V-45 Head

This project started as a rebuild of a 5f6a Bassman, but all that changed when the original power transformer primary shorted (no clear indication why). At that point a complete rebuild emerged as the favored option since it offered the opportunity to try some new ideas and to make a more versatile amp out of the now dead chassis. A dead chassis is a terrible thing to waste.

The V-45 is a more ambitious and versatile design than a Bassman. The 20" wide chassis panel has room for an array of controls over signal levels, EQ, and stage behaviors. I considered a multi-channel preamp, but I opted for a single channel where very different amp characters can be created by switches. It comes down to a trade-off between allocating panel space to fewer but redundant controls for a true multi-channel amp, or using the same space for more control options in a single channel. The single channel approach won the day after sketching out both options.

This amp is designed for control over tone shaping and gain options. The v1.4 iteration had 6 switches to configure the preamp. I found that fun for a while, but too bothersome in the long run. The new version (v1.6) has only 2 switches.

Update 3/14/21:

The schematic and amp is now at version v1.6. This is a keeper! As described below, it's highly influenced by the Solo-100 described in my new book (GAD), so it's earned its affiliation with the high-five of Tommy-Tube and Semi-Sam (below).

The text below has more to say about how the book ideas and approaches influence this design.

In writing my new book (Guitar Amplifier Design), I thought a lot about the many lessons learned in designing and building amps over many years. The writing effort was a very productive time to think and organize all of the information I had gathered in my head. There’s no question that I do things differently now than I did even 2-3 yrs ago. Armed with new insights and understanding, I am motivated to revisit old projects that no longer reflect my current thinking and approaches to amps. I chose the V45 project as the first to update.

The v1.4 design was a useful learning experience, but I never considered the amp as reaching it's potential. The mojo was missing. In the early versions, I focused on new preamp clipping and overdrive circuits. Those circuits worked, but the final amp was missing something. I spent a lot of time analyzing each part of the amp and the amp as a whole, but it was clear that simple changes did not impact the problems, so I shelved the project for a while.

The tests and analysis did show that the v1.4 power amp works really well. I tested it with other preamps (via the FX jack) and the tremolo and reverb effects also work well. The variable screen voltages produce different characters in the power amp, and the low screen-voltage option sounds best to me. So, the power supply and power amp and effects are kept just as they were in the v1.4 design.

The v1.5 and v1.6 redesigns focused on the preamp. That means everything from the input jack to the FX jack is open to change. Two 12ax7 tubes (4 triodes) are available. The question is how best to use them. Four triodes can provide lots of gain, so preamp OD is possible and desirable. A single triode for the OD section (or "color stage," as I call it in the GAD book) is OK, but it produces a somewhat uninteresting OD character. More interest is created by more triodes.

The GAD book describes a warm and cold triode pair (1 tube) as a color stage and that could be replicated in the V45, leaving the other two triodes for gain stages. This strategy would be relatively risk-free and likely produce a similar result to that obtained in the Solo-100. However, why do the same thing twice unless you are intentionally replicating something? I prefer to try something with a bit of risk and more reward potential. If two triodes produce a better OD than a single triode, how about 3 or 4 triodes? This project seems like a good opportunity to consider this question.

I worked through some options for using 3 triodes for OD, and using one for the input stage. The problem is that if an OD triode is in the same bottle as the input triode, the ~2pf capacitance between the triode plates becomes a problem. Once the OD triode hits cutoff, that sharp transition couples to the input stage as a significant distortion. So, the remaining option is to use all 4 triodes for the color stage and make the bulk of the preamp solid-state. Why not?

As described in GAD, there are reasons to make the SS stages very clean and place the color stage at the end of the preamp. The tone stack is moved to the middle, rather than after the OD stages, as in the v1.4 design. (See GAD for reasons and benefits.)

With the above constraints, the redesign takes shape. After some Spice simulations and iterations, the following stages comprise the new preamp.

The input stage (Q1, Q2) is clean and has high gain (~75) and high headroom. A buffer (Q3) drives a switch-selectable (SW2) tilt network (more about that in GAD) to emphasize string harmonics by 6, 10, or 15 dB.

A gain control (P1) and gain stage (Q4) determines the level of drive signal into the tone stack and later color stage. A second buffer (Q5, Q6) drives the tone stack cleanly at low and high-drive signal levels.

The tone stack has 3 selectable scoop-options (SW5) and traditional Bass, Mid, and Treble controls. The output of the tone stack is buffered (Q8) and DC-coupled to drive the 4-triode color stage without coupling caps. The DC coupling idea was first tried in v1.4 and further developed for the Solo-100. No caps means no bias-shifts and no blocking problems. Since the color stage is driven HARD, as in the Solo-100, DC-coupling matters a lot and it allows the full amp bandwidth to pass into and through the OD stage.

