Embedded Files
This project evolved around the tweed era 5e3 and 5f11 amps. As the project matured, it also took on ideas from Trainwreck and Deluxe Reverb designs. Every design has its inspirations. The challenge in this project was to fuse these ideas into a small footprint without compromising the essence of the inspirations.
This project evolved around the tweed era 5e3 and 5f11 amps. As the project matured, it also took on ideas from Trainwreck and Deluxe Reverb designs. Every design has its inspirations. The challenge in this project was to fuse these ideas into a small footprint without compromising the essence of the inspirations.
The project evolved incrementally. It was not planned as a complete amp and then executed. Sometimes incremental projects take unexpected twists and turns - and this is one of those.
The project evolved incrementally. It was not planned as a complete amp and then executed. Sometimes incremental projects take unexpected twists and turns - and this is one of those.
The project started with a 5e3 chassis and a damaged 5e3 cabinet. Those formed the physical constraints for the project and gave it an early "junkyard dog" flavor. It wasn't going to be a looker, but I wanted it to have a bite. The cab wound up with a dark "aged" finish simply because that hid all the defects. The tube compliment was also set as two 12ax7s, two 6V6s, and a 5Y3.
The project started with a 5e3 chassis and a damaged 5e3 cabinet. Those formed the physical constraints for the project and gave it an early "junkyard dog" flavor. It wasn't going to be a looker, but I wanted it to have a bite. The cab wound up with a dark "aged" finish simply because that hid all the defects. The tube compliment was also set as two 12ax7s, two 6V6s, and a 5Y3.
Next, I shopped for transformers. For reasons I don't fully recall I selected a Classictone - 40-18087 OT, which has a 6.6k primary and output taps at 4/8/16. I do recall thinking that multiple taps would give me design flexibility since I picked the OT before I knew exactly what I was building. I also picked a Classictone 40-18016 PT, which is rated for 120ma @ 660v ct. These are not a typical transformer pair for 6V6 amps, but they're both in the ballpark and rated for more than I expected to need.
Next, I shopped for transformers. For reasons I don't fully recall I selected a Classictone - 40-18087 OT, which has a 6.6k primary and output taps at 4/8/16. I do recall thinking that multiple taps would give me design flexibility since I picked the OT before I knew exactly what I was building. I also picked a Classictone 40-18016 PT, which is rated for 120ma @ 660v ct. These are not a typical transformer pair for 6V6 amps, but they're both in the ballpark and rated for more than I expected to need.
Borrowing the PI and output topology from the tweed amps, a 6V6 output stage is driven by a cathodyne PI, and an unbypassed gain stage applies NFB at its cathode. By switching NFB in/out, I had the essence of the 5e3 and 5f11 (as well as other tweed amps).
Borrowing the PI and output topology from the tweed amps, a 6V6 output stage is driven by a cathodyne PI, and an unbypassed gain stage applies NFB at its cathode. By switching NFB in/out, I had the essence of the 5e3 and 5f11 (as well as other tweed amps).
One 12ax7 was left for the preamp. This is where the Trainwreck ideas come in. The tone stack of the Express Kelly has two features I like. First it has a very low freq scoop (around 200Hz) and it has two bright cap options. I tried both in some breadboard experiments I liked them so they became the frequency response shaping core.
One 12ax7 was left for the preamp. This is where the Trainwreck ideas come in. The tone stack of the Express Kelly has two features I like. First it has a very low freq scoop (around 200Hz) and it has two bright cap options. I tried both in some breadboard experiments I liked them so they became the frequency response shaping core.
I like bias trem and a touch of spring reverb in my amps. These are Deluxe Reverb (or Princeton Reverb) features, but with no tubes left, these had to be done with SS circuits. Since both effects are essentially clean, SS is a good option and saves precious space in the small chassis I had to work with.
I like bias trem and a touch of spring reverb in my amps. These are Deluxe Reverb (or Princeton Reverb) features, but with no tubes left, these had to be done with SS circuits. Since both effects are essentially clean, SS is a good option and saves precious space in the small chassis I had to work with.
