Deluxe Plus

Project Overview:

The classic mid-60's Deluxe Reverb is wildly popular and loved for a variety of reasons, including clean scoop tone; a nice compromise on weight, size, and power; a pedal-friendly preamp, and all-tube tremolo and reverb. That said, it has some nits, including thin bass response and too much brightness for some. There are well known mods to address these and other issues, but basically the amp has survived intact for about 50 yrs in various releases. The current DRRI “reissue” model is almost exactly the same as the original AB763 circuit.

While mulling over some new project ideas last year, it occurred to me that not only was the DR a classic amp, but it could also be a great platform for an amp project. A rebuild is however, not what I was after. The chassis is big enough for some circuit and control panel expansion and all the basics are already there – a complete clean channel amp with transformers, cabinet, etc.

Adding new features is simplified, especially if you want to keep the essential DR features anyway. I thought about a list of feature to explore in a new design. Some of them didn’t make it past the idea stage, some were tried and discarded, but many were feasible and worked out nicely. The core idea was to keep the DR reverb channel for its signature sound and features, and to replace the second channel with something different.

I spent a lot of time thinking about what to do and how to do it. I also wanted to clean up some of the DR nits and add some features I like in any amp. The final list of features (below) is long, but each issue was worthwhile (IMO) and contributes something of value to the final package.

While I pursued all of these features as parts of the same project, many of the features are independent and most can be implemented with only modest effort and non-destructively (as reversible mods) on existing DR (or similar) amps.

Most of the gripes I’ve heard about the stock Deluxe Reverb are addressed in this design. There is a story to tell about this amp, so the following text is a bit long, but there are pictures and sound clips at the end.

After reading this page, check out Version 2 of this amp and the hunt for 5e3 tone.

Project Goals:

- DR classic AB763 circuit core with something “special” for the second channel (options like Pentode or Cascode circuits.)

- Switchable mixing of preamp stages: A / B / A+B (both channels with independent vol and tone)

- Reverb derived from clean ChB only, so ChB becomes “Reverb FX channel” when ChA-only is selected

- Switchable 2^nd ChA hi/low gain – fat tone vs vivid tone

- Add switched scoop-shift to new ChA tone stack for low, med, high scoop freq

- Add two additional 9-pin tube sockets for new ChA preamp section

- Switched bright caps on both channels (hi/lo/off)

- A master volume control - pre or post Phase Inverter (PI)

- Switched Hi-Cut (cut more, cut less, off) to PI output as post-master tone control

- Increase PI input cap from 0.001 to 0.015 for increased bass response – as in Pro Reverb and Blues Deluxe

- Ground PI tail and remove DC from output transformer and NFB loop

- Increase NFB path impedance (x100) to allow (smaller) 0.1uf cap for presence

- Add NFB control for loose / norm / presence

- Switched optical trem (classic DR) or bias trem (Pro Reverb or Princeton)

- Use audio-taper control for reverb level

- Add grid stoppers where overdrive can occur in DR channel (ChB)

- Increase 6v6 grid stoppers and screen resistors for blocking and failure insurance

- Change footswitch jacks to isolated phone (mono) jacks

- Trem default is “enabled” by switched foot pedal jack and upgrade to noiseless DR-reissue soft-switching circuit

- Add bias pots (level and balance) to chassis holes for convenient access

- Add bias test points and bias test/play switch to rear panel

- Add power supply filter caps, a separate preamp filter chain, and local decoupling caps for noise and crosstalk reduction

- Use shielded cable for all hi-z preamp connections (e.g., grid connections)

These features were decided over 3-4 months of consideration, reading, and some prototyping. The panel controls were determined by compromising on what I wanted as controls and what the DR chassis would allow. In the end, there isn’t really much that’s missing. Everything I really wanted was fit somehow.

Mechanical Construction

Once the amp features and control panels were decided, the actual construction strategy was considered. The options were 1) buy a DR (working or not) as cheaply as possible and then gut it and rebuild it, 2) buy DR components from varied sources and build from scratch, or 3) buy a DR kit and modify it as needed.

I opted for the latter option and I bought the Weber 6a20 kit. Used DRs are not cheap, and they can be cosmetically challenged. It’s a bit more money to buy a kit, but it saves a lot of time hunting for parts from different sources and then hoping they’ll fit together. I opted for the basic Weber transformers and parts package – but I went with their recommended 12F150-25w speaker upgrade.

The desired features required front and rear panel changes and some new chassis holes. The panel changes were minimized by sticking with the DR layout wherever possible. I scanned the Weber panels and used them as templates for laying out the new panels. I created PDF files from Powerpoint and had the new metal panels made by Plaquemaker. I've used them before and they do a great job.

Once the layouts were done, the steel chassis was modified. This is painful since I have limited machine tools, but I work slowly and use a drill, files, a dremel tool, and a reamer. The two new tube socket holes were made with a chassis punch. It turned out that the chassis was not as spacious as I had hoped and it took some effort to get everything to fit without obstructions.

The image below shows the final chassis top layout with the two added tube sockets (on the right). The entire row of three sockets on the right is allocated for the new preamp. Early on, I contemplated a small second chassis dedicated just for the new preamp tubes. That would save the trouble of adding tube sockets, and leave more space in the DR chassis but once I realized how many other panel and chassis changes were needed, there was little to gain by going that route. (Here is a link describing an EF86 preamp extension chassis for an amp.)

