Embedded Files
Update (Oct 2018):
Update (Oct 2018):
The Sugar Bee 1x12 combo amp is based on the AOT design described below. The amp, made by Texas-based Honeydripper Amps, was "Editors Pick" in Guitar Player magazine (Oct 3 2018) and "VG Approved" by Vintage Guitar in January 2019. Video clips here. Audio clips here.
The Sugar Bee 1x12 combo amp is based on the AOT design described below. The amp, made by Texas-based Honeydripper Amps, was "Editors Pick" in Guitar Player magazine (Oct 3 2018) and "VG Approved" by Vintage Guitar in January 2019. Video clips here. Audio clips here.
Princeton-AOT:
Princeton-AOT:
This project includes several new ideas and circuits in music amplifiers. The goal is a low-weight (<30 lbs) mid-power (~40w) combo amp that has a range of good clean and overdrive tones. This is a "go-to amp" that could be used for home, practice sessions, and small venues. That covers just about 99% of what I do with an amp.
This project includes several new ideas and circuits in music amplifiers. The goal is a low-weight (<30 lbs) mid-power (~40w) combo amp that has a range of good clean and overdrive tones. This is a "go-to amp" that could be used for home, practice sessions, and small venues. That covers just about 99% of what I do with an amp.
My Bassman has 45w, the Deluxe has about 20w. Those are good power ranges, but they're both heavy (over 40 lbs). While moving those amps around, I find myself grousing, why does a good sounding amp have to weigh so much? I know, the answer is sort of obvious... they are tube amps, and tube amps require big iron for the power supply transformer and the output transformer.
My Bassman has 45w, the Deluxe has about 20w. Those are good power ranges, but they're both heavy (over 40 lbs). While moving those amps around, I find myself grousing, why does a good sounding amp have to weigh so much? I know, the answer is sort of obvious... they are tube amps, and tube amps require big iron for the power supply transformer and the output transformer.
Once you have heavy transformers, you also need a steel chassis and a solid cabinet to hold things together. All of that adds up to 40-50 lbs in most cases. Light-weight amps exist. 10-15w amps are available based on the 5e3 (Tremolux) or similar circuits, which start around 25Lbs. The 15w Blues Jr. weighs in at 32 Lbs. These are fine amps, but they don't have the range of tone and they don't have 40w of power and headroom.
Once you have heavy transformers, you also need a steel chassis and a solid cabinet to hold things together. All of that adds up to 40-50 lbs in most cases. Light-weight amps exist. 10-15w amps are available based on the 5e3 (Tremolux) or similar circuits, which start around 25Lbs. The 15w Blues Jr. weighs in at 32 Lbs. These are fine amps, but they don't have the range of tone and they don't have 40w of power and headroom.
So the question is, what design is feasible for a flexible 40w amp without requiring big iron in the power supply or output transformer? First consider the need for an output transformer. Over approximately a century of vacuum tube and audio amplifier development, only three approaches have emerged to provide the functions of a push-pull tube amplifier output transformer.
So the question is, what design is feasible for a flexible 40w amp without requiring big iron in the power supply or output transformer? First consider the need for an output transformer. Over approximately a century of vacuum tube and audio amplifier development, only three approaches have emerged to provide the functions of a push-pull tube amplifier output transformer.
First and most common is to use a magnetic core output transformer. This is the approach used in most guitar and bass tube amps to date, and a main contributor to their weight. Examples and variations are described in any textbook, including the well known text, Valve Amplifiers by Morgan Jones(4th Ed) 2012.
First and most common is to use a magnetic core output transformer. This is the approach used in most guitar and bass tube amps to date, and a main contributor to their weight. Examples and variations are described in any textbook, including the well known text, Valve Amplifiers by Morgan Jones(4th Ed) 2012.
The second approach is often called "OTL" or Output Transformer-Less. This approach configures tubes to drive a low-impedance load, such as a loudspeaker or headphones directly, without any transformer. These circuits are limited by the output current available directly from tubes. An example is included in the Morgan text, and the OTL approach is the subject of US Patent #2,773,136, Futterman, Dec. 1959. The OTL approach is impractical for more than a few watts due it's gross inefficiency.
The second approach is often called "OTL" or Output Transformer-Less. This approach configures tubes to drive a low-impedance load, such as a loudspeaker or headphones directly, without any transformer. These circuits are limited by the output current available directly from tubes. An example is included in the Morgan text, and the OTL approach is the subject of US Patent #2,773,136, Futterman, Dec. 1959. The OTL approach is impractical for more than a few watts due it's gross inefficiency.
The third approach employs an active circuit often referred to as a Solid State OutPut Transformer (SSOPT), or ZOTL. An SSOPT chops the input signal at a high frequency, and passes is through a transformer before filtering it to reconstruct the original signal. This approach reduces the size and weight of the transformer, however, a transformer is still required, and it adds the requirement for switching circuits to chop the signal. US Patent No. 5612646 describes a tube amplifier using SSOPT technology. To my knowledge, only Milbert Amplifiers makes a line of these and they're quite complex to design and build and also expensive.
The third approach employs an active circuit often referred to as a Solid State OutPut Transformer (SSOPT), or ZOTL. An SSOPT chops the input signal at a high frequency, and passes is through a transformer before filtering it to reconstruct the original signal. This approach reduces the size and weight of the transformer, however, a transformer is still required, and it adds the requirement for switching circuits to chop the signal. US Patent No. 5612646 describes a tube amplifier using SSOPT technology. To my knowledge, only Milbert Amplifiers makes a line of these and they're quite complex to design and build and also expensive.
I'm introducing a new and novel fourth option. The design described here is an "Active Output Transformer" or AOT. The AOT circuit module mimics the transformer functions in a push pull audio tube amplifier circuit. The AOT requires no transformer and no switching circuits or chopping or reconstruction of the signal. The AOT is an active solid-state circuit implementation of the output transformer's function. The AOT comprises a unique active circuit topology to perform the current gain and phase inversion functions of a magnetic output transformer.
I'm introducing a new and novel fourth option. The design described here is an "Active Output Transformer" or AOT. The AOT circuit module mimics the transformer functions in a push pull audio tube amplifier circuit. The AOT requires no transformer and no switching circuits or chopping or reconstruction of the signal. The AOT is an active solid-state circuit implementation of the output transformer's function. The AOT comprises a unique active circuit topology to perform the current gain and phase inversion functions of a magnetic output transformer.
If you're interested in AOT technical details, read on. Otherwise, skip to the Project Description below or just listen to the clips at the very end of this page.
If you're interested in AOT technical details, read on. Otherwise, skip to the Project Description below or just listen to the clips at the very end of this page.
AOT Details
AOT Details
Vacuum tube amplifiers are used widely for musical instruments and high fidelity applications due to their desirable sonic characteristics. The output stages of these amplifiers are commonly push-pull configurations that require an output transformer to drive the low-impedance load presented by a loudspeaker. Output transformers are heavy and expensive components. Therefore, a substitute for output transformers is desirable and much sought after.
Vacuum tube amplifiers are used widely for musical instruments and high fidelity applications due to their desirable sonic characteristics. The output stages of these amplifiers are commonly push-pull configurations that require an output transformer to drive the low-impedance load presented by a loudspeaker. Output transformers are heavy and expensive components. Therefore, a substitute for output transformers is desirable and much sought after.
The Active Output Transformer (AOT) circuit replaces a conventional passive output transformer in push-pull vacuum tube audio amplifiers. New amplifier designs can use an AOT to achieve cost savings and lower weight. Existing amplifiers can replace the output transformer with an AOT to reduce weight and improve performance.
The Active Output Transformer (AOT) circuit replaces a conventional passive output transformer in push-pull vacuum tube audio amplifiers. New amplifier designs can use an AOT to achieve cost savings and lower weight. Existing amplifiers can replace the output transformer with an AOT to reduce weight and improve performance.
Conventional passive transformers are used in push-pull configurations such as those shown in Figures 1 and 2. Figure 1 shows two output tubes (110, 111) in a push-pull configuration, driving an output transformer (120). Two input signals (100, 101) are fed to the tube grids. Nominally, these inputs are out of phase with each other, as illustrated by the waveforms (100, 101) in Figure 1. Figure 1 shows the output transformer (120) driven by currents from the tube anodes. This configuration requires a positive supply voltage (+V) connected to the center tap of the output transformer primary.