With Trem and Reverb, a typical tweed 2 or 3 knob control layout was not going to work. Instead, there are two rows of three small knobs (6 total) and 5 mini toggle switches for the front panel. Of course the front also has to have an input jack, power switch, fuse, and pilot light. Surprisingly, these all fit nicely into a standard 5e3 chassis by drilling a few new holes. A cover plate hides any excess and unused holes.
With Trem and Reverb, a typical tweed 2 or 3 knob control layout was not going to work. Instead, there are two rows of three small knobs (6 total) and 5 mini toggle switches for the front panel. Of course the front also has to have an input jack, power switch, fuse, and pilot light. Surprisingly, these all fit nicely into a standard 5e3 chassis by drilling a few new holes. A cover plate hides any excess and unused holes.
An FX jack is added on the back panel between the 12ax7s. It's perfect for an external reverb card - again saving space in the chassis.
An FX jack is added on the back panel between the 12ax7s. It's perfect for an external reverb card - again saving space in the chassis.
The main circuit stages were now settled and the controls were in place. A turret board was trimmed slightly and mounted on 1/4" standoffs. It's tempting to build real circuits, but first, it's time for Spice simulations to home in on signal levels, control operation, etc. It's very productive to do all the tweaks in Spice before building anything.
The main circuit stages were now settled and the controls were in place. A turret board was trimmed slightly and mounted on 1/4" standoffs. It's tempting to build real circuits, but first, it's time for Spice simulations to home in on signal levels, control operation, etc. It's very productive to do all the tweaks in Spice before building anything.
The design takes on concrete shape and form in Spice. That said, the weakness of Spice lies in modeling the output stage. The simulated anode and screen currents at high power operation are just approximations so I need to build the actual power supply and output stage and then tweak/test it to be sure it performs safely and sounds right. This is where problems and surprises await.
The design takes on concrete shape and form in Spice. That said, the weakness of Spice lies in modeling the output stage. The simulated anode and screen currents at high power operation are just approximations so I need to build the actual power supply and output stage and then tweak/test it to be sure it performs safely and sounds right. This is where problems and surprises await.
Before delving into the PS and output stage, a note about construction. I used a turret board, but the construction is really point-to-point as much as possible using the turrets for anchors as needed. Note that most components that connect to the tube sockets are actually soldered directly to the socket pins. This keeps leads *really* short for most of the signal path.
Before delving into the PS and output stage, a note about construction. I used a turret board, but the construction is really point-to-point as much as possible using the turrets for anchors as needed. Note that most components that connect to the tube sockets are actually soldered directly to the socket pins. This keeps leads *really* short for most of the signal path.
The phase inverter (V2b) is a tricky layout area since it's right next to the V1 circuits and the signal amplitudes are among the highest in the amp. Unless care is taken, unintended feedback will corrupt the amp tone. There are unavoidable long leads between the PI and the 6V6 grids, so the PI output leads are twisted together to minimize their signal radiation. The lines from the bias resistors (R20, R21) to the 6V6 grid stoppers are shielded to prevent radiation and stray pickup. The lines to and from the FX jack are also carefully routed to minimize pickup.
The phase inverter (V2b) is a tricky layout area since it's right next to the V1 circuits and the signal amplitudes are among the highest in the amp. Unless care is taken, unintended feedback will corrupt the amp tone. There are unavoidable long leads between the PI and the 6V6 grids, so the PI output leads are twisted together to minimize their signal radiation. The lines from the bias resistors (R20, R21) to the 6V6 grid stoppers are shielded to prevent radiation and stray pickup. The lines to and from the FX jack are also carefully routed to minimize pickup.
Update 5/4/19: The schematic is now v1.2 and the text below reflects this new version.
Update 5/4/19: The schematic is now v1.2 and the text below reflects this new version.