I also had to consider the circuit board. The stock DR eyelet board just doesn’t support significant added circuitry. I found a dense eyelet board from Weber. It was about the same size as the DR board, but it had lots of eyelets. The stock DR backing board was used for stiffness and insulation and the combination was supported on ¼” standoffs so wires could be tucked or crossed under the board. Below is a picture showing the eyelet board with some of the initial power wiring.

Once the chassis was modified, the new aluminum panels were mounted. The panel material is thin so it’s easily cut. The back panel was drilled and then glued directly to the chassis with a thin coat of spray adhesive. The connectors and controls provide solid support.

The new front panel was first glued to the Weber panel for strength and then both were drilled together and mounted with the controls. The final panels are solid and fit well. The below images show the panels soon after they were mounted.

There are many chassis changes and they are summarized below. Most are visible in the images above.

Front panel changes

- change both bright switches to spdt co toggle

- remove and cover vibrato input jacks

- add preamp A/B/A+B dpdt co toggle sw

- add fat/vivid 4pdt toggle

- add 4pdt co toggle for ChA tone-stack shift (hi-norm-low)

- add bias/opto trem dpdt toggle

- add MV control pot (either post PI vox-style, or pre-PI)

- add dpdt co switch for hi-cut on PI output (off, more-cut, less-cut)

Rear panel changes

- add bias level pot

- add bias test points

- add test/play dpdt switch

- add dpdt co switch for NFB norm/off/presence

- use chassis-isolated RCA reverb in/out jacks

- change reverb pedal to isolated phone jack

- change vibrato pedal to isolated phone jack

- remove ground switch and cover hole

- add line out control

- add line out jack (TRS)

Other non-panel chassis changes

- add two tube sockets to chassis for ChA preamp

- add filament transformer (6.3v 1amp) for filament supply for the two new ChA tubes

- use Weber high-density eyelet board to mount all circuitry – elevate on standoffs

- add bias-balance pot between output tube sockets

- add 8 ohm dummy-load resistor

- add filament hum-balance pot

High Level Circuit Design and Construction

Once the mechanical construction was done, the circuit construction could begin, however, I was still debating ChA preamp design options. I had some ideas by this point. I was leaning towards two cascode stages, but I didn’t want to decide on a final configuration until I could hear the results. So I needed to build the DR sections of the amp first and then try out my ChA ideas.

Even without a final design for ChA, the rest of the amp needed to be designed to include all the features described above. Even with careful consideration and calculation, it’s unrealistic to expect to get everything right the first time.

The initial schematics were the best I could come up with, but there were many changes as I tested and optimized things. Those post-design changes and decisions are really what leads to success or failure in a project like this. Some changes are simple resistor or cap value tweaks while others are more significant circuit changes that arise from a need to change circuit behavior or tonal character.

There is no way a paper design can fully anticipate everything. Actual construction and tweaking with a critical ear is irreplaceable. I’ll describe the main decisions and design twists in the circuit descriptions below.

The final amp schematic is here.

To help explain the sections, a colored block overlay is shown here.

Keep in mind that the ChA preamp was only a vaguely defined block when I designed and constructed the rest of amp. I reserved three tube sockets for ChA, but I wasn’t quite sure if I needed all of them or how I would use them. The whole point of ChA was to do something “different” and add something worthy of the classic DR core tone. More about that later, since first I had to build the basic DR core.

The following sections describe each of the colored circuit blocks.

Power Supply

Construction started with the AC power supply, carefully grounding the green wire to chassis and routing power to the power switch, fuse, and transformers. A second filament transformer was added for the ChA tubes to avoid loading the stock DR transformer with two additional tubes of filament current, and to allow for an elevated filament supply, if needed (and it was) for a cascode design.

Initially, AC power was routed to all the filaments of V4-V11, the tubes needed for the DR functions. Later the filament circuit was modified to create a DC supply for V4 and V2. The DC supply reduced hum in V4, and it was essential for V2 in ChA. The hum-balance pot P16 minimized hum when AC was used for V4 filaments. Later, with the DC supply, the pot had no audible effect, but it’s still there to provide a ground reference.

A standard DR high-voltage supply was built, using slightly higher value capacitors, since they are available and fit easily under the supply cover. Note the ChA preamp has its own filter chain coming off the choke. Those caps also fit under the supply cover. The elevated DC filament reference for V1 and V3 is controlled by P15. This was added once the ChA design was done. The DC filament voltages are fine-tuned by the 1ohm resistors (R76,77), which drop about 0.5 volts each to hit the target 6.3vdc.

Most of the power supply is complete in this image. The 8 ohm 25watt dummy load resistor is shown mounted near the choke.

The bias supply is simple and adjustable through P14, which is accessible on the rear panel. The front-panel image above shows the bias supply on the right end of the eyelet board. (Note the Styrofoam supports on the chassis ends. They are carved from shipping material and work great to support the chassis during construction.)

Power Amp

The power amp is a simple classic design with a long-tail phase inverter (PI) and fixed-bias output stage. I’ve made several changes, but the only intended change to the stock DR tone arises from a bigger input cap C29 to improve bass response. The remaining changes add convenience or tone options, but are not intended to alter the stock DR tone. I believe I’ve preserved the DR signature tone while adding flexibility and robustness.

