Newbie messing with a Tube Screamer overdrive circuit

[ExCx]

Hi there.

I'm a newbie trying to understand and modify a simplified clone of Tube Screamer.



Although I had assembled an amp and a fuzz pedal successfully before, I was just blindly following the schematics. Now I want to learn about the inner workings and hopefuly design my own. I'm just a hobbyist with no electrical background but I do know some theory on a simple level and I did some reading on building analogue circuitry. Still, I'm way behind in practice and have LOTS of questions about this simple project. I'll highlight the parts for each question I have, so you can follow me better here.



1- I understood that this resistor is setting the input impedance for the circuit. From what I read, it's always better to make this as high as possible. I also stumbled upon a pedal maker bragging about his designs and why he puts a 10M input resistor instead of 1M so the signal loss is minimal. So I also changed this to 10M and I *think* I heard some subtle difference though I'm not really sure. What I want to ask is, if this is the right way, why don't every pedal in existence have a 10M (or higher?) input resistor to transmit pickup signal as good as possible? Is there a catch here, like over-sensitivity?

2- When I first assembled the circuit (on breadboard) with the exact values written on schematic, there was an unbearable high pitched noise whenever the guitar is not playing. Looking for a solution, I found that unintended oscillation is pretty common in these kind of high gain feedback circuitry. So I tried some proposed solutions like putting caps of various values between Vref & ground, adding a small resistor before output cap or adding a RC pair between polar inputs of the opamp but none of these work. Then I hopelessly examined similar screamer schematics and noticed something different in this one. Here, the filter of the first gain stage is biased into Vref instead of the ground. So, I just gave it a shot and poof, the noise is gone! Pedal is working as expected. I really don't know what I have done here. So yeah, this ground connection on (2) is Vref for me. Can anyone explain why this works and why would the other one is noisy?

3- This here should be a low-pass filter for the gain stage am I right? When I calculate the cutoff freq is 268Hz. So does that mean only the frequencies lower than 268Hz are amplified? Isn't it a bit low? Also, shouldn't an overdrive pedal amplify the higher freqs instead? What I'm missing here? Oh also, referring to my previous question, did my modification of the ground to Vref  may have cancelled or altered this filter in any way?

4- This RC pair works as a high-pass filter for input, right? Or is it not because it's biased with Vref instead of ground and R here is only for limiting current and C is just for decoupling purposes? When calculated as HP, the cutoff freq is 7Hz which looks kind of absurd. I'm really confused.

5- I do understand simple workings of a non-inverting amp. So this stage is amplifying the signal by [1 + (51000 / 2700) = 19.88] when the gain pot is at zero and [1 + (551000 / 2700) = 205.07] when the gain pot is at max. Is this range wide enough? Because I barely hear much difference between both ends. Or did I also broke this stage somehow ?

6- Before starting this project, I knew the large number of parts on the breadboard will make it hard to build, troubleshoot and even harder to modify. So I prepared my power supply and glued I/O jacks onto a seperate breadboard. I assembled the supply according to this little project. So, now the R and C values are different than the screamer schematic. I wonder if this would make any difference. R's here are for dividing the voltage so could they be 100Ω or 1M as long as they are of the same value? Are the values of C's related to R's or totally independent? Why is it 47u in one and 22u in the other? I believe they're there to smooth out voltage peaks and prevent external RF noise (right?). What's stopping me from adding more of them with various values so my pedal works smooth as a baby's bottom? I'm probably very wrong here but enlighten me please.

7- This RC pair functions as a LP filter right? I calculated the cutoff as 7234Hz. Although I know the guitar's frequency ceiling is about 1200Hz, I guess this is a fair value considering the harmonics added by the first gain stage. No questions here, I just need an affirmation 

8- This output cap and volume knob would create another HP filter I believe. I'm not sure how to calculate this, though at R=100k cutoff is 15.91Hz which means no hearable effect on sound, as it should be. Still, please tell me if I'm wrong and what should I be aware of if I ever need to change values of these. The volume pot also sets the output impedance I assume. So doesn't that mean changing the volume changes both the impedance and the HP filter's cutoff? How is this not a problem?

