Embedded Files
I obtained a used Flexi-50 some 5+ years ago and I've always been impressed by its build and tone. It's a beast, mechanically. Between tube swaps and the front panel controls, the range of tone is impressive. That said, it's always had a peaky and hairy top end that is part of the signature THD sound. Since I now mostly play the amps I build, the Flexi has only seen occasional recent use. Whenever I do play it, I'm often left with a sense that I'll get into it someday and figure out how to tone down the treble and hair just a bit. Well, the day has come and I'll describe what I found and what I did about it.
I obtained a used Flexi-50 some 5+ years ago and I've always been impressed by its build and tone. It's a beast, mechanically. Between tube swaps and the front panel controls, the range of tone is impressive. That said, it's always had a peaky and hairy top end that is part of the signature THD sound. Since I now mostly play the amps I build, the Flexi has only seen occasional recent use. Whenever I do play it, I'm often left with a sense that I'll get into it someday and figure out how to tone down the treble and hair just a bit. Well, the day has come and I'll describe what I found and what I did about it.
The amp I have is PCB Version 1_1. The schematic can be found online here:
The amp I have is PCB Version 1_1. The schematic can be found online here:
A later Version 1_21, has a few changes in the power tube bias and master volume circuits. Nothing I've done impacts or is impacted by those changes, so what I describe below applies to both versions.
A later Version 1_21, has a few changes in the power tube bias and master volume circuits. Nothing I've done impacts or is impacted by those changes, so what I describe below applies to both versions.
I'll describe two mods. One to reduce treble and some of the hairy distortion character, and one that fixes the FX loop, since the stock loop doesn't work. Before I start with mods, it's worth some time to review the first stages of the preamp. The Flexi is not simply a cloned Fender or Marshall circuit. Its unique aspects are not described anywhere in manuals or online descriptions I've found, so I'm making some comments here to help people appreciate what it is. Whether you like it or not, the Flexi is technically a rebel - pursuing some unconventional ideas in amp design.
I'll describe two mods. One to reduce treble and some of the hairy distortion character, and one that fixes the FX loop, since the stock loop doesn't work. Before I start with mods, it's worth some time to review the first stages of the preamp. The Flexi is not simply a cloned Fender or Marshall circuit. Its unique aspects are not described anywhere in manuals or online descriptions I've found, so I'm making some comments here to help people appreciate what it is. Whether you like it or not, the Flexi is technically a rebel - pursuing some unconventional ideas in amp design.
I'll start with the first stage of the preamp. I didn't modify this stage, but it has behaviors that few people seem to understand. The Flexi first stage (V1a) is grid leak biased. This is kind of cool and unique since modern amps rarely use this configuration.
I'll start with the first stage of the preamp. I didn't modify this stage, but it has behaviors that few people seem to understand. The Flexi first stage (V1a) is grid leak biased. This is kind of cool and unique since modern amps rarely use this configuration.
An argument against its use is that it's not stable and tube type and tube age can impact the bias and therefore the tone qualities of the stage. A counter-argument for grid leak bias is that the tube type and tube age can impact the bias and therefore the tone qualities of the stage. It's a case of different people seeing this through different lenses (like tone in general). Frankly, I like the fact that tube changes can make a difference and I enjoy trying different tubes. Although tubes differ, I have no trouble finding tubes that sound good to me. For the tests and the tweaking I've done, I use a 5751 as V1. It has a slightly lower gain than a 12ax7, but otherwise they're very similar.
An argument against its use is that it's not stable and tube type and tube age can impact the bias and therefore the tone qualities of the stage. A counter-argument for grid leak bias is that the tube type and tube age can impact the bias and therefore the tone qualities of the stage. It's a case of different people seeing this through different lenses (like tone in general). Frankly, I like the fact that tube changes can make a difference and I enjoy trying different tubes. Although tubes differ, I have no trouble finding tubes that sound good to me. For the tests and the tweaking I've done, I use a 5751 as V1. It has a slightly lower gain than a 12ax7, but otherwise they're very similar.
The V1a stage is pretty clean for input signals around 200mV peak. That's the level I use to characterize a telecaster pickup. A G&L puts out more like 300mV peak, and hot humbuckers will output even more. A grid leak bias is fairly small (again, depending on the specific tube you have) so the larger your pickup output, the more variation you will hear from tube to tube. There is no "right" answer to this. Just try different tubes with your favorite guitar until you find something you like. My 5751 tube produces a stage gain of about 40 from V1a. With the guitar volume at max, the 200mV pickup signal becomes about 8V peak of signal at the V1a plate.
The V1a stage is pretty clean for input signals around 200mV peak. That's the level I use to characterize a telecaster pickup. A G&L puts out more like 300mV peak, and hot humbuckers will output even more. A grid leak bias is fairly small (again, depending on the specific tube you have) so the larger your pickup output, the more variation you will hear from tube to tube. There is no "right" answer to this. Just try different tubes with your favorite guitar until you find something you like. My 5751 tube produces a stage gain of about 40 from V1a. With the guitar volume at max, the 200mV pickup signal becomes about 8V peak of signal at the V1a plate.
Something worth noting is that the first stage also has a uncommonly high input capacitance. C3 is a 22pf capacitor, which with a stage gain of 40 becomes equivalent to a 880pf capacitor to ground. Add to that the approximately 70pf of internal plate to grid capacitance and you get a total of ~950pf from the grid to ground. The 68k input resistor (R1) and the 950pf create a high frequency (HF) roll-off at about 2.5kHz. No one will argue the Flexi has too little high-end, so clearly that grid capacitance and HF roll-off are not an issue.
