Tube Screamer feedback loop order?

[artofharmony]

Hi all! I am working on building a Tube Screamer from Jack Orman's Son of Screamer schematic. I wanted the gain to be permanently fixed at minimum, so I've removed the gain pot completely. In my PCB design, I accidentally put the 47k resistor below the diodes and the cap in the gain stage feedback loop. Would this have a negative effect on the gain? I ask because my PCB isn't producing as much gain as I would expect it to. Link to the schematic here: https://www.muzique.com/tech/scream.htm

[GGBB]

Can you post a diagram of what you designed? "Below" is a relative term, and also you haven't said which cap. A picture is worth a thousand words.

[antonis]

If you mean something like below, it's OK..



Gain is 11X (20.8dB) but you can raise it according to your taste by making R3 bigger, following A = 1 + R3/R2 formula

P.S.
You might need to make C3 bigger to retain original LPF cut-off frequency, or modify it according to  f = 0.159/(C3*R3) corner frequency formula..

[artofharmony]

Ah my bad lol referring to antonis's pic above, move R3 to between D2 and IC1a. Including a screenshot of a portion of the PCB design; the red lines from the 47k resistor form the feedback loop. That sort of illustrates what I'm talking about.

[marcelomd]

Unless I royally misunderstood the questions...

If you mean you put the resistor physically closer to the opamp instead of the cap or diodes, it doesn't matter.

The feedback loop should be as short and close to the opamp as possible, but the "order" of the steps don't matter.

[artofharmony]

Unless I royally misunderstood the questions...

If you mean you put the resistor physically closer to the opamp instead of the cap or diodes, it doesn't matter.

The feedback loop should be as short and close to the opamp as possible, but the "order" of the steps don't matter.

Nope, that was the question! Thanks!

[Mark Hammer]

I like to explain op-amps thusly...

Imagine you had a car whose gas pedal was fixed at pedal-to-the-metal, and the only way to adjust its speed was to vary how hard you stepped on the brakes.  Op-amps are designed in a similar fashion.  The way one adjusts their gain is to adjust the amount of negative feedback applied to the inverting (-) pin.  When more of that negative feedback is preserved, gain is reduced more.  When less of it is preserved, gain is increased.

The 47k feedback resistor, and 4k7 on the ground leg, form a sort of "volume control" for the negative feedback.  Make the 47k higher in value (like the stock gain pot maxed) OR the 4k7 lower in value, and more of the negative feedback is bled off to ground.  So, 500k feedback and 4k7 ground leg gets you a gain of 107x, but so would 47k feedback and 430R to ground (well, a little more than that, but I'm assuming standard resistor values).  And of course, if the resistance of the ground leg is infinite, then ALL negative feedback is preserved.  This is why you'll see unity-gain buffer stages with an output-to-input feedback connection, but NO ground leg.

As you've drawn it, 47k feedback and 4k7 ground leg gets you a gain of 11x.  That's not nothing, but it is highly unlikely to get you much audible clipping, except for when you slam a power chord into it, and then only on the initial peaks, which may pass too quickly to be heard as anything more than a momentary brightness.  Use of silicon diodes means the amplified signal will need to be greater than around +/-600mv to yield clipping.  If they were Schottky diodes, that often have forward voltages of around 200mv, then a gain of 11x would likely be sufficient to produce something you could hear as clipping.  But then the amplitude of your final output would be unacceptably low.  So, while there is nothing "incorrect" about having fixed gain, I think you will want to use a higher feedback resistor value, like 150k, which will yield a gain of 33x.  That won't get you any serious grind or sustain, but will add more coloration than your current setup.

Note as well that the basic TS circuit was really designed to achieve more consistent clipping over the whole fingerboard/neck.  Because wound strings provide a hotter signal than unwound ones, it achieved that consistency by attenuating the bass in the clipping stage.  The .047uf/4k7 pair rolls off bass, beginning around 720hz.  If you make the .047uf cap a higher value, like .1uf or .22uf, you feed the clipping stage more bass, and get more clipping for the same amount of gain.

[artofharmony]

Imagine you had a car whose gas pedal was fixed at pedal-to-the-metal, and the only way to adjust its speed was to vary how hard you stepped on the brakes.  Op-amps are designed in a similar fashion.  The way one adjusts their gain is to adjust the amount of negative feedback applied to the inverting (-) pin.  When more of that negative feedback is preserved, gain is reduced more.  When less of it is preserved, gain is increased.