Grid stoppers for the output tubes (R56,57) are increased to 22k. This helps ensure that the improved bass response doesn’t create blocking problems. The screen resistors (R60,61) are increased to 1K as a bit of insurance against the catastrophic failures reported with the 470 ohm standard values. Others report using 1.5K with negligible impact on tone or power.

Three test points are added to the back panel to allow convenient bias measurement and control. These measure the currents in output tube cathode circuits. For convenience, 1 ohm, 1% resistors are used so that milliamp (mA) currents translate directly to millivolt (mV) measurements.

SW6 is added to allow measurements when it’s open, or to short these resistors and remove them from the output circuit. While such test points are common, the switch is not. A hard driven 6V6 output tube has a peak cathode current of 70-80ma or more. That means there are 70-80mv peaks across these test resistors during high-power operation. That voltage appears as a negative feedback relative to the grid signal and it may impact tone, particularly at the onset of clipping or breakup of the power stage.

Since the stock DR doesn’t have these resistors, I added the switch to preserve the “pure” 6V6 behavior. I’ve since played with the switch settings and honestly, I can’t say I hear any difference, but it’s low-cost “tone insurance” since I may use different output tubes in the future and they may behave differently than the ones I have now.

The fixed-bias network includes P13, a bias-balance control, to adjust for tube variations. The single bias control (P14) relies on matched tubes. Any differences in the tubes create a net dc current flowing through the output transformer. This is bad for several reasons, and sever mismatches can seriously impact tone, power, and hum.

A better option is to provide separate bias control for each output tube (as in the Pro-Reverb and many other amps). I opted for a bias-balance control (P13) to shift the bias voltage provided by the level control (P14) between tubes. It works like stereo volume and balance controls. Bias is set to 15 ma for each tube. With 360v on the plates, that runs the tubes at ~5.4w each or ~45% rated dissipation. It sounds good to me and doesn’t stress the tubes. The bias-balance pot (P13) is visible on the top chassis image above, protruding between the two output tubes.

The network of bias resistors also provides inputs for bias-tremolo. A low frequency signal modulates both tube bias voltages to create the tremolo effect. Bias trem has a different (I’d say richer) tone than the stock DR optical trem, and both are available as switched options. In this design, the bias trem option impacts all signals from all sources that make it to the output stage, while optical trem only impacts ChB. Bias trem is used widely, for example in Vox AC30 and AC15 amps and in Fender designs like the Princeton and Pro Reverb.

SW7 provides two high freq cut options. If overdrive distortion or excessive brightness is created after the tone controls, it’s useful to have some way to tame these high frequencies. Vox amps, the Flexi-50, and others provide a “high cut” control at this stage. Some designs just put a fixed cap in this position. In a concession to limited control panel space, I used a three position switch. This network takes the edge off shrill or “ice pick” tones that are difficult to subdue with tone controls. The two “on” positions cut more or less high frequencies, and in the center-off position, the switch has no effect, preserving the standard DR bright spanky tone.

In almost all guitar amps, negative feedback (NFB) to a long-tailed phase inverter (PI) feeds the inverting input (through C30) and the long tail resistor R41 (which is a common mode input). I have R41 grounded rather than connected to R44 as the classic Fender design requires.

While this connection has some historic significance (see Belcowe’s discussion about this in his excellent second-edition preamp design book), it creates some undesirable circuit conditions. The classic connection of R41 to R44 creates a small but significant dc voltage at the connection point. That voltage feeds dc current through R62 and the output transformer secondary.

Why anyone would want that current to flow is beyond me. Any signal coupled to R41 is a common-mode signal so it has minimal impact on the PI outputs. I opted to break this connection by grounding R41 and keeping only R42 as the input for negative feedback from R62.

Two good things (IMO) result from this change. First, no dc currents flow in the negative feedback circuit and transformer secondary. Also, the NFB network is now independent of resistors in the PI tail. That means I can increase or decrease the NFB network impedance without impacting the PI tail current.

With that independence, I added the SW8 network to control the feedback path without creating clicks and pops and other undesired side-effects. I increased the NFB network resistances (R62, R38) 100-fold to allow for a convenient cap value (C28) to create a “presence” setting. SW8 settings can eliminate NFB (loose), reduce it at high frequencies (presence) or leave it alone (off). This switch is mounted on the rear chassis as a concession to limited front panel space. These resistors and cap can be tweaked for taste – without impact to the PI.

(See more details about long tail PI operation and optimization here.)

The output transformer drives speakers jacks, with the main jack wired to connect an 8 ohm resistor (25w) when no speaker is connected. The stock DR simply shorts the output to ground, which may be OK as far as amp survival is concerned, but it’s not difficult to add the 8 ohm dummy load and it allow the amp to operate more safely without a speaker. The load resistor is mounted outside the chassis – near the power transformer. The DR bias eyelet board was used for this purpose.

A balanced line output is added for recordings. With the dummy load, the amp can even be driven hard without a speaker, while the line out is used to record and/or monitor. T3 is a signal transformer that creates a balanced line out. I bought a balun sold for this purpose by Weber, but I had problems with it and eventually switched over an old mic-input transformer I had in a parts bin.

There is nothing special about this transformer, other than it has moderate (few K) input and output impedances. It’s important to keep impedances of this circuit relatively low to minimize hum pickup. After all, the transformer is in the amplifier chassis where strong hum and signal fields abound.