Something worth noting is that the first stage also has a uncommonly high input capacitance. C3 is a 22pf capacitor, which with a stage gain of 40 becomes equivalent to a 880pf capacitor to ground. Add to that the approximately 70pf of internal plate to grid capacitance and you get a total of ~950pf from the grid to ground. The 68k input resistor (R1) and the 950pf create a high frequency (HF) roll-off at about 2.5kHz. No one will argue the Flexi has too little high-end, so clearly that grid capacitance and HF roll-off are not an issue.
I mention it here because it can become an issue if you dial down your guitar volume. When you turn down your guitar volume pot, it effectively inserts additional resistance into the grid circuit. If the guitar volume pot adds 68K, which is just a slight turn of most volume pots, the roll-off frequency halves to under 1.5K. You can easily hear the impact of that. So, whether it was intentional or not (I suspect it was), the Flexi input stage is highly sensitive to the guitar volume pot or any other series resistance in the input circuit (like a pedal output pot). If you know this is happening (and now you do) you can use this behavior to your advantage, since it gives you another treble control at the very input of the amp. This can be really useful for taming overdrive/distortion from pedals that have output volume controls. Using a 12ax7 as V1 produces higher gain, which means the impact of the 22pf cap is increased and the HF roll-off occurs at an even lower frequency.
I mention it here because it can become an issue if you dial down your guitar volume. When you turn down your guitar volume pot, it effectively inserts additional resistance into the grid circuit. If the guitar volume pot adds 68K, which is just a slight turn of most volume pots, the roll-off frequency halves to under 1.5K. You can easily hear the impact of that. So, whether it was intentional or not (I suspect it was), the Flexi input stage is highly sensitive to the guitar volume pot or any other series resistance in the input circuit (like a pedal output pot). If you know this is happening (and now you do) you can use this behavior to your advantage, since it gives you another treble control at the very input of the amp. This can be really useful for taming overdrive/distortion from pedals that have output volume controls. Using a 12ax7 as V1 produces higher gain, which means the impact of the 22pf cap is increased and the HF roll-off occurs at an even lower frequency.
After V1a, there is a network of resistors and capacitors that work with the Input Sensitivity switch (SW1). This network boosts treble and controls how much signal gets to the volume pot (POT1) and the second stage (V1b).
After V1a, there is a network of resistors and capacitors that work with the Input Sensitivity switch (SW1). This network boosts treble and controls how much signal gets to the volume pot (POT1) and the second stage (V1b).
With sensitivity (SW1) set to Hi, the relay R1 closes and shorts out R6 and C6. That means C7 works as a classic "bright cap" bypassing highs around the volume control (POT1). When volume is at max, the bright cap has no effect, so there is no high-boost. As the volume setting decreases, the bright cap adds more highs to the signal at the grid of V1b. To reduce treble boost, you can turn up the volume pot.
With sensitivity (SW1) set to Hi, the relay R1 closes and shorts out R6 and C6. That means C7 works as a classic "bright cap" bypassing highs around the volume control (POT1). When volume is at max, the bright cap has no effect, so there is no high-boost. As the volume setting decreases, the bright cap adds more highs to the signal at the grid of V1b. To reduce treble boost, you can turn up the volume pot.
With SW1 set to Lo, the R6, C6 network reduces the signal level to roughly 1/3, but adds a small treble boost (a shelf at ~2.6kHz). The effect of this added treble boost also depends on the volume setting. In general, maximum treble boost is obtained with SW1 in the Hi position, and the Lo setting produces lower signal levels with less boost. Since C7 is the main contributor to treble boost in the first stages of the Flexi, that's the effect I chose to minimize in the mod below.
With SW1 set to Lo, the R6, C6 network reduces the signal level to roughly 1/3, but adds a small treble boost (a shelf at ~2.6kHz). The effect of this added treble boost also depends on the volume setting. In general, maximum treble boost is obtained with SW1 in the Hi position, and the Lo setting produces lower signal levels with less boost. Since C7 is the main contributor to treble boost in the first stages of the Flexi, that's the effect I chose to minimize in the mod below.
MOD1:
MOD1:
The mod to cut treble boost is simple. A resistor in series with C7 will shelve or limit the treble boost C7 produces. The Flexi PC board is hard to remove, so unsoldering C7 is not the best option. Fortunately, the Vol pot (POT1) and the V1 socket have exposed wire connections above the PCB. The PC board connects C7 across the exposed connections to POT1. The mod will leave C7 on the PC board, but route a separate connection from the POT1 wiper to the grid (pin 7) of V1b. The original wire from the PCB to the POT1 wiper is where we add a series resistor, effectively in series between C7 and the POT1 wiper.
The mod to cut treble boost is simple. A resistor in series with C7 will shelve or limit the treble boost C7 produces. The Flexi PC board is hard to remove, so unsoldering C7 is not the best option. Fortunately, the Vol pot (POT1) and the V1 socket have exposed wire connections above the PCB. The PC board connects C7 across the exposed connections to POT1. The mod will leave C7 on the PC board, but route a separate connection from the POT1 wiper to the grid (pin 7) of V1b. The original wire from the PCB to the POT1 wiper is where we add a series resistor, effectively in series between C7 and the POT1 wiper.
