Evolution of an OD Stage
Overdrive tones are often distinctive guitar amplifier qualities, yet little is written about how to achieve or control overdrive (OD) tone qualities. There is an aura of inevitability about how a particular circuit or system of circuits sounds. Specific amplifiers are often associated with a characteristic OD tone even though the amp actually produces a wide range of tones, some more, and some less desirable than others. Players often search for combinations of control settings that achieve a desired tone, often also controlling their playing for optimal effects.
Overdrive tones are often distinctive guitar amplifier qualities, yet little is written about how to achieve or control overdrive (OD) tone qualities. There is an aura of inevitability about how a particular circuit or system of circuits sounds. Specific amplifiers are often associated with a characteristic OD tone even though the amp actually produces a wide range of tones, some more, and some less desirable than others. Players often search for combinations of control settings that achieve a desired tone, often also controlling their playing for optimal effects.
Circuit designers know that many variables affect the tone of even one overdriven gain stage. The number of variables grow with every added circuit component and OD stage. Iterations of tweaking/testing are difficult since everything seems to impact everything else. The parameter search space is large, making it hard to optimize. Adding to the complexity, time-varying circuit effects like bias shifts mean that past circuit conditions impact present circuit behavior.
Circuit designers know that many variables affect the tone of even one overdriven gain stage. The number of variables grow with every added circuit component and OD stage. Iterations of tweaking/testing are difficult since everything seems to impact everything else. The parameter search space is large, making it hard to optimize. Adding to the complexity, time-varying circuit effects like bias shifts mean that past circuit conditions impact present circuit behavior.
In addition to circuit variations, actual playing conditions are also highly variable. In the hands of a skilled musician the guitar signal has far more dynamics and frequency content than a simple repetitive signal. An OD stage is exposed to single notes vs. 6-string chords; high vs. low note frequencies; close vs. distant intervals; and soft vs. hard string plucking. How an amp responds and sounds under all these conditions determines its ease of use. Ideally, pleasing tones are achieved under a wide range of playing conditions, although some amps are valued for a specific tone they can only achieve under limited conditions.
In addition to circuit variations, actual playing conditions are also highly variable. In the hands of a skilled musician the guitar signal has far more dynamics and frequency content than a simple repetitive signal. An OD stage is exposed to single notes vs. 6-string chords; high vs. low note frequencies; close vs. distant intervals; and soft vs. hard string plucking. How an amp responds and sounds under all these conditions determines its ease of use. Ideally, pleasing tones are achieved under a wide range of playing conditions, although some amps are valued for a specific tone they can only achieve under limited conditions.
The instrument also varies significantly. Guitars and pickups vary widely in signal amplitude and frequency content. A given amp and distortion channel may produce sweet single-coil tones, yet produce muddy raspy tones from a high-output humbucker. Players quickly learn what amps are likely to work well for their instrument and tone preferences. The challenge to players is to search for those "magic" combinations of guitar, amp control settings, and speakers that produce a desired tone and playing experience.
The instrument also varies significantly. Guitars and pickups vary widely in signal amplitude and frequency content. A given amp and distortion channel may produce sweet single-coil tones, yet produce muddy raspy tones from a high-output humbucker. Players quickly learn what amps are likely to work well for their instrument and tone preferences. The challenge to players is to search for those "magic" combinations of guitar, amp control settings, and speakers that produce a desired tone and playing experience.
Amp designers are also challenged since classic design theory and calculations are basically useless to predict OD tone. Load lines, bias, impedance, RC network response curves are all modeled on linear operation. These don't easily extend or apply to non-linear OD conditions. Simulations (Spice) can reveal first order cause-effect relationships and parameter impacts, but component models lack precision in nonlinear conditions so simulations of actual audio are not reliable.
Amp designers are also challenged since classic design theory and calculations are basically useless to predict OD tone. Load lines, bias, impedance, RC network response curves are all modeled on linear operation. These don't easily extend or apply to non-linear OD conditions. Simulations (Spice) can reveal first order cause-effect relationships and parameter impacts, but component models lack precision in nonlinear conditions so simulations of actual audio are not reliable.
Fortunately, there is a body of "prior-art" that is very helpful, at least as a start point. Many OD designs are copies or variations of existing circuits with known tone qualities. Merlin Blencow's 2nd edition preamp design book has a very useful chapter discussing the process of designing overdrive preamps along with some excellent examples. Merlin's chapter stands out as the most in-depth treatment of the topic I could find.
Fortunately, there is a body of "prior-art" that is very helpful, at least as a start point. Many OD designs are copies or variations of existing circuits with known tone qualities. Merlin Blencow's 2nd edition preamp design book has a very useful chapter discussing the process of designing overdrive preamps along with some excellent examples. Merlin's chapter stands out as the most in-depth treatment of the topic I could find.
A number of circuits have been explored and classified with terms like cold-clippers, diode clippers, warm distortion, etc. These terms relate to circuits that are known to produce certain OD tone qualities. Discussions of common circuits are found in places like Rob Robinette's overdrive web page, Internet forum articles, etc. Well known examples include an unbypassed high-bias resistor (produces low bias current) gain stage (a cold-clipper) and a direct-coupled cathode follower that produces "warm" distortion. It's not hard to find examples of such categories, however, they are all only points in a huge space of possibilities where relatively small changes to component values or circuits can sound very different.
