How many gain stages til it chugs?

[KarenColumbo]

I'm pretty firm now with those Mu Amp stages. Still a lot to ingest, but I can breadboard a decent distortion almost by myself.
Except it sounds like toejam
But I digress.

What do I (the guitar signal) need to have this saturated, thick, almost bursting compressed sound that's not fuzz, but has an almost kickdrum style bass content?

It seems to me that afer three carefully balanced gain stages there's the end of the line. It cant get more dostorted (which noone needs, really), but it gets ,,flabby" and oversaturated, no matter how ,uch filtering takes place.

I took all your wisdom and tried different approaches with pre-, post- and inter-gain EQing, but I cant seem to get this ultra heavy chug/djent bass content when hitting those hand-damped power chord attacks.

I tried ,,sub bass" tricks, I mojofied baxandall bass/treble stacks,somehow it only gets thick, but not massive.

Could you walk me through a thing like Openhaus distortion, only with jfets?

[Mark Hammer]

Generally speaking, you have stumbled on what other better minds than mine have concurred.  Much the way that a piece of paper cannot be folded more than seven times, more than 3 gain stages is rather pointless.

Why?  Part of that is because there is only so much gain that can be applied to a guitar signal averaging around 100mv, using a 9VDC supply.  Another aspect is that if those gain stages are not absolutely pristine clean, and add harmonic content, once you've hit 3 stages of doing just that, notes tend to get lost in the "harmonic haze".  It's not just the inevitable compression and loss of dynamics, but the inability to actually hear distinct notes amidst all the added crap.

Is there a way around this limit?  Probably.

1) Increase the supply voltage so that one doesn't run up against headroom limitations quite so soon/easily.
2) Decrease the amount of gain demanded of any single stage and find some way to attenuate the output of earlier stages, keeping in mind that gain is multiplicative. A gain of 10x, followed by a gain of 10x, followed by another gain of 10x is a total gain of 1000x, in the absence of any interstage attenuation.
3) Tame the two tails of the frequency spectrum.  Harmonic content accumulates over stages, and harmonics of harmonics of harmonics sounds awful.  So use your lowpass.  Most of the signal "lives in the basement", so one will reach the limits of headrooma little more gently when low end is trimmed appropriately.

[PRR]

Raising the supply just moves the problem down the tracks. You have the *EXACT* same problem in 300V tube amplifiers. More than about-three stages is hard to handle and generally no better.

Trimming the ends has merit.

Attenuating between distorter stages is THE next path to follyXXX follow. One of the Big Names puts 6 or 8 tube stages together (only 110V rail), but R-C networks between to both reduce and shape the signal. As a starter target, each stage may run a total gain like "2". The 6th stage maybe distorts at 10mV input. The 5th stage dirts for 20mV input, the 4th stage for 40mV input, etc. But they should not all be the same size steps or same for all frequencies. I suspect a lot of "research" (hard jamming) went into that product. It works but so do simpler rigs.

[thehallofshields]

Turns out it was 3.

[KarenColumbo]

I set up three identical gain stages on the board (it will be five, just to have it done):
(there's of course a connection from C1 to "STAGE_IN" - just wanted to clarify that I don't repeat the input thing C1/R1/R2 in every stage)


There's some questions (of course), before I proceed ...