The process is simple. First disconnect the wire from the POT1 wiper (center lug) that goes to the PCB. Leave the PCB end connected. Then disconnect the short bare wire from the PCB to V1 pin 7. You can just clip out the bare wire, or desolder it. Now run a shielded cable from the POT1 wiper to pin 7 of the V1 socket.
The process is simple. First disconnect the wire from the POT1 wiper (center lug) that goes to the PCB. Leave the PCB end connected. Then disconnect the short bare wire from the PCB to V1 pin 7. You can just clip out the bare wire, or desolder it. Now run a shielded cable from the POT1 wiper to pin 7 of the V1 socket.
I suggest tying the shield to the ground side of POT1 (see pics below). The cable shield at the V1 socket doesn't connect to anything, but it can safely be soldered to the center pin (unused) of the V1 socket to provide mechanical support for that end of the cable. Now both ends of the cable are rigidly held to stabilize the connections against vibration.
I suggest tying the shield to the ground side of POT1 (see pics below). The cable shield at the V1 socket doesn't connect to anything, but it can safely be soldered to the center pin (unused) of the V1 socket to provide mechanical support for that end of the cable. Now both ends of the cable are rigidly held to stabilize the connections against vibration.
Lastly, attach a resistor to the floating wire end that used to connect to the POT1 wiper. Solder the other end of the resistor to the POT1 wiper (where the shielded cable also connects).
Lastly, attach a resistor to the floating wire end that used to connect to the POT1 wiper. Solder the other end of the resistor to the POT1 wiper (where the shielded cable also connects).
The question you're asking is "what value is that resistor?" The answer is "it's up to you." This resistor reduces the bright cap action - it's still a bright cap, just less so. The higher the resistor value, the lower the bright cap boost, the lower the resistor value, the greater the boost.
The question you're asking is "what value is that resistor?" The answer is "it's up to you." This resistor reduces the bright cap action - it's still a bright cap, just less so. The higher the resistor value, the lower the bright cap boost, the lower the resistor value, the greater the boost.
Consider the extremes. A zero ohm resistor is the case where you simply connect the wire back to the POT1 wiper, and you'd be back to stock conditions. If you used an infinite resistor, which is like leaving the connection open, you'd effectively remove C7 from the circuit and there would be no bright cap at all. While either of these options is perfectly OK, by using a resistor, you can obtain varying degrees of bright cap boost.
Consider the extremes. A zero ohm resistor is the case where you simply connect the wire back to the POT1 wiper, and you'd be back to stock conditions. If you used an infinite resistor, which is like leaving the connection open, you'd effectively remove C7 from the circuit and there would be no bright cap at all. While either of these options is perfectly OK, by using a resistor, you can obtain varying degrees of bright cap boost.
I've settled on a 1.2M resistor. That value gives me just a trace of bright cap boost - probably about 20% of the original boost. Values around 680K provide about 40% of the boost. Lower resistance values provide even more boost. You're free to experiment to find the degree of boost you like. While there are dangerous voltages in the amp - and you need to be careful of what you touch - there are no high voltages on the resistor connections we're discussing, so you can use clip leads to try different resistor values while the amp is on, without danger. Another option is to clip in a 1M pot and vary the settings until you get the boost you like. Then measure that pots resistance and hard wire a close value resistor.
I've settled on a 1.2M resistor. That value gives me just a trace of bright cap boost - probably about 20% of the original boost. Values around 680K provide about 40% of the boost. Lower resistance values provide even more boost. You're free to experiment to find the degree of boost you like. While there are dangerous voltages in the amp - and you need to be careful of what you touch - there are no high voltages on the resistor connections we're discussing, so you can use clip leads to try different resistor values while the amp is on, without danger. Another option is to clip in a 1M pot and vary the settings until you get the boost you like. Then measure that pots resistance and hard wire a close value resistor.
Lastly, I want to talk about the interaction of the treble boost circuits and the second stage of the Flexi preamp. The bias on V1b is very low (~.9V) due to the low value of R8. That means that clipping starts for any signals over ~0.8V peak (depending slightly on what tube used as V1). A 200mV input produces about 8V at the output of V1a, so visible clipping occurs once the volume (POT1) is set to ~ 2pm for Lo sensitivity or ~12pm for Hi sensitivity. That means the preamp remains clean only for volume settings below those.
Lastly, I want to talk about the interaction of the treble boost circuits and the second stage of the Flexi preamp. The bias on V1b is very low (~.9V) due to the low value of R8. That means that clipping starts for any signals over ~0.8V peak (depending slightly on what tube used as V1). A 200mV input produces about 8V at the output of V1a, so visible clipping occurs once the volume (POT1) is set to ~ 2pm for Lo sensitivity or ~12pm for Hi sensitivity. That means the preamp remains clean only for volume settings below those.
Unfortunately, the C7 bright cap boost increases at lower volume settings. That makes the clean range of volume settings very bright. Using a Hi sensitivity setting boost highs by increasing the effect of C7 and forcing a lower volume setting to keep the preamp clean. Basically, it's impossible to get low distortion (clean tone) in the stock amp in the Hi sensitivity setting without also getting a lot of treble boost. That's where the above mod can really help most. With the added resistor, the lower volume settings needed to keep distortion levels down, can be used without adding too much treble boost. Some boost is usually good, the mod lets you decide how much you want.