Conventional passive transformers are used in push-pull configurations such as those shown in Figures 1 and 2. Figure 1 shows two output tubes (110, 111) in a push-pull configuration, driving an output transformer (120). Two input signals (100, 101) are fed to the tube grids. Nominally, these inputs are out of phase with each other, as illustrated by the waveforms (100, 101) in Figure 1. Figure 1 shows the output transformer (120) driven by currents from the tube anodes. This configuration requires a positive supply voltage (+V) connected to the center tap of the output transformer primary.
Figure 2 shows the same input signals as shown in Figure 1, but in Figure 2, the output transformer is driven by currents from the tube cathodes (210, 211). In this relatively uncommon configuration, the center tap of the output transformer primary is connected to a low voltage supply, or ground.
Figure 2 shows the same input signals as shown in Figure 1, but in Figure 2, the output transformer is driven by currents from the tube cathodes (210, 211). In this relatively uncommon configuration, the center tap of the output transformer primary is connected to a low voltage supply, or ground.
In both of the push pull configurations (Figures 1 and 2), a center-tapped primary provides a phase inversion for combining the two phases of the output tube signals. In both configurations, two out-of-phase signals drive the two transformer inputs. These two transformer inputs route their currents in opposite directions through the primary coil, generating inverted magnetic fields in the core. Their combined field in the magnetic core produces a single phase aggregate output signal. Mathematically, the output transformer implements the following relationship,
In both of the push pull configurations (Figures 1 and 2), a center-tapped primary provides a phase inversion for combining the two phases of the output tube signals. In both configurations, two out-of-phase signals drive the two transformer inputs. These two transformer inputs route their currents in opposite directions through the primary coil, generating inverted magnetic fields in the core. Their combined field in the magnetic core produces a single phase aggregate output signal. Mathematically, the output transformer implements the following relationship,
O = G x (A-B) [1]
O = G x (A-B) [1]
where O is the output signal current into a load, G is the current gain constant dependent on the transformer design, A is the signal current fed into one of the inputs, and B is the signal current fed into the other input. Both current polarities are determined by whether the currents flow into or out of their respective input terminals.
where O is the output signal current into a load, G is the current gain constant dependent on the transformer design, A is the signal current fed into one of the inputs, and B is the signal current fed into the other input. Both current polarities are determined by whether the currents flow into or out of their respective input terminals.
The Active Output Transformer (300, 400) replaces a magnetic core output transformer with a unique active circuit. The AOT implements Equation 1, and therefore it replaces the transformer in push pull tube amplifier configurations as shown in Figures 3 and 4. Figure 3 shows an anode drive configuration, similar to Figure 1. Figure 4 shows a cathode drive configuration, similar to Figure 2. Because the AOT is an active circuit, it requires a DC power source, as noted by connections for a positive voltage (+B) source and a low voltage return (-B).
The Active Output Transformer (300, 400) replaces a magnetic core output transformer with a unique active circuit. The AOT implements Equation 1, and therefore it replaces the transformer in push pull tube amplifier configurations as shown in Figures 3 and 4. Figure 3 shows an anode drive configuration, similar to Figure 1. Figure 4 shows a cathode drive configuration, similar to Figure 2. Because the AOT is an active circuit, it requires a DC power source, as noted by connections for a positive voltage (+B) source and a low voltage return (-B).
Figure 5 shows the equivalence between the magnetic transformer and the AOT. Both have primary inputs A and B and a primary center tap. Both have two output terminals to provide current to the load and a return current path. The AOT requires a DC power supply providing a positive voltage (VH) and a relative negative voltage (VL), or return current path to the power supply. A normal transformer can work with any tube type. Common output tubes are triodes, tetrodes, or pentodes. In fact, different tube types can be used for the two output tubes in the same amplifier. Similarly, the AOT can work with any tube type, or mixed tube types, in a push pull configuration.
Figure 5 shows the equivalence between the magnetic transformer and the AOT. Both have primary inputs A and B and a primary center tap. Both have two output terminals to provide current to the load and a return current path. The AOT requires a DC power supply providing a positive voltage (VH) and a relative negative voltage (VL), or return current path to the power supply. A normal transformer can work with any tube type. Common output tubes are triodes, tetrodes, or pentodes. In fact, different tube types can be used for the two output tubes in the same amplifier. Similarly, the AOT can work with any tube type, or mixed tube types, in a push pull configuration.
Project Description:
Project Description:
Jan 16, 2016 Note - version 2 (v2) circuit described lower on this page
Jan 16, 2016 Note - version 2 (v2) circuit described lower on this page
March 31, 2016 Note - version 3 (v3) circuit described at the end of this page
March 31, 2016 Note - version 3 (v3) circuit described at the end of this page
The AOT circuit was developed and tested prior to this project. The AOT approach eliminates the output transformer. It also reduces the need for a high-power high-voltage supply. Basically, the tube portion of an AOT design is a preamp, a phase inverter and a low power push-pull stage to drive the AOT module.
The AOT circuit was developed and tested prior to this project. The AOT approach eliminates the output transformer. It also reduces the need for a high-power high-voltage supply. Basically, the tube portion of an AOT design is a preamp, a phase inverter and a low power push-pull stage to drive the AOT module.
Since these require little power from a high-voltage power supply, only a small high-voltage power transformer is needed for an AOT amplifier. This leads to dramatic weight loss, and it allows use of a light-weight aluminum chassis, and a lighter cabinet.
Since these require little power from a high-voltage power supply, only a small high-voltage power transformer is needed for an AOT amplifier. This leads to dramatic weight loss, and it allows use of a light-weight aluminum chassis, and a lighter cabinet.
The project goal is for ~40w output into a full sized 12" speaker. This requires a cabinet about the size of a Princeton Reverb. So, the Princeton cab became the project target. Princeton cabs and blank chassis are readily available.
The project goal is for ~40w output into a full sized 12" speaker. This requires a cabinet about the size of a Princeton Reverb. So, the Princeton cab became the project target. Princeton cabs and blank chassis are readily available.
A 4 ohm speaker is needed for max power and Jenson offers a 12" Neo Jet in a 4 ohm version. It's rated at 100w RMS and its ~4.5 Lbs weight is great for this project. The AOT output stage requires a DC supply. A 24-volt, 4-amp switching supply provides the needed power. Such supplies are widely available as a laptop chargers. I purchased one for around $10 on ebay.
A 4 ohm speaker is needed for max power and Jenson offers a 12" Neo Jet in a 4 ohm version. It's rated at 100w RMS and its ~4.5 Lbs weight is great for this project. The AOT output stage requires a DC supply. A 24-volt, 4-amp switching supply provides the needed power. Such supplies are widely available as a laptop chargers. I purchased one for around $10 on ebay.
So, the cab and speaker are selected and the AOT output stage and its DC power supply are known. The major remaining issues relate to the preamp design and overall chassis construction. This preamp design incorporates several features from my prior designs and a few new ones. The schematic is below. I'll work backwards through the design starting with the power supply, output stage, phase inverter, and finally the preamp.
So, the cab and speaker are selected and the AOT output stage and its DC power supply are known. The major remaining issues relate to the preamp design and overall chassis construction. This preamp design incorporates several features from my prior designs and a few new ones. The schematic is below. I'll work backwards through the design starting with the power supply, output stage, phase inverter, and finally the preamp.
Power Supply
Power Supply
As described above, the power supply requirements for an AOT design are a bit different from traditional tube amps. There is no high-current, high-power, high voltage needed for the output tubes. It's basically only a preamp power supply. This amp design settled on 7 tubes, for which 1.8 amps of filament current are needed. A preamp stage doesn't need 400V for B+ to work properly or sound good. A relatively low 200V supply is fine. The design target was 25mA at 200V, and the final circuit runs at slightly lower current, consuming below 5 watts from the 200V supply. That makes the total power consumption below 25w, so a 25-30VA power transformer would be adequate. The smallest I could find was a 50VA toroid, so I have a 100% safety margin and reserves for filament startup surges. It's no surprise that the power transformer runs stone cold.
As described above, the power supply requirements for an AOT design are a bit different from traditional tube amps. There is no high-current, high-power, high voltage needed for the output tubes. It's basically only a preamp power supply. This amp design settled on 7 tubes, for which 1.8 amps of filament current are needed. A preamp stage doesn't need 400V for B+ to work properly or sound good. A relatively low 200V supply is fine. The design target was 25mA at 200V, and the final circuit runs at slightly lower current, consuming below 5 watts from the 200V supply. That makes the total power consumption below 25w, so a 25-30VA power transformer would be adequate. The smallest I could find was a 50VA toroid, so I have a 100% safety margin and reserves for filament startup surges. It's no surprise that the power transformer runs stone cold.