A capacitor (C35) is added to shape the output tone to taste. The image below shows the power supply and power amp sections completed and ready for test. The test points are visible on the back panel to the left of the power switches and the NFB control switch is to the right of the speaker jacks.

ChB Preamp

Channel B is basically the stock DR reverb channel. Slightly lower input resistors (R2,3) are used to reduce noise. These resistors are mounted on the input jacks and a shielded cable runs to the V4a grid.

The tone stack and volume control are standard, but two bright caps and a switch are added. A grid stopper (R8) is added in case high drive from pedals leads to overdrive of V4b.

Note that the output of V4a is used to drive ChA through C53. That means the guitar signal drives only one tube and we have gain and inversion before the signal gets to ChA. The DC filament supply to V4 makes this channel relatively hum-free.

To save circuit board eyelets and to reduce needless wire lengths that pick up hum and noise in the high impedance tone stack, the associated caps (C3,4,5) and resistors(R6,7) are mounted directly to the pots. That keeps exposed leads short and away from other circuit board signals. Bright caps are also mounted directly between the SW1 and P3.

The remaining circuitry for ChB takes up only a few eyelets on the circuit board, which leaves much-needed space for ChA. Note also that where possible, cathode bias/bypass networks (e.g., R5, C2) are built by soldering the caps directly across the resistor bodies, so lead lengths are short and only two leads need supporting eyelets – thereby also saving board space and eyelets.

Mixer Stage

The mixer stage controls the passage of the two preamp signals and the reverb signal to the power amp. SW5b mutes either or none of the preamp signals. That allows for A/B/A+B options. When ChA is selected, the ChB signal is muted. However, due to R11, the ChB signal still feeds the reverb section. In this mode, the ChB preamp does nothing but feed the reverb tank so the levels and tone controls allow a lot of flexibility in shaping the reverb tone.

The reverb output signal passes into the mixer via the P7 control where it is mixed with ChA signal. This arrangement means that ChB always provides reverb, regardless of what ChA is set to do. It means that reverb is a clean signal even when ChA is in high-gain (Fat) mode. This keeps reverb more useful with Fat ChA tones.

Mixing of ChA and ChB occurs in the resistor network around SW5a. This switch helps maintain consistent output levels between the A/B/A+B options. For the values shown, there is a slight volume increase with the A+B setting, but R34 and R36 can be tweaked to taste.

The master volume (P12) controls the mixed signal and passes it to the power amp. A small bright cap elevates the highs a bit at mid and low master settings.

Note that the ChA and ChB signals are in-phase when mixed. There are three inversions (V4a, V4b, and V6a) in the ChB path to the mix point. Likewise, there are three inversions in the ChA path created by V4a, V2b, and V2a.

Reverb

The Reverb section is unchanged from the stock DR, however it is driven directly from the anode of V4b, rather than the 3.3M resistor (R18). This allows SW5b to mute ChB without affecting the reverb input signal.

Reverb output is controlled by P7, an audio-taper pot. I find the gain of the reverb section a bit too high, so I may add a series resistor between C12 and P7 (~100k) next time I open the chassis. Other options would be to reduce R16 to about half its current value (47k), or double the value of R17 to 1M.

I think the Weber-supplied reverb tank sounds pretty good, but clearly it could be replaced to suit any taste. The footswitch is the stock DR circuit although I used chassis isolate jacks for both footswitches to reduce pops and noise.

Tremolo

The tremolo oscillator (V7a) and its speed control (P8) are unaltered. The oscillator signal drives the standard DR tremolo circuit (V7b) and the optical element (OPT1). The OPT1 element is a simply neon bulb placed next to a light-sensitive resistor. When the bulb lights, the resistance of the resistor element drops. That variable resistance feeds into P9a, which presents that variation, or some fraction of it, to the anode of V6a (through C26). The resistance variation in the optical element creates a variable signal level at the ChB input of the mixer stage.

To allow switchable tremolo (optical or bias), the connection to the variable resistance is switched by SW6 (upper pole). Opto trem is enabled by making the connection between P9a and C26. Optical trem is disabled (when bias trem is enabled) by switching C26 to a fixed resistance (R33). P9a and R33 are both ~50k so that the signal level at C26 is kept consistent.

The bias trem circuit uses the same oscillator signal. The bias trem input network to the 6V6 tubes is relatively low impedance, so R29 is used to isolate the oscillator from the bias network. The level of bias tremolo is controlled by P9b and the AC bias variations are coupled to the bias network by C24 and C25.

Note that P9a and P9b are a dual-pot control. This was difficult to implement since I don’t know anyone that builds dual pots with two different resistances. If they exist, I couldn’t find them.

The problem arises by the constraints of the OPT1 element. This element is selected to work with a 50K RA pot. RA (Reverse Audio taper) pots are relatively rare, and the likelyhood of finding a dual pot with an RA element is nil. I tried making an RA pot from a 250K linear pot (see The Secret Life of Pots), and it worked, but the behavior was not exactly what I’d hoped for.

I finally made a custom dual pot by opening the back of an Alpha dual 250K L pot and replacing the back element with one taken from a single Alpha 50K RA pot. It's fortunate that the Alpha pot components are interchangeable. It wasn’t that hard, and only took about 30 mins, but it required some dremel grinding and then a drop of superglue to hold the wiper on the ground-down shaft. In the end, I got exactly what I wanted - a dual control with one 50K RA element and a 250K L element.