A number of circuits have been explored and classified with terms like cold-clippers, diode clippers, warm distortion, etc. These terms relate to circuits that are known to produce certain OD tone qualities. Discussions of common circuits are found in places like Rob Robinette's overdrive web page, Internet forum articles, etc. Well known examples include an unbypassed high-bias resistor (produces low bias current) gain stage (a cold-clipper) and a direct-coupled cathode follower that produces "warm" distortion. It's not hard to find examples of such categories, however, they are all only points in a huge space of possibilities where relatively small changes to component values or circuits can sound very different.
In the end, nothing fully predicts the actual sound of a circuit under playing conditions and it may be impossible to produce an optimal OD circuit that pleases everyone. Design often boils down to picking a target instrument(s) and a circuit type (e.g., cold clipper) and then tweaking it to produce a desired or pleasing tone within the context of the surrounding amplifier circuits, speakers, cabinets, etc. The OD design process is therefore an iterative playing-listening search. The search is aided and guided by a designer's instincts, prior experience, and circuit knowledge. If time and resources are available, a relatively comprehensive search leads to a "best-possible" configuration that meets the tone goals. A "best-possible" (or locally optimal) solution is generally acceptable since search time (time usually costs money) is limited and the number of circuit-variable combinations is essentially unlimited - especially for multi-stage OD circuits.
In the end, nothing fully predicts the actual sound of a circuit under playing conditions and it may be impossible to produce an optimal OD circuit that pleases everyone. Design often boils down to picking a target instrument(s) and a circuit type (e.g., cold clipper) and then tweaking it to produce a desired or pleasing tone within the context of the surrounding amplifier circuits, speakers, cabinets, etc. The OD design process is therefore an iterative playing-listening search. The search is aided and guided by a designer's instincts, prior experience, and circuit knowledge. If time and resources are available, a relatively comprehensive search leads to a "best-possible" configuration that meets the tone goals. A "best-possible" (or locally optimal) solution is generally acceptable since search time (time usually costs money) is limited and the number of circuit-variable combinations is essentially unlimited - especially for multi-stage OD circuits.
Focusing on just a single OD stage, what do the components do to the tone and what variations can be expected? My book on amplifier overdrive explores the behavior of overdriven tube circuits, however, it lacks a chapter specifically dealing with an OD stage and shaping its tone qualities. This document is an effort to address that. An evolved version may appear as a chapter in a future book edition, however, for now this is available online to readers of the existing book or anyone interested in the topic.
Focusing on just a single OD stage, what do the components do to the tone and what variations can be expected? My book on amplifier overdrive explores the behavior of overdriven tube circuits, however, it lacks a chapter specifically dealing with an OD stage and shaping its tone qualities. This document is an effort to address that. An evolved version may appear as a chapter in a future book edition, however, for now this is available online to readers of the existing book or anyone interested in the topic.
The following discussion develops a single OD stage through a series of iterations and refinements in hopes that this account informs or guides others in similar pursuits. The process also focuses on a novel OD circuit that provides new and added controls over the distortion components produced by the stage.
The following discussion develops a single OD stage through a series of iterations and refinements in hopes that this account informs or guides others in similar pursuits. The process also focuses on a novel OD circuit that provides new and added controls over the distortion components produced by the stage.
A few words about the context of this work. I recently designed and built the Bassman Lite (BML) amp. It uses the second preamp triode to generate OD tones. The first version (v1.1) has a simple OD circuit that was developed with relatively few iterations and tests. This describes the evolution of that circuit for the next version of the BML preamp. The first stage triode produces a signal of about 2-10V peak, depending on the guitar and pickups used. In my world a telecaster is important and that guitar produces about 2-4V peak at the output of the first stage. I also use a G&L S-500. Those pickups are hotter and they can produce 5-10V from the first stage. So, the OD stage has to contend with 2-10V peak signals. I expect a gradual increase in distortion over that range. This OD stage is not a high-gain square-wave generator. I want it to add character and thickness, but still retain playing dynamics - like a slightly overdriven output-stage. I expect the guitar volume control to control the OD distortion level. I want coloration and added harmonics for single notes and small chords without losing the notes themselves. Bass notes should remain distinctive without farting or blocking. I like treble notes sweet without fizz or raspy decays. These preferences complement the clean tone context of the BML amplifier. This is not a general recipe I expect everyone to like and they're not the goals for all my projects. They're just something to keep in mind as the motivation behind the tuning process described below.
A few words about the context of this work. I recently designed and built the Bassman Lite (BML) amp. It uses the second preamp triode to generate OD tones. The first version (v1.1) has a simple OD circuit that was developed with relatively few iterations and tests. This describes the evolution of that circuit for the next version of the BML preamp. The first stage triode produces a signal of about 2-10V peak, depending on the guitar and pickups used. In my world a telecaster is important and that guitar produces about 2-4V peak at the output of the first stage. I also use a G&L S-500. Those pickups are hotter and they can produce 5-10V from the first stage. So, the OD stage has to contend with 2-10V peak signals. I expect a gradual increase in distortion over that range. This OD stage is not a high-gain square-wave generator. I want it to add character and thickness, but still retain playing dynamics - like a slightly overdriven output-stage. I expect the guitar volume control to control the OD distortion level. I want coloration and added harmonics for single notes and small chords without losing the notes themselves. Bass notes should remain distinctive without farting or blocking. I like treble notes sweet without fizz or raspy decays. These preferences complement the clean tone context of the BML amplifier. This is not a general recipe I expect everyone to like and they're not the goals for all my projects. They're just something to keep in mind as the motivation behind the tuning process described below.