About the rest of the parts; I'm using a TL072 op-amp, I think I understood about the diode clipping, smoothing out the huge gain so the output would not be a pure square signal. Pico valued caps inside feedback loop are to stabilize and prevent the unwanted noises caused by clipping (so I read). I still wonder what would happen if I change their value.

You see, I'm pretty clueless about nearly every aspect of this simple circuitry Or you can say I know little enough to make uneducated guesses about every part of it  . Anyways thanks for reading and thanks in advance for your valuable responses.

[idy]

Welcome to the forum.
You  are asking great questions, informed by good basic grasp of principles.

http://www.geofex.com/article_folders/tstech/tsxtech.htm

Has the now classic breakdown of the original TS; the 'reamer is missing the buffers and has a few "improvements."

I am not one of the top guys here, but I will take a swipe at a few of these.

1) Pull down resistors are "as needed." Too high and the cap may not be pulled down. Too low and the impedance is dragged down. The original screamer had an impedance of 512k (more or less). This one ends up the same with the pull down and bias resistor.(?) Tube amps are usually quoted as 1m. The choice of a lower impedance with slight loading is aesthetic, it produces a slight filtering some people liked. Lower input impedence means loss of treble, maybe "warming." A mosfet booster at 10m sounds dazzling and sparkly to me, cold hard and too bright to others.

3) That "low pass" is attached to the negative feedback loop. It gives a bass heavy signal to the negative input, so limiting the gain of that signal... and producing a mid/high boost.

4) That "high pass" lets most everything through. The resistor is there to bias the opamp, the cap keeps the DC inside so that plugging a guitar in doesn't send dc up into your axe making the pots crackle and messing with the bias. I think the filtering here is negligible for guitar.

5) You should hear a change in gain with the pot. From 20 to 200 you should hear.

6) Values in the vref section you can vary. Resitors too high and maybe noise, too low and you are wasting current. The values of the electro caps... in these things whatever you have will be ok. An engineer will have a better answer. One may chime in. There are finer points, but yes they smooth out supply noise. There are ways to do more of this, like a small series resistor from the 9v in. Or small caps in parallel to the big ones...

7) That low pass is pitched 10x higher than the tube screamer. So it really only slopes gently off the high "sizzle." Just because the guitar's highest fundamental is around 1k doesn't mean a filter above 1k won't change the sound. It hits the harmonics. But that is a pretty high point, 7k.... The original TS had a cut at 723. Meaning the really cut frequencies were well above that. 6db per octave is gentle slope. When you turn your guitar tone pot down you are moving the start of that slope lower and lower.

This is the crude, down and dirty way of putting a volume control. The fancy Screamer has a buffer avoiding the problems you mention. The 'reamer is for people who don't worry about things like that. Bufferless TS variants are a whole class of pedals unto themselves.

The Runoffgroove article doesn't really explain much...in general this is simpler and has less of the "midhump" the original has...more bass and more highs...more "transparent" maybe... The traditional midhump (from the Rangemaster to the TS) means that as you go up the neck, the notes get livelier and more intense, supporting the common rhetorical device of rising pitch = rising intensity.

[antonis]

Hi & Welcome..

>8- This output cap and volume knob would create another HP filter I believe. I'm not sure how to calculate this, though at R=100k cutoff is 15.91Hz which means no hearable effect on sound, as it should be. Still, please tell me if I'm wrong <

You're right on your calculation but wrong on the filter action..

[ExCx]

Thanks for the warm welcome and the clear answer idy Let's reiterate:

1- So, increasing the pull-down resistor value makes the overall sound brighter and colder. Is that because that R and the input cap forming another filter there? That's something I want to avoid so I'll leave the R value as it is then. As you imply, a 10M ohms of impedance is better suited for a treble booster circuit anyway. I also forgot about the bias resistor contributing to the impedance value so thanks for pointing that out too. Now excuse my ignorance but what do you mean by "the cap may not be pulled down"? I know it would get a longer time for a cap to charge/discharge with a bigger R so is that an issue?

