Local Feedback in a Mu Amp Stage Question

[KarenColumbo]

Dear experts,

This is a classic for me: I'm trying to do a mild overdrive for Bass guitar, y'know, add a bit of "Rock Grumbling" to an otherwise perfectly civilized bass guitar voice.

After years (really) of occasional experimenting I found that totem pole circuitry has the best balance between gain reserve, pleasant colouring and ease of breadboarding (no trim pots) for this kind of goal.

So I ended up with the usual suspect - 2 gain stages, a FMV tone stack after them and an output buffer. here's gain stage 1:



It sounds quite alright. I tried with J201, 2N5457 and 5 of preciously few MPF 4393 I bought just in time a few years ago (they're unobtainium now, only certain "business men" on eBay sell them for ridiculous amounts of money).

Since "Mu Amps" (in this case, I guess, the "real" version thanks to the 1k R4 in the correct position) have LOADS of gain, a bit of Local Negative Feedback seemed to be in order (R5 & C5 in my schematic). I just chose the 47ish values according to some circuits that utilize this kind of thing (C***inbread SFT II for example).

But somehow I can't get the sound quite "right". It varies between fuzzy and doom clean. And I don't mean the tone stack.

Instead of sticking in a 1M pot instead of R5 and tiddling around on it I'd like to know: How can I calculate the physicially correct value for R5 to have the perfect balance between signal gain, noise reduction and treble roll-off?

[antonis]

http://www.geofex.com/Article_Folders/modmuamp/modmuamp.htm



NFB is a very good thing but, IMHO, not a nostrum/panaceia..
(for a mild overdrive, there isn't any reason for huge gain and respectively huge NFB..)

[KarenColumbo]

Ah, thx for the graphics, I knew I had exactly this one in my archive but didn't think to look there.

Well, I tried with single Fet stages, i tried with Mosfets, I tried with BJTs and I treid the config on the pic you posted, i.e. without load resistance between the two transistors - but they all didn't "purr" just right. This totem pole things give me the amount and kind of pleasant dirt I was trying to achieve. Toning the gain down by fiddling with the source resistor and bypass cap didn't come to fruition either. That's why ... and the dynamics are just right now, if I dial in the right amount of gain, i can bring a certain growl into the sound if I really lean into it while "normal" notes are almost completely clean.

[antonis]

OK.. Start by deleting C2..

[KarenColumbo]

Okay, done. Had it there to cut some noise, but I don't really miss it.

Concerning LNF: If I decrease R5 i get more gain, correct? It's still a bit much drive now - gotta find a value that 1. doesn't overdrive a bass guitar too much, but 2. makes the thing still attractive for guitar use. I'll try 680k/6.8n next

[amptramp]

The top of C5 should connect to the source of the upper FET (the other end of R4) because that is where the output is.  That will give you feedback from the output (which is what you want to clean up) rather than the intermediate point at the drain of the lower FET.

[Kipper4]

We will require a demo Andreas.
How are you chum?

[KarenColumbo]

The top of C5 should connect to the source of the upper FET (the other end of R4) because that is where the output is.  That will give you feedback from the output (which is what you want to clean up) rather than the intermediate point at the drain of the lower FET.

Yeah, i think this is what I wanted to do, when I accidentally (I do this mostly while working my day job) connected it as depicted an rather liked the result. Never thought of doing it the way you described, which I will rectify in a jiffy.
(Edit) Connected it directly @ the source - sounds fine to me, less noise, more treble, tighter sound. Thx! Gotta try this with bass soon.

[KarenColumbo]

We will require a demo Andreas.
How are you chum?

Sick as a dog with an autumn bug right now, so I can't hit the rehearsal room in the cellar 3 floors below. Will do a.s.a.p. Thanks for asking, mate. It's the annual antivirus update, I guess

[KarenColumbo]

Incidentally (I swear!) there's a thing called "Tweed Man" by pedalpcb that does more or less what I try to accomplish here. Only more refined and with a rather complicated 4PDT switch thing for extra gain which, disconnects my (their) local feedback circuit. Gotta build me one of those soon. And they have the same LF connection as I did. If I look at the SFT2 schem from the same builder, it's the same thing. Maybe this adds some grit to the whole mess. Anyway, I will keep mine like Mr. amptramp told me to.

I may be wrong of deaf already but the switch I included (not in the schem) that toggles bypass cap between 1uF and 33uF (there's 22 uF inthe schem, what a mess) now doesn't seem to make as much difference as before. Cool and interesting ...

Heres the SFT2 schem with the LF cap/resistor connected at the drain of the lower fet instead of upper source, btw.

[amptramp]

One thing to keep in mind with the µ-amp: without a load, it will distort because the input swing allows more current through the lower FET and less through the upper FET when it goes positive and less in the lower FET and more in the upper FET when it swings negative.