The images below show the open rear pot element and black plastic wiper. The shaft that holds the wiper had to be ground down to remove the original element, but superglue easily holds the wiper in place after exchanging the element. The final assembled pot is shown with the C 50K label showing on the new rear element.

The footswitch control over the trem circuit copies the new Fender DR Reissue amp. A few extra parts makes for pop-free “soft-switching” of trem. The footswitch jack defaults to “tremolo enabled” when no switch is plugged in.

DR Tuning and ChA Design Overview

As the above sections were completed, I could play the amp and fine tune the various features I added. I fixed a few grounding issues and the amp was ready for “tuning”.

The “tuning” process requires extensive listening and playing while observing signals in all the stages. This process is invaluable to creating a “musical” amplifier. No numerical circuit analysis can replace time spent listening and observing the actual signals at different levels and playing conditions. This process leads to dozens of tweaks to values and configurations, resulting in the well behaved “musical” circuits shown in the posted schematics. This is, of course, completely subjective. Different people will "tune" with different results.

I should stress, the posted schematic is not where I started – it’s the end point reached through a process of considered design and careful construction, followed by critical tuning. Once the DR sections were all tuned and tweaked, I had a complete “extended DR” core. It sounded beautiful. It had all the character I hoped for and the new features worked as intended to extend the amp's range and capabilities.

Early in the design process, I focused on unusual, but musically-useful features. I think it was Blencowe who planted the idea in my head for shifting the scoop frequencies in a tone stack for musical effect. I’ve tweaked a lot of tone control sections, and I find that many options sound good, but usually I can implement only one. Why not allow a switched range of tone-control options? That gave rise to the idea of the scoop-shift circuit and switch. More about that later – but it was a commitment I made early since I put the switch on the front panel. Likewise, I wanted both clean and distorted tones from the ChA preamp. I also committed to the idea of the “Vivid / Fat” modes by putting the switch on the front panel. It was important and a challenge to me to make ChA support both of these features.

Now, I had to make ChA real and make it live up to expectations. I managed to leave a fair amount of eyelet-board space open so I could consider something “big” and unique.

The image below shows the complete DR core with all the enhancements during the initial testing and tuning. Many changed followed, but you can see that even with the mods and additions to the DR core, it fits into a relatively small area. The whole right-side of the chassis, including the bare eyelet board, ChA controls, and three tube sockets, is reserved for ChA.

I was intrigued by the idea of a cascode stage due to its dynamics. See Merlin Blencowe’s book and discussion and Randall Aiken’s post about this circuit. There are many hifi instances of cascode circuits (see Morgan Jones), but I could not find a commercial guitar amp circuit. (Ken’s Preamp is posted but I don’t know if it’s commercially available.)

The cascode is highly regarded by knowledgeable people, yet it’s uncommon in commercial guitar amps – these were both attractive qualities. In addition, the cascode circuit has a tube designed specifically for it, the ECC88 or 6922. While a standard tube in hifi applications, I could find no trace of its use in guitar amps – again, uniqueness made it attractive. Different tubes lead to different tone, particularly in overdrive.

The grid-leak bias cascode configuration described by Blencowe and used by Ken has dynamic compression behavior. I like the idea of a responsive circuit – one that behaves dynamically in response its input. Output stage screen-grid compression and power supply sag have long been desired musical amp characteristics. The perception of an amp “breathing” is due to a dynamic circuit response. The cascode offers dynamic circuit behavior in a preamp. I was compelled to explore this idea.

The fact that cascodes achieve high gain was also attractive since a two stage cascode preamp was sufficient to get clipping, especially since I had the output of V4a to use as input to ChA. Two inverting stages for ChA also matched phase with ChB, simplifying their combination in the mixing stage.

The high output impedance of a cascode was a concern since I needed to drive a tone stack. I had a spare tube socket available, and I also like the tone characteristics of a direct coupled cathode follower (DCCF). This lead naturally to the idea of coupling a DCCF to each cascode stage.

The DCCF provides a low impedance source to drive a tone stack, and it offers opportunities for “warming” the tone. A DCCF is a key element of the tone characteristics of the Bassman, Univalve, and countless other amplifiers. (See Blencowe’s discussion about this.) Since the stock DR doesn't include a DCCF stage, adding one extends the tone-scape of the DR package.

The configuration of ChA started to take shape. Two gain stages, each with cascode-DCCF, and a tone stack in the middle. One other thought occurred to me – to bootstrap the cascode stage in high gain (Fat) mode to get both added gain (in case I wanted it) and additional control over stage distortion characteristics. A bootstrapped stage distorts differently due to the gain (positive feedback) impacts of the bootstrap network.

Now I had a two-stage ChA design with many controls over each stage's distortion characteristics. I could vary the bias, gain, the bootstrap network, and the DCCF characteristics. These are a rich set of options for tuning the final tone. With so many variables, the tuning process for ChA took more time than the tuning of the entire DR core. I literally rebuilt ChA about three times to get it right. Fortunately, eyelet boards allow easy modification.