The AC mains drives a switched AC socket that's handy for an FX unit or pedal power. A Molex connector feeds switched mains power to the 24V DC supply. That DC supply is mounted outside the chassis. The 24V power is fed back into the chassis on the same connector. An external supply makes sense to save chassis space and allow complete freedom in choosing a supply since supplies do vary in size and shape. Any 24V 4A supply would work.
The AC mains drives a switched AC socket that's handy for an FX unit or pedal power. A Molex connector feeds switched mains power to the 24V DC supply. That DC supply is mounted outside the chassis. The 24V power is fed back into the chassis on the same connector. An external supply makes sense to save chassis space and allow complete freedom in choosing a supply since supplies do vary in size and shape. Any 24V 4A supply would work.
The 200V supply circuit uses a standard bridge and RC filter. Due to the relatively low voltage and low current consumption, big caps are not needed. The ones I used fit easily into the chassis. Two outputs are provided. B+ is used to power the phase inverter and output tubes. These stages are relatively immune to ripple, so the residual ripple on B+ is not critical. The C+ output powers the preamp where ripple matters a great deal since the preamp stages are high gain and cascaded. There is no ripple (<5mV) that I can measure on the C+ line.
The 200V supply circuit uses a standard bridge and RC filter. Due to the relatively low voltage and low current consumption, big caps are not needed. The ones I used fit easily into the chassis. Two outputs are provided. B+ is used to power the phase inverter and output tubes. These stages are relatively immune to ripple, so the residual ripple on B+ is not critical. The C+ output powers the preamp where ripple matters a great deal since the preamp stages are high gain and cascaded. There is no ripple (<5mV) that I can measure on the C+ line.
The filaments are elevated to the 24V supply. Raising them adds zero cost and can only help reduce preamp hum. There is a single diode rectifier and large filter cap driven by the 6.3V filament supply. This develops about 7-8 volts to power a small fan. Low cost plastic 12V fans for PC cooling are widely available. A small (lightweight) ~4" fan runs slowly (quietly) to cool the chassis mounted AOT output transistors so there is no need for an additional heat sink, saving that cost and weight.
The filaments are elevated to the 24V supply. Raising them adds zero cost and can only help reduce preamp hum. There is a single diode rectifier and large filter cap driven by the 6.3V filament supply. This develops about 7-8 volts to power a small fan. Low cost plastic 12V fans for PC cooling are widely available. A small (lightweight) ~4" fan runs slowly (quietly) to cool the chassis mounted AOT output transistors so there is no need for an additional heat sink, saving that cost and weight.
AOT Output Stage
AOT Output Stage
The AOT output stage is basically two unity-gain current amplifiers driving a speaker in differential mode. The important functions of this circuit are: 1) the transistors provide current gain to drive the low impedance speaker, and 2) the outputs follow the voltages and currents produced by the output tubes V7a and V7b. This circuit implements the function of the output transformer as described in Equation 1.
The AOT output stage is basically two unity-gain current amplifiers driving a speaker in differential mode. The important functions of this circuit are: 1) the transistors provide current gain to drive the low impedance speaker, and 2) the outputs follow the voltages and currents produced by the output tubes V7a and V7b. This circuit implements the function of the output transformer as described in Equation 1.
The AOT transistors operate in their linear range so there is no clipping or overdrive that colors the waveforms produced by V7. Each amplifier only has to swing about +/- 8 volts peak to produce 16V peak signals across the speaker for a 40w output. Bias in each amplifier is adjusted to ~25mA using P12 and P13. V7a and b operate as the output tubes whose cathodes drive out of phase signals into the AOT amplifier circuits. R75 and R84 are hand-selected to trim the two V7 cathode bias currents to ~3mA. The AOT output transistors (Q9, 11, 13, 15) are chassis mounted with the cooling fan directly above them. Switch5 can bypass a 4-ohm resistor that cuts the power to the speaker by 75%, so the speaker output is either 40w or 10w RMS. The 4-ohm 25w resistor is also chassis mounted near the fan. Note that neither speaker terminal is grounded so an insulated jack is needed for the speaker connection.
The AOT transistors operate in their linear range so there is no clipping or overdrive that colors the waveforms produced by V7. Each amplifier only has to swing about +/- 8 volts peak to produce 16V peak signals across the speaker for a 40w output. Bias in each amplifier is adjusted to ~25mA using P12 and P13. V7a and b operate as the output tubes whose cathodes drive out of phase signals into the AOT amplifier circuits. R75 and R84 are hand-selected to trim the two V7 cathode bias currents to ~3mA. The AOT output transistors (Q9, 11, 13, 15) are chassis mounted with the cooling fan directly above them. Switch5 can bypass a 4-ohm resistor that cuts the power to the speaker by 75%, so the speaker output is either 40w or 10w RMS. The 4-ohm 25w resistor is also chassis mounted near the fan. Note that neither speaker terminal is grounded so an insulated jack is needed for the speaker connection.
Phase Inverter
Phase Inverter
This PI employs an active current source (Q5) with a cascode element (Q6). It's effective impedance is high... probably 10's of megohms. I have not bothered to calculate or measure it. Two pentodes (V5 and V6) share the current source and their output balance is quite good. On a scope image of the two output signals, their amplitudes are equal. The pentode grids are biased with a +12V reference to provide head room for the current source.
This PI employs an active current source (Q5) with a cascode element (Q6). It's effective impedance is high... probably 10's of megohms. I have not bothered to calculate or measure it. Two pentodes (V5 and V6) share the current source and their output balance is quite good. On a scope image of the two output signals, their amplitudes are equal. The pentode grids are biased with a +12V reference to provide head room for the current source.
The PI outputs are DC coupled to the AOT output tubes. Ideally, the quiescent PI outputs sit at the DC point midway between the PI clipping levels. On a scope, the clipped outputs swing roughly between 35V and 175V. P10 adjusts the current source, and therefore the DC anode voltages. P11 is used to compensate for any imbalance between the tubes and components. The gain of each PI output is about 30, due to the relatively low values for R46 and R53. High gain is not needed in the PI stage since the AOT stage has unity gain, and only 8 volt peak signals are needed on the V7 grids to produce a 40w output.
The PI outputs are DC coupled to the AOT output tubes. Ideally, the quiescent PI outputs sit at the DC point midway between the PI clipping levels. On a scope, the clipped outputs swing roughly between 35V and 175V. P10 adjusts the current source, and therefore the DC anode voltages. P11 is used to compensate for any imbalance between the tubes and components. The gain of each PI output is about 30, due to the relatively low values for R46 and R53. High gain is not needed in the PI stage since the AOT stage has unity gain, and only 8 volt peak signals are needed on the V7 grids to produce a 40w output.
The screen resistors and caps to ground form an RC time constant of a few hundred milliseconds. The whole point of using pentodes for the PI is for the dynamic screen voltage shifts that occur during overdrive. Screen current starts increasing at input levels of about 2 volts peak. With a cathode current source, the only free variable is cathode voltage, so the effect of increased screen current (and reduced screen voltage) is for the cathode voltage to drop, warming the bias of the tube. These dynamic changes to the tube operating parameters alter the tone slightly, and the screen RC time constant controls the attack and decay rates of these changes. A lower cost PI implementation could use simple triodes (12at7), and do without the screen dynamics.
The screen resistors and caps to ground form an RC time constant of a few hundred milliseconds. The whole point of using pentodes for the PI is for the dynamic screen voltage shifts that occur during overdrive. Screen current starts increasing at input levels of about 2 volts peak. With a cathode current source, the only free variable is cathode voltage, so the effect of increased screen current (and reduced screen voltage) is for the cathode voltage to drop, warming the bias of the tube. These dynamic changes to the tube operating parameters alter the tone slightly, and the screen RC time constant controls the attack and decay rates of these changes. A lower cost PI implementation could use simple triodes (12at7), and do without the screen dynamics.
It's worth noting that there are no coupling caps between the PI grids and the speaker output caps. DC coupling in these stages gives a tighter and deeper bass at high volumes than most amps offer. It's an immediately noticeable difference. The amp sounds "bigger", and feels more responsive. The low strings also sound more articulate - they're not lost in "mush".
It's worth noting that there are no coupling caps between the PI grids and the speaker output caps. DC coupling in these stages gives a tighter and deeper bass at high volumes than most amps offer. It's an immediately noticeable difference. The amp sounds "bigger", and feels more responsive. The low strings also sound more articulate - they're not lost in "mush".