The OD stage follows a simple gain stage (V1) as shown below. The values shown for the V1 gain stage are optimized for high clean gain. The OD stage (V2) is simply shown as a common cathode (CC) stage at this point.
The OD stage follows a simple gain stage (V1) as shown below. The values shown for the V1 gain stage are optimized for high clean gain. The OD stage (V2) is simply shown as a common cathode (CC) stage at this point.
The stage 1 output (S1out) of 2-10V peak signals is sufficient to cause nonlinear V2 operation, but it is too low to completely overwhelm the V2 stage since it operates from a relatively high power supply voltage of ~335V. A bypassed 12ax7 stage produces a gain of ~60 so 2V inputs will barely clip if the V2 stage is center biased. I begin test iterations with the small signals produced by a telecaster. A low bias (warm) clipping tone is not to my taste, and besides, a direct-coupled cathode follower is the next stage in the BML amp and that produces warm tone distortions already. By iteration I find that a bias resistor in the 2-5K range and 100K plate load for V2 give me a good tone and a region to start exploring. These values are a slightly high bias, so the tone has a touch of metallic harmonics - but not something you'd call an extreme cold-clipper.
The stage 1 output (S1out) of 2-10V peak signals is sufficient to cause nonlinear V2 operation, but it is too low to completely overwhelm the V2 stage since it operates from a relatively high power supply voltage of ~335V. A bypassed 12ax7 stage produces a gain of ~60 so 2V inputs will barely clip if the V2 stage is center biased. I begin test iterations with the small signals produced by a telecaster. A low bias (warm) clipping tone is not to my taste, and besides, a direct-coupled cathode follower is the next stage in the BML amp and that produces warm tone distortions already. By iteration I find that a bias resistor in the 2-5K range and 100K plate load for V2 give me a good tone and a region to start exploring. These values are a slightly high bias, so the tone has a touch of metallic harmonics - but not something you'd call an extreme cold-clipper.
The high bias puts the anode operating point at ~250-300V, nearer to cutoff than grid clipping. Stage gain, however, is lowered by the unbypassed cathode, so it's tempting to bypass the cathode resistor to recover gain. Bypass introduces grid clipping, since the cathode voltage becomes fixed and that completely changes the tone. A solution is to use a smaller capacitor or partial bypass to recover some gain and maintain control over grid clipping. Iterations produce the circuit below with more gain and more harmonics, while maintaining the slightly-edgy metallic tone of the unbypassed stage.
The high bias puts the anode operating point at ~250-300V, nearer to cutoff than grid clipping. Stage gain, however, is lowered by the unbypassed cathode, so it's tempting to bypass the cathode resistor to recover gain. Bypass introduces grid clipping, since the cathode voltage becomes fixed and that completely changes the tone. A solution is to use a smaller capacitor or partial bypass to recover some gain and maintain control over grid clipping. Iterations produce the circuit below with more gain and more harmonics, while maintaining the slightly-edgy metallic tone of the unbypassed stage.
The AC impedance of the V2 cathode circuit is worth understanding and exploring. Note that the cathode circuit impedance is not the same as the cathode signal output impedance. With a high (3k) impedance cathode circuit, grid clipping rarely occurs. Cold clipper circuits often use unbypassed cathode resistors of 10K or more. Grid clipping rarely occurs since the cathode can easily follow positive grid signal swings. Bypassing the cathode with a capacitor lowers the cathode circuit impedance, increasing gain and keeping the cathode voltage relatively fixed so grid clipping occurs more easily on positive grid signal swings. The partial bypass above is a means of controlling the cathode circuit impedance and therefore the degree of grid clipping that occurs for a given input signal. Controlling this impedance is a means to controlling the distortion tone. While a scope can show the presence/absence of grid clipping, only your ears tell you how much or little is desired.
The AC impedance of the V2 cathode circuit is worth understanding and exploring. Note that the cathode circuit impedance is not the same as the cathode signal output impedance. With a high (3k) impedance cathode circuit, grid clipping rarely occurs. Cold clipper circuits often use unbypassed cathode resistors of 10K or more. Grid clipping rarely occurs since the cathode can easily follow positive grid signal swings. Bypassing the cathode with a capacitor lowers the cathode circuit impedance, increasing gain and keeping the cathode voltage relatively fixed so grid clipping occurs more easily on positive grid signal swings. The partial bypass above is a means of controlling the cathode circuit impedance and therefore the degree of grid clipping that occurs for a given input signal. Controlling this impedance is a means to controlling the distortion tone. While a scope can show the presence/absence of grid clipping, only your ears tell you how much or little is desired.