Two color-stage triodes (V2a, V2b) are arranged in a DCCF configuration, and two triodes are arranged as cool (V1a) and cold-clippers (V1b). Their color stage outputs are summed and buffered (Q9, Q10) to drive the volume control and FX jack. The differences between v1.5 and v1.6 are minor tweaks in the driver and clipper stages. The schematics for v1.6 are below.

The results are rewarding in many ways. First, the 4-triode color stage (quad-OD or QOD) does sound crazy good – very rich and interesting at all drive levels. At low drive, it's spanky clean and high drive produces great sustain, octave feedback, crunch, sizzle, ... you name it.

Tone and drive settings achieve a wide range of cool and musical tones. It takes bass really well and literally screams at high treble levels. The absence of dissonant or sour harmonics reinforces the theory explained in GAD about harmonic creation in amplifiers.

The SS preamp sounds and performs great. It provides tons of clean gain and variable EQ to drive the color stage. It stays neutral and lets the color-stage triodes do what they do best.

The preamp and power amp schematic is shown below. Everything up to the FX jack (J3) is new for v1.6. There are no significant changes (from v1.4) to the power amp.

Update 5/4/19:

The schematic and the text below reflects the (now old) v1.4 version. This section should be seen as a reference and start point for the v1.6 version described above.

The schematic covers two sheets. Page 1 is essentially the signal flow from input jack to spkr, while page 2 shows, power supply, switching, and effects. I'll describe the signal flow on page 1 first.

The v1.5 power supply has minor changes from v1.4 and there is no need for the slow-switching that was in v1.4. The v1.6 supply changes are in how the FX cards are powered.

The bias supply is a voltage-doubler with S10a selecting 3 ranges of bias for different screen voltages selected by S10b. The idea is to scale the bias voltage to approximately track screen voltages. While this can't be done perfectly for all tubes, this arrangement tracks well enough that a bias setting at full screen voltage with be reasonable (conservative) for lower screen voltages as well.

Screen voltage options are created by a series string of fifteen 10v 5w zener diodes. SW10b bypasses all of them for full screen voltage, or bypasses 5 diodes for ~100 V reduction. In the mid (open) position, all 15 diodes reduce the screen voltage by almost 150 volts.

In summary, the switch options for the output stage provide ~20 volts of plate voltage variation from the rectifier choice, and a ~100 or ~150 volt decrease in screen voltage. In play testing, I've found the best tone is obtained with -150 V screen setting and the SS rectifier.

Output stage behaviors for these variations are simulated in Nick B's simulator at the bottom of this page. Visually, the most linear output transfer function arises from the settings I favor. I don't think that's a coincidence.

A visual indication of screen current is provided by a front panel LED (D19). As screen current increases, the LED lights in proportion to average screen current. To ensure the LED doesn't burn out, it's current is limited (at ~30 mA) by shunting higher currents through Q5. In practice, this indicator is useful to show when the output stage is pushed to clipping. It also provides some indication of tube health since worn tubes may draw excessive screen current. The LED and zener circuits are all bypassed by C58 to ensure that the screen supply free of any rectified signals created by zener or LED impedances.

From the screen supply, the preamp circuits are powered through a series of RC filters that reduce voltage and decouple the stages. A separate RC filter network is provided by R83 and C54 for the effects cards. These cards (mounted on the chassis top) provide reverb and two tremolo effects. The reverb card (HiH Reverb) is made by Amp-Aid and drives a standard 8-10 ohm spring pan. The card has unity gain for the input (dry) signal, and reverb is added to the dry signal (similar to the Fender 6G15 system). The reverb level is controlled by P6. The reverb card output is always driving the input of the harmonic tremolo card.

The harmonic tremolo card (H Trem) is also made by Amp-Aid and it implements a high-quality sine wave tremolo oscillator and a harmonic tremolo function (based on the Fender Concert 6G12a and Bandmaster 6G7a). This card overcomes the thumping issues that were common with the original circuits. The card's oscillator sine wave is buffered and output for external use.

The sine wave output is used with the circuits of Q6-8 to produce bias tremolo in the phase inverter. A Villard circuit (D26 and C58) shifts the osc sine wave to drive Q6 so that a sinusoidal current drawn by Q7 and Q8 through the LTPI bias and tail resistors modulate PI gain and headroom to create the tremolo effect. Since this LTPI tremolo approach does not directly impact power tube bias, it doesn't stress the output tubes and the power supply the way a traditional bias-tremolo would.

The TRS1 plug interfaces the effects cards with the amp. With the plug inserted into J3, the reverb and harmonic tremolo effects are in the signal path. Unplugging the TRS1 plug removes the cards from the amp and makes the J3 jack available for other line-level effects. Note that bias tremolo always remains active, even with TRS1 unplugged. Since there are no gain controls after the FX jack, the signal levels and dynamic range of the effects loop are determined by the PI/output stage gain and dynamic range. That means the effects cards can be designed (and they are) to NEVER overload or clip until after the output stage is well into clipping.

