Tubescreamer clean and clip Project

[preciousmolina666]

To the OP - it's still hard to say what exactly you're asking.  The part of the signal that exceeds the diode drop is not really filtered, but if you take out the diodes, that signal will be.  Whether that's what you want is completely up to you.  I think it's more likely they will sound more similar if the "clean" one is filtered, but.

Now if in understand correctly tubescreamer drive section is parralel of original signal and clipped signal

Tubescreamer drive section:
1st  input -> unity gain(unfiltered) -> output
2nd input -> opamp clipped -> low pass ->high pass -> diode to ground -> output
or
2nd input -> opamp clipped -> diode to ground -> low pass ->high pass -> output

[EBK]

To the OP - it's still hard to say what exactly you're asking.  The part of the signal that exceeds the diode drop is not really filtered, but if you take out the diodes, that signal will be.  Whether that's what you want is completely up to you.  I think it's more likely they will sound more similar if the "clean" one is filtered, but.

Now if in understand correctly tubescreamer drive section is parralel of original signal and clipped signal

Tubescreamer drive section:
1st  input -> unity gain(unfiltered) -> output
2nd input -> opamp clipped -> low pass ->high pass -> diode to ground -> output
or
2nd input -> opamp clipped -> diode to ground -> low pass ->high pass -> output

Getting closer, I think.  I believe I understand what you have been trying to say.  I'll try to explain a bit more later today.

[ashcat_lt]

Now if in understand correctly tubescreamer drive section is parralel of original signal and clipped signal

Tubescreamer drive section:
1st  input -> unity gain(unfiltered) -> output
2nd input -> opamp clipped -> low pass ->high pass -> diode to ground -> output
or
2nd input -> opamp clipped -> diode to ground -> low pass ->high pass -> output

Well that's not exactly how it works, but it has about the same effect.  The drive section is completely in series with the signal.  There isn't actually a parallel "dry mix" happening in the circuit itself.  The interactions between the opamp, diodes, caps and resistors makes it act like there is, but...

[EBK]

Think of pin 3 of that op amp as v_in.  That gets copied to pin 2 via op amp "magic".

Now, consider the 4.7k resistor and 0.047uF cap.  These establish the current that will get pulled through the combination of parts between pin 2 and the output of the op amp.  This current can be thought of a high passed representation of v_in.  Ignoring the diodes for now, you have in that feedback loop a resistor in parallel with a capacitor.  Together, they act like a resistor whose value decreases as frequency goes up.  That current I mentioned earlier flows through that combination, which acts as a low pass filter, to produce a voltage that is effectively a bandpassed version of v_in.  This bandpassed voltage is referenced to the voltage at pin 2, which is an unfiltered copy of v_in, so to arrive at the final output voltage you get the bandpassed voltage plus the unfiltered copy of v_in. 

The things I glossed over (other than ignoring the diodes): there is voltage gain associated with the component values, and the corner frequency of the LP portion varies with the drive pot setting.

I'll try to add more to this explanation later, but to sum up a partial answer to your question, the unity gain clean signal is not affected by the filters.  Only the amplified, and potentially clipped portion is.

[ashcat_lt]

I'll try to add more to this explanation later, but to sum up a partial answer to your question, the unity gain clean signal is not affected by the filters.  Only the amplified, and potentially clipped portion is.

I feel like this is going backwards and not exactly right and kind of not helping.  I'm sure you mean well, but it's just not getting there.  Among other things, there is no "clean signal".  The portion that of the signal which exceeds the diode drop is mostly unfiltered, and it works out about the same as mixing the clean input with a filtered-and-hard-clipped copy, but that ain't really how it works.

Also, both filters are shelving types (as I said above) in the BizzaroWorld of a negative feedback loop such that the opamp output is amplified by the amount that the filter is attenuating.

I never actually use 1 + R1/R2, but rather figure the divider ratio - R2/(R1+R2) - and then invert.  It's important to remember that those Rs should really be Zs, and their values are dependent both on frequency and (with the diodes) level.