With good bass response in the PI and AOT stages, there is also no need for negative feedback to the PI from the output stage. The pentodes and dc-coupled cathode follower output tubes add their own colors to the tone and their own distortions under high-signal drive. Negative feedback would remove much of that character and alter the distortion tone as well. The PI-AOT stages perform best when operated open loop.
With good bass response in the PI and AOT stages, there is also no need for negative feedback to the PI from the output stage. The pentodes and dc-coupled cathode follower output tubes add their own colors to the tone and their own distortions under high-signal drive. Negative feedback would remove much of that character and alter the distortion tone as well. The PI-AOT stages perform best when operated open loop.
Preamp
Preamp
The preamp has three gain stages. Each can operate with or without distortion. This design includes attenuation and tone control between all stages. The first stage is a conventional triode V1a with a gain of ~50. It's biased warm and fully bypassed to minimize heater hum and maximize gain. The V1a output is clean with plenty of headroom. This signal drives a second triode V1b that has an unbypassed colder bias. The cascaded triode gain easily produces clipping in V1b (due mainly to cutoff) with a distinctive metallic aggressive tone. The guitar volume control determines how aggressive V1b clipping gets. SW1 (voice) selects between the V1a (warm) or V1b (aggressive) signals. Note that the clipped V1b signal is attenuated (R7 and R11) so it's level at the switch is about the same as the V1a signal.
The preamp has three gain stages. Each can operate with or without distortion. This design includes attenuation and tone control between all stages. The first stage is a conventional triode V1a with a gain of ~50. It's biased warm and fully bypassed to minimize heater hum and maximize gain. The V1a output is clean with plenty of headroom. This signal drives a second triode V1b that has an unbypassed colder bias. The cascaded triode gain easily produces clipping in V1b (due mainly to cutoff) with a distinctive metallic aggressive tone. The guitar volume control determines how aggressive V1b clipping gets. SW1 (voice) selects between the V1a (warm) or V1b (aggressive) signals. Note that the clipped V1b signal is attenuated (R7 and R11) so it's level at the switch is about the same as the V1a signal.
The V1a output also drives a built in tuner. This is a module made by Crate. Power is controlled by a micro toggle on the front panel. I don't have any specs on the tuner module itself, but its made to work with a 9V battery, so I provide about 8V via Q1. Measured current consumption varies from a few milliamps to ~100mA depending on its state. It's handy to have a tuner in the amp - less to carry and no cables and batteries to fuss with.
The V1a output also drives a built in tuner. This is a module made by Crate. Power is controlled by a micro toggle on the front panel. I don't have any specs on the tuner module itself, but its made to work with a 9V battery, so I provide about 8V via Q1. Measured current consumption varies from a few milliamps to ~100mA depending on its state. It's handy to have a tuner in the amp - less to carry and no cables and batteries to fuss with.
The voice switch (SW1) feeds the selected signal into a 3 option EQ section. These options are flat response (Full), bass attenuation (Firm), or bass attenuation and high boost (Brite). These EQ options are useful for shaping the signal that drives the next two stages, which may run clean or create heavy distortions. The three response curves are plotted below.
The voice switch (SW1) feeds the selected signal into a 3 option EQ section. These options are flat response (Full), bass attenuation (Firm), or bass attenuation and high boost (Brite). These EQ options are useful for shaping the signal that drives the next two stages, which may run clean or create heavy distortions. The three response curves are plotted below.
The EQ output feeds the Drive control (P1) which controls how hard the next stage is driven. In addition, the P1 output also feeds an FX output buffered by Q2 and Q3. The FX output level is ~1.5V peak when the Drive control is max'd at 10. A shorting jack for the FX input passes this signal to the FX input by default. Note that the FX loop bypasses the tone control section.
The EQ output feeds the Drive control (P1) which controls how hard the next stage is driven. In addition, the P1 output also feeds an FX output buffered by Q2 and Q3. The FX output level is ~1.5V peak when the Drive control is max'd at 10. A shorting jack for the FX input passes this signal to the FX input by default. Note that the FX loop bypasses the tone control section.
The Drive control feeds the second preamp stage V2, which is a pentode. This stage is cathode biased and fully bypassed. The screen RC time constants are again set to be a few hundred milliseconds to control the attack and decay times of the overdrive screen shifts. In this stage, bias voltage and anode current vary as the screen voltage drops due to overdrive. The stage gain is ~100, so it enters clipping with ~0.5V peak inputs, which are typical once the Drive control exceeds 5 or 6 (with a telecaster-level input signal).
The Drive control feeds the second preamp stage V2, which is a pentode. This stage is cathode biased and fully bypassed. The screen RC time constants are again set to be a few hundred milliseconds to control the attack and decay times of the overdrive screen shifts. In this stage, bias voltage and anode current vary as the screen voltage drops due to overdrive. The stage gain is ~100, so it enters clipping with ~0.5V peak inputs, which are typical once the Drive control exceeds 5 or 6 (with a telecaster-level input signal).
The V2 output is dc coupled to V3a, a cathode follower that feeds a four-control tone stack. The direct-coupled cathode follower has known tone coloring qualities, and the tone stack is shamelessly lifted from Merlin Blencowe's book (2nd Ed). It offers a wide range of useful EQ curves that shape the tone at this point in the preamp signal flow.
The V2 output is dc coupled to V3a, a cathode follower that feeds a four-control tone stack. The direct-coupled cathode follower has known tone coloring qualities, and the tone stack is shamelessly lifted from Merlin Blencowe's book (2nd Ed). It offers a wide range of useful EQ curves that shape the tone at this point in the preamp signal flow.
The tone stack output feeds a Volume control (P6), which controls the signal level to a cascode mixer stage. V4b is warm biased and fully bypassed. It's anode current is summed with the current from V4a, the FX input, at the cathode of V3b. The anode of V3b produces an output voltage for the combined currents.
The tone stack output feeds a Volume control (P6), which controls the signal level to a cascode mixer stage. V4b is warm biased and fully bypassed. It's anode current is summed with the current from V4a, the FX input, at the cathode of V3b. The anode of V3b produces an output voltage for the combined currents.
While cascode stages can easily achieve high gains of 200 or more, (see my Deluxe Plus preamp) this stage uses a low anode load resistor (R31, 32) for a gain of ~40. Even at this modest gain, the stage is easily over-driven when the Volume control is set beyond 5 or 6. An important feature of the cascode configuration is the RC time constant for R33 and C16. These control the attack and decay of overdrive parameter changes. The overdrive behaviors of this cascode configuration are described in Blencowe's book and in my book (GAO), so I won't repeat that discussion, but the point is, this stage also has a dynamic response to being over-driven.
While cascode stages can easily achieve high gains of 200 or more, (see my Deluxe Plus preamp) this stage uses a low anode load resistor (R31, 32) for a gain of ~40. Even at this modest gain, the stage is easily over-driven when the Volume control is set beyond 5 or 6. An important feature of the cascode configuration is the RC time constant for R33 and C16. These control the attack and decay of overdrive parameter changes. The overdrive behaviors of this cascode configuration are described in Blencowe's book and in my book (GAO), so I won't repeat that discussion, but the point is, this stage also has a dynamic response to being over-driven.
The FX input signal is buffered and amplified by Q4. This transistor provides a gain of ~20 and a phase inversion to match the signal levels and phase at the Volume control. When no FX cables are plugged in, the FX (P7) signal level roughly matches the Volume (P6) levels. V4a is center biased and fully bypassed to roughly equal the gain of V4b in the cascode mixer.
The FX input signal is buffered and amplified by Q4. This transistor provides a gain of ~20 and a phase inversion to match the signal levels and phase at the Volume control. When no FX cables are plugged in, the FX (P7) signal level roughly matches the Volume (P6) levels. V4a is center biased and fully bypassed to roughly equal the gain of V4b in the cascode mixer.
SW3 (the Mid switch) works with the Tilt control (P8) to provide a final stage of EQ control prior to the power amp. The Tilt control raises one end of the spectrum while lowering the other. The Tilt middle and two extreme responses are shown below for the case when SW3 is in the Flat position.
SW3 (the Mid switch) works with the Tilt control (P8) to provide a final stage of EQ control prior to the power amp. The Tilt control raises one end of the spectrum while lowering the other. The Tilt middle and two extreme responses are shown below for the case when SW3 is in the Flat position.