Since the FET's are in series, the difference in current has to go somewhere and that somewhere is the load.  This is not an amp you want to run into a high-impedance stage unless you actually want the distortion you get from trying to drive the upper and lower stages to the rails.

The SFT ii schematic shows loads that are too light for each stage although the negative feedback connection (either the one in the schematic or the one I suggested) is a load that can reduce the need for a load in the subsequent stage.

[KarenColumbo]

One thing to keep in mind with the µ-amp: without a load, it will distort because the input swing allows more current through the lower FET and less through the upper FET when it goes positive and less in the lower FET and more in the upper FET when it swings negative.

Since the FET's are in series, the difference in current has to go somewhere and that somewhere is the load.  This is not an amp you want to run into a high-impedance stage unless you actually want the distortion you get from trying to drive the upper and lower stages to the rails.

The SFT ii schematic shows loads that are too light for each stage although the negative feedback connection (either the one in the schematic or the one I suggested) is a load that can reduce the need for a load in the subsequent stage.

I see (I think). What about this here? That's the "Tweed Man" from pedalpcb, allegedly a traced Catalin "Formula 5F6":


I've seen this kind of system in almost countless variations now and had quite a few on breadboard. Doesn't the buffer at the end "suck up" this high impedance and iron it down? And the quite lossy tone stack has some kOhms impedance, too, hasn't it?

[tubegeek]

The top of C5 should connect to the source of the upper FET (the other end of R4) because that is where the output is.  That will give you feedback from the output (which is what you want to clean up) rather than the intermediate point at the drain of the lower FET.

What happens to C4 after this change? Where is it connected - or does it get removed?

[KarenColumbo]

C4 got to stay in in any case. I quote R.G.:



What happens with signal? If a small upward voltage is connected to the gate of the lower JFET, it turns the lower JFET on a bit, and the lower drain sucks in more current. The current can't come from the capacitor, as the capacitor is effectively in series with a 5M resistor. It can only come out of the source of the upper JFET. The source voltage then drops, and so the capacitor pulls the gate of the upper JFET down as well. This has the effect of turning the upper JFET off more, which lets the voltage on the source drop further. In effect, the upper JFET is actively contributing to the voltage swing on the lower JFET's drain. The reverse happens when there is a negative-going signal on Vin.

[antonis]

Doesn't the buffer at the end "suck up" this high impedance and iron it down?

Of course it does it.. But this action concerns the output stage..
(tonestack and next effect input..)

And the quite lossy tone stack has some kOhms impedance, too, hasn't it?

Of course (again) it has.. But that impedance should be considered in parallel with Q3 Drain load (refer on Tweed Man), which load is considered higher than, or at least comparable to, tonestack impedance..
(it dominates significantly Q3/Q4 stage gain..)

[KarenColumbo]

Ah, I see - well thanks for that - seems I must learn to read a schematic the proper way.

[antonis]

Well said, Andreas..!!
(but, STF2 & Tweed Man aren't the most proper circuits for an "elementary" diagram reader - from the signal flow point of view, at least..)

[amz-fx]

Andreas,

There are a few other schematics that you might like to review in one of my recent blogs:

http://www.muzique.com/news/amp-in-a-box/

Enjoy!

Best regards, Jack

[amptramp]

The top of C5 should connect to the source of the upper FET (the other end of R4) because that is where the output is.  That will give you feedback from the output (which is what you want to clean up) rather than the intermediate point at the drain of the lower FET.

What happens to C4 after this change? Where is it connected - or does it get removed?

C4 stays where it is, connecting the gate of the upper FET to the drain of the lower FET and most importantly, to the bottom of R4.  If R4 is equal to the reciprocal of the transconductance of the upper FET, the change in current is symmetrical - if the lower transistor current swings upward by one milliamp, the upper transistor current swings downward by one milliamp.  Since the currents are in series, the difference in current has to go to the load.  If the load resistance is too high, the amp will hit the upper or lower rail or possibly both if it is overdriven.

You will see designs where there is no R4, it is replaced by a short circuit / zero ohm resistor.  The operation of this circuit is different.  In this case, the upper FET acts like a high impedance current source.  It is not a µ-amp.  The real µ-amp is also referred to as SRPP (shunt regulated push-pull) in some of the literature.

[KarenColumbo]

Andreas,

There are a few other schematics that you might like to review in one of my recent blogs:

http://www.muzique.com/news/amp-in-a-box/

Enjoy!

Best regards, Jack

Hey Jack, thx, those circuits are regularly inhaled by me. Somehow when I'm breadboarding the blues away, I always return to this kind of circuit - this diode-clipping world is not really for me. I like the hint of raw power behind these design and love to taunt it with various fet combinations: weaker ones in the first stage to keep them from fuzzing out, big trolls like 201s in the second to give the output (and the amp after it) something to work with. An inexhaustable source of different flavours in those