JFET Biasing (or not) question

[KarenColumbo]

I can't get my head around this one.



I breadboarded BWs Mu Amp variation (first stage is right side of image above). Then I had a look at the (alleged) Catalinbread SFTII schem (images left).

They are "more or less" identical. So I just put the 10k source resistor (upper jfet) of the SFT into the Mu Amp. Worked. Then I thought: "Well, the SFTII has no biasing pot at it's drain, this should work in the mu amp, too." But it didn't.

It's just hand-on-tinkering with things I try to understand.

What's the difference between the two circuits?

Is it

1. The R/C combination at the input?
2. The different jfets?

Could you point me in the right direction?

EDIT: Basically it's this schem (circled), isn't it?

[antonis]

Could you point me in the right direction?

Hmmm...

Maybe you should face it from the point of both upper FETs act as adjustable current sources for lower Fets..
(current adjustment is obtained by R6 value on left circuit & by Trimpot setting on right circuit..)

Current of right upper FET (J2) is more unpredictable than the left one (Q2) because of wide I_DSS spread..
On the other hand, self-biasing resistor R6 back-biases the Gate by I_DxR6, reducing I_D and bringing Q2 closer to pinch-off..

I leave the rest on you.. 

EDIT: Basically yes..
(but it's simplified, using upper FET D-S channel resistance as lower FET drain resistor..)

[KarenColumbo]

Could you point me in the right direction?

Hmmm...

Maybe you should face it from the point of both upper FETs act as adjustable current sources for lower Fets..
(current adjustment is obtained by R6 value on left circuit & by Trimpot setting on right circuit..)

Current of right upper FET (J2) is more unpredictable than the left one (Q2) because of wide I_DSS spread..
On the other hand, self-biasing resistor R6 back-biases the Gate by I_DxR6, reducing I_D and bringing Q2 closer to pinch-off..

I leave the rest on you.. 

EDIT: Basically yes..
(but it's simplified, using upper FET D-S channel resistance as lower FET drain resistor..)

HMMMMMMM .....

My new reading glasses (thanks to the wife) showed me that in the left (SFT) schem there is a 1 uF polcap, in the right a 0.1 uF.
I realized at one point that with jfet circuits caps somehow regulate gain like resistors. At the moment I don't know why (there's a loooong reading list ahead of me), but:

I changed to 1 uF, got rid of the biasing pots in stage 1 and 2 and voilà -> it works and doesn't sound half bad.

Now for me it's back to reading and finding out what it acually means what you said in your answer, dear sir. And how I can achieve the same with J5 in the schem below, which is still in need of a biasing pot. Anyway: A hearty Ευχαριστώ!

[KarenColumbo]

To make matters complete: Here's the SFTII schem I'm perusing

[antonis]

@Andreas: the capacitance value difference between C4 & C8 (in your first scheme) isn't so critical..

What should be bother you more is the absence(?) of C9 (22μF)  - ruled out by red X - capacitor..
(it by-passes J1 Source resistor (R1) for AC, effectivelly turning J1 to grounded Source amplifier - like C1* & C2* at 2nd & 3rd stage of bwmuamp in your 3rd scheme..)

Maybe you should compare Q1 & J1 for their individual gain setting..

Anyway: A hearty Bitte schön..!!

P.S.
Having 4 scheme windows open, I'm a bit confused..
Which is the final non-working circuit..??

[KarenColumbo]

@Andreas: the capacitance value difference between C4 & C8 (in your first scheme) isn't so critical..

What should be bother you more is the absence(?) of C9 (22μF)  - ruled out by red X - capacitor..
(it by-passes J1 Source resistor (R1) for AC, effectivelly turning J1 to grounded Source amplifier - like C1* & C2* at 2nd & 3rd stage of bwmuamp in your 3rd scheme..)

Maybe you should compare Q1 & J1 for their individual gain setting..

Anyway: A hearty Bitte schön..!!

P.S.
Having 4 scheme windows open, I'm a bit confused..
Which is the final non-working circuit..??

Hehe. The final non-non-working circuit (it works actually) would be this one (now with my changes):




I omitted C9 and C1 because Brian Wampler wrote in his book:

"... Notes: C9, C1, and C2 all boost the gain of the clipping circuit, smaller caps will yield less gain in conjunction with R1.  If you want a lot of distortion, make R1 a 1k or so. If you want more of an overdrive, make R1 a 4.7k – 22k (in both locations shown). Alternatively, you can remove C9, C1, AND C2 and have a smooth overdrive, especially ifyou make R1 a 10k or larger resistor."