Unfortunately, the C7 bright cap boost increases at lower volume settings. That makes the clean range of volume settings very bright. Using a Hi sensitivity setting boost highs by increasing the effect of C7 and forcing a lower volume setting to keep the preamp clean. Basically, it's impossible to get low distortion (clean tone) in the stock amp in the Hi sensitivity setting without also getting a lot of treble boost. That's where the above mod can really help most. With the added resistor, the lower volume settings needed to keep distortion levels down, can be used without adding too much treble boost. Some boost is usually good, the mod lets you decide how much you want.
Hopefully this discussion helped you understand how the guitar volume, Sensitivity switch, and amp volume impact tone. This is a bit more complex than the way most amp input stages work. It can be harder to dial in settings that you like since the settings interact. Now that I took the time to understand how they work, I find it easier to reach my tone targets.
Hopefully this discussion helped you understand how the guitar volume, Sensitivity switch, and amp volume impact tone. This is a bit more complex than the way most amp input stages work. It can be harder to dial in settings that you like since the settings interact. Now that I took the time to understand how they work, I find it easier to reach my tone targets.
MOD2:
MOD2:
The FX loop in the Flexi is simply a break in the signal path between the tone stack and V2b. The Return jack is a shorting type that simply passes the signal along when nothing is plugged in. That's a simple FX loop, but it's OK and I don't see any reason to go over the top to improve things. Normally, it's advisable to buffer the Send output so that lower impedance loads (some solid-state effects systems) don't load down the tone stack. As long as you plug in high-impedance effects systems (>1M impedance), the effect is minimal. If you're curious about the loading impact of your effects unit, just plug it's input into the Send jack (without plugging anything into the Return) and listen for any difference. You'll hear a drop in Flexi volume if the effects unit is loading the Send signal.
The FX loop in the Flexi is simply a break in the signal path between the tone stack and V2b. The Return jack is a shorting type that simply passes the signal along when nothing is plugged in. That's a simple FX loop, but it's OK and I don't see any reason to go over the top to improve things. Normally, it's advisable to buffer the Send output so that lower impedance loads (some solid-state effects systems) don't load down the tone stack. As long as you plug in high-impedance effects systems (>1M impedance), the effect is minimal. If you're curious about the loading impact of your effects unit, just plug it's input into the Send jack (without plugging anything into the Return) and listen for any difference. You'll hear a drop in Flexi volume if the effects unit is loading the Send signal.
The bigger issue with the FX loop is related to the Return jack. If you plug an effects unit output into the Return jack, you break the connection between the tone stack and the grid of V2b. While that's exactly what should happen, there is a slightly horrible side effect, since the V2b grid depends on the return jack connection for its DC grid leak path.
The bigger issue with the FX loop is related to the Return jack. If you plug an effects unit output into the Return jack, you break the connection between the tone stack and the grid of V2b. While that's exactly what should happen, there is a slightly horrible side effect, since the V2b grid depends on the return jack connection for its DC grid leak path.
To work properly, a tube grid needs a DC path to ground (or some fixed voltage). Normally, the V2b grid leak path snakes through the Return jack, R13, R12, the Bass pot, and R11 (for a total of 540k - 1.54M, depending on the Bass setting). That works as long as that path exists. When you plug a cable into the Return jack, you're breaking that path. The V2b grid then depends on the external effects unit for a path to ground. Most effects units will drive their output through an output capacitor, which provides no DC path. So, the V2b grid floats up slowly, saturating V2b and its gain drops very low or it ceases to pass signal altogether. Basically, most effects units will cause the Flexi to appear to die. At best, the output is low and distorted.
To work properly, a tube grid needs a DC path to ground (or some fixed voltage). Normally, the V2b grid leak path snakes through the Return jack, R13, R12, the Bass pot, and R11 (for a total of 540k - 1.54M, depending on the Bass setting). That works as long as that path exists. When you plug a cable into the Return jack, you're breaking that path. The V2b grid then depends on the external effects unit for a path to ground. Most effects units will drive their output through an output capacitor, which provides no DC path. So, the V2b grid floats up slowly, saturating V2b and its gain drops very low or it ceases to pass signal altogether. Basically, most effects units will cause the Flexi to appear to die. At best, the output is low and distorted.
Fortunately, the solution is simple. A hard-wired solution is to add a resistor from the V2b grid to ground. A high value resistor prevents loading of the tone stack. The resistor value must also be low enough to avoid impacting the bias of V2b. I used a 2.2M resistor, which works with a 12ax7 as V2. A 1.5M resistor may be better to avoid bias problems when 12at7 or other tubes are used as V2. I tied the resistor from the PCB connection of V2 pin7 to the ground end of R18 (the end near R16 on the PCB). My 2.2M jumps over C19 as shown in the photos below.
Fortunately, the solution is simple. A hard-wired solution is to add a resistor from the V2b grid to ground. A high value resistor prevents loading of the tone stack. The resistor value must also be low enough to avoid impacting the bias of V2b. I used a 2.2M resistor, which works with a 12ax7 as V2. A 1.5M resistor may be better to avoid bias problems when 12at7 or other tubes are used as V2. I tied the resistor from the PCB connection of V2 pin7 to the ground end of R18 (the end near R16 on the PCB). My 2.2M jumps over C19 as shown in the photos below.
