this control (HPF) on the Orange/Matamp "FAC" control, which selects the bypass capacitor on the output of the first triode stage.  Now, as you can see, I have changed to a dual-gang 6-pos switch, and the switch also changes the cathode circuit of the first triode.  There are three choices of cathode circuit, each with two choices of bypass capacitor: the "regular" value, and a lower capacitance (formed by two caps in series), which will give a higher cutoff frequency for the low-end.  The three cathode circuits correspond to Marshall 1987 low-gain input (positions 1-2), Fender input (3-4), and Marshall 1987/2203 high-gain input (5-6).  These are arranged in order of higher cutoff, i.e., decreasing low-end, as the knob is turned clockwise.  Normally, positions towards CCW would be used for "clean" tones, and positions towards CW would be used for "dirty", but of course the reason for the separate control is so that one can choose otherwise!  (E.g., one might wish to use a more-CCW position for a highly-distorted sound, if one were then going to dial-in a significantly non-flat response in the following active EQ module (eq1); the input stage would pass a full-range signal, which the EQ would then shape as desired.)

The second triode stage is an inverting stage with negative feedback; I've shown gain of -1, but it could be set a little higher if the loss through the passive components upstream is too high.  Most of the time, we'll want to keep the input gain set for clean headroom, and then use the switchable gain stages to produce overdrive; but we want to have enough gain to produce some reasonable overdrive with just the input knob, when desired (i.e., for the closest thing to "Tweed" sound that this amp can produce).

Hopefully, the inverting-NFB stage as above, will work well as the output stage for several of these modules.  Each module output may have to drive any of the switchable modules as input, or the inputs of the power amp if no intervening modules are enabled.  Getting all of the module gains and drive levels set right, so that the system works as intended in each of the possible configurations of module in/out (6 relays == 64 combinations total), will probably take quite a bit of experimentation, measurement, and component-swapping.