ChA Preamp

The input levels to ChA are controlled by SW4a. There is a 10x difference in input signal levels between clean (Vivid) and drive (Fat) modes. The added gain provided by V4a is essential for getting optimal levels in this stage.

I tuned the stage levels with a Telecaster. It’s a low output guitar and I set the first stage levels just short of clipping in Vivid mode while getting strong clipping in Fat mode.

Bias in the first stage (V2b) is set so that the anode of Va1 is ~100v or just under mid-supply voltage (240v). Due to the way the cascode works (see Blencowe’s discussion), the V2b anode voltage drops, while the V1a anode voltage rises as the input signal level increases. These voltage changes give rise to the dynamic cascode behavior I want to exploit. The changing anode voltages lead to dynamic stage gain and clipping characteristics.

The V1a anode voltage increases as much as 80-100v under high-drive. As the signal decays, so does the anode voltage, which means the stage distortion behavior responds to what’s played. Since the bias point of the stage is approx centered (given the anode voltage variations), the bootstrap signal and output signal have high unclipped headroom, which is utilized for Vivid mode.

When in Fat mode, increased input signal and stage gain lead to strong clipping. However, by using the guitar volume level, the stage can produce a smooth softly-clipped waveform for low-volume settings, while higher guitar volume levels produce a progressively asymmetric clip with a sharper tone.

The actual room loudness doesn't change much since the signal level is clipped for most of the guitar volume-setting range and the cascode stage has intrinsic compression due to the dynamics. So, the amount of gain and distortion varies dynamically and the circuit blends smoothly between gain and distortion levels in response to what's played.

All I'll say at this point is that ChA has a very different feel to it, (compared to ChB) which was one of the reasons for doing this project in the first place. (I'll have more to say about tone quality after I post some clips.)

The cathode bypass cap is small in Fat mode (C56) to cut bass to taste. Vivid mode switches in C57 with SW3c for full bass. SW3b disables bootstrap in vivid mode, and enables it for Fat mode. Bootstrap approximately doubles the gain of the stage. Vivid gain is measured at ~250 for this first stage. Careful tuning was required to control the stage levels and gains and layout care is needed to prevent oscillations.

The DCCF (V1b) has a small grid stopper (R90) and a relatively small cathode bias resistor (R92). These were tuned for tone since they affect clipping characteristics. Reducing R90, smooths the positive clipping transition, while reducing R92 extends and smooths the negative output swing before clipping. A grid-cathode protection network (D2, R91) is used for both V1 and V3.

At this point I need to mention the filament supply. V2 has a filament supply referenced to ground, while V1 and V3 need an elevated supply. Initially I used AC supplies for V1 and V2, but the high stage gain produced far too much hum.

The 6922 tube is spec’d at ~5-10 microamps of leakage current, but this turns into volts of hum at the output since V1 sees a high impedance at its cathode. A hum-balance pot helped, but not enough. So the first iteration of the circuit revealed that the filament supplies had to be modified into a DC supply.

The DC bridges and filter caps were added and the elevated supply was set to ~75v which is safely in the range of what V1 and V3 experience at their cathodes. Hum issues were basically solved by the DC supply. Note that V1b and V3b experience generally higher cathode voltages than V1a and V3a, and all of them are dynamically changing, so a compromise elevated level is needed.

A related digression is the presence of local bypass caps like C58 at various points in the amp. These provide some local high-frequency supply filtering and crosstalk reduction. They’re easily overlooked, but they make a significant difference, particularly with the high signal voltages seen in ChA. The output of the first stage (V1b) swings 140vpp while clipping in Fat mode. That signal can produce a lot of noise on the supply and ground lines and a local bypass cap helps reduce this noise and its propagation into other sections of the amp.

The ChA tone stack looks complicated but it’s a standard Fender circuit with switchable options. The scoop shift switch (SW4) moves the dip (scoop) in the frequency response higher or lower than the DR standard dip used in ChB. The shift is less than an octave, but it produces fairly dramatic results. The table below provides the values that result from the three different switch positions.

Tone stack switching

These values can be plugged in the Duncan TS simulator to get the following three curves that show the shift up and down in the scoop. The bass and treble controls were set at 5 (middle) in all cases to visualize the curves.

The idea is shift the dip to favor different pickups and guitar tones. From the curves it’s apparent that a low shift moves the dip almost an octave lower, thereby actually reducing the low frequencies (or cutting bass). Likewise a high-shift has the effect of reducing high-frequencies.

These shifts provide different tone palates that can’t be created by simply adjusting tone controls. Tone controls have little effect on the frequencies near the dip, which are in a perceptually important musical range. By shifting the dip, different frequency blends or "tone-colors" are possible.

After using the shift control for a while, I liken it to switching PUs on a guitar. The change is definitely significant - in fact the effect is as great or greater than actual PU switching. I often use the high-shift for bright bridge PUs, and low-shift for added bite from a high-gain tone - these produce unique "pickup" tones that are not accessible without the shift. The shift control is a unique extension to the amps control over tone. I liken it to a 3-way variation of each guitar pickup. So, think of the 2-pickup telecaster becoming a 6-tone (virtual pickup) instrument.

