SS jFet (pre)amp

[Steben]

Hi guys, I posted this on the Facebook page, but feel free to comment here as well.
the idea behind the circuit is to combine FET's as harmonic (comparable to fetzer stages) and overdriven element (using mu amp) and the use of an Class AB BJT stage to enable variable current draw, creating power sag.

[Fancy Lime]

Hi Steven,

very nifty! I would be very interested in hearing this thing. I played with a similar idea for making a very saggy fuzz but never quite got it working as I wanted.

Cheers,
Andy

[Steben]

Hi Steven,

very nifty! I would be very interested in hearing this thing. I played with a similar idea for making a very saggy fuzz but never quite got it working as I wanted.

Cheers,
Andy

Did you use Class AB stages?

[Fancy Lime]

One class AB stage to drive a class A stage. Sounded weak and farty and I quickly got distracted and went in a different direction with that project. Maybe I should give the concept another go. I do like me some seriously saggy fuzz.

Andy

[antonis]

Is there a missing Input cap..?? 

[TejfolvonDanone]

One thing might be good to consider is to have the tone stack after the mu-amp (with an added source/emitter follower) so you can roll off some high from the distorted signal. This way the output signal very well may be too harsh and there is no option to roll that off. Maybe a simple low pass filter after the mu amp would do the trick.

I would also put some low value resistor in the class AB stage's transistors emitter circuit. This adds a bit of negative feedback to the transistors and helps to have a stable bias.

[Steben]

Is there a missing Input cap..?? 

Such input stages gates are biased to ground, just as in many Tube amps.
Caps are only to be added for protection against DC, which normally shouldn't be there.

[Steben]

One thing might be good to consider is to have the tone stack after the mu-amp (with an added source/emitter follower) so you can roll off some high from the distorted signal. This way the output signal very well may be too harsh and there is no option to roll that off. Maybe a simple low pass filter after the mu amp would do the trick.

I would also put some low value resistor in the class AB stage's transistors emitter circuit. This adds a bit of negative feedback to the transistors and helps to have a stable bias.

Good thinking. Are you talking about 10 to 100 ohms as in some power amps?  Doen't it interfere a bit with the crossover induction idea?

[TejfolvonDanone]

Good thinking. Are you talking about 10 to 100 ohms as in some power amps?  Doen't it interfere a bit with the crossover induction idea?

100 Ohm should do.
Crossover distortion only depends on the bias current. But wasn't your plan to create power supply sag with the AB stage?

[Steben]

Good thinking. Are you talking about 10 to 100 ohms as in some power amps?  Doen't it interfere a bit with the crossover induction idea?

100 Ohm should do.
Crossover distortion only depends on the bias current. But wasn't your plan to create power supply sag with the AB stage?

Yes, it was. Yet a sag would lower the bias at the bases as well, leading to cut off and class B. Which can be intentional. I wonder whether the 100 ohm resistor interferes.

[TejfolvonDanone]

Yes, it was. Yet a sag would lower the bias at the bases as well, leading to cut off and class B. Which can be intentional. I wonder whether the 100 ohm resistor interferes.

When the class AB stage draws more current the supply starts to sag. But then on of the transistor is already turned on therefore won't increase the crossover distortion. That's why i said the crossover distortion is only dependent of the bias voltage when there is no signal present.

[Steben]

Yes, it was. Yet a sag would lower the bias at the bases as well, leading to cut off and class B. Which can be intentional. I wonder whether the 100 ohm resistor interferes.

When the class AB stage draws more current the supply starts to sag. But then on of the transistor is already turned on therefore won't increase the crossover distortion. That's why i said the crossover distortion is only dependent of the bias voltage when there is no signal present.

I am afraid i think it works differently. Power sag is much slower than a wave fragment. That's why it compresses, instead of just adding clipping. Just as .... in a compressor. The voltage drop only occurs after an amount of current draw raise, when the capacitor is getting discharged. This attack and decay of the sag is determined by the impedance and capacitance time constant. A zero power source impedance woud lead to no sag at all. An infinite capacitance would lead to infinite time constant (no current draw would change the voltage held by the cap).
in other words: the voltage drop occurs after the signal draws more current and this with a certain time span. This changes the headroom and the bias of all stages as well. If the voltage drops, the voltage between the two bases of the class AB stage drops as well, possibly below 2 x Vbe(cut off).....