A partial bypass produces a varied cathode impedance at different frequencies. This is a means of boosting gain and increasing levels of grid clipping at high frequencies. In the circuit above, less gain and grid clipping occurs for bass frequencies, and more gain and grid clipping occurs for treble frequencies. This results in less, but cleaner bass (smaller and less waveform distortion and harmonics) and more gain and asymmetric grid-clipping distortions for treble frequencies. The cap and resistors in the cathode circuit control the variations.
A partial bypass produces a varied cathode impedance at different frequencies. This is a means of boosting gain and increasing levels of grid clipping at high frequencies. In the circuit above, less gain and grid clipping occurs for bass frequencies, and more gain and grid clipping occurs for treble frequencies. This results in less, but cleaner bass (smaller and less waveform distortion and harmonics) and more gain and asymmetric grid-clipping distortions for treble frequencies. The cap and resistors in the cathode circuit control the variations.
So now we have a CC stage with partial bypass. The bias and load resistor set the operating point and therefore the relative tendency for cut-off (cold) clipping or grid (warm) clipping. The partial bypass allows control over gain and the degree of grid clipping as functions of frequency. There is interaction between all of these effects so test iterations are needed to fine tune the final tone. This circuit already has a lot of parameters to set and tune. Only fine tuning will reveal the "sweet spots" that sound best to you. This circuit complexity is typical of many OD stages. A circuit like this provides sufficient control to produce desired results in many designs.
So now we have a CC stage with partial bypass. The bias and load resistor set the operating point and therefore the relative tendency for cut-off (cold) clipping or grid (warm) clipping. The partial bypass allows control over gain and the degree of grid clipping as functions of frequency. There is interaction between all of these effects so test iterations are needed to fine tune the final tone. This circuit already has a lot of parameters to set and tune. Only fine tuning will reveal the "sweet spots" that sound best to you. This circuit complexity is typical of many OD stages. A circuit like this provides sufficient control to produce desired results in many designs.
However, an issue that often creates difficulty for the simple circuit shown above is input signal variations. Many amps employ internal trim pots to precisely control signal amplitudes in their OD circuits, and often players have to carefully set controls and levels for good results. Fine tuning can produce good tones with the above circuit for limited (2-4V) inputs, but surprising or undesirable effects may occur for larger (10V) inputs. Signal amplitudes under player control invariably have wide ranges. In the BML amp the only control over the signal amplitude fed to the OD stage is the guitar volume control.
However, an issue that often creates difficulty for the simple circuit shown above is input signal variations. Many amps employ internal trim pots to precisely control signal amplitudes in their OD circuits, and often players have to carefully set controls and levels for good results. Fine tuning can produce good tones with the above circuit for limited (2-4V) inputs, but surprising or undesirable effects may occur for larger (10V) inputs. Signal amplitudes under player control invariably have wide ranges. In the BML amp the only control over the signal amplitude fed to the OD stage is the guitar volume control.
At higher input levels, an OD stage clips harder, meaning it spends a greater percentage of time (duty cycle) in cutoff and/or grid clipping. One effect of higher clipping duty cycles is loss of high frequencies since the clipped portions of a waveform lose all their high-frequency variations. This "chopping" of the high freq waveform also increases intermodulation distortion components that may sound good or bad when multiple notes are played together. If distinct notes and less fizzy harmonics are desired (as by me) a means of reducing hard clipping of higher frequencies is needed.
At higher input levels, an OD stage clips harder, meaning it spends a greater percentage of time (duty cycle) in cutoff and/or grid clipping. One effect of higher clipping duty cycles is loss of high frequencies since the clipped portions of a waveform lose all their high-frequency variations. This "chopping" of the high freq waveform also increases intermodulation distortion components that may sound good or bad when multiple notes are played together. If distinct notes and less fizzy harmonics are desired (as by me) a means of reducing hard clipping of higher frequencies is needed.
Recall that grid clipping occurs when a gain stage with low cathode circuit impedance is driven by a high source impedance input signal. When a positive input swing equals or exceeds the bias voltage, grid current flows, lowering the grid impedance and clipping the top of the waveform. A typical anode output of a gain stage has an impedance in the range of 40-60k, so it is unable to drive a grid positive - even less so for large signals since grid impedance falls with increased signal levels. To impose unclipped high signal levels on the grid, a low source impedance is needed. The circuit below is described in the GAO book as a Low Impedance Overdrive (LIO) circuit. This version uses a MOSFET follower, but tubes or transistors can also be used as followers.
Recall that grid clipping occurs when a gain stage with low cathode circuit impedance is driven by a high source impedance input signal. When a positive input swing equals or exceeds the bias voltage, grid current flows, lowering the grid impedance and clipping the top of the waveform. A typical anode output of a gain stage has an impedance in the range of 40-60k, so it is unable to drive a grid positive - even less so for large signals since grid impedance falls with increased signal levels. To impose unclipped high signal levels on the grid, a low source impedance is needed. The circuit below is described in the GAO book as a Low Impedance Overdrive (LIO) circuit. This version uses a MOSFET follower, but tubes or transistors can also be used as followers.