The power supply voltages are higher than usual for a 6V6 amp. The switchable tube/SS rectifier adds another 30-40 V to the supply (in SS mode). After several days of measuring and testing I settle on a solution that drops the screen supply almost 100 V using nine 10 V zener diodes. Fifty volts are dropped in the main supply chain, and another 40 V are dropped in the screen circuit only. The 6V6 anodes operate at ~440-470 volts, with the screens at ~350-370 volts.
The power supply voltages are higher than usual for a 6V6 amp. The switchable tube/SS rectifier adds another 30-40 V to the supply (in SS mode). After several days of measuring and testing I settle on a solution that drops the screen supply almost 100 V using nine 10 V zener diodes. Fifty volts are dropped in the main supply chain, and another 40 V are dropped in the screen circuit only. The 6V6 anodes operate at ~440-470 volts, with the screens at ~350-370 volts.
By using 10 V diodes I have flexibility in tweaking the screen voltage and power dissipation is spread among all the diode packages. At continuous clipping, the diodes are barely warm to the touch since they only carry preamp and screen currents.
By using 10 V diodes I have flexibility in tweaking the screen voltage and power dissipation is spread among all the diode packages. At continuous clipping, the diodes are barely warm to the touch since they only carry preamp and screen currents.
The combination of high anode voltage and low screen voltage puts the load line far below the knee in the 6V6 curves. Basically, the 6.6k OT primary impedance is too low. The original 4 ohm OT tap is therefore used for an 8 ohm load, thereby doubling the OT primary impedance to 13.2k.
The combination of high anode voltage and low screen voltage puts the load line far below the knee in the 6V6 curves. Basically, the 6.6k OT primary impedance is too low. The original 4 ohm OT tap is therefore used for an 8 ohm load, thereby doubling the OT primary impedance to 13.2k.
The process of finding the right screen voltage and primary impedance is an iterative one (in this case) which requires measuring 6V6 voltages and currents at clipping, while driving full power into an 8 ohm resistive load. In the end, the 6V6 anode clips (saturates) at ~50 V, which is approx in the knee of the plate curves, with cathode currents peaks of ~220ma. Allowing for ~60ma peaks of screen current, the 6V6 anodes are delivering ~ 160ma peaks. Actual speaker impedance is typically higher, so in actual operation, the load line will drop below the knee with slightly lower anode currents.
The process of finding the right screen voltage and primary impedance is an iterative one (in this case) which requires measuring 6V6 voltages and currents at clipping, while driving full power into an 8 ohm resistive load. In the end, the 6V6 anode clips (saturates) at ~50 V, which is approx in the knee of the plate curves, with cathode currents peaks of ~220ma. Allowing for ~60ma peaks of screen current, the 6V6 anodes are delivering ~ 160ma peaks. Actual speaker impedance is typically higher, so in actual operation, the load line will drop below the knee with slightly lower anode currents.
These clipping conditions provide ~25w RMS at clipping into an 8 ohm load. This performance is demanding of a 6V6GT and exceeds the published specs. However, such demands are not that unusual in 6V6 guitar amps and the JJ6V6S has a reputation for surviving over-spec 6V6GT conditions.
These clipping conditions provide ~25w RMS at clipping into an 8 ohm load. This performance is demanding of a 6V6GT and exceeds the published specs. However, such demands are not that unusual in 6V6 guitar amps and the JJ6V6S has a reputation for surviving over-spec 6V6GT conditions.
With tube/SS rectifier switching (S6), it's important to have separate bias controls (P6 and P7) for each rectifier option. Both bias controls are set to idle the tubes at ~20ma per tube.
With tube/SS rectifier switching (S6), it's important to have separate bias controls (P6 and P7) for each rectifier option. Both bias controls are set to idle the tubes at ~20ma per tube.
Since I expect to drive the output stage hard, I added an LED in the screen circuit (in series with R47). This LED brightness provides a visual indication of how much screen current is being consumed. It's just simple visual indicator that lets me know how hard I'm driving the tubes.
Since I expect to drive the output stage hard, I added an LED in the screen circuit (in series with R47). This LED brightness provides a visual indication of how much screen current is being consumed. It's just simple visual indicator that lets me know how hard I'm driving the tubes.