Flipping S3 to the Peak or Scoop position impacts the middle spectrum (~350Hz) by either boosting or scooping it out as shown in the response below for the case where the Tilt control is in its middle position. The combination of SW3 and the Tilt control allow an important opportunity to control the post-distortion equalization of the preamp output signal. The Master volume control P9, provides the usual control over how much preamp signal is fed to the power amp.
Flipping S3 to the Peak or Scoop position impacts the middle spectrum (~350Hz) by either boosting or scooping it out as shown in the response below for the case where the Tilt control is in its middle position. The combination of SW3 and the Tilt control allow an important opportunity to control the post-distortion equalization of the preamp output signal. The Master volume control P9, provides the usual control over how much preamp signal is fed to the power amp.
EQ and Dynamics
EQ and Dynamics
With level and tone adjustments between four gain stages, there is complete control over which stages run clean and which are over-driven. The V1b triode, the V2 pentode, the V4 cascode, and the V5,6 phase inverter all can be selectively, or in combination, pushed to overdrive. This provides a wide range of tone and feel qualities since each stage distorts and sounds different. Also, as described above, three of these stages are explicitly designed so their operating parameters shift when over-driven. So the combination of EQs, levels, distortion sources, and dynamic parameter shifts provide a HUGE palette of tone and feel characteristics to this amp. When needed, an FX unit (see the spring reverb page) plugged into the FX jacks adds a nice effect.
With level and tone adjustments between four gain stages, there is complete control over which stages run clean and which are over-driven. The V1b triode, the V2 pentode, the V4 cascode, and the V5,6 phase inverter all can be selectively, or in combination, pushed to overdrive. This provides a wide range of tone and feel qualities since each stage distorts and sounds different. Also, as described above, three of these stages are explicitly designed so their operating parameters shift when over-driven. So the combination of EQs, levels, distortion sources, and dynamic parameter shifts provide a HUGE palette of tone and feel characteristics to this amp. When needed, an FX unit (see the spring reverb page) plugged into the FX jacks adds a nice effect.
Conclusions
Conclusions
IMO the amp is hugely successful in terms of all my goals. The range of tone is very satisfying. Good clean tone is always important and this amp has sparkling clean tone, and distortion-galore is just a knob tweak away. It's really quiet too, despite all its gain stages. Oh yeah, it's also really loud. It competes easily with the 6L6-pair amps I've encountered. Last but not least, it weighs under 24 lbs. This amp is easy to carry, fun to play, and there are no output tubes to worry about burning out. The tubes that are used are operating so far under their limits, I consider them basically permanent components.
IMO the amp is hugely successful in terms of all my goals. The range of tone is very satisfying. Good clean tone is always important and this amp has sparkling clean tone, and distortion-galore is just a knob tweak away. It's really quiet too, despite all its gain stages. Oh yeah, it's also really loud. It competes easily with the 6L6-pair amps I've encountered. Last but not least, it weighs under 24 lbs. This amp is easy to carry, fun to play, and there are no output tubes to worry about burning out. The tubes that are used are operating so far under their limits, I consider them basically permanent components.
Construction
Construction
Below are a series of pictures showing the build sequence and some of the details of component layout. Planning is key to building something from scratch. I do all the layouts to scale in PowerPoint to verify clearances on everything.
Below are a series of pictures showing the build sequence and some of the details of component layout. Planning is key to building something from scratch. I do all the layouts to scale in PowerPoint to verify clearances on everything.
The planning took far longer than the build - but that's the way you want it. Finding and fixing problems in planning is easy, not so during the build. I'm pleased that everything worked out as planned and there were no last minute D'ohs.
The planning took far longer than the build - but that's the way you want it. Finding and fixing problems in planning is easy, not so during the build. I'm pleased that everything worked out as planned and there were no last minute D'ohs.
Cut all the needed holes in the aluminum chassis and mount AOT transistor sockets and circuit board standoffs.
Cut all the needed holes in the aluminum chassis and mount AOT transistor sockets and circuit board standoffs.
Mount the transformer and rear panel parts.
Mount the transformer and rear panel parts.
A 1/8-inch glass-epoxy board provides stiffness and insulation. The transistor socket leads poke through holes for access.
A 1/8-inch glass-epoxy board provides stiffness and insulation. The transistor socket leads poke through holes for access.
The eyelet board sits on the glass-epoxy board and the transistor socket lugs are exposed by drilling their pattern into the eyelet board. Tube sockets are in place. Molex connectors for the DC supply and fan are press-fit.
The eyelet board sits on the glass-epoxy board and the transistor socket lugs are exposed by drilling their pattern into the eyelet board. Tube sockets are in place. Molex connectors for the DC supply and fan are press-fit.
Top view of chassis with connectors for fan and AOT DC supply. The gold resistor is R66.
Top view of chassis with connectors for fan and AOT DC supply. The gold resistor is R66.
Wire the AC mains and 200v supply. Wire connector to DC supply. Wire filaments to all sockets.
Wire the AC mains and 200v supply. Wire connector to DC supply. Wire filaments to all sockets.
Mount transistors to chassis and mount fan above them using standoffs. The output tube V7 and AOT module is complete.
Mount transistors to chassis and mount fan above them using standoffs. The output tube V7 and AOT module is complete.
The AOT circuitry is built around transistor sockets and output tube socket (V7).
The AOT circuitry is built around transistor sockets and output tube socket (V7).
Completed AOT and phase inverter circuitry. The green caps are the speaker output caps. Front panel components are mounted.
Completed AOT and phase inverter circuitry. The green caps are the speaker output caps. Front panel components are mounted.
Preamp stages (V1, V2) are completed.
Preamp stages (V1, V2) are completed.
All circuitry is completed and tested and tuned by listening and playing it for about 2 weeks. (The NanoVerb2 FX unit is visible)
All circuitry is completed and tested and tuned by listening and playing it for about 2 weeks. (The NanoVerb2 FX unit is visible)
Chassis and Jensen Jet fit into the cabinet as planned.
Chassis and Jensen Jet fit into the cabinet as planned.
Front view looks good, but deceptively tame and harmless. This thing roars when you turn it up.
Front view looks good, but deceptively tame and harmless. This thing roars when you turn it up.
I've pulled a few clips from a session I recorded with a Tascam DR-05 (an inexpensive portable recorder). I just put the recorder in front of the amp about two feet out and to the side. I played random licks and song segments and set the controls to a variety of positions. There's no system to it, but below are some examples of what it sounded like.
I've pulled a few clips from a session I recorded with a Tascam DR-05 (an inexpensive portable recorder). I just put the recorder in front of the amp about two feet out and to the side. I played random licks and song segments and set the controls to a variety of positions. There's no system to it, but below are some examples of what it sounded like.
You can tell I was playing pretty easy most of the time. The room level was modest. It would have been easy to have a conversation while I was playing. I played a stock US Telecaster with the vol/tone at max and the neck pickup. A NanoVerb2 FX unit is adding a bit of plate-reverb.
You can tell I was playing pretty easy most of the time. The room level was modest. It would have been easy to have a conversation while I was playing. I played a stock US Telecaster with the vol/tone at max and the neck pickup. A NanoVerb2 FX unit is adding a bit of plate-reverb.
UPDATE 10/21/15
UPDATE 10/21/15
I've added an outboard spring reverb described here. It's mounted in the cab and "permanently" connected to the effects loop. That module provides a high quality reverb.
I've added an outboard spring reverb described here. It's mounted in the cab and "permanently" connected to the effects loop. That module provides a high quality reverb.
UPDATE 1/16/16: P-AOT v2
UPDATE 1/16/16: P-AOT v2
After many months of playing the initial P-AOT, a few behaviors began to bother me, mainly at high volume levels. After a some investigation, I made several changes to produce P-AOTv2 (version 2). It's not that version 1 was bad, just that version 2 is better for my tastes and needs. It also incorporates some of the ideas I wrote about in the Guitar Amplifier Overdrive (GAO) book.
After many months of playing the initial P-AOT, a few behaviors began to bother me, mainly at high volume levels. After a some investigation, I made several changes to produce P-AOTv2 (version 2). It's not that version 1 was bad, just that version 2 is better for my tastes and needs. It also incorporates some of the ideas I wrote about in the Guitar Amplifier Overdrive (GAO) book.