[antonis]

Is there a missing Input cap..?? 

Such input stages gates are biased to ground, just as in many Tube amps.
Caps are only to be added for protection against DC, which normally shouldn't be there.

Glad to hear you are totally ensured for DC absence of any previous coupled stage output.. 

[Steben]

Is there a missing Input cap..?? 

Such input stages gates are biased to ground, just as in many Tube amps.
Caps are only to be added for protection against DC, which normally shouldn't be there.

Glad to hear you are totally ensured for DC absence of any previous coupled stage output.. 

true, it is how the fetzer valve is built up by copying old tube amps.

Example of kits available that use the same input concept:

[antonis]

true, it is how the fetzer valve is built up by copying old tube amps.

Of course, a simple circuit copy of an old tube amp for a FET one hides sometimes design clues..

e.g. R2 (on your scheme) plays a dual role on a tube pre-amp. It's used to avoid "blocking distortion" and also to form a low pass filter in conjunction with Grid/Cathode interelectode capacitance..
Its first use is needless for FET (I presume so.. ) and its second use could be obtained by omiting C1 (FETs also exhibit interelectode capacitance..)

P.S.
I hope you accede the above as a benevolent "academic" discussion..   

[teemuk]

OP, a word of warning: You will not be able to build this circuit with consistent results. Inherent and fairly large variation of FET characteristics ensures that you "trim" every drain load to match the particular FET characteristics and preferred bias point. Effectively you will end up with different drain load for each FET stage.

Why is it an issue? Because the drain load practically defines both voltage gain and output impedance of the gain stage in question. Gain variation will effect clipping characteristics (namely amount of overdrive) and headroom. Output Z in turn will affect the turnover frequency of coupling. With variable drain resistance these parameters, that highly affect the perceived tone. will be all over the place.

Any sensible design aims to bias the FETs in a method that provides more consistent results than widely varying drain resistance. You practically never see trimmer resistors in commercial designs due to inconsistency they introduce.

Second, don't be too impressed by seeming attempts to emulate a triode gain stage. The "Runoffgroove" -method totally ignores vital design parameters of the stage, such as: gain vs. headroom, drain bias voltage and overall asymmetry vs. symmetry of clipping distortion, frequency response, etc. I fail to see how these things "emulate" anything when every important parameter is basically dependent on FET characteristics and drain load resistance.

Most Runoffgroove designs suggest a "center bias", which leaves equal headroom for both negative and positive signal swings and produces largely symmetric clipping. On the other hand, in practice many tube gain stages are biased for asymmetric signal swing and asymmetric clipping. 

FETs can be tweaked to produce similar distortion to triode tubes, or at least when they operate at their "linear" region. The circuit from which "Fetzer Valve" was designed could do so. The guy who designed the Fetzer, on the other hand, completely missed the point of the original design and just designed yet another widely inconsistent FET gain stage that is particularly not too different from other generic common source amps.

Not surprisingly newer "runoffgroove" designs are based on generic opamps and diodes, which provide more consistency than the horrible FET gain stage with trimmer resistor load. Nope, this doesn't decrease "realism" of the tube emulation by slightest bit. On a contrary, it allowed to design a better and more realistic tube emulator circuit that actually behaves more like circuit it tries to emulate.

Finally, if you aim towards characteristics like "sag" then do note that chances for the phemonenon are slim with circuits that draw very little current (e.g. preamp circuits). To achieve sag you need to create enough current flow to the power supply, and also introduce a resistance across which the power supply voltage can "sag". DC batteries alone are poor for that, until their charge has nearly died.

With virtually resistanceless, low impedance power supply, sag probably never happens in any worthwhile magnitudes.

..And regarding that power supply overall: DC batteries are "forgiving" to some extent, but it is good design practice to introduce some interstage decoupling in order to prevent signals from bleeding across stages via power supply rails. With insufficient decoupling you quickly run into weird issues with frequency response, or alternatively find out you built yourself a motorboating oscillator.

[Steben]

true, it is how the fetzer valve is built up by copying old tube amps.