This LED feature is useful, but it adds a rectified screen current signal into both tubes - meaning it generates 2nd harmonics (and some higher ones). This was unintended, and contributed to some of the unique tone quality of the amp. It's a question of taste and after about a year of playing the amp that way I've decided to bypass the LED (C27) to remove the screen signal. The tone is more "classic" now - a bit cleaner, and with with more attack.
This LED feature is useful, but it adds a rectified screen current signal into both tubes - meaning it generates 2nd harmonics (and some higher ones). This was unintended, and contributed to some of the unique tone quality of the amp. It's a question of taste and after about a year of playing the amp that way I've decided to bypass the LED (C27) to remove the screen signal. The tone is more "classic" now - a bit cleaner, and with with more attack.
UPDATE 9/20/18
UPDATE 9/20/18
I recently found a wonderful tool for visualizing tube curves and load lines and other circuit parameters. I highly recommend you check it out: http://bmamps.com/ivds.html. My thanks to the author, Nick, for making this available.
I recently found a wonderful tool for visualizing tube curves and load lines and other circuit parameters. I highly recommend you check it out: http://bmamps.com/ivds.html. My thanks to the author, Nick, for making this available.
I set the tool for the parameters of the 6V6 output stage and the results are shown at the bottom of this page. The tool predicts operation that is very close to what I actually measured during testing and tweaking.
I set the tool for the parameters of the 6V6 output stage and the results are shown at the bottom of this page. The tool predicts operation that is very close to what I actually measured during testing and tweaking.
It took far longer to build, test, and tweak this output stage than it takes to simulate it - so there's the lesson. Simulation can save a lot of time and it provides a level of confidence that things are working as intended.
It took far longer to build, test, and tweak this output stage than it takes to simulate it - so there's the lesson. Simulation can save a lot of time and it provides a level of confidence that things are working as intended.
With the output stage completed and tested, the preamp and controls are built and tested. The preamp supply voltage is kept ~250 V, as in many tweed amps.
With the output stage completed and tested, the preamp and controls are built and tested. The preamp supply voltage is kept ~250 V, as in many tweed amps.
SW1 controls the tone stack scoop frequency. In the center-off position, the normal TW response puts the scoop ~200Hz. With a 1M resistor switched in, the scoop moves up to ~350Hz, and the 330K resistor moves it to ~500Hz, which is basically a Blackface response.
SW1 controls the tone stack scoop frequency. In the center-off position, the normal TW response puts the scoop ~200Hz. With a 1M resistor switched in, the scoop moves up to ~350Hz, and the 330K resistor moves it to ~500Hz, which is basically a Blackface response.
There is no mid control since there is no panel space for one. SW2 selects one of the two bright caps (or none). The V1b stage is unbypassed (as similar TW stages).
There is no mid control since there is no panel space for one. SW2 selects one of the two bright caps (or none). The V1b stage is unbypassed (as similar TW stages).
Just prior to J2 (the FX jack), SW3 provides two high-cut options (or none) to help control fret and string noise that gets excessive with high-treble boost settings. I find this control of HF roll-off to be vital. The treble control and bright caps all boost highs. Highs are vital to guitar tone, but there also has to be some control of high freq roll-off to prevent shrill or ice-pick tones.
Just prior to J2 (the FX jack), SW3 provides two high-cut options (or none) to help control fret and string noise that gets excessive with high-treble boost settings. I find this control of HF roll-off to be vital. The treble control and bright caps all boost highs. Highs are vital to guitar tone, but there also has to be some control of high freq roll-off to prevent shrill or ice-pick tones.
V2a is a pre-PI gain stage that also accepts the NFB signal from the OT. SW4 controls NFB on/off and adds some presence in the third position.
V2a is a pre-PI gain stage that also accepts the NFB signal from the OT. SW4 controls NFB on/off and adds some presence in the third position.
To boost signal amplitude in V2a and impart more coloration on the signal (by V2a), the PI stage (V2b) is driven by a voltage divider. The PI drives the output tubes through 1.5K grid stoppers. The output tubes have independent 1.5K screen stoppers, and share an ~800 ohm screen supply impedance. All of this helps control and limit screen current, which is vital for tube longevity.