1. The pot ("GAIN") of course does not really set the gain, it's a resistive divider, and part of a filter in conjunction with those caps. It WILL influence the next stage, though, by controlling how much juice I hit it with. I guess I'd need R3 to be a variable resistor to set the REAL (nominal?) gain of the stage. Am I right?
2. I put in C10 because I saw this somewhere - I think in the Dr. Boogey schematic. It emulates something I forgot the name of. Transcapacitance? Can't remember. Is it necessary/will it shape the waveform?
3. The same goes for C2. In another place I saw a diode there. I can't remember what the cap does. Is there a possibility of sound shaping here? I tried different values <1n, not much difference, at least to the ear. Will have a look another days through the scope. Or should I ditch the cap and never look back?
4. R6: As I understand it, the setup with the low ohmage voltage divider (R4 & R5) plus R6 is the so called "low noise biasing" technique. I tried to make R6 200k, as I did with the particular resistor in the next stage and put a 500k pot with the wiper to VCC/2 there and wiggled it to "balance between those stages. I heard some difference: when the pot was in the exact middle, the sound had the most punch, but seemed "fuzz-y" to me. In either extremes, left or right, I thought the sound became more transparent. What did I do there with the biasing?
5. C3: As I read along many texts about mu amp/SRPP I realized this cap is essential for the circuit to function properly/at all. If I play around with the value (I've seen everything from 22n to 470n there), what does it change? I couldn't really extract this information from the texts, except it couples both jfets so the upper one actively "helps" driving the load. But what does a change in capacitance result in?
6. If I change the "center voltage" between R4 and R5, say, to 1/3 to 2/3 in the first stage, then 2/3 to 1/3 in the next, will this unbalance to waveform in any way? Like, clip the lower swing more, then the upper in the other stage? Is this a sensible area of experimenting?

I know it's all been done and asked about before - and you experienced and long-time members must have read (and answered!) those questions a zillion times before. I still dare to ask them again, all in one place. Hope you don't mind!

[PRR]

> gain stages

Looks to me like a Source Follower. Unity gain.

[KarenColumbo]

> gain stages
Looks to me like a Source Follower. Unity gain.

Yes Forgot the magic resistor. Now it's 4 stages and the tone control of "Spitfire".

[KarenColumbo]

Some mods.

- An R/C filter in the first stage from upper jfet's gate to lower jfet's gate. Saw this @ Catalinbread SFTII. Tones down the whole setup a bit
- Generally upped the capacitance between stages
- Voltage follower between stage 4 and the tone pot



With careful tuning of the gain pots it almost sounds digestable. Very raw and tightly woven. Surprisingly quiet - only slight hum (Stratocaster)

Next step: inter-stage filtering. Will be a tight fit.

[Mark Hammer]

I think you're going to want to remove those 150pf bypass caps.  At least the first 3.  Unless your plan is to leave each of the "Gain" pots up near close to full.

My own strategy of cascaded gain stages is to gradually spool out top end in later stages.  The reasoning is simple.  If the gain stage adds harmonic content, then you want to avoid harmonics of harmonics of harmonics, and keep the focus on enhancng the lower-order harmonics.

[duck_arse]

I can see from your breadboard that you have followed your diagram, but your diagram(s) has that top row of 1nF caps to Vcc, which are all shorted out. you can't really bypass a wire link to any good effect.

a question - how come the first three stages output via the lower fet source, but the fourth stage outputs via the upper fet source?

[KarenColumbo]

I can see from your breadboard that you have followed your diagram, but your diagram(s) has that top row of 1nF caps to Vcc, which are all shorted out. you can't really bypass a wire link to any good effect.

a question - how come the first three stages output via the lower fet source, but the fourth stage outputs via the upper fet source?

Well, I've been trying to get my head around two TubeCad blog articles concerning SRPP. I remember one or two of the schematics had a resistor from the upper jfet's drain to VCC. This resistor has to be calculated meticulously, I read, so I didn't put anything in at the moment. The 1n caps are in fact waiting for their bypass resistors

The 4th stage's upper out put was a mistake - I forgot to put in the "connection" resistors between the jfet pairs, so all was well. Then I put in those 1ks, in case of the fourth stage I put it in BELOW the voltage follower connection. Didn't sound worse than before, so I left it that way.

Will of course try and correct those errors.

Edit: The configuration(s) with drain resistors at the upper jfets is an entirely different story, so away with those caps! Tried the 4th stage's output at the lower fet: No difference to the ear, so I leave it this way, since it's easier on the breadboard.

I think you're going to want to remove those 150pf bypass caps.  At least the first 3.  Unless your plan is to leave each of the "Gain" pots up near close to full.
My own strategy of cascaded gain stages is to gradually spool out top end in later stages.  The reasoning is simple.  If the gain stage adds harmonic content, then you want to avoid harmonics of harmonics of harmonics, and keep the focus on enhancng the lower-order harmonics.