Anyway, I still don't know what the OP actually wants!  I thought the idea was to make two pedals - one a normal TS and the other the same thing without the diodes.  I'd figure you'd want to have all the filtering exactly as is for both.  It would then be a TS and essentially a high mid/treble booster and they would tend to sound more like each other than if the "clean" one was unfiltered.

If, on the other hand, we're trying to "fake" a TS by mixing the clean signal with a diode-to-ground version of a TS, you'd want to move the diodes out of the feedback and use a simple flat unity gain buffer for the clean signal.

Note that the big cap to ground is necessary unless that leg actually goes to a Vref, so it can be tough to get actual flat gain out of this type of circuit.  The smaller cap is also pretty important, but is easier to size small enough to not touch the audio spectrum.

[EBK]

I'll try to add more to this explanation later, but to sum up a partial answer to your question, the unity gain clean signal is not affected by the filters.  Only the amplified, and potentially clipped portion is.

I feel like this is going backwards and not exactly right and kind of not helping.  I'm sure you mean well, but it's just not getting there.  Among other things, there is no "clean signal".  The portion that of the signal which exceeds the diode drop is mostly unfiltered, and it works out about the same as mixing the clean input with a filtered-and-hard-clipped copy, but that ain't really how it works.

Also, both filters are shelving types (as I said above) in the BizzaroWorld of a negative feedback loop such that the opamp output is amplified by the amount that the filter is attenuating.

I agree with most of what you said.  Part of the reason I said I would explain more later was precisely because the strained analogies are far from perfect....

I never actually use 1 + R1/R2, but rather figure the divider ratio - R2/(R1+R2) - and then invert.

If you want to write R2/R2 (or Z2/Z2) instead of 1, go right ahead. 

[preciousmolina666]

I'll try to add more to this explanation later, but to sum up a partial answer to your question, the unity gain clean signal is not affected by the filters.  Only the amplified, and potentially clipped portion is.

Thanks

Anyway, I still don't know what the OP actually wants!  I thought the idea was to make two pedals - one a normal TS and the other the same thing without the diodes.  I'd figure you'd want to have all the filtering exactly as is for both.  It would then be a TS and essentially a high mid/treble booster and they would tend to sound more like each other than if the "clean" one was unfiltered.
If, on the other hand, we're trying to "fake" a TS by mixing the clean signal with a diode-to-ground version of a TS, you'd want to move the diodes out of the feedback and use a simple flat unity gain buffer for the clean signal.

i going to emulate "TS drive section" but seperate control as close as possible
Now i want to know if the signal is filtered before it clipped by diode/opamp

[EBK]

At the risk of annoying ashcat_lt with this simplified characterization, the portion of the signal that is affected by the filtering/clipping network gets high pass filtered before it gets clipped.

The important thing to keep in mind is that we've been referring to a unity gain unfiltered portion of the input that is mathematically separable, but not physically separable within that feedback loop.  In this regard, ashcat_lt is quite right that this "clean signal" does not exist, per se.

Add a separate op amp to have a separate clean path that you can process separately. I think that has been your plan all along.

Here is what I am assuming you are asking for:
one complete tube screamer circuit that has two parallel copies of the drive op amp subcircuit, one drive subcircuit being exactly like the original, the other drive subcircuit being similar, but without the clipping diodes and potentially with different filtering.  I believe the nature of your question referred somewhat to whether some of the other components were only there for the benefit of shaping the clipping itself or, conversely, were only there to shape the sub-clipping portion.

It's important to remember that those Rs should really be Zs, and their values are dependent both on frequency and (with the diodes) level.

very important, indeed.  This reminds me that I also glossed over the issue of phase....  I was quite reluctant to bring up Laplace domain analysis or complex impedance for fear of creating additional confusion (believe it or not, I was also actually a bit indecisive about whether it would be potentially clearer to call sqrt(-1) i or j as part of a lengthier explanation  ).