With TRS1 plugged into J3, the reverb effect is always active and SW7 selects the tremolo effect to be used. With S7 in the "bias-tremolo" position, the reverb card output is fed to TRS1 by SW7a (and the harmonic tremolo output is not used), and SW7b routes the osc signal to the transistors producing PI bias tremolo. In the harmonic-tremolo position, SW7a routes the harmonic tremolo output signal to TRS1 while SW7b disables the PI bias tremolo.

Lastly, the Q9 and Q10 circuits merit an explanation. SW3 selects the clipping or bias options for V1b. If the OF1 and OF2 circuits were directly switched, the change in V1b bias would be instantaneous, producing 100 volt (or more) transients at the plate of V1b. Such transients would produce a huge output "pop" and make switching SW3 unpleasant and possibly dangerous to the amp, spkrs, and anyone near the amp.

The optical FETs have a linear transfer characteristic, so they can gradually transition from high to low resistance. If the OFs are turned on/off gradually, the bias transition (and plate voltage change) of V1b is also gradual. In fact, if the transitions occur over a second or more, they become inaudible due to the limited low freq response of the amp. The Q9 and Q10 circuits are a simple means of producing a slowly changing current that ramps up and down to switch OF1 and OF2 on/off slowly. SW3 can turn on either OF1 or OF2, or turn them both off safely and inaudibly. This is an alternative to a mute circuit. It eliminates the cause of a thump, rather than just masking it. All V-45 amp switches are completely "pop-free" without any need for a mute function.

Here are a few chassis pictures and comments about construction. The 5f6a chassis was a perfect fit for the PT and OT. Circuits are now on a turret board. Many new holes were drilled to accommodate the controls and components. Two printed pages covered with plastic film form the front panel. A custom cabinet is on order.

The first image shows the back panel wiring. Most of the signal path is directly wired point-to-point between the tube sockets and the nearest turrets. This approach creates dense clusters of parts caps and resistors at the sockets. So be it. It's worth the extra time and care it takes IMO. Short wires make for clean signals and no surprises (oscillations, cross-coupling, noise, etc).

A small perf board near the rectifier supports circuits for the plate and screen voltage switching. The 4 jacks in the center are the FX jack and 3 speaker outputs.

This next image shows the control panel wiring. Here again, most of the signal path is directly wired to the controls and switches using 1/4w resistors and film caps. Shielded cables are used for signals that run more than a few inches between controls or between the front and back panels. Most of the components actually mounted on the turret board are the power supply and effects circuits shown on page 2 of the schematic. The string of 15 zeners for the screen supply are visible at the right end of the board.

The chassis top is populated by the PT, OT, and choke, as well as the 24 V power supply and the two effects cards. Shielded cables are used for all the signal paths to/from these boards.

The effects of plate and screen voltage variations from SW10 and SW11 are illustrated below. Nick B's simulator (http://bmamps.com/ivds.html) clearly shows the variations in linearity and high-output waveform compression caused by the switch settings. The plots below show KT66 behaviors for low (-150 V) and medium (-100 V) and full screen voltages. The full screen voltage case uses the tube rectifier, while the other cases use the SS rectifier. All three options are biased at about 30-35 mA per output tube.

This first graphic shows the full screen voltage case. Note that the load line lies well below the knee of the Vgk=0 curve. This occurs since the load impedance is higher than commonly used for these plate and screen voltages. The effects are slightly reduced power, but it produces early and aggressive compression (clipping) in the output stage as shown by the transfer function inflections (bending horizontal) at high outputs.

Although not captured in the simulation, these conditions also produce high screen currents so they stress the screens. However, if you want that compressed tone, this gets it done.

This second graphic shows the medium screen voltage case (reduced by ~100 V). Note that the load line passes just below the knee of the Vgk=0 curve. The reduction in screen voltage pulls the curves down to more optimally match the load impedance.

The effects are a slight increase in output power and less compression (later clipping) in the output stage. The transfer function inflections (bending horizontal) are reduced at high outputs. These conditions also reduce screen currents (and screen stress), however the load line now passes further inside the max plate dissipation area, increasing the stress on the plates.

These conditions are moderate and lead to moderate behaviors for breakup/clipping, screen dissipation, and plate dissipation. These are the settings and conditions I use as default.

This third graphic shows the low screen voltage case (reduced by ~150 V). Note that the load line intersects near the knee of the Vgk=0 curve. This added reduction in screen voltage produces a slight decrease in output power. This condition produces very abrupt compression (clipping). The transfer function inflections (bending horizontal) are slight, even as Vgk approaches zero.

These conditions reduce screen currents further (and screen stress), however the load line now passes even further inside the max plate dissipation area, increasing the stress on the plates. These conditions produce maximum headroom and hard breakup/clipping at high signal levels.