Thanks for chiming in, Mark. I removed all bypass caps. About those harmonics: Haven't started with inter-stage filtering yet, but I sure will take extra care to tame the beast. But I must say, the bottom end is coming along nicely. Good solid bass now - hope I don't screw it up.

New iteration:

[PRR]

> New iteration:

[KarenColumbo]

> New iteration:

Oh man. I really shouldn't tinker with electronics while working (proof-reading badly written general interest print magazines). This is extremely embarassing Thank you for pointing this out. Can't believe I'm this inattentive. I'm glad I'm not doing something others depend on ...
Edit: It's "only" the schematic that's wrong. Nevertheless: sloppy work.

[merlinb]

What do I (the guitar signal) need to have this saturated, thick, almost bursting compressed sound that's not fuzz, but has an almost kickdrum style bass content?

You need a soft clipping threshold that tracks the signal amplitude (over three or four stages). This something tube amps tend to produce naturally, although it needs critical tweaking of course. With solid-state circuits it can be more tricky as they often are more inherently linear (so it's either clean or clipped with nothing in between), they clip harder, and the clipping often doesn't track the signal amplitude (esp with most opamp clippers). Simple JFET 'copies' of tube stages can emulate tube clipping to some extent -which is more or less what you're doing- but you need some interstage dividers (adjust on test) or series gate stoppers so you can control the onset of clipping.

[rockhorst]

One thing to also consider is what you're feeding the distortion into. Are you driving a relatively transparent power amp, or does it go into a commercial preamp, with it's own EQ and clipping? This also makes a huge difference. The mother of mu-amp designs, in pedal for at least, the mini booster will distort my amps clean channel when level and gain are high. However, feeding it into a power amp, it makes a nice sounding clean channel without too much distortion. Something to consider.

Mu amps are fun to experiment with but for some reason I rarely like their distorted sound. An exception to this is the V1 Dirty Little Secret by Catalinbread. Nice, well documented circuit, worth checking out.

[KarenColumbo]

What do I (the guitar signal) need to have this saturated, thick, almost bursting compressed sound that's not fuzz, but has an almost kickdrum style bass content?

You need a soft clipping threshold that tracks the signal amplitude (over three or four stages). This something tube amps tend to produce naturally, although it needs critical tweaking of course. With solid-state circuits it can be more tricky as they often are more inherently linear (so it's either clean or clipped with nothing in between), they clip harder, and the clipping often doesn't track the signal amplitude (esp with most opamp clippers). Simple JFET 'copies' of tube stages can emulate tube clipping to some extent -which is more or less what you're doing- but you need some interstage dividers (adjust on test) or series gate stoppers so you can control the onset of clipping.

Could you elaborate eventually? I'd be a most eager listener! I intended those Gain" pots as dividers - am I on the wrong track here? Gate stoppers? I will google this - if you have any insider info, I'd be thrilled

One thing to also consider is what you're feeding the distortion into. Are you driving a relatively transparent power amp, or does it go into a commercial preamp, with it's own EQ and clipping? This also makes a huge difference. The mother of mu-amp designs, in pedal for at least, the mini booster will distort my amps clean channel when level and gain are high. However, feeding it into a power amp, it makes a nice sounding clean channel without too much distortion. Something to consider.

Mu amps are fun to experiment with but for some reason I rarely like their distorted sound. An exception to this is the V1 Dirty Little Secret by Catalinbread. Nice, well documented circuit, worth checking out.

Thanks, I will sure have a look at this Secret! And the output goes straight into the (very) clean channel of this beast: https://en.audiofanzine.com/tube-guitar-amp-head/hughes-kettner/trilogy/medias/pictures/a.play,m.415320.html

[PRR]

> badly written general interest

Also KISS. When you want a lot of gain stages, use *simple* gain stages. You'll still have over-abundant gain. Your simpler distortions will still add-up to hard mangling. You save a few bucks. And the plans are easier to proof-read at a glance.