An alternative to the hard-wired solution is to create a special cable for use as the Return signal. Add a 1M resistor from tip to ground under the screw-on shell of a phone plug. Mark that end and always plug that end into the Return jack. The resistor provides the needed DC path without modifying the Flexi.
An alternative to the hard-wired solution is to create a special cable for use as the Return signal. Add a 1M resistor from tip to ground under the screw-on shell of a phone plug. Mark that end and always plug that end into the Return jack. The resistor provides the needed DC path without modifying the Flexi.
These two simple mods address the two problems I've described. They are completely reversible if you want to undo them. To me the impacts of the mods are simple... I enjoy playing the amp more. The amp controls have more useful range. I have a Celestion Creamback in a pine cab and it sounds great with the mod'd Flexi. A NanoVerb adds a bit of reverb and echo now that the FX loop is working. I'm already thinking about building a dedicated spring reverb into the pine cab since that worked so well for the P-AOT combo.
These two simple mods address the two problems I've described. They are completely reversible if you want to undo them. To me the impacts of the mods are simple... I enjoy playing the amp more. The amp controls have more useful range. I have a Celestion Creamback in a pine cab and it sounds great with the mod'd Flexi. A NanoVerb adds a bit of reverb and echo now that the FX loop is working. I'm already thinking about building a dedicated spring reverb into the pine cab since that worked so well for the P-AOT combo.
Here are some pics showing the mods.
Here are some pics showing the mods.
The above picture shows the shielded cable connection to V1 pin 7.
The above picture shows the shielded cable connection to V1 pin 7.
The shield is soldered to the center pin of the socket to provide support for the cable.
The shield is soldered to the center pin of the socket to provide support for the cable.
The above image shows the volume control (POT1) with the shielded cable connection to the wiper (center lug). The cable shield is soldered to the POT1 ground connection (the lug with the white wire).
The above image shows the volume control (POT1) with the shielded cable connection to the wiper (center lug). The cable shield is soldered to the POT1 ground connection (the lug with the white wire).
A 1.2M resistor is connected between the POT1 wiper and the center wire (of the three wires to POT1). That's the resistor that controls how much treble boost is created by C7.
A 1.2M resistor is connected between the POT1 wiper and the center wire (of the three wires to POT1). That's the resistor that controls how much treble boost is created by C7.
The image above shows the 2.2M grid leak resistor added between the grid of V2b (pin 7) and the grounded end of R18. The resistor crosses over C19 since that was convenient. This resistor isn't needed if you add a resistor to the Return cable.
The image above shows the 2.2M grid leak resistor added between the grid of V2b (pin 7) and the grounded end of R18. The resistor crosses over C19 since that was convenient. This resistor isn't needed if you add a resistor to the Return cable.
Update 2/8/16: Flexi-Verb
Update 2/8/16: Flexi-Verb
Well, it didn't take long before I tired of the Nanoverb again. To be fair, the Nanoverb is only part of the problem. The Flexi FX loop has very high signal levels (>5V peak), which most FX units have problems with. Regardless, I decided to build a dedicated spring reverb for the Flexi (a Flexi-Verb) based on the design for the P-AOT reverb.
Well, it didn't take long before I tired of the Nanoverb again. To be fair, the Nanoverb is only part of the problem. The Flexi FX loop has very high signal levels (>5V peak), which most FX units have problems with. Regardless, I decided to build a dedicated spring reverb for the Flexi (a Flexi-Verb) based on the design for the P-AOT reverb.
The Flexi has a serial effects loop, so whatever you insert in the loop potentially alters the dry signal, even if the effects mix is 100% dry. The Flexi FX loop takes the output of the tone stack and feeds it to the grid of V2b. In designing an FX unit, two issues need to be considered: 1) the interaction of the FX unit input with the tone stack that drives it, and 2) the interaction of the FX unit output with the V2b grid impedance.
The Flexi has a serial effects loop, so whatever you insert in the loop potentially alters the dry signal, even if the effects mix is 100% dry. The Flexi FX loop takes the output of the tone stack and feeds it to the grid of V2b. In designing an FX unit, two issues need to be considered: 1) the interaction of the FX unit input with the tone stack that drives it, and 2) the interaction of the FX unit output with the V2b grid impedance.
Ideally, with the FX unit plugged in and set to 100% dry signal, the V2b grid signal is identical to the signal with the FX unit unplugged. This may sound simple, but it's not. The output impedance of the tone stack interacts with the input impedance of the V2b grid. To preserve the dry signal tone qualities, that interaction between the tone stack and the V2b grid has to be preserved, but with the FX unit in the signal path.
Ideally, with the FX unit plugged in and set to 100% dry signal, the V2b grid signal is identical to the signal with the FX unit unplugged. This may sound simple, but it's not. The output impedance of the tone stack interacts with the input impedance of the V2b grid. To preserve the dry signal tone qualities, that interaction between the tone stack and the V2b grid has to be preserved, but with the FX unit in the signal path.
There are essentially two ways to approach the problem. First consider the interaction from the tone stack perspective. Breaking the FX loop (by plugging into the Return jack) leaves the tone stack feeding only the Send jack, which is the FX unit input. To mimic the normal operation of the tone stack, the FX unit should have an input impedance that is similar to the input impedance of the V2b grid.