2- I hope somebody here got an answer about this mystery 

3- If I understand correctly, filters attached to the negative feedback will function in the opposite way it normally would. Am I right?

4- Ok I get this now.

5- You're right. That may be because I don't have a log pot though. I'll test more and get back about this later.

6- Understood. Thx.

7- I guess I'll play with the C value and observe the change.

8- I want my pedal to be suitable for any setup and prevent it from messing up with the chain. So if this substraction was only for simplicity's sake, I guess it would be wise to re-add the buffer stages. But I want to know if it has any other benefits than lesser part count. Btw, antonis, if my calculation was right then what am I wrong about here? Can you elaborate?

[iainpunk]

2- When I first assembled the circuit (on breadboard) with the exact values written on schematic, there was an unbearable high pitched noise whenever the guitar is not playing. Looking for a solution, I found that unintended oscillation is pretty common in these kind of high gain feedback circuitry. So I tried some proposed solutions like putting caps of various values between Vref & ground, adding a small resistor before output cap or adding a RC pair between polar inputs of the opamp but none of these work. Then I hopelessly examined similar screamer schematics and noticed something different in this one. Here, the filter of the first gain stage is biased into Vref instead of the ground. So, I just gave it a shot and poof, the noise is gone! Pedal is working as expected. I really don't know what I have done here. So yeah, this ground connection on (2) is Vref for me. Can anyone explain why this works and why would the other one is noisy?

there could be several reasons for this, but the most likely is bootstrapping, which is a slight gain reduction thru common mode signal (signal that appears on both input pins), especially high frequency's seem to be '''victim''' of this gain reduction, while i find that a bit illogical/counter-intuitive.
this combined with the Vref resistors reducing the gain because they pose a resistance in the feedback loop.

cheers, Iain

[idy]

The big resistor you see "outside" the input cap of effects pedals is called a "pull down" resistor. Caps have leakage (or something) and a cap can develop a charge if it has no path to ground and one of its legs has voltage on it and the other end is just "hanging." Since it has some charge, it will pop when you hit the stomp switch. The resistor allows any charge on that cap to trickle off, "pulling" it dow;  in a perfect world to zero.

[Echan42]

great post! The information you presented helped me get a tighter grasp on overdrives, I'm currently trying to solve a whistle very similar to what you mention on your third point.

The diagram you mention has me puzzled, all TS808's and tube screamer diagrams I've found bias to ground with exception of one, electosmash close look at the tubescreamer, or so I thought, but it was just the way the diagram was drawn, take a look at R4 https://www.electrosmash.com/images/tech/tube-screamer/tube-screamer-block-diagram.png

Anyway, here's my diagram https://king.seedhost.eu/echan42/Schematic.PNG I think R6 and c4 is your circle numbered 2 so I might try biasing to vin to see if it solves my unintended oscillation. Keep up the good work!

[antonis]

Biasing which part to what voltage..?? 

The only one bias point is first amp (U1a) non-inverting input (via 1M to Vref)..
Second amp (U1b) is directly biased from U1a output due to DC unity gain and the absence of coupling capacitor..

[iainpunk]

Biasing which part to what voltage..?? 

The only one bias point is first amp (U1a) non-inverting input (via 1M to Vref)..
Second amp (U1b) is directly biased from U1a output due to DC unity gain and the absence of coupling capacitor..

he doesn't mean biasing, he's talking about the termination of a NFB (Negative FeedBack-loop), either to ground or an un-buffered Vref

[antonis]

he doesn't mean biasing, he's talking about the termination of a NFB (Negative FeedBack-loop), either to ground or an un-buffered Vref

Ahaaa..
And probably he doesn't mean Vin but Vref..

I might try biasing to vin

P.S.
We better use adverbs (here, there, up, down, etc) with respective arrow pointers instead of terminology and acronyms.. 

[Echan42]

he doesn't mean biasing, he's talking about the termination of a NFB (Negative FeedBack-loop), either to ground or an un-buffered Vref

Ahaaa..
And probably he doesn't mean Vin but Vref..