The follower output is AC coupled into a low impedance 10K load that also acts as the grid leak path for V2. The coupling cap needs to be rather large to pass bass frequencies to a low impedance load, but a low-Z load ensures that blocking isn't a problem. A resistor R10 is added to create a controlled source impedance, which in turn controls the degree of grid clipping for V2. A low value (1K) produces slight clipping, while higher values (10k-100k) produce progressively sharper grid clipping. There are now two controls over grid clipping, R10 and the V2 cathode impedance. Of course they interact to some degree.
The follower output is AC coupled into a low impedance 10K load that also acts as the grid leak path for V2. The coupling cap needs to be rather large to pass bass frequencies to a low impedance load, but a low-Z load ensures that blocking isn't a problem. A resistor R10 is added to create a controlled source impedance, which in turn controls the degree of grid clipping for V2. A low value (1K) produces slight clipping, while higher values (10k-100k) produce progressively sharper grid clipping. There are now two controls over grid clipping, R10 and the V2 cathode impedance. Of course they interact to some degree.
The motivation for the low-Z grid drive is to preserve high frequencies during grid clipping. A low value of R10 achieves the goal, however it impacts all frequencies equally. So, to selectively reduce high frequency clipping, R10 needs to be high enough to clip low frequencies, and a bypass cap (C5) is needed to allow high frequencies to drive the grid. After some iterations to adjust R10, C5 and the cathode network, the result is shown below.
The motivation for the low-Z grid drive is to preserve high frequencies during grid clipping. A low value of R10 achieves the goal, however it impacts all frequencies equally. So, to selectively reduce high frequency clipping, R10 needs to be high enough to clip low frequencies, and a bypass cap (C5) is needed to allow high frequencies to drive the grid. After some iterations to adjust R10, C5 and the cathode network, the result is shown below.
Now, you may wonder why a source impedance of 47k is created this way when that's similar to the source impedance of the first stage anode output. The reason is that this impedance can be bypassed with a capacitor to create a lower source impedance for high frequencies. High frequencies drive the grid positive with less clipping while lower frequencies clip harder. This provides another means of control over the harmonics generated in the OD stage. More low and mid frequency harmonics and fewer high harmonics are generated. Note that this is not the same effect produced by simply adding a shunt capacitor to the stage output to roll off high frequencies. A shunt cap at the output reduces both the original signal and any generated harmonics. They are both present in the output signal and can't be separated The method above maintains the signal amplitudes, only reducing the higher harmonics generated by overdrive.
Now, you may wonder why a source impedance of 47k is created this way when that's similar to the source impedance of the first stage anode output. The reason is that this impedance can be bypassed with a capacitor to create a lower source impedance for high frequencies. High frequencies drive the grid positive with less clipping while lower frequencies clip harder. This provides another means of control over the harmonics generated in the OD stage. More low and mid frequency harmonics and fewer high harmonics are generated. Note that this is not the same effect produced by simply adding a shunt capacitor to the stage output to roll off high frequencies. A shunt cap at the output reduces both the original signal and any generated harmonics. They are both present in the output signal and can't be separated The method above maintains the signal amplitudes, only reducing the higher harmonics generated by overdrive.
The LIO method also reduces the chopping of high frequencies during strong clipping and the resulting intermodulation distortion components. The effect is to reduce the fizzy sounding harmonics. It's impact on the final tone is hard to overstate - especially over a wide range of input levels. The tuning of R10 and C5 are important as they have strong impacts on the tone at all input levels. Values of 10k-100k and 33nf-220nf are ranges worth exploring and I found several combinations that produce distinctive and good tones. Switched options could be useful for these components.
The LIO method also reduces the chopping of high frequencies during strong clipping and the resulting intermodulation distortion components. The effect is to reduce the fizzy sounding harmonics. It's impact on the final tone is hard to overstate - especially over a wide range of input levels. The tuning of R10 and C5 are important as they have strong impacts on the tone at all input levels. Values of 10k-100k and 33nf-220nf are ranges worth exploring and I found several combinations that produce distinctive and good tones. Switched options could be useful for these components.
While the bypassed source impedance provides control over grid clipping, it does little for hard cutoff. High input levels can produce abrupt tone changes. The addition of diodes and a resistor simulate grid clipping behavior for negative signal swings. (A similar idea has been used to control output stage blocking.) The challenge is to soften cutoff somewhat without destroying its distinctive tone contributions. For this circuit, 2-4 diodes in series and a resistance provide a range of cut-off control. Varying the number of diodes and the resistor value moves the diode clipping curve closer or farther from actual cutoff, thereby increasing or decreasing actual cutoff contribution to the tone. Once again, playing iterations are needed and signal amplitude needs to be varied for each test. I settled on three diodes and the R11 value shown below as producing the best tone for my taste.
While the bypassed source impedance provides control over grid clipping, it does little for hard cutoff. High input levels can produce abrupt tone changes. The addition of diodes and a resistor simulate grid clipping behavior for negative signal swings. (A similar idea has been used to control output stage blocking.) The challenge is to soften cutoff somewhat without destroying its distinctive tone contributions. For this circuit, 2-4 diodes in series and a resistance provide a range of cut-off control. Varying the number of diodes and the resistor value moves the diode clipping curve closer or farther from actual cutoff, thereby increasing or decreasing actual cutoff contribution to the tone. Once again, playing iterations are needed and signal amplitude needs to be varied for each test. I settled on three diodes and the R11 value shown below as producing the best tone for my taste.