There are essentially two ways to approach the problem. First consider the interaction from the tone stack perspective. Breaking the FX loop (by plugging into the Return jack) leaves the tone stack feeding only the Send jack, which is the FX unit input. To mimic the normal operation of the tone stack, the FX unit should have an input impedance that is similar to the input impedance of the V2b grid.
The V2b grid is operating with a cathode bias of about 3.37V, created by R18. (R18, is incorrectly shown as 47K on the schematics. The actual resistor in the circuit is 4.7K.) With that bias, the grid impedance is many 10's of Meg Ohms. However, C21 and R17 provide some negative feedback to the V2b grid, which is magnified by the stage gain (similar to the negative feedback of the input stage). Using a ball park gain of 50 for the stage, the grid impedance of V2b becomes 135K in series with 11nF. This relatively low grid impedance is the load seen by the tone stack, and it's effect is noticeable when the FX loop is broken.
The V2b grid is operating with a cathode bias of about 3.37V, created by R18. (R18, is incorrectly shown as 47K on the schematics. The actual resistor in the circuit is 4.7K.) With that bias, the grid impedance is many 10's of Meg Ohms. However, C21 and R17 provide some negative feedback to the V2b grid, which is magnified by the stage gain (similar to the negative feedback of the input stage). Using a ball park gain of 50 for the stage, the grid impedance of V2b becomes 135K in series with 11nF. This relatively low grid impedance is the load seen by the tone stack, and it's effect is noticeable when the FX loop is broken.
Observing the Send signal before and after breaking the FX loop path shows a substantial increase in tone stack output (10-20%) with the FX loop open. An FX unit could use a 150K resistor in series with a 10nF capacitor at its input to mimic the load the tone stack normally feeds. Then the FX unit would pass on a buffered dry signal to V2b to minimize the impact of V2b's grid impedance. This approach effectively makes the FX unit part of the V2b stage. The impedance interaction is modeled at the input of the FX unit.
Observing the Send signal before and after breaking the FX loop path shows a substantial increase in tone stack output (10-20%) with the FX loop open. An FX unit could use a 150K resistor in series with a 10nF capacitor at its input to mimic the load the tone stack normally feeds. Then the FX unit would pass on a buffered dry signal to V2b to minimize the impact of V2b's grid impedance. This approach effectively makes the FX unit part of the V2b stage. The impedance interaction is modeled at the input of the FX unit.
My approach is an alternative that conceptually makes the FX unit part of the tone stack by creating a very high impedance at the FX unit input and then reproducing the impedance interaction at the FX unit output, where it drives the V2b grid. In either approach, what matters is to preserve the dry signal integrity at the grid of V2b, whether the FX unit is plugged in or not. This is a challenge (and problem) with serial effects loops since these Send and Return impedances can vary a great deal from amp to amp. This design is tailored to the Flexi, although the approach may be viable for other amps as well.
My approach is an alternative that conceptually makes the FX unit part of the tone stack by creating a very high impedance at the FX unit input and then reproducing the impedance interaction at the FX unit output, where it drives the V2b grid. In either approach, what matters is to preserve the dry signal integrity at the grid of V2b, whether the FX unit is plugged in or not. This is a challenge (and problem) with serial effects loops since these Send and Return impedances can vary a great deal from amp to amp. This design is tailored to the Flexi, although the approach may be viable for other amps as well.
A test signal is applied to the Flexi input. The test signal (shown below) is a 200mV burst of 220Hz over a low-level continuous 2200Hz background signal. The Flexi Tone controls are set to midpoints, and the input sensitivity switch is Lo.
A test signal is applied to the Flexi input. The test signal (shown below) is a 200mV burst of 220Hz over a low-level continuous 2200Hz background signal. The Flexi Tone controls are set to midpoints, and the input sensitivity switch is Lo.
The V2b anode signal is shown in the following series of images, since this reflects the signal after the FX loop, whether an FX unit is plugged in or not. The first image below shows the anode signal with the Volume set to its midpoint with no FX unit. The signal amplitude is almost 20V peak during the burst. A major design goal for the FX unit is that the V2b anode signal should be identical to this when the FX unit is plugged in and set to 100% dry. In other words, inserting the FX unit should have minimal impact on the passage of the dry signal.
The V2b anode signal is shown in the following series of images, since this reflects the signal after the FX loop, whether an FX unit is plugged in or not. The first image below shows the anode signal with the Volume set to its midpoint with no FX unit. The signal amplitude is almost 20V peak during the burst. A major design goal for the FX unit is that the V2b anode signal should be identical to this when the FX unit is plugged in and set to 100% dry. In other words, inserting the FX unit should have minimal impact on the passage of the dry signal.
Before explaining the circuit itself, here are the results. The image below shows the signal at the V2b anode with Volume set to midpoint and the FX unit plugged in and set to 100% dry. The dry signal is clearly well-preserved by the FX unit circuitry.
Before explaining the circuit itself, here are the results. The image below shows the signal at the V2b anode with Volume set to midpoint and the FX unit plugged in and set to 100% dry. The dry signal is clearly well-preserved by the FX unit circuitry.
Another case worth examining is when the Volume is set to Full. The high volume setting creates distortions in V1b and V2b due to nonlinear tube operation. The most serious concern is the nonlinear grid impedance of V2b as grid current flows on positive grid signal peaks. The image below shows the aggregate distortions at the V2b anode without an FX unit. When the FX unit is plugged in and set to 100% dry, this signal should not change.