I've already described the spring reverb in the 10/21/15 update above. While it's external to the P-AOT amp chassis, it was designed and constructed as part of the version 2 effort. I've also added tremolo in version 2, although I'm not overjoyed at the outcome. The tremolo tone is not the problem - it's actually pretty good. The problem is control. The pots I used don't provide smooth control - they're not tapered as advertised. They have sudden jumps in resistance. If you implement this circuit with standard Alpha pots (or similar) it will probably be fine. I used mini pots since I had to mount the controls on the rear panel due to lack of front panel space. Basically, I'm cramming too much into this chassis to do it right. That's a hazard of incremental mods.
I've already described the spring reverb in the 10/21/15 update above. While it's external to the P-AOT amp chassis, it was designed and constructed as part of the version 2 effort. I've also added tremolo in version 2, although I'm not overjoyed at the outcome. The tremolo tone is not the problem - it's actually pretty good. The problem is control. The pots I used don't provide smooth control - they're not tapered as advertised. They have sudden jumps in resistance. If you implement this circuit with standard Alpha pots (or similar) it will probably be fine. I used mini pots since I had to mount the controls on the rear panel due to lack of front panel space. Basically, I'm cramming too much into this chassis to do it right. That's a hazard of incremental mods.
The P-AOTv2 input stages are completely different from version 1. In version 2, V1a (the warm channel) is colder biased and produces a cleaner, more vintage tone. It's higher gain is partially offset by the split-load resistors, R3 and R4. The aggressive channel distortion created by V1b in the version 1 circuit is replaced by a LIO circuit. LIO stands for Low-Impedance Overdrive. LIO is something I describe in the GAO book, where Ch4 has more details about it. The LIO stage uses Q17 as a follower to provide low impedance drive to V1b.
The P-AOTv2 input stages are completely different from version 1. In version 2, V1a (the warm channel) is colder biased and produces a cleaner, more vintage tone. It's higher gain is partially offset by the split-load resistors, R3 and R4. The aggressive channel distortion created by V1b in the version 1 circuit is replaced by a LIO circuit. LIO stands for Low-Impedance Overdrive. LIO is something I describe in the GAO book, where Ch4 has more details about it. The LIO stage uses Q17 as a follower to provide low impedance drive to V1b.
The distortion character is strongly controlled by bias and R85. In this case, a 2.2k resistor gives V1b a very gradual and symmetric distortion. The tone is hard to describe since I don't know of any LIO circuits in common use. That's too bad, IMO, since it sounds very pleasing to me. If you try a LIO stage, I encourage you to play with R85 to find the value(s) most pleasing to you. Values from 0 to 100K will give wide variations in distortion character. A reduced value for R11 matches the final warm and aggressive channel signal levels.
The distortion character is strongly controlled by bias and R85. In this case, a 2.2k resistor gives V1b a very gradual and symmetric distortion. The tone is hard to describe since I don't know of any LIO circuits in common use. That's too bad, IMO, since it sounds very pleasing to me. If you try a LIO stage, I encourage you to play with R85 to find the value(s) most pleasing to you. Values from 0 to 100K will give wide variations in distortion character. A reduced value for R11 matches the final warm and aggressive channel signal levels.
The V2 stage is also biased colder to provide more clean headroom. The R23 and C10 screen circuit is modified to increase the screen shift effects (also described in GAO), making the pentode stage more responsive when playing at the edge of overdrive, which occurs with the P1 volume set around 5 or 6 when playing a telecaster. For higher-output PUs, I'd suggest setting P1 lower, around 2, for maximum sensitivity. As the drive (P1 level) increases, V2 distorts and compresses in a manner reminiscent of a VOX AC-15.
The V2 stage is also biased colder to provide more clean headroom. The R23 and C10 screen circuit is modified to increase the screen shift effects (also described in GAO), making the pentode stage more responsive when playing at the edge of overdrive, which occurs with the P1 volume set around 5 or 6 when playing a telecaster. For higher-output PUs, I'd suggest setting P1 lower, around 2, for maximum sensitivity. As the drive (P1 level) increases, V2 distorts and compresses in a manner reminiscent of a VOX AC-15.
The V4 cascode mixing stage is also biased colder than in version 1. A shared cathode resistor (R35) and bypass cap (C18) biases both halves of V4 to about 3 times the version 1 bias. Again, this increases headroom. The V4 tube is also changed to a 12au7 for the same reason.
The V4 cascode mixing stage is also biased colder than in version 1. A shared cathode resistor (R35) and bypass cap (C18) biases both halves of V4 to about 3 times the version 1 bias. Again, this increases headroom. The V4 tube is also changed to a 12au7 for the same reason.
The phase inverter, V5 and V6, have smaller screen caps (C28, 29) to increase the effect of screen voltage shift during high signal conditions. The current source, Q5, is now modulated by the new tremolo oscillator (Q18,19), which modulates the gain of the PI stage The P14 speed pot has sudden jumps in tremolo rate, which I'm unhappy about. Otherwise, the circuit works nicely, although I'm using it less than expected. Maybe that's because the controls are on the back panel - out of sight, out of mind.
The phase inverter, V5 and V6, have smaller screen caps (C28, 29) to increase the effect of screen voltage shift during high signal conditions. The current source, Q5, is now modulated by the new tremolo oscillator (Q18,19), which modulates the gain of the PI stage The P14 speed pot has sudden jumps in tremolo rate, which I'm unhappy about. Otherwise, the circuit works nicely, although I'm using it less than expected. Maybe that's because the controls are on the back panel - out of sight, out of mind.
Basically, the version 2 circuit has more clean headroom. The aggressive channel distortion is softer - I'd say less aggressive - due to the LIO implementation. There is still plenty of distortion, but it occurs at higher drive, volume, and master control settings.
Basically, the version 2 circuit has more clean headroom. The aggressive channel distortion is softer - I'd say less aggressive - due to the LIO implementation. There is still plenty of distortion, but it occurs at higher drive, volume, and master control settings.
UPDATE 3/31/16: P-AOT v3
UPDATE 3/31/16: P-AOT v3
The third iteration of this amp is complete. Version 3 is no longer the simple practice amp that was the limited goal of the original AOT design and build. This version is a much more complete and mature package. The evolution has occurred simply because I like these features in an amp I take out to play and I missed them in my outings with the prior versions. I also like to explore new features like the compression and emitter follower circuits described below. Both are good examples of optional add-ons - they're not really critical - but they are unique and useful at times and certainly add a tone dimension that's fun to play with.
The third iteration of this amp is complete. Version 3 is no longer the simple practice amp that was the limited goal of the original AOT design and build. This version is a much more complete and mature package. The evolution has occurred simply because I like these features in an amp I take out to play and I missed them in my outings with the prior versions. I also like to explore new features like the compression and emitter follower circuits described below. Both are good examples of optional add-ons - they're not really critical - but they are unique and useful at times and certainly add a tone dimension that's fun to play with.
The size of the Princeton package has become a limiting factor in many ways. Obviously the initial controls consumed the entire face of the amp, so the only option is to add new controls to the back panel. If I were to build another version of this amp from scratch, I'd use a larger chassis (the panel size of a Deluxe Reverb is looking about right) so the controls all fit in the front.
The size of the Princeton package has become a limiting factor in many ways. Obviously the initial controls consumed the entire face of the amp, so the only option is to add new controls to the back panel. If I were to build another version of this amp from scratch, I'd use a larger chassis (the panel size of a Deluxe Reverb is looking about right) so the controls all fit in the front.
Another important issue is the speaker and cab. The Jensen Jet potential is handicapped by the cab size. The Princeton cab is a bit too small to produce the robust tone the electronics produce (no surprise, really). While it doesn't matter much in band settings where bass is cut anyway to reduce overall mud, it is noticeable when playing solo or in sparser settings. As a comparison, the amp tone is fuller and smoother when it's driving the Creamback cab I built for the Flexi. In fact, it's surprising to me how similar both amps can sound when driving that cab. So, I'm mulling over options. I could move the P-AOT chassis and Jensen speaker to a bigger combo cab. Or, I could move the P-AOT chassis to a smaller head cabinet that gets used with a separate (bigger) cab. Decisions, decisions, ... No rush to do any of this since it's quite good as is and an external cab option already exists.
Another important issue is the speaker and cab. The Jensen Jet potential is handicapped by the cab size. The Princeton cab is a bit too small to produce the robust tone the electronics produce (no surprise, really). While it doesn't matter much in band settings where bass is cut anyway to reduce overall mud, it is noticeable when playing solo or in sparser settings. As a comparison, the amp tone is fuller and smoother when it's driving the Creamback cab I built for the Flexi. In fact, it's surprising to me how similar both amps can sound when driving that cab. So, I'm mulling over options. I could move the P-AOT chassis and Jensen speaker to a bigger combo cab. Or, I could move the P-AOT chassis to a smaller head cabinet that gets used with a separate (bigger) cab. Decisions, decisions, ... No rush to do any of this since it's quite good as is and an external cab option already exists.