To boost signal amplitude in V2a and impart more coloration on the signal (by V2a), the PI stage (V2b) is driven by a voltage divider. The PI drives the output tubes through 1.5K grid stoppers. The output tubes have independent 1.5K screen stoppers, and share an ~800 ohm screen supply impedance. All of this helps control and limit screen current, which is vital for tube longevity.
The SS tremolo oscillator modulates the output tube bias voltages, just as in Princeton or Deluxe Reverb circuits. A 24 V power supply is needed to power the external chassis reverb card. It also provides clean DC filament current for V1 and V2, ensuring low-noise preamp performance and reduces filament current drawn from the PT.
The SS tremolo oscillator modulates the output tube bias voltages, just as in Princeton or Deluxe Reverb circuits. A 24 V power supply is needed to power the external chassis reverb card. It also provides clean DC filament current for V1 and V2, ensuring low-noise preamp performance and reduces filament current drawn from the PT.
The reverb card and the 24V power supply are mounted on the outside of the chassis next to the transformers. Foil shielding ensures that any 24V PS switching noise is blocked and therefore inaudible. The card (5e3Verb at Amp-Aid) uses JFETs for low noise and a completely isolated signal path for the dry signal.
The reverb card and the 24V power supply are mounted on the outside of the chassis next to the transformers. Foil shielding ensures that any 24V PS switching noise is blocked and therefore inaudible. The card (5e3Verb at Amp-Aid) uses JFETs for low noise and a completely isolated signal path for the dry signal.
An 8 or 10 ohm pan is driven by a dedicated power amp on the card. The card could be hard wired, but I connect it to the FX jack with a TRS plug. The reverb control on the front panel is hard-wired to the card.
An 8 or 10 ohm pan is driven by a dedicated power amp on the card. The card could be hard wired, but I connect it to the FX jack with a TRS plug. The reverb control on the front panel is hard-wired to the card.
Most new projects become the favorite amp for a while. That's certainly true for this amp. Its size and tone fit my needs very well. It sounds great with the Weber Vintage (ceramic) speaker. For a long while I used a Celestion Gold, and most recently, I use a Weber Chicago 10" Alnico. The all sound good, but the Weber is the current favorite.
Most new projects become the favorite amp for a while. That's certainly true for this amp. Its size and tone fit my needs very well. It sounds great with the Weber Vintage (ceramic) speaker. For a long while I used a Celestion Gold, and most recently, I use a Weber Chicago 10" Alnico. The all sound good, but the Weber is the current favorite.
The amp has the sweetest highs of any amp I've played. It's almost impossible to set it for "too much" treble, it just stays clear and rings like a bell. Compression and transient response varies as expected with tube or SS rectifier selection.
The amp has the sweetest highs of any amp I've played. It's almost impossible to set it for "too much" treble, it just stays clear and rings like a bell. Compression and transient response varies as expected with tube or SS rectifier selection.
At clean (lower) power levels, the tube rectifier seems more responsive, while the SS setting gives better transients at higher OD levels. Either way, breakup is smooth. NFB smooths the freq response and tightens the breakup transition, as it should. The EQ options are wide and useful. I can dial in a tone I like/want in any room or mix. That's the main thing I want from amp controls.
At clean (lower) power levels, the tube rectifier seems more responsive, while the SS setting gives better transients at higher OD levels. Either way, breakup is smooth. NFB smooths the freq response and tightens the breakup transition, as it should. The EQ options are wide and useful. I can dial in a tone I like/want in any room or mix. That's the main thing I want from amp controls.
The screen capture below is from the simulation system posted at: http://bmamps.com/ivds.html
The screen capture below is from the simulation system posted at: http://bmamps.com/ivds.html
It simulates the output stage and it agrees closely with actual amp measurements.
It simulates the output stage and it agrees closely with actual amp measurements.
Page updated
Google Sites
Report abuse