Compute (or measure) the gain of the simple stage and figure the interstage networks to loose a lot of that. For low-current JFET stages with gain of 30, take 10:1 minimum loss, 1Meg into a 100K pot. Recall that the start-dart I gave was gain of 2 in each stage. Trying to get 15:1 loss with just a pot is fiddly, and you never need most of the pot range.

[KarenColumbo]

Thx very much, Paul (& Sirs!), for bearing with me!
Of course you're right. It'd be much easier to get rid of at least half of the components to get a very good sounding contraption that's much more controllable (and computable). It's this: I wanted to take 2 or 3 steps at once so I ...
1.) ... get a grip on Mu Amp/Srpp by combining them (I printed out some of the math and am trying to internalize the most inportant formulas - been some decades since I was at school, I'm very rusty, never needed it). K.I.S.S. will be the way to go, I humbly admit. Still I HAVE to have tried it at least once, haven't I? E.g. there's this particular layout I'd like to try out (digging in the dark):

2.) This is where I'm probably off the track - because it gets homeopathic: In my head it's "the more parts for a gain stage, the more sound-colouring takes place." This probably is bullshit, isn't it? But that's why I have those totem poles instead of single fet stages there. 
3.) I WILL replace those "Gain" pots with fixed resistors as a first step towards simplifying - I see now that it's next to impossible to get them in a position where all four of them sound balanced and I'm hearing what one does to the other three. And this still does not incorporate the GAIN of the stage itself, where I also will try to follow your advice!
( 4.) I ordered better breadboards. The ones I'm using - bargain, bargain - are of very poor quality, many contacts have been practically unusable from the start.)

Plus: I should take my time and think for myself before I post here - to not bore you all to death 

[merlinb]

2.) This is where I'm probably off the track - because it gets homeopathic: In my head it's "the more parts for a gain stage, the more sound-colouring takes place." This probably is bullshit, isn't it?

Yes it is. Often all the extra parts are there to iron out all the imperfections of the simpler, no-frills gain stage. You don't find clever-dick gain stages in guitar amps for good reason: the simple resistor-loaded gain stage or cathode follower is king. They produce a little distortion when 'clean' which increases pleasantly until you hit clipping which is soft(ish), and becomes harder as more stages become overloaded and join in the mix. There is also the 'self rectification effect' at the grids of the tubes as the coupling caps charge up, which forces the signal envelope to hug the clipping threshold (can't really explain this in a quick forum post!), which has the same end result as a clipping threshold that follows the signal envelope. There may also be PSU sag but that is less important IMO.

FETs have large manufacturing spreads that mean they often need adjustment on test. The advantage of a mu-amp is that it can set the idle current automatically without you having to adjust any bias resistors on test. But you pay for this convenience with very different gain/overdrive bahaviour. With the amount of tweaking you're going to have to do anyway, I think you're better off ignoring the mu amp and going with simple stages instead:
http://runoffgroove.com/fetzervalve.html

[KarenColumbo]

2.) This is where I'm probably off the track - because it gets homeopathic: In my head it's "the more parts for a gain stage, the more sound-colouring takes place." This probably is bullshit, isn't it?

Yes it is. Often all the extra parts are there to iron out all the imperfections of the simpler, no-frills gain stage. You don't find clever-dick gain stages in guitar amps for good reason: the simple resistor-loaded gain stage or cathode follower is king. They produce a little distortion when 'clean' which increases pleasantly until you hit clipping which is soft(ish), and becomes harder as more stages become overloaded and join in the mix. There is also the 'self rectification effect' at the grids of the tubes as the coupling caps charge up, which forces the signal envelope to hug the clipping threshold (can't really explain this in a quick forum post!), which has the same end result as a clipping threshold that follows the signal envelope. There may also be PSU sag but that is less important IMO.

FETs have large manufacturing spreads that mean they often need adjustment on test. The advantage of a mu-amp is that it can set the idle current automatically without you having to adjust any bias resistors on test. But you pay for this convenience with very different gain/overdrive bahaviour. With the amount of tweaking you're going to have to do anyway, I think you're better off ignoring the mu amp and going with simple stages instead:
http://runoffgroove.com/fetzervalve.html

Ok, that was pretty clear. I have to rethink this