As for the v3 circuit changes, the highlights of the v3 circuit are described below. These are major changes or additions since v2. It's easy to see that most of the amp has been altered in some way, making this circuit more of a new model than simply a modification of the original or v2 circuits.
As for the v3 circuit changes, the highlights of the v3 circuit are described below. These are major changes or additions since v2. It's easy to see that most of the amp has been altered in some way, making this circuit more of a new model than simply a modification of the original or v2 circuits.
The V1 input stage has been tweaked again. This is mostly a matter of taste, but I think I've converged on a configuration I'll be pleased with for the long term. V1a is still a clean gain stage. The addition of C3 rolls off high frequencies, thus reducing hiss and RFI. With C3 in place, R2 is increased to 47k to tame the high-freq peak (3-5kHz) produced by the LC resonance of pickups. Further reductions of this peak are obtained by turning down the guitar volume slightly. Even a few K-ohms of series resistance from the guitar volume pot has significant impact on that resonance peak. With the C3 and R2 changes, the choice of pickups and guitar volume setting have more impact on the high-freq response than in the original or v2 circuit versions.
The V1 input stage has been tweaked again. This is mostly a matter of taste, but I think I've converged on a configuration I'll be pleased with for the long term. V1a is still a clean gain stage. The addition of C3 rolls off high frequencies, thus reducing hiss and RFI. With C3 in place, R2 is increased to 47k to tame the high-freq peak (3-5kHz) produced by the LC resonance of pickups. Further reductions of this peak are obtained by turning down the guitar volume slightly. Even a few K-ohms of series resistance from the guitar volume pot has significant impact on that resonance peak. With the C3 and R2 changes, the choice of pickups and guitar volume setting have more impact on the high-freq response than in the original or v2 circuit versions.
The LIO circuit (Q17 and V1b) provides a range of distortion characters that are strongly impacted by R10 and V1 tube type. The lower the value of R10, the more impact a particular tube has on the tone. After trying about a dozen different tubes, I've settled on the 510 ohm value as the option that sounds good for all of the tubes. The Sovtek 12ax7 sounds best to me so I'm using it for now. The frequency response of this stage is re-calibrated to cut both bass and treble at the extremes, which smooths out the LIO distortion. The Full/Firm/Brite network has also been tweaked to taste.
The LIO circuit (Q17 and V1b) provides a range of distortion characters that are strongly impacted by R10 and V1 tube type. The lower the value of R10, the more impact a particular tube has on the tone. After trying about a dozen different tubes, I've settled on the 510 ohm value as the option that sounds good for all of the tubes. The Sovtek 12ax7 sounds best to me so I'm using it for now. The frequency response of this stage is re-calibrated to cut both bass and treble at the extremes, which smooths out the LIO distortion. The Full/Firm/Brite network has also been tweaked to taste.
The pentode stage, V2, is largely unchanged, however the Direct-Coupled Cathode Follower (V3a) or the Emitter Follower (Q23) can now buffer the V2 output to drive the tone stack. Many of the changes in v2 and v3 increase headroom of the amp's clean tone. This is another example.
The pentode stage, V2, is largely unchanged, however the Direct-Coupled Cathode Follower (V3a) or the Emitter Follower (Q23) can now buffer the V2 output to drive the tone stack. Many of the changes in v2 and v3 increase headroom of the amp's clean tone. This is another example.
A cathode follower is often used to drive a tone stack, and it adds some nice overdrive distortion (think 5F6A Bassman). However, what if you want to drive a low impedance tone stack without added distortion? (Think a cleaner Bassman.) Q23 offers is a simple and elegant solution. A switch (SW8) selects either the cathode follower (CF V3a) or the emitter follower (EF Q23). Both options provide good drive for the tone stack, but the CF adds distortion, while the EF does not. This is a simple viable option for increasing clean headroom in many amps that use DCCF circuits (the many derivatives of the 5F6A Bassman).
A cathode follower is often used to drive a tone stack, and it adds some nice overdrive distortion (think 5F6A Bassman). However, what if you want to drive a low impedance tone stack without added distortion? (Think a cleaner Bassman.) Q23 offers is a simple and elegant solution. A switch (SW8) selects either the cathode follower (CF V3a) or the emitter follower (EF Q23). Both options provide good drive for the tone stack, but the CF adds distortion, while the EF does not. This is a simple viable option for increasing clean headroom in many amps that use DCCF circuits (the many derivatives of the 5F6A Bassman).
Q23 is trivial to add since the transistor and R30 can mount directly on the tube socket pins if the socket center pin is used for the emitter (see image below). Of course, Q23 must tolerate the voltages in a given amp. Fortunately, high voltage transistors are available. A STX616-AP comes in a TO-92 package and handles up to 500v B+ preamp voltages at a cost of about 50 cents. Although I used a BJT, a MOSFET should also work. The image below shows Q23 mounted on the V3 tube socket and SW8 nearby on the rear panel (lower right in image).
Q23 is trivial to add since the transistor and R30 can mount directly on the tube socket pins if the socket center pin is used for the emitter (see image below). Of course, Q23 must tolerate the voltages in a given amp. Fortunately, high voltage transistors are available. A STX616-AP comes in a TO-92 package and handles up to 500v B+ preamp voltages at a cost of about 50 cents. Although I used a BJT, a MOSFET should also work. The image below shows Q23 mounted on the V3 tube socket and SW8 nearby on the rear panel (lower right in image).
The original tone stack was copied from Blencowe's book. The v3 version is a complete redesign, based on the Fender stack. I used a 1/3 octave graphic EQ to find pleasing EQ curves for my telecasters and G&L guitars. Spice simulations allowed me to determined the optimal ranges of Bass, Treble, Mid, and Shift controls, and the pot values and tapers to achieve similar EQ curves. My initial impressions are highly favorable. I think the effort was worthwhile and I've finally found a tone stack that works for me with my guitars and tone goals. At minimum, this is a huge improvement over the original tone stack and probably makes the biggest overall impact to the v3 amp sound.
The original tone stack was copied from Blencowe's book. The v3 version is a complete redesign, based on the Fender stack. I used a 1/3 octave graphic EQ to find pleasing EQ curves for my telecasters and G&L guitars. Spice simulations allowed me to determined the optimal ranges of Bass, Treble, Mid, and Shift controls, and the pot values and tapers to achieve similar EQ curves. My initial impressions are highly favorable. I think the effort was worthwhile and I've finally found a tone stack that works for me with my guitars and tone goals. At minimum, this is a huge improvement over the original tone stack and probably makes the biggest overall impact to the v3 amp sound.
The mixer stage (V4 and V3b) is unchanged, but the Tilt control has a resistor removed to slightly improve bass response.
The mixer stage (V4 and V3b) is unchanged, but the Tilt control has a resistor removed to slightly improve bass response.
The FX loop is completely redesigned. The old versions picked up the signal from the Drive control. The new design taps the signal after the tone stack and Volume control. The SEND signal is buffered by a JFET (Q2) to eliminate any loading of the V4b grid signal. The FX input is now double-inverted by Q3 and Q4 so the bypass signal arrives at the V4a grid in the correct phase. The new loop still works with the reverb system described in v2, only it now it tracks the tone stack settings, which is an improvement, IMO.
The FX loop is completely redesigned. The old versions picked up the signal from the Drive control. The new design taps the signal after the tone stack and Volume control. The SEND signal is buffered by a JFET (Q2) to eliminate any loading of the V4b grid signal. The FX input is now double-inverted by Q3 and Q4 so the bypass signal arrives at the V4a grid in the correct phase. The new loop still works with the reverb system described in v2, only it now it tracks the tone stack settings, which is an improvement, IMO.
Changes to the tremolo oscillator and current source for the phase inverter stage make the tremolo cleaner over the whole range of control settings. Q6 is added to modulate the current source (Q5) only on positive swings of the oscillator waveform. That makes the tremolo function only attenuate the phase inverter's signal. The v2 circuit attenuated and boosted the PI signal, which created problems for high Depth settings. The tremolo oscillator is also simplified by the removal of the 15v regulator. Component value changes (optimized in SPICE) make the oscillator waveform cleaner to remove any tendency of thumping.