Another case worth examining is when the Volume is set to Full. The high volume setting creates distortions in V1b and V2b due to nonlinear tube operation. The most serious concern is the nonlinear grid impedance of V2b as grid current flows on positive grid signal peaks. The image below shows the aggregate distortions at the V2b anode without an FX unit. When the FX unit is plugged in and set to 100% dry, this signal should not change.
This last image below shows the V2b anode signal with Volume set to Full and the FX unit plugged in with the mix set to 100% dry. While there are subtle differences with the no-FX waveform above, the waveforms are strikingly similar, indicating the success of the design approach.
This last image below shows the V2b anode signal with Volume set to Full and the FX unit plugged in with the mix set to 100% dry. While there are subtle differences with the no-FX waveform above, the waveforms are strikingly similar, indicating the success of the design approach.
While simple waveform comparisons don't reveal the whole story, these signals do match pretty well. The balance between the 220Hz and 2200Hz signal components is preserved. The waveshapes at a range of volume settings are preserved. Even the low-frequency shifts are preserved. In fact, the tone and dynamics of the amp played with and without the FX unit plugged in is indistinguishable to me. (Audio clips are below.)
While simple waveform comparisons don't reveal the whole story, these signals do match pretty well. The balance between the 220Hz and 2200Hz signal components is preserved. The waveshapes at a range of volume settings are preserved. Even the low-frequency shifts are preserved. In fact, the tone and dynamics of the amp played with and without the FX unit plugged in is indistinguishable to me. (Audio clips are below.)
So, now that the dry signal is preserved, the wet reverb signal should be mixed in (or added) on top of the dry signal. However, this is also challenging, since the dry signal should not be impacted as varying amounts of wet signal are added. (An opamp summing amplifier can also achieve this mixing, but no opamps are used in this design.) The details of my implementation are explained below and the circuit is shown below.
So, now that the dry signal is preserved, the wet reverb signal should be mixed in (or added) on top of the dry signal. However, this is also challenging, since the dry signal should not be impacted as varying amounts of wet signal are added. (An opamp summing amplifier can also achieve this mixing, but no opamps are used in this design.) The details of my implementation are explained below and the circuit is shown below.
A JFET follower input stage (Q1) provides a very high input impedance to eliminate any loading of the tone stack signal (from the Send jack). The input stage is followed by a driver amp similar to the one described for the P-AOT project. A few values in the driver amp were tweaked to optimize it for the higher signal levels from the Flexi FX Send. Clipping in the driver amp doesn't occur until the Flexi Send signal level is high and already heavily distorted in the early preamp stages. The ECG transistors are what I had on hand. Later versions of this circuit use TIP29 and TIP30 transistors.
A JFET follower input stage (Q1) provides a very high input impedance to eliminate any loading of the tone stack signal (from the Send jack). The input stage is followed by a driver amp similar to the one described for the P-AOT project. A few values in the driver amp were tweaked to optimize it for the higher signal levels from the Flexi FX Send. Clipping in the driver amp doesn't occur until the Flexi Send signal level is high and already heavily distorted in the early preamp stages. The ECG transistors are what I had on hand. Later versions of this circuit use TIP29 and TIP30 transistors.
The recovery amplifier uses a high impedance JFET (Q6) as the first stage for low noise. The Q6 output drives the mixing stage with the wet signal. SW1 is a 3 position toggle switch that selects high-cut, high-boost, or neutral tone options for the wet signal. The mixing stage uses a high-impedance drive (Q8) for the wet signal and a low impedance drive (Q7) for the dry signal. (Note the Q1 JFET input stage feeds the buffered dry signal to the Q7 mixing stage.)
The recovery amplifier uses a high impedance JFET (Q6) as the first stage for low noise. The Q6 output drives the mixing stage with the wet signal. SW1 is a 3 position toggle switch that selects high-cut, high-boost, or neutral tone options for the wet signal. The mixing stage uses a high-impedance drive (Q8) for the wet signal and a low impedance drive (Q7) for the dry signal. (Note the Q1 JFET input stage feeds the buffered dry signal to the Q7 mixing stage.)
The dry signal is maintained at R22 by Q7, regardless of how much wet signal is added. Since Q8 is a high impedance source, the dry signal remains constant for any setting of the mix control (P1b). The mix control simply varies how much wet signal is added to the constant dry signal level. The mix control (P1) is a dual pot. P1a varies how much wet signal gets to the mixing stage, and P1b varies how much of the wet signal is mixed with the dry. The two pots are linear taper, but since they feed each other in series, they behave together as an audio taper pot, and the amount of perceived wetness varies smoothly over the whole range of the mix control.
The dry signal is maintained at R22 by Q7, regardless of how much wet signal is added. Since Q8 is a high impedance source, the dry signal remains constant for any setting of the mix control (P1b). The mix control simply varies how much wet signal is added to the constant dry signal level. The mix control (P1) is a dual pot. P1a varies how much wet signal gets to the mixing stage, and P1b varies how much of the wet signal is mixed with the dry. The two pots are linear taper, but since they feed each other in series, they behave together as an audio taper pot, and the amount of perceived wetness varies smoothly over the whole range of the mix control.
The output RC network is important since it has to mimic the output impedance of the tone stack. Recall, this FX unit is effectively within the tone stack and it has to drive the V2b grid impedance as the tone stack would. Rather than pursuing a complex analysis (Spice or otherwise), I simply selected R24 to produce the required amplitude for the 220Hz signal, and then C12 was selected to produce the required level of the 2200Hz signal. With those values set at the relative low and high end of the guitar range, the output network approximates the tone stack output impedance.