Thank you for this explanation! Short, sweet, and to the point. I have a feeling I'll be referring back to it 👍

As you've drawn it, 47k feedback and 4k7 ground leg gets you a gain of 11x.  That's not nothing, but it is highly unlikely to get you much audible clipping, except for when you slam a power chord into it, and then only on the initial peaks, which may pass too quickly to be heard as anything more than a momentary brightness.

That's kind of the point! I use Tube Screamers almost exclusively as boosts. Except in rare cases, my TS is set at max volume, min gain, and tone to taste. I slam the front of a transparent drive with the TS, which then slams the front of whatever amp I'm using.

This TS circuit is built to be the Boost section of a 2-in-1, transparent drive/mid boost combo. I took out the gain knob for two reasons. 1. I rarely use it. 2. It saves space for the four other knobs and order toggle that I wanted on the front of the pedal. It's basically my own signature overdrive 🙂

[Mark Hammer]

OK understand better now.
Given your intended application, I'll suggest a slightly different route.  I like the approach adopted by the Proco Rat and Nobels ODR-1.  Instead of a simple bass rolloff, like the TS and similar pedals use, a "dual ground leg" approach is used, in which there is modest gain for the full bandwidth, without any bass rolloff, and a second ground leg is used to increase gain for content above some frequency.  By this means there is no sacrifice of bass and lower mids content, but still has that greater sensitivity to upper freqs the TS strives for.

In your case, assuming you're content with a gain of around 11x, I'd recommend the following.  One ground leg would be 5k6 and 100nf (instead of 4k7/47nf).  This yields a gain of 9.4x for the entire spectrum, down to 56hz.  A second ground leg of 3k9 and 39nf provides a gain of 13x for content over 1khz.  In this way, you still maintain roughly the same overall high-to-low balance, but you don't have to sacrifice the bottom and "guts" to do so, as you would with a progressive bass rolloff.  A gain of 9.4x might seem like audibly less than 11x, but bear in mind a stock TS ground leg would cut the heard amplitude of content at 360hz by 6db, by 6db again at 180hz, and another 6db at 90hz.  So the gain might seem less in what I've suggested, but the actual bass loss is much smaller.

I've illustrated the impact here.  This pedal is a multi-drive I made for myself, with a TS on the right.  The two toggles alter the clipping diodes and the tone shaping of the clipping section.  The leftmost toggle selects between a standard TS bass rolloff, and my implementation of an ODR-1 dual-ground leg.  It's probably set to more gain than you intend to employ but you can hear the difference between the two different rolloffs.  When the toggle is pushed up, it's the stock TS rolloff, and when pushed down, it's the ODR mod.  There's a clear difference in "guts".

The suggested component values aren't compulsory.  If you used 39nf/4k7, instead of 39nf/3k9, that would provide a gain of 11x for content above around 870hz, which may yield the sort of tonal balance you're aiming for.  The overarching goal is to have a bit of "bite" to the boost, without having to forfeit girth/guts.

[Rob Strand]

Nope, that was the question! Thanks

Parallel components don't care about the order on the schematic or on the PCB.

FWIW, series components don't care about the order either.   For example if you take antonis's schematic and changed the positions of C2 and R2 you still have C2 in series with R2 and there is not change in behaviour.

There are cases where you might switch components with a switch, say the C2 cap value, and you might find one ordering is less susceptible to noise than the other.   That's a more advanced topic and you it shouldn't hinder the idea that series and parallel connected components don't care about order.

[artofharmony]

FWIW, series components don't care about the order either.   For example if you take antonis's schematic and changed the positions of C2 and R2 you still have C2 in series with R2 and there is not change in behaviour.

Rob, thanks for that explanation! I had wondered that too.

[antonis]

if you take antonis's schematic and changed the positions of C2 and R2 you still have C2 in series with R2 and there is not change in behaviour.

That said (and done), it should be a good idea to connect R1 to R2/C2 junction..
(not too much effective R1 bootstrapping due to C2 relatively low value but better than none..)

P.S.
Only if one persist to unnecessarily "pollute" Vref instead of connecting R2/C2 branch to circuit's GND..

[m4268588]