Changes to the tremolo oscillator and current source for the phase inverter stage make the tremolo cleaner over the whole range of control settings. Q6 is added to modulate the current source (Q5) only on positive swings of the oscillator waveform. That makes the tremolo function only attenuate the phase inverter's signal. The v2 circuit attenuated and boosted the PI signal, which created problems for high Depth settings. The tremolo oscillator is also simplified by the removal of the 15v regulator. Component value changes (optimized in SPICE) make the oscillator waveform cleaner to remove any tendency of thumping.
The single transistor current source (Q5) for the phase inverter stage is simpler than the prior two transistor version. The phase inverter itself has C41 added to roll off high freq's. The RC networks feeding the output stage (V7) grids are added to maintain the optimal DC levels for AOT operation, while providing greater signal amplitude and clean headroom. The AOT stage is not changed, although P12 and P13 are now set for higher bias current (45ma) for a total of ~2 watt quiescent dissipation.
The single transistor current source (Q5) for the phase inverter stage is simpler than the prior two transistor version. The phase inverter itself has C41 added to roll off high freq's. The RC networks feeding the output stage (V7) grids are added to maintain the optimal DC levels for AOT operation, while providing greater signal amplitude and clean headroom. The AOT stage is not changed, although P12 and P13 are now set for higher bias current (45ma) for a total of ~2 watt quiescent dissipation.
Lastly, a new circuit is added to simulate the time-varying circuit behaviors of power supply sag and signal compression. Q20 creates a rectified version of the Mixer stage output signal. Q21 thresholds that signal and drives an LED (on the front panel) whenever the Mixer stage exceeds about 50% of its clean output level. An RC network filters that signal, creating the envelope of the signal from Q21. The envelope level turns on Q22, which pulls down the V2 pentode screen voltage in proportion to the size of the envelope signal and the setting of the Stiffness control (P16).
Lastly, a new circuit is added to simulate the time-varying circuit behaviors of power supply sag and signal compression. Q20 creates a rectified version of the Mixer stage output signal. Q21 thresholds that signal and drives an LED (on the front panel) whenever the Mixer stage exceeds about 50% of its clean output level. An RC network filters that signal, creating the envelope of the signal from Q21. The envelope level turns on Q22, which pulls down the V2 pentode screen voltage in proportion to the size of the envelope signal and the setting of the Stiffness control (P16).
A high Stiffness setting (~ CW 1/3 of the pot rotation) produces a high emitter resistance which means that Q22 has very little impact on the V2 screen and the amp operates "normally" as if this circuit didn't exist. For lower Stiffness settings, Q22 will pull down the V2 screen in proportion to the Mixer stage output amplitude. Lower screen voltages reduce the gain of V2, so this modulation of the screen voltage acts to compress the preamp signal.
A high Stiffness setting (~ CW 1/3 of the pot rotation) produces a high emitter resistance which means that Q22 has very little impact on the V2 screen and the amp operates "normally" as if this circuit didn't exist. For lower Stiffness settings, Q22 will pull down the V2 screen in proportion to the Mixer stage output amplitude. Lower screen voltages reduce the gain of V2, so this modulation of the screen voltage acts to compress the preamp signal.
The intent is to mimic the behavior of compression due to power supply sag at high volumes. This circuit senses signal levels before the Master volume control, so compression is achieved based on how "hard" the preamp drives the Master control. Actual room volume is controlled by the Master, so compression can be obtained for any room volume.
The intent is to mimic the behavior of compression due to power supply sag at high volumes. This circuit senses signal levels before the Master volume control, so compression is achieved based on how "hard" the preamp drives the Master control. Actual room volume is controlled by the Master, so compression can be obtained for any room volume.
The RC network (R57, 58, 59 and C30, 31) values control the attack and decay of the compression. Current values produce ~100ms attack and ~150ms decay times, which sound and feel OK to me. Reduce R57 if you want faster attack. At the extreme low end of the Stiffness control, the compression is ridiculously aggressive and it creates it's own unique "effect". At all levels its clean and free of any thumping artifacts.
The RC network (R57, 58, 59 and C30, 31) values control the attack and decay of the compression. Current values produce ~100ms attack and ~150ms decay times, which sound and feel OK to me. Reduce R57 if you want faster attack. At the extreme low end of the Stiffness control, the compression is ridiculously aggressive and it creates it's own unique "effect". At all levels its clean and free of any thumping artifacts.
The image below shows the v3 front panel and circuitry. The compression level LED is visible near the Master control. Two perf-boards are visible on the rear panel. The left board is the tremolo circuit, the smaller board to right is the compression circuit. Both boards are supported by the control pots associated with each circuit.
The image below shows the v3 front panel and circuitry. The compression level LED is visible near the Master control. Two perf-boards are visible on the rear panel. The left board is the tremolo circuit, the smaller board to right is the compression circuit. Both boards are supported by the control pots associated with each circuit.
The final image shows the back panel with the additional Stiffness control, two tremolo controls, and SW8 (from left to right along the back panel). The cabinet view also shows the 24v power supply (right side), as well as the reverb electronics and pan (lower surface). It's a tight package that makes it easy to pickup and go.
The final image shows the back panel with the additional Stiffness control, two tremolo controls, and SW8 (from left to right along the back panel). The cabinet view also shows the 24v power supply (right side), as well as the reverb electronics and pan (lower surface). It's a tight package that makes it easy to pickup and go.
UPDATE:
UPDATE:
I've been playing this v3 amp for several months in varied sessions and I am pleased to report that the new effects are much cooler features than I had expected them to be. The v3 tremolo is really great. It's not like Fender tremolo - it's a swirling-phasing effect that adds more than just amplitude modulation. I'm using it all the time, even for stuff I don't think of as tremolo material. It adds harmonic interest to single note melodies and it makes even simple chords shimmer. I had hoped for something like a good bias trem effect, but v3 has an unexpectedly cool and rich quality to it that goes beyond bias trem. Even at low depth settings where it's hard to hear a modulation effect, it produces a richer tone quality.
I've been playing this v3 amp for several months in varied sessions and I am pleased to report that the new effects are much cooler features than I had expected them to be. The v3 tremolo is really great. It's not like Fender tremolo - it's a swirling-phasing effect that adds more than just amplitude modulation. I'm using it all the time, even for stuff I don't think of as tremolo material. It adds harmonic interest to single note melodies and it makes even simple chords shimmer. I had hoped for something like a good bias trem effect, but v3 has an unexpectedly cool and rich quality to it that goes beyond bias trem. Even at low depth settings where it's hard to hear a modulation effect, it produces a richer tone quality.
The compression is also a nice surprise. I've used external compressors before (like the TLA 5051), but they didn't grow on me. This v3 compressor is different. Perhaps it has something to do with the way the signal sensing and compression are sequenced. Maybe it's the relatively longer time constants used since the compressor is trying to mimic the way power supply sag creates compression.
The compression is also a nice surprise. I've used external compressors before (like the TLA 5051), but they didn't grow on me. This v3 compressor is different. Perhaps it has something to do with the way the signal sensing and compression are sequenced. Maybe it's the relatively longer time constants used since the compressor is trying to mimic the way power supply sag creates compression.
In the v3 circuit the signal is sensed at the Master volume and the compression actually happens upstream at the V2 pentode. So, the compression impacts the sensed signal. I believe this also happens with sag compression since the power supply sags based on the current drawn by the output tubes, while some of the compression produced by the sag is created in the preamp and the power tube screens - both effectively reducing the current that causes the sag in the first place. There may be something about that feedback that is vital to the touch and feel. Regardless of exactly how it happens, the v3 compression is really addictive. I use it all the time. The "firmness" control works beautifully, with the middle range being the most useful to me. Extreme compression (min firmness) is a bit over the top and unnatural, but it's a cool effect that may be useful at times.
In the v3 circuit the signal is sensed at the Master volume and the compression actually happens upstream at the V2 pentode. So, the compression impacts the sensed signal. I believe this also happens with sag compression since the power supply sags based on the current drawn by the output tubes, while some of the compression produced by the sag is created in the preamp and the power tube screens - both effectively reducing the current that causes the sag in the first place. There may be something about that feedback that is vital to the touch and feel. Regardless of exactly how it happens, the v3 compression is really addictive. I use it all the time. The "firmness" control works beautifully, with the middle range being the most useful to me. Extreme compression (min firmness) is a bit over the top and unnatural, but it's a cool effect that may be useful at times.
Page updated
Google Sites
Report abuse