[ashcat_lt]

If you want to write R2/R2 (or Z2/Z2) instead of 1, go right ahead. 

There's no need.  Figure it, then flip it.     I'm not at all saying my way is right, but I think a lot of people see that equation for opamp gain and think it's some new special thing, and never quite realize that it's really just the standard divider equation that we all know and love (or should, because "everything useful is a voltage divider").  I think it's important to keep this in mind at all times in order to understand anything that's happening in the negative feedback loop. 

The opamp doesn't use magic, and it's not all that complicated really.  Whatever you do between the output and the inverting input has to be undone at the output so that what actually reaches that input is exactly the same as the non-inverting.

I'm pretty sure we've answered the OP's answer a couple times no matter what they actually want, but I'll put it out here one more time:

A) If you want a "clean boost that sounds like a tubescreamer" just pull out the diodes and go.  It wants the same filtering.

B) If you want to emulate a tubescreamer by mixing a clean signal with a hard clipped signal, then the clipped side should be filtered and the clean side should be flat.

I'd love it if you'd tell us whether A or B is what you're trying to achieve, but I'm not that worried about it. 

[EBK]

The opamp doesn't use magic, and it's not all that complicated really.  Whatever you do between the output and the inverting input has to be undone at the output so that what actually reaches that input is exactly the same as the non-inverting.

Hey! You're not supposed to reveal a magician's secrets! 

I'm pretty sure we've answered the OP's answer a couple times no matter what they actually want....

Perhaps the only thing left is to write the answer as a haiku to finish where we started. 

Modded tube screamer
Lift diodes or add more parts
What is it you want?

[preciousmolina666]

the portion of the signal that is affected by the filtering/clipping network gets high pass filtered before it gets clipped.

B) If you want to emulate a tubescreamer by mixing a clean signal with a hard clipped signal, then the clipped side should be filtered and the clean side should be flat.

my fault id confused you

https://www.dropbox.com/s/w7hgyejawaa7hzm/tubescreamer%20haiko.jpg?dl=0

[ashcat_lt]

B) If you want to emulate a tubescreamer by mixing a clean signal with a hard clipped signal, then the clipped side should be filtered and the clean side should be flat.

my fault id confused you

https://www.dropbox.com/s/w7hgyejawaa7hzm/tubescreamer%20haiko.jpg?dl=0

Ooh good yes this helps!  The image doesn't actually show up in the msg on my phone, but I clicked through and I see what you're going for. 

A couple of things:
1)  It looks like you've illustrated a "mix pot" at the output.  That sort of arrangement is actually more like a crossfade.  It turns one side down as the other goes up.  In a TS that doesn't happen.  You want a static, even mix.  The proportion of one side to the other should only depend on gain in the clipped leg.

B)  In a TS, any opamp clipping that happens is on the "mix" of the two.  In fact, the "clipped side" can't hit the rails on its own.  The way you've drawn it, the opamp clipping will be mostly lost in the diode clipping anyway.  Ideally you'd give both sides plenty of headroom so that neither one clips any opamps, then mix them, then add your opamp clipping section.  Unfortunately, I think the gain section of the clipped side would want hundreds of volts for rails.

So yeah, your unity side is just a clean flat buffer.  Your dirty side is basically the drive section of the TS with the diodes moved out of the feedback loop and place as a to-ground clipper after.  Then a 1:1 mixer which will probably drop about half the voltage from both sides so you have a final opamp with enough gain to compensate for that which will give you the opamp clipping.  If you can run everything but that last off of 18V, would be better.

I tend to think that the real "magic" - the part that's going to be tweakiest and have the biggest impact on the authenticity of the result - is going to be the resistances around the diodes-to-ground.  There needs to be some series resistance after the opamp going into them, and there will be the mix resistors as a load parallel, and that will changes the "curve".  IDK for sure how to pick appropriate values there.  Part of me thinks the series one should be the same as in the ground leg of the feedback loop, but I really can't say.

[EBK]

I think you should consider placing the LP after the clipping (or add another separate LP after the clipping).