This final OD stage is smooth over a wide range of input levels and provides many controls over its tone and behavior. Without a disciplined approach to tuning, the number of variables easily becomes overwhelming. Many of the parameters interact, but after spending some hours (lots actually) on tuning, I've developed an intuition and feel for the tone impacts of many of the parameters. It seems both flexible and manageable to me.
This final OD stage is smooth over a wide range of input levels and provides many controls over its tone and behavior. Without a disciplined approach to tuning, the number of variables easily becomes overwhelming. Many of the parameters interact, but after spending some hours (lots actually) on tuning, I've developed an intuition and feel for the tone impacts of many of the parameters. It seems both flexible and manageable to me.
Admittedly, this complexity may be overkill for some. Many people are happy with a simple cathode follower or cold clipper stage. Sometimes lack of choices leads to happiness. I suspect many players would rather focus on their playing skills rather than tuning circuits. There's nothing wrong with any of that. I personally found the design and tuning of this stage as enjoyable as playing it. Having explored many of the options and possibilities also makes me appreciate the outcome. In my case, the knowledge and choices explored are part of the reward - although I should probably focus more on my playing skills too.
Admittedly, this complexity may be overkill for some. Many people are happy with a simple cathode follower or cold clipper stage. Sometimes lack of choices leads to happiness. I suspect many players would rather focus on their playing skills rather than tuning circuits. There's nothing wrong with any of that. I personally found the design and tuning of this stage as enjoyable as playing it. Having explored many of the options and possibilities also makes me appreciate the outcome. In my case, the knowledge and choices explored are part of the reward - although I should probably focus more on my playing skills too.
To illustrate the stage behavior, here are some scope traces at different signal levels. The signal used is the same one used in the GAO book - a burst mix of 220Hz and 2200Hz. The signals shown are the stage1 output (cyan) and the stage2 output (gold). The spectrum plot is for the stage2 output. The first spectrum is for the complete circuit, the second spectrum is with C5 removed. A word of caution about reading too much into these waveforms and plots. There is no simple perceptual mapping from either of these signal visualizations to what you actually hear. These mainly serve to illustrate relative signal/spectrum trends and relationships.
To illustrate the stage behavior, here are some scope traces at different signal levels. The signal used is the same one used in the GAO book - a burst mix of 220Hz and 2200Hz. The signals shown are the stage1 output (cyan) and the stage2 output (gold). The spectrum plot is for the stage2 output. The first spectrum is for the complete circuit, the second spectrum is with C5 removed. A word of caution about reading too much into these waveforms and plots. There is no simple perceptual mapping from either of these signal visualizations to what you actually hear. These mainly serve to illustrate relative signal/spectrum trends and relationships.
The first cluster of three images below are produced with a 25mv peak burst input to the first stage. This is a level about that of a gentle telecaster string pluck. Stage 1 produces about 1V peak and there is just a hint of OD distortion audible in the stage 2 output signal. Only a few IM components of the 2200Hz signal are visible in the first spectrum and their energies are at least 30db below the 2200Hz signal component. As a comparison, the lower spectrum image shows the OD stage output after removing C5. Even without visible grid clipping at this signal level, the removal of C5 elevates the levels of distortion components. At low signal levels the OD stage is therefore slightly cleaner with C5. While this may not be desirable, the audible effect is minor and this must be traded off against benefits at higher signal levels.
The first cluster of three images below are produced with a 25mv peak burst input to the first stage. This is a level about that of a gentle telecaster string pluck. Stage 1 produces about 1V peak and there is just a hint of OD distortion audible in the stage 2 output signal. Only a few IM components of the 2200Hz signal are visible in the first spectrum and their energies are at least 30db below the 2200Hz signal component. As a comparison, the lower spectrum image shows the OD stage output after removing C5. Even without visible grid clipping at this signal level, the removal of C5 elevates the levels of distortion components. At low signal levels the OD stage is therefore slightly cleaner with C5. While this may not be desirable, the audible effect is minor and this must be traded off against benefits at higher signal levels.
A 50mv peak input is applied in the next example, producing a 2V peak signal to drive the OD stage. That is about the level produced by a normal pluck of a telecaster string. No hard clipping is visible yet at the output, but the relative IM components in the spectrum are about 7-8db higher than in the first example. Note also that only the only the two closest IM components have significant energy. Higher components are suppressed.
A 50mv peak input is applied in the next example, producing a 2V peak signal to drive the OD stage. That is about the level produced by a normal pluck of a telecaster string. No hard clipping is visible yet at the output, but the relative IM components in the spectrum are about 7-8db higher than in the first example. Note also that only the only the two closest IM components have significant energy. Higher components are suppressed.
As a comparison, the lower spectrum shows the OD stage output without C5. Again, removing C5 effectively adds additional harmonic and IM components and elevates their levels. While OD output is cleaner with C5, it's also clear that the main effect of adding C5 is the reduction of higher order distortion components. This sharp cut in higher order distortion components continues as the input level increases further.
As a comparison, the lower spectrum shows the OD stage output without C5. Again, removing C5 effectively adds additional harmonic and IM components and elevates their levels. While OD output is cleaner with C5, it's also clear that the main effect of adding C5 is the reduction of higher order distortion components. This sharp cut in higher order distortion components continues as the input level increases further.