The tone stack feeds a divider to control the level presented to the second stage (V2a). SW3c works with R100,101 and P10 to provide a perceived consistent level between Vivid and Fat modes. It’s impossible to measure these signals in a perceptually accurate way, since one is usually clipped and one is not, so P10 is simply adjusted to make a Vivid tone sound as loud as a Fat tone. SW3 is switched back and forth while P10 is set. The small P10 pot on the circuit board is set to ~25K now, but it’s easily tweaked if needed. The ChA volume control and bright caps are the same as in ChB.

The second stage of ChA (V2a, V3) is similar in topology to the first stage, but its parameters are significantly different. With so many tone options in each stage to choose from, it’s hard to explore them all, and some guiding principle is needed. Mine was to make the second stage relatively unbalanced and therefore different from the first.

With that as a guiding tenet, the bias resistor (R105) was increased (relative to the first stage), to produce a lower current in V2a and V3a. Also the load resistors (R102, 103) are higher than in the first stage, and split unevenly. The result is higher gain and a lower DC voltages at the anodes of V2a and V3a. Although the cascode dynamics increase the V3a anode voltage, it is far from center bias, so the second stage headroom and clipping characteristics are quite different from the first stage. Note the V3a grid leak and "breathing" time-constant resistor (R104) is reduced slightly to make the dynamic variations track the input slightly faster than in the first stage.

The bootstrap network (SW3d) provides a full bootstrap signal for Vivid mode and mainly a high-frequency signal for Fat mode. That produces a slight bass cut in Fat mode. A high-value grid stopper (R107) is used for V3b to increase the distortion effects of the DCCF. A higher impedance, but split load (R109,110) is used for V3b to provide a full amplitude bootstrap signal, as well as a reduced signal to the mixer to match the output level from ChB.

Despite intentional bias imbalance, this stage produces 220 vpp at the DCCF output. It also has a measured Vivid-mode gain of ~637. A clean 37 vpp output is produced from 0.058 vpp @ 1KHz. This is largely due to the bootstrap circuit increasing gain by a factor of ~2.5X. Without the bootstrap, the gain is ~250.

Although this stage has lower headroom, the cascode dynamics produce progressively larger swings with progressively stronger clipping. Clipping starts early (~50 vpp) at the bottom of the waveform and the circuit voltages shifts by well over 100v at the V3a anode to allow larger swings under high drive.

Clipping is gentle at mid volume (~5) levels for both vivid and Fat modes. This stage is not intended to produce the Fat vs Vivid variations created by the first stage. Rather the second stage clips both Vivid and Fat mode signals, starting at moderate levels (volume = 5) and produces strong clipping for both modes at high volume (10) settings.

So, by setting ChA volume to 10, both Vivid and Fat modes produce lots of clipping in the second stage, and that’s a very different tone (darker and more grit) than produced by the first stage. Actual room volume in all cases is controlled by the Master (P12), so high-drive of either or both stages can be achieved at any room volume.

In summary, ChA really has 4 gain and tone modes. Vivid and Fat (with mid volume settings) and Vivid and Fat at high volume settings. All four are different due to the intended variations in the two ChA stages. And smooth blending between them is afforded by control over the guitar volume and ChA volume controls. There’s endless fun in exploring these options.

About Switching

There is a lot of switching in ChA (and elsewhere) in this design. The switching networks are carefully designed to minimize clicks and pops. I hate hum, but I hate clicks and pops more. No DC levels are switched anywhere. Any switched capacitors have DC leakage paths (except for bright caps since they are very small). Care is taken so that even when switches are in transition (between positions) no pops or other ill-effects occur.

The ground on SW5b must be a clean preamp ground or crosstalk will occur. The switch panel and circuit layout favors short wire paths, but any lengthy wires to switches (especially near the power supply or transformers) are connected with shielded cables.

The design uses common (low cost) multipole toggle switches rather than an array of relays. Clearly, if footswitch control is desired, a relay system could replace the toggle switches.

Conclusions

I hope you found this posting interesting and informative. I try to address a broad audience. This was a big project and I realize most readers are unlikely to pursue anything like this, HOWEVER, many pieces of this project are independent and can be used relatively easily and without permanent change (drilling holes, etc) to a DR or similar amp.

Forget ChA for the moment, some relatively simple changes to any DR include:

- improve bass response with increased C29 and increased R56,57.

- add a hi-cut switch or control (if you can spare the ground switch hole)

- add a second internal bias pot – to provide separate control for each tube.

- convert to bias trem – this isn’t hard if you abandon the opto trem.

- add an 8 ohm dummy load (mount it anywhere outside the chassis) for safe operation with a line out jack mounted in any chassis hole - it doesn't need to be on the back panel.

- add a master control (if you can find a hole for the control).

Any of these changes adds something useful to the stock DR capabilities and don’t alter anything permanently in case you want to revert to stock configuration in the future.

Since you made it this far, (congratulations) you probably have a DR or similar amp, so consider the idea of ChA-lite. The two stock DR channels are essentially the same, which seems like a waste of tube circuits to me. A ChA-lite can be created by rewiring the two triodes of the Normal channel into the same signal flow positions I used for my two ChA stages. Sure, they are simple triodes, not cascodes, but they can be fed from the Reverb channel input stage (like my ChA) to provide plenty of high gain possibilities.

Two cascaded triode stages have a gain of ~60x60=3600! Caps and resistors can be added or changed to tune the results. There is no need for switches since the Normal channel volume and tone controls become the high-gain channel controls. The Normal output can be mixed with the Reverb channel by using fixed resistors to a desired balance, and mixing is easy since the two channels are in the same phase.