I might try biasing to vin

P.S.
We better use adverbs (here, there, up, down, etc) with respective arrow pointers instead of terminology and acronyms.. 

You're correct, I'll try wording my post better. I think I solved my issue and I'll be posting a writeup on my original post as soon as I test it.

[amptramp]

Welcome to the forum!  It looks like you have the basic concepts correct but there are some things that are not obvious at first glance.

1. The input resistor in parallel with the other 1 Meg resistor beside it in your question 4 sets the input impedance.  The input impedance in the passband (set by the resistors and the capacitor between them) is 1 Meg in parallel with 1 Meg or 500 k.  If the input resistor goes to 10 Meg, the parallel combination is 909K.  If you are feeding this pedal directly from a guitar, there are several things to consider.  The guitar has a pickup with something on the order of 7k to 15 K resistance and 3 to 5 Henries of inductance going to the volume control on the pot that may be anywhere from 250K to 1 Meg.  As the resistance goes down, the inductance of the pickup reacts to form a lowpass filter that cuts the highs as the resistance goes down.  This means you have three resistors in parallel, the pot on the guitar, the input resistor and the bias resistor for the op amp beside it.  Raising the input resistor will have some effect but it may be buried in the parallel resistances.

There is a resistive noise set by the parallel input resistors.  If the 22 nF input capacitor is reduced, it decouples the input from the bias resistor of question 4 meaning the noise equivalent resistance increases because it is decoupled from the parallel resistances of the pulldown resistor of question 1 and the guitar or previous stage.

In general, it is a good idea to design with a high input impedance, a low output impedance and buffers between input and output to eliminate interaction between the controls between the buffers and other pedals in the pedal chain.

2. If grounds behaved like real grounds then life would be easy.  You have the input resistor, the lowpass filter of question 7, the output pot and the op amp power all referenced to the same ground.  If the ground has any resistance or inductance along its length, any current will modulate the ground voltage by multiplying current times ground resistance and this can get back into input stages to cause oscillation.  By taking the feedback to the same point as the bias resistor of the first stage, you at least have a common point that does not contain feedback from other stages.  We often use a star ground where all grounds are connected to a common point to avoid this.  Sometimes, it even works, dependent on the behaviour of the power input.

3. When you put a capacitor in a feedback circuit like this, in the passband the circuit will behave as if the capacitor was a dead short.  At low frequencies, it will behave as if it was open, in which case, the amplifier has unity gain.  Your gain is increased at higher frequencies.  This has an effect on the muddiness of the signal.  Imagine you have a low-frequency signal of one volt superimposed (added to) a high-frequency signal of 0.1 volt.  Without the capacitor, as the waveform goes above the clipping point, the high-frequency signal disappears because the total signal is beyond the clipping level.  This kills the highs for a large portion of the low-frequency waveform.  Reducing the amplification of the lows limits this effect.

4. Without the capacitor, the input of the op amp would vary with the resistance it sees to ground.  This capacitor should be kept large enough to avoid isolating the resistor connected to Vref as this would increase low-frequency noise.  7 Hz is a reasonable number.

5. Once the diodes start clipping, the gain goes to unity so changing the pot setting varies the point at which the waveform is clipped.  You may not notice a great deal of sound level difference.

6. The value of the Vref resistors is not that important but they do form a time constant with the capacitor in parallel with the lower resistor.  In an ideal world, the capacitor would eliminate all the noise by referring it to ground.  The resistor values can be made higher to limit current drain.

7. The original tube screamer had a tone control here but this is a simple rolloff.  The clipping process produces a lot of higher harmonics so the lowpass is needed to keep the output from being any more screechy than necessary.  The cutoff looks high and 1K is a big load for a TL072, so I would at least double this resistance.

8. These are fairly common values since any pedal designed to operate directly from a guitar has to operate with the guitar level pot set to the midrange where a 500K pot would look like 250K to ground in parallel with 250K to the pickup which has inductance and low resistance, giving 125K.  These are not bad values but I prefer all controls to be located within buffers so the response is consistent no matter what you are connected to.

If you have any more questions, I'll write another novel.