A 100mv peak input is about the level of a strummed chord on a telecaster. The stage1 output shown below is still clean at about 4V peak. However, the OD stage output is showing slight cutoff clipping. Recall that the diodes soften cut-off clipping and low-Z drive make high-frequency variations still visible on the compressed positive wave swings. The spectrum now shows strong harmonics of the 220Hz signal (about 22db down) and stronger IM harmonics of the 2200Hz signal (about 15db down). Note again the sharp drop of higher harmonic and IM components. This is where hiss and fizz live - out beyond the nearby harmonic and IM components of the original signal. These higher components are at least 30db down from the input signals. A cluster of components around the 2nd harmonic of the 2200Hz signal is also visible, about 30db down from the fundamental, at about the threshold of aural significance.
A 100mv peak input is about the level of a strummed chord on a telecaster. The stage1 output shown below is still clean at about 4V peak. However, the OD stage output is showing slight cutoff clipping. Recall that the diodes soften cut-off clipping and low-Z drive make high-frequency variations still visible on the compressed positive wave swings. The spectrum now shows strong harmonics of the 220Hz signal (about 22db down) and stronger IM harmonics of the 2200Hz signal (about 15db down). Note again the sharp drop of higher harmonic and IM components. This is where hiss and fizz live - out beyond the nearby harmonic and IM components of the original signal. These higher components are at least 30db down from the input signals. A cluster of components around the 2nd harmonic of the 2200Hz signal is also visible, about 30db down from the fundamental, at about the threshold of aural significance.
As a comparison, the lower spectrum shows the OD stage output without C5. This shows increased relative amplitudes of the higher order components. In addition, the amplitudes of both original signals are slightly reduced. These effects both can contribute to loss of note articulation as the distortion components begin to overwhelm the ear's ability to discern the fundamentals.
As a comparison, the lower spectrum shows the OD stage output without C5. This shows increased relative amplitudes of the higher order components. In addition, the amplitudes of both original signals are slightly reduced. These effects both can contribute to loss of note articulation as the distortion components begin to overwhelm the ear's ability to discern the fundamentals.
A 250mv peak signal is unlikely from a telecaster, but it's well within the range of a G&L S-500 or humbucker pickups. At this level, the traces below show a still-clean 10V peak output from stage1 driving the OD stage. The OD output now shows sharp cutoff clipping with essentially no high frequencies. Reducing R11 can soften cut-off clipping and recover more highs, but that also changes the OD character. The spectrum shows more energy in higher frequencies, but note that relative to the 2200Hz levels, components drop to about 30db down at around 4kHz and they decrease from there. This is a heavy overdrive tone and the tighter the spectrum energy about the input notes, the more articulate it is likely to sound.
A 250mv peak signal is unlikely from a telecaster, but it's well within the range of a G&L S-500 or humbucker pickups. At this level, the traces below show a still-clean 10V peak output from stage1 driving the OD stage. The OD output now shows sharp cutoff clipping with essentially no high frequencies. Reducing R11 can soften cut-off clipping and recover more highs, but that also changes the OD character. The spectrum shows more energy in higher frequencies, but note that relative to the 2200Hz levels, components drop to about 30db down at around 4kHz and they decrease from there. This is a heavy overdrive tone and the tighter the spectrum energy about the input notes, the more articulate it is likely to sound.
For comparison, the lower spectrum shows the spectrum without C5. Note that the 220Hz signal and its nearby harmonics are similar with and without C5. This illustrates the intention behind adding C5 to reduce the distortion components created by higher frequency signals. The 2200Hz note is clearly visible with C5, but it's becoming hard to distinguish without C5 due to the spread and height of component energies. Without C5 the 2200Hz signal itself is reduced about 5db - probably due to higher clipping duty cycle, and the point where IM components are 30db lower isn't reached until ~6kHz.
For comparison, the lower spectrum shows the spectrum without C5. Note that the 220Hz signal and its nearby harmonics are similar with and without C5. This illustrates the intention behind adding C5 to reduce the distortion components created by higher frequency signals. The 2200Hz note is clearly visible with C5, but it's becoming hard to distinguish without C5 due to the spread and height of component energies. Without C5 the 2200Hz signal itself is reduced about 5db - probably due to higher clipping duty cycle, and the point where IM components are 30db lower isn't reached until ~6kHz.
A last image shows a 10V peak signal at S1out (cyan) and the V2 grid (gold). With a normal anode output driving V2, this signal level would cause significant grid clipping and bias shifts - possibly even blocking. The low impedance source drives the V2 grid about 8V positive through C5 and R10. The 220Hz burst is slightly compressed while the 2200Hz signal is clearly visible at both peaks. Bias shifts are small and recover quickly.
A last image shows a 10V peak signal at S1out (cyan) and the V2 grid (gold). With a normal anode output driving V2, this signal level would cause significant grid clipping and bias shifts - possibly even blocking. The low impedance source drives the V2 grid about 8V positive through C5 and R10. The 220Hz burst is slightly compressed while the 2200Hz signal is clearly visible at both peaks. Bias shifts are small and recover quickly.