Of course, a simple circuit copy of an old tube amp for a FET one hides sometimes design clues..

e.g. R2 (on your scheme) plays a dual role on a tube pre-amp. It's used to avoid "blocking distortion" and also to form a low pass filter in conjunction with Grid/Cathode interelectode capacitance..
Its first use is needless for FET (I presume so.. ) and its second use could be obtained by omiting C1 (FETs also exhibit interelectode capacitance..)

P.S.
I hope you accede the above as a benevolent "academic" discussion..   

It is based on the fetzer. This circuit tries to copy indeed a classic Fender input stage, including some frequency responses. Whether all parts are necessary or not is something I haven't thought about deeply actually. It all will probably end in modding anyway.

[Steben]

OP, a word of warning: You will not be able to build this circuit with consistent results. Inherent and fairly large variation of FET characteristics ensures that you "trim" every drain load to match the particular FET characteristics and preferred bias point. Effectively you will end up with different drain load for each FET stage.

Why is it an issue? Because the drain load practically defines both voltage gain and output impedance of the gain stage in question. Gain variation will effect clipping characteristics (namely amount of overdrive) and headroom. Output Z in turn will affect the turnover frequency of coupling. With variable drain resistance these parameters, that highly affect the perceived tone. will be all over the place.

Any sensible design aims to bias the FETs in a method that provides more consistent results than widely varying drain resistance. You practically never see trimmer resistors in commercial designs due to inconsistency they introduce.

Second, don't be too impressed by seeming attempts to emulate a triode gain stage. The "Runoffgroove" -method totally ignores vital design parameters of the stage, such as: gain vs. headroom, drain bias voltage and overall asymmetry vs. symmetry of clipping distortion, frequency response, etc. I fail to see how these things "emulate" anything when every important parameter is basically dependent on FET characteristics and drain load resistance.

Most Runoffgroove designs suggest a "center bias", which leaves equal headroom for both negative and positive signal swings and produces largely symmetric clipping. On the other hand, in practice many tube gain stages are biased for asymmetric signal swing and asymmetric clipping. 

FETs can be tweaked to produce similar distortion to triode tubes, or at least when they operate at their "linear" region. The circuit from which "Fetzer Valve" was designed could do so. The guy who designed the Fetzer, on the other hand, completely missed the point of the original design and just designed yet another widely inconsistent FET gain stage that is particularly not too different from other generic common source amps.

I agree, yet it works. And it does add second order harmonics. For personal use the trim idea is actually fun on its own, since it allows for the non-ideal triode modes just as good.

Not surprisingly newer "runoffgroove" designs are based on generic opamps and diodes, which provide more consistency than the horrible FET gain stage with trimmer resistor load. Nope, this doesn't decrease "realism" of the tube emulation by slightest bit. On a contrary, it allowed to design a better and more realistic tube emulator circuit that actually behaves more like circuit it tries to emulate.

Yet all it does is clipping. I haven't seen non-linear opamp stages (yet).

Finally, if you aim towards characteristics like "sag" then do note that chances for the phemonenon are slim with circuits that draw very little current (e.g. preamp circuits). To achieve sag you need to create enough current flow to the power supply, and also introduce a resistance across which the power supply voltage can "sag". DC batteries alone are poor for that, until their charge has nearly died.
With virtually resistanceless, low impedance power supply, sag probably never happens in any worthwhile magnitudes.
..And regarding that power supply overall: DC batteries are "forgiving" to some extent, but it is good design practice to introduce some interstage decoupling in order to prevent signals from bleeding across stages via power supply rails. With insufficient decoupling you quickly run into weird issues with frequency response, or alternatively find out you built yourself a motorboating oscillator.

The idea is not to use batteries... an 18W power amp won't run long. There is a pot allowing for impedance by which the voltage drop is controllable.

[Steben]

Another point to consider regarding opamp/diode based distortion stages is the fixed clipping point. When the power supply alters... more precisely: if the supply voltage is ment to change ... the headroom isn't affected unless the opamp's headroom is lower than that of the clipper diodes. An exception is MOSfet inverter based circuits. But these do not sound good at high voltage supplies.

[teemuk]

Clipping threshold of diodes is fixed only if their voltage reference is fixed as well.