[KarenColumbo]

I just put C1 and C9 back in - 1 uF each. Now the distortion is quite heavy, but the circuit is extremely noisy - I think it buggers the biasing, at least it sounds like badly biased jfets now Wonders above wonders. Mine is gonna be a reading vacation

A quick one: Is it really necessary to have those voltage divider resistors in 1M? Would 10k be sufficient? I ask because the circuit is quite noisy (murmuring).

[antonis]

I just put C1 and C9 back in - 1 uF each. Now the distortion is quite heavy, but the circuit is extremely noisy - I think it buggers the biasing,

Maybe bias has some nasty habits but, for sure, Source resistors by-pass caps aren't intruded..
(they set the gain like in a BJT CE amplifier with Emitter degeneration..)

More responsible for any bias "bugger" is upper FET Drain resistor / lower FET drain resistor / lower FET Source resistor - in that order..

Is it really necessary to have those voltage divider resistors in 1M? Would 10k be sufficient? I ask because the circuit is quite noisy (murmuring).

You can lower them as far as you like but let us see the results:
You'll lower total output impedance (which is good but unnecessary due to next stage very high input impedance) but you'll simultaneously lower J1 & J3 gain (R_D/R_S) (which is bad).
For signal, 1M voltage divider resistors are "seen", through C8 & C7 caps, like been grounded in parallel so total Drain resistance is actually 1M//1M//10k(or Trimpot)//whatever comes next.. 

Also, significally lοwering those resistors values isn't a battery friendy gesture..

[KarenColumbo]

I just put C1 and C9 back in - 1 uF each. Now the distortion is quite heavy, but the circuit is extremely noisy - I think it buggers the biasing,

Maybe bias has some nasty habits but, for sure, Source resistors by-pass caps aren't intruded..
(they set the gain like in a BJT CE amplifier with Emitter degeneration..)

More responsible for any bias "bugger" is upper FET Drain resistor / lower FET drain resistor / lower FET Source resistor - in that order..

Is it really necessary to have those voltage divider resistors in 1M? Would 10k be sufficient? I ask because the circuit is quite noisy (murmuring).

You can lower them as far as you like but let us see the results:
You'll lower total output impedance (which is good but unnecessary due to next stage very high input impedance) but you'll simultaneously lower J1 & J3 gain (R_D/R_S) (which is bad).
For signal, 1M voltage divider resistors are "seen", through C8 & C7 caps, like been grounded in parallel so total Drain resistance is actually 1M//1M//10k(or Trimpot)//whatever comes next.. 

Also, significally lοwering those resistors values isn't a battery friendy gesture..

Thanks! Much to learn i have!

[amptramp]

In the first post, the diagram on the left is a µ-amp but the diagram on the right is not.  This is further complicated by the fact that the gate return for the lower FET is missing.  There should be some resistance to ground from the lower FET gate.  The circuit on the right is a FET amp with a high-impedance / constant current load.  They behave in an entirely different fashion.

The one on the left has a resistor from the lower drain to the upper source and the gate connection is AC coupled to the drain of the lower device.  As the lower drain current goes up, the resistor current is higher and the bias on the upper transistor gets more negative, tending to shut it off.  So as one device draws more current, the other draws less and vice versa.  But these currents are in series so the difference in current has to go to the load.  The µ-amp (or as it is also called, SRPP for shunt-regulated push-pull) must have a load with a low enough impedance that the difference in upper and lower transistor current can flow into the load.  If the load is too high, the stage will act as a fuzz-style limiter with the output going to either the upper rail or lower rail.  The value of the resistor  between the lower drain and upper source should be the reciprocal of the upper transistor transconductance (meaning the 10K shown in the picture is too high) so that the current swing in the upper stage is equal to the current swing in the lower stage but in the opposite direction.  The resistor value is not all that critical because the current swings do not have to be equal but the correct resistor value tends to give best dynamic range.  If you want to follow the same design done with tubes, see here:

http://www.tubecad.com/may2000/

The design on the right is NOT a µ-amp.  It is a normal grounded source amp with a constant current source for a load.  This gives a very high stage gain but unlike the µ-amp, this current does not change much.  The resistance of the following stage tends to set the gain since it is in parallel with the drain load (which is effectively well into the megohms).

Both types of amplifier have their uses.  You just have to choose what characteristics you want.  Hope this helps.

[POTL]

I can't get my head around this one.



I breadboarded BWs Mu Amp variation (first stage is right side of image above). Then I had a look at the (alleged) Catalinbread SFTII schem (images left).

They are "more or less" identical. So I just put the 10k source resistor (upper jfet) of the SFT into the Mu Amp. Worked. Then I thought: "Well, the SFTII has no biasing pot at it's drain, this should work in the mu amp, too." But it didn't.

It's just hand-on-tinkering with things I try to understand.