Now you have a DR with a blendable high gain channel with separate volume and tone controls. No holes are drilled, and all the wiring changes can be undone, assuming a hand wired chassis. Although the changes can be made on newer DR versions with PCB construction, these amps are not easily altered without leaving permanent scars. I wouldn’t be surprised to hear from people that have done this already. If you’ve done it or tried something similar, I'd like to hear from you and get your comments on the results.

Final Pictures

Here are two chassis shots (the top one is high-res) and two full cab shots showing the final amp. At first glance, it looks like a simple DR... but when you look closer, you can see many differences. Of course, if you play it... the difference is clear... (sound clips below).

Sound Clips

The recording setup is simple, but differs from what I've usually done for amp clips on these pages. These clips are recorded with a pencil condenser mic - a NANDY CM-100. The amp is in a stand about 18" above ground, with a slight tilt so it faces a bit upwards. The mic is held about 12" away, and directly facing and centered on the speaker.

I tried a number of mic locations, but they all sound relatively dull. This position captures the closest to what I hear while I'm playing as long as I'm within about 45-degrees of the speaker axis. The bright "ping-y" attack of a clean note was the hardest to capture and very sensitive to mic placement. Playing these clips on a good hifi system sounds about right to me.

All the clips are recorded in one session, and to minimize the number of variables between clips, I use these settings on all clips: Bass = 5, Treble = 5, Reverb = 3, Speed = 7, Intensity = 3, Bias Tremolo is selected, No Boost (bright) is selected, No Cut is selected, and the rear panel NFB switch is set to Normal. The Master Volume is set around 2.5, but reduced slightly (~2) for the clips with ChA volume at 10 (the latter 3 clips) to keep the room and mic levels roughly constant.

I didn't touch the mic gain during the entire session and room volume was a comfortable practice level (for me), even with others in the house. The guitar is my Amer Stand Telecaster with its volume and tone at max. Forgive my playing nits - I'm trying for tone range and effects.

DP_ChB - Only ChB is enabled, so this is the Deluxe Reverb core by itself. Vol = 5. The clip starts with the neck PU and then I switch to the bridge PU. Sounds like a classic clean DR to me.

DP_ChA.v - Only ChA is enabled, Vol = 5, Mode = Vivid. The clip starts with the bridge PU and Shift = Norm. Then it switches to the neck PU with the three shift options; first Shift = NS, then HS, then LS. Then the bridge PU is selected and we switch shifts from HS, to LS, and wind up in NS. The final clip strums are with the neck PU and Shift = NS. ChA at low volumes (1-5) in vivid mode has a clean sound quality that is similar to ChB. However, there is some compression and the scoop-shift switch in ChA behaves much like a guitar PU switch. The ChA tele now effectively has 6 PUs - not two - for more tone range.

DP_ChA.f - Only ChA is enabled. VTB = 5, Mode = Fat. The clip starts with the neck PU with Shift = NS. I switch PUs and Shifts throughout the clip, but I can't recall the sequence. I just wanted to capture some variations. The low end sounds very "brown" to me - like an old Bassman, but high notes have nice harmonics and a bite to them. This is the tone produced with the first stage of ChA clipping.

DP_ChA.vmax - Only ChA is enabled. Vol = 10, TB = 5, Mode = Vivid. I switch PUs and Shifts throughout the clip to capture variations. Note we're in Vivid mode, so this is the sound of the second PhA stage clipping. It's still fairly clean, but it's not a sweet clean tone anymore. The tone varies - it's mellow when you lay off the attack, but it steps up nicely when you hit it hard. The bridge PU can get a nice ping and bite tone.

DP_ChA.fmax - Only ChA is enabled. Vol = 10, TB = 5, Mode = Fat. I switch PUs and Shifts throughout the clip to capture variations. This is Fat mode, so both ChA stages are clipping hard. This produces a more classic rock distortion quality.

DP_ChA+B.fmax - Both ChA and ChB is selected. All TB = 5, ChA vol = 10, ChB vol = 5. Two channels together produce a wider range than either alone. This is just one example of a mix that sounds good to me. I especially like the tone of the lick at ~0:40 secs into the clip. Remember, the amp is just purring along at room volume (MV = 2). There is no power amp distortion. These tones come only from switching between neck and bridge PUs and Scoop-Shift settings.

A last comment about tone clips... they're great but only part of the story. The clips can't convey the feel of the guitar - which is noticeably different to me in ChA, compared to ChBalone or my other amps. Its hard to know exactly why, but the ChA compression and distortion dynamics likely mimic the effects of playing a maxed-out amp with all that interaction of screen-current compression, supply sag, and stage clipping.

Of course you can get sustain and some nice feel from boost pedals too. However, ChA is more than that. It feels like something "soft" and "pliable" and "stretching". If I overdrive ChB with the MV down, I also get a nice distortion tone, but it's "hard" - it has a very consistent feel and tone - as do most amps and pedals at low volumes.

I like choice and variations. I enjoy flipping that A/B switch and getting something unique and different under my fingers. It's just fun to play - which is the whole point, as far as I'm concerned. Anyway, hope you get some info, ideas, and inspiration from all of this. Enjoy.

Since you got this far - check out Version 2 of this amp and the hunt for 5e3 tone.