A final comment - this circuit is just a point in a huge space of possibilities. It sounds uniquely smooth to me, but I lay no claim as the "greatest ever" in OD stages. Tastes and needs vary, so what's more important than any one tone is to have control over OD tone. I can see myself toying with this basic circuit again and again. I expect my tastes and needs will change and I expect to discover new tones I like. I find this LIO OD stage provides many controls over circuit behavior and therefore I'm able to create a greater range of tone options.
A final comment - this circuit is just a point in a huge space of possibilities. It sounds uniquely smooth to me, but I lay no claim as the "greatest ever" in OD stages. Tastes and needs vary, so what's more important than any one tone is to have control over OD tone. I can see myself toying with this basic circuit again and again. I expect my tastes and needs will change and I expect to discover new tones I like. I find this LIO OD stage provides many controls over circuit behavior and therefore I'm able to create a greater range of tone options.
An obvious next step is to consider additional stages - whether LIO or otherwise. Since my current focus is on the BML amp, I'll breadboard this OD circuit and a DC cathode follower (as in the BML amp) and see where that goes.
An obvious next step is to consider additional stages - whether LIO or otherwise. Since my current focus is on the BML amp, I'll breadboard this OD circuit and a DC cathode follower (as in the BML amp) and see where that goes.
NEW: If you read this far, you must have some interest in OD stages. When you're done here, have a look at the OD system for the V-45 versions. The first version uses a novel variable-bias cold clipper followed by a modified DC-coupled cathode follower (DCCF). The latest version uses four parallel triodes in an OD stage. These are some novel examples of many options for OD circuits. Details are posted on the V-45 site.
NEW: If you read this far, you must have some interest in OD stages. When you're done here, have a look at the OD system for the V-45 versions. The first version uses a novel variable-bias cold clipper followed by a modified DC-coupled cathode follower (DCCF). The latest version uses four parallel triodes in an OD stage. These are some novel examples of many options for OD circuits. Details are posted on the V-45 site.
Also, a LIO OD stage similar to what's described here, is used in the Honeydripper Sugarbee amp. See reviews and hear clips on the Honeydripper site. Also, a new page has been added HERE describing methods for tuning the harmonics generated by the DCCF circuit.
Also, a LIO OD stage similar to what's described here, is used in the Honeydripper Sugarbee amp. See reviews and hear clips on the Honeydripper site. Also, a new page has been added HERE describing methods for tuning the harmonics generated by the DCCF circuit.
Lastly, some clips from a telecaster and the S-500. The above OD circuit is bread-boarded on a bench. The output is attenuated and sent to the BML amp input, which is set for clean tone (no internal OD) and bassman EQ settings. Basically, this is what the BML amp should sound like if this OD stage were built into it. Recordings are edited clips from one 30 min session. Room volume throughout is a low level that would easily allow people to have a conversation.
Lastly, some clips from a telecaster and the S-500. The above OD circuit is bread-boarded on a bench. The output is attenuated and sent to the BML amp input, which is set for clean tone (no internal OD) and bassman EQ settings. Basically, this is what the BML amp should sound like if this OD stage were built into it. Recordings are edited clips from one 30 min session. Room volume throughout is a low level that would easily allow people to have a conversation.
A Tascam DR-05 stereo MP3 recorder is about 1-2 foot off-axis of the Celestion Gold and Weber Neo cab. The amp and recorder settings are not changed during the session.
A Tascam DR-05 stereo MP3 recorder is about 1-2 foot off-axis of the Celestion Gold and Weber Neo cab. The amp and recorder settings are not changed during the session.
The first clip is the Tele neck pickup played at max guitar volume and tone settings directly into the BML amp. This gives a reference for what the clean BML guitar/amp sound is like without an OD stage.
The first clip is the Tele neck pickup played at max guitar volume and tone settings directly into the BML amp. This gives a reference for what the clean BML guitar/amp sound is like without an OD stage.
The second clip has the OD stage inserted while playing the same Tele neck pickup at max guitar volume and tone settings. The OD is slight and smooth - after listening for a while it becomes "normal" and you can forget it's the OD tone - it's just a thicker tone.
The second clip has the OD stage inserted while playing the same Tele neck pickup at max guitar volume and tone settings. The OD is slight and smooth - after listening for a while it becomes "normal" and you can forget it's the OD tone - it's just a thicker tone.
The last clip is the S-500 played mostly with the mid and neck pickups at increasing volume settings. The guitar volume starts at about 70% and increases to 100% as the clip progresses. The guitar tone settings are max'd. The OD quality is still smooth, but now it gets more edgy as guitar volume increases and strings are hit harder. Peaks have a slightly cold cutoff ring to them, but the transitions between tone characters are smooth. Compression is noticeable as you're playing, and harmonics are sensitive to fingering.
The last clip is the S-500 played mostly with the mid and neck pickups at increasing volume settings. The guitar volume starts at about 70% and increases to 100% as the clip progresses. The guitar tone settings are max'd. The OD quality is still smooth, but now it gets more edgy as guitar volume increases and strings are hit harder. Peaks have a slightly cold cutoff ring to them, but the transitions between tone characters are smooth. Compression is noticeable as you're playing, and harmonics are sensitive to fingering.