What's the difference between the two circuits?

Is it

1. The R/C combination at the input?
2. The different jfets?

Could you point me in the right direction?

EDIT: Basically it's this schem (circled), isn't it?

Hi
I see that you, as well as I work on breadboard boards and work my skills

What you specify is called Mu Amp and does not require bias and trimmers
I recommend that you pay attention to the basic version

Here you can read in detail about this scheme
http://www.muzique.com/amz/mini.htm
In principle, the scheme is minimal changes that will be useful
The larger the value of the resistor in the drain of the transistor 2, the smaller the gain, for greater benefit it is possible to connect the drain of the transistor 2 directly to ground
Capacitors 1 2 and 3 are responsible for the number of low frequencies that pass before distortion, during distortion and after distortion respectively
Condenser 6 will add low frequencies and loudness
You can also refuse from it
I went through it a few months ago and want to save your time and give you a hint. The effects of Catalinbread and Wampler (with all due respect to Brian), though they have a good sound but in general do not very well copy the sound of the amplifiers.
My advice is to pay attention to the Ramle FX Marvel Drive

For comparison, the amplifier circuit


How do you see the difference a little
1) The drain resistors are replaced by trim
2) The second tube is replaced by Mu Amp, of course you can fully copy the circuit to the bass reflex (this will work well), but as my experience showed, Mu Amp gives a better result than the classic use of the second lamp in old marshalls (the sound is bigger, louder, More gain)


In general, I recommend to start from the Marvel Drive scheme, and not Wampler / Catalinbread sound will be closer to the amplifier that you want to build.
I collected more than 15 classic and modern amplifiers from Fender Twin to Mesa Mark V, everything works fine (only a pair of Peavey 5150/6505 / 5150III, Mesa Boogie Rectifier amplifiers) did not give me the desired result, but most likely the problem is that I do not have The required quality of J201 and I use 2N5485 with them

[KarenColumbo]

In the first post, the diagram on the left is a µ-amp but the diagram on the right is not.  This is further complicated by the fact that the gate return for the lower FET is missing.  There should be some resistance to ground from the lower FET gate.  The circuit on the right is a FET amp with a high-impedance / constant current load.  They behave in an entirely different fashion.

The one on the left has a resistor from the lower drain to the upper source and the gate connection is AC coupled to the drain of the lower device.  As the lower drain current goes up, the resistor current is higher and the bias on the upper transistor gets more negative, tending to shut it off.  So as one device draws more current, the other draws less and vice versa.  But these currents are in series so the difference in current has to go to the load.  The µ-amp (or as it is also called, SRPP for shunt-regulated push-pull) must have a load with a low enough impedance that the difference in upper and lower transistor current can flow into the load.  If the load is too high, the stage will act as a fuzz-style limiter with the output going to either the upper rail or lower rail.  The value of the resistor  between the lower drain and upper source should be the reciprocal of the upper transistor transconductance (meaning the 10K shown in the picture is too high) so that the current swing in the upper stage is equal to the current swing in the lower stage but in the opposite direction.  The resistor value is not all that critical because the current swings do not have to be equal but the correct resistor value tends to give best dynamic range.  If you want to follow the same design done with tubes, see here:

http://www.tubecad.com/may2000/

The design on the right is NOT a µ-amp.  It is a normal grounded source amp with a constant current source for a load.  This gives a very high stage gain but unlike the µ-amp, this current does not change much.  The resistance of the following stage tends to set the gain since it is in parallel with the drain load (which is effectively well into the megohms).

Both types of amplifier have their uses.  You just have to choose what characteristics you want.  Hope this helps.

Well ... thanks, Sir! I'm not sure I follow - yet But I already have a reading list for the next two weeks I will spend in wonderful warsaw - now it's been revised and optimized. I will try and find the optimized value for the resistor between the two fets in the two stages. Which is just in time cause I planned on putting in a third stage to "power" some sort of tone control - maybe a current sucker like a 3-band. This is extremely fascinating stuff, if a bit far over my head at the moment. Well ... nothing like cold water to learn to swim.

[KarenColumbo]

I can't get my head around this one.



I breadboarded BWs Mu Amp variation (first stage is right side of image above). Then I had a look at the (alleged) Catalinbread SFTII schem (images left).

They are "more or less" identical. So I just put the 10k source resistor (upper jfet) of the SFT into the Mu Amp. Worked. Then I thought: "Well, the SFTII has no biasing pot at it's drain, this should work in the mu amp, too." But it didn't.

It's just hand-on-tinkering with things I try to understand.

What's the difference between the two circuits?

Is it

1. The R/C combination at the input?
2. The different jfets?

Could you point me in the right direction?

EDIT: Basically it's this schem (circled), isn't it?