The output RC network is important since it has to mimic the output impedance of the tone stack. Recall, this FX unit is effectively within the tone stack and it has to drive the V2b grid impedance as the tone stack would. Rather than pursuing a complex analysis (Spice or otherwise), I simply selected R24 to produce the required amplitude for the 220Hz signal, and then C12 was selected to produce the required level of the 2200Hz signal. With those values set at the relative low and high end of the guitar range, the output network approximates the tone stack output impedance.
Admittedly, the tone stack does not have a fixed impedance, it varies as the tone controls settings change, but this network is a good approximation for mid-tone settings and seems adequate overall. The R23 resistor is also noteworthy, since it (with R24) approximates the V2b grid leak path through the tone stack. With this FX unit, the mods that create a DC path (MOD2 described above) are not necessary. However, since I like the option of using other FX systems, I'm leaving the 2.2M resistor mod in the Flexi in place.
Admittedly, the tone stack does not have a fixed impedance, it varies as the tone controls settings change, but this network is a good approximation for mid-tone settings and seems adequate overall. The R23 resistor is also noteworthy, since it (with R24) approximates the V2b grid leak path through the tone stack. With this FX unit, the mods that create a DC path (MOD2 described above) are not necessary. However, since I like the option of using other FX systems, I'm leaving the 2.2M resistor mod in the Flexi in place.
Build photos
Build photos
As with any new design, it comes to life as a prototype. A proto board holds the whole design and allows simple testing and tuning.
As with any new design, it comes to life as a prototype. A proto board holds the whole design and allows simple testing and tuning.
After the circuit is fully debugged and finalized, it's built on a perf board. The plastic enclosure is shielded by laying copper foil over the whole box, including the joint where the case cavity and cover meet. Plastic is lighter and easier to machine than metal. The component side of the board faces down into the case cavity, so some headers allow simple connections to the 4 jacks.
After the circuit is fully debugged and finalized, it's built on a perf board. The plastic enclosure is shielded by laying copper foil over the whole box, including the joint where the case cavity and cover meet. Plastic is lighter and easier to machine than metal. The component side of the board faces down into the case cavity, so some headers allow simple connections to the 4 jacks.
The wiring side of the board has a DIP socket and a header is wired to make all the interconnections to the controls and LED on the case lid. Four standoffs mount the board to the lid.
The wiring side of the board has a DIP socket and a header is wired to make all the interconnections to the controls and LED on the case lid. Four standoffs mount the board to the lid.
The completed enclosure has 4 signal jacks (FXin, FXout, Pan Drive, and Pan Recovery), as well as a 2 pin Molex connector at the top of the case (not visible) for power (24V and ground).
The completed enclosure has 4 signal jacks (FXin, FXout, Pan Drive, and Pan Recovery), as well as a 2 pin Molex connector at the top of the case (not visible) for power (24V and ground).
The FX unit is recessed into the back panel of the cab. The reverb mix knob and hi-cut/boost toggle lie below the cab panel so the cab can lie flat on it's back without breaking anything. The Creamback speaker and reverb pan are visible through the semi-open back cover. The pan is mounted on some foam for acoustic isolation.
The FX unit is recessed into the back panel of the cab. The reverb mix knob and hi-cut/boost toggle lie below the cab panel so the cab can lie flat on it's back without breaking anything. The Creamback speaker and reverb pan are visible through the semi-open back cover. The pan is mounted on some foam for acoustic isolation.
A single AC connector provides power both to the reverb and a loop through connector to power the Flexi. A small 24V (16 watt) DC power supply is mounted inside the back panel (not visible) to power the FX unit. A recessed slide switch controls reverb power. My 24V supply is a model APV-16-24 made by MEAN WELL. It's under $10 from varied sources.
A single AC connector provides power both to the reverb and a loop through connector to power the Flexi. A small 24V (16 watt) DC power supply is mounted inside the back panel (not visible) to power the FX unit. A recessed slide switch controls reverb power. My 24V supply is a model APV-16-24 made by MEAN WELL. It's under $10 from varied sources.
Update 2/12 - Increased mid cut
Update 2/12 - Increased mid cut
After running some simulations of the Flexi tone stack, (which I'm appreciating more and more) I added a 68k resistor in parallel with R15 (initially 68k) to make it 34k. This cuts another 3-4 db of mids (~550hz) when the mid control is at minimum. The mod has insignificant impact at other mid control settings, so it really just allows a greater "scoop" in the mid frequencies at the control minimum. I find this helpful to soften the G&L pickups, whose strong mid-range output can be hard to control at times. The plot below shows the response curve with bass and treble at mid settings and the mid control at minimum (after the mod).
After running some simulations of the Flexi tone stack, (which I'm appreciating more and more) I added a 68k resistor in parallel with R15 (initially 68k) to make it 34k. This cuts another 3-4 db of mids (~550hz) when the mid control is at minimum. The mod has insignificant impact at other mid control settings, so it really just allows a greater "scoop" in the mid frequencies at the control minimum. I find this helpful to soften the G&L pickups, whose strong mid-range output can be hard to control at times. The plot below shows the response curve with bass and treble at mid settings and the mid control at minimum (after the mod).
Page updated
Google Sites
Report abuse