Hi
I see that you, as well as I work on breadboard boards and work my skills

What you specify is called Mu Amp and does not require bias and trimmers
I recommend that you pay attention to the basic version

Here you can read in detail about this scheme
http://www.muzique.com/amz/mini.htm
In principle, the scheme is minimal changes that will be useful
The larger the value of the resistor in the drain of the transistor 2, the smaller the gain, for greater benefit it is possible to connect the drain of the transistor 2 directly to ground
Capacitors 1 2 and 3 are responsible for the number of low frequencies that pass before distortion, during distortion and after distortion respectively
Condenser 6 will add low frequencies and loudness
You can also refuse from it
I went through it a few months ago and want to save your time and give you a hint. The effects of Catalinbread and Wampler (with all due respect to Brian), though they have a good sound but in general do not very well copy the sound of the amplifiers.
My advice is to pay attention to the Ramle FX Marvel Drive

For comparison, the amplifier circuit


How do you see the difference a little
1) The drain resistors are replaced by trim
2) The second tube is replaced by Mu Amp, of course you can fully copy the circuit to the bass reflex (this will work well), but as my experience showed, Mu Amp gives a better result than the classic use of the second lamp in old marshalls (the sound is bigger, louder, More gain)


In general, I recommend to start from the Marvel Drive scheme, and not Wampler / Catalinbread sound will be closer to the amplifier that you want to build.
I collected more than 15 classic and modern amplifiers from Fender Twin to Mesa Mark V, everything works fine (only a pair of Peavey 5150/6505 / 5150III, Mesa Boogie Rectifier amplifiers) did not give me the desired result, but most likely the problem is that I do not have The required quality of J201 and I use 2N5485 with them

Hehe, yeah, i noticed, you're into jfet/mosfet drives, too You're quite some bit ahead of me, too, maybe years So thank you very specially for giving me a direction. I will do as you suggested - feels like this is a good way to dive into this.

[KarenColumbo]

Well, I threw those things together on the breadboard and tried the result with 3 different guitars. It's basically the "Mu Amp" example from one of Mr. Wampler's books (which isn't really a Mu Amp, as I learned) with some parts of the SFT II with the tone stack and presence R/C as a switch taken from a dr. boogie schem I found.



Not bad at all. Sounds like fuzz or 80ies hair metal, depending on the instrument:

. Yamaha Pacifica power strat: Stoner rock, very disillusioned rock sound, bit fuzzy, lots of balls
. Some Les Paul lookalike with passive EMGs: warm, tubey, can get a bit shrill quite easily, lots of bass
. Peavey Custom (2 HBs), high output PUs: dry, saturated rock sounds, very high gain, lots of different characters while fiddling with the tonestack

The boost switch is absolute mayhem - twice the gain, lots of treble and presence. Not really a "Lead" sound, but definitely boosted.

Tried with different fets, settled for J201 in the first stage, rest are 5457.

I decided to call this one "Magnetite" - a bit of metal, a bit of stoner rock. I'm sure there's a lot of beginner's catastrophies in there (if you spot some obvious idiotic mistakes please tell me), but I definitely like the sound.

I will box this asap before I mod it to death. Had a very nice bass overdrive two days ago, growly and dry, but it disappeared after tinkering without writing down the changes I made Will be the next one, I think.

[duck_arse]

recheck the wiring of your mids pot against der doctor circuit. yours seems to be floating above ground.

[KarenColumbo]

recheck the wiring of your mids pot against der doctor circuit. yours seems to be floating above ground.

will do. thanks!

[KimJongFun]

I have some questions regarding your schematic
R8 R14 and R20 are voltage dividers to bias the fets to 4,5 volts, right?
Wouldn't a lower value also reduce noise? Or is 100k a critical value?

Then something else:
I've read an article about mu-amp using mosfets but it used two 10meg resistor instead of 1meg (so pretty noisy in comparison) and to reduce noise the author made a voltage divider using two 10k resistors and simply removed the 10meg resistor to 9v, and reduced the other resistor to 4.7M (you can leave 1M I guess) and plugged it into the 4,5v (and he also added another cap (10uf) like your C4 but from the 4,5v to ground), maybe you can do that if you still have noise issues.

[antonis]

On a brute aproximation, noise is proportional to resistor value..
(on a less brute aproximation, like in case of shot noise, it's proportional to (square root of) current flowing through the resistor..)

Low resistor values are preferable for low noise level and "stiff" voltage dividers..

High resistor values are preferable for low current consumption and high impendances..
(like in case of MOsFet (or BJT) bias..)

Generally, is prefered to form a low value resistive voltage divider and feed anything via a high value resistor when current need is very low (or negligible..)