TubeScreamer: Clipping Questions, diodes, etc.

[N Orlo]

Hi, I recently modded my TS-9 to 808 specs. I have become interested in the difference between aysemetrical and symeterical clipping. The way I understand it is the aysemetrical is a "smoother" distortion compared to symetrical, is that correct? I have a schematic from tonepad.com that reccomends the 1N914 diodes, are those the most ideal? Any thoughts or suggestions are appreciated.

Thanks,
Nick

[Kornell]

With all due respect. 

[QSQCaito]

I certainly believe that Tube Screamer, and diode clipping is one of the most talked topics. As my mate said, and for if you didn't understand:



No mean to offend.

Here are some discussions:
http://www.diystompboxes.com/smfforum/index.php?topic=2595.0
Diodes and how they sound:
http://www.diystompboxes.com/smfforum/index.php?topic=2665.0

[tonefreak]

IMO, the experimentation of clipping diodes has reached a point of exhaustion.  There are combinations that certainly provide a difference in tone, but many of the combinations start sounding the same.  The TS808 is a simple circuit, but there are a lot of other things about it that will make a more drastic difference in the tone than standard clipping diode variation experiment.

[MartyMart]

I do agree that this has come up almost weekly for the past three years ! BUT here's a list
of changes that are WELL worth doing - since I've now modified around 40 of these !!

Diodes : remove one of the 914's and replace with one 1N4001 = assym clipping and more
harmonicly rich .
IC : Whatever dual IC "floats your boat" but good 'ole 4558 or 4559 are just fine.
Gain : reduce the 4k7 off FB loop pin 2 to a 2k2 and increase the 0.047uf here to 0.1uf for less bass loss.
You can also change the 500k pot to a 1M but it's not worth it IMO.
Reduce the 51k to drive pot's wiper to a 20k/22k to keep the lower gain range value stable.
Tone : Reduce the 0.22uf tantalum between pins 1 & 5 to ground, to a 0.1uf or even 0.068uf, this will
brighten things a bit and make the tone pot more useful too.
Make this a poly cap and also the one from the tone pot's wiper - another 0.22uf tantalum as stock.
Change the two electro 'NP" 1uf caps to 1uf "poly" caps - a Keeley "HiFi" trick but it's not absolutely required.
Enjoy !
MM.

[Mark Hammer]

Hi, I recently modded my TS-9 to 808 specs. I have become interested in the difference between aysemetrical and symeterical clipping. The way I understand it is the aysemetrical asymmetrical s a "smoother" distortion compared to symmetrical, is that correct?

Common misconception.  If I cranked the gain on ANY pedal set up for "asymmetrical" clipping, and fed that pedal with a heavily boosted signal, you'd have a pure square wave for each half cycle (pos and neg peaks), except that one half cycle would be a higher amplitude than the other.  Not exactly "smoother", just a different tone.

If the gain is substantially reduced and a "normal" signal is fed in, then - depending on how hot the signal is, relative to the clipping threshold - you'll get less clipping for one half-cycle than for the other.  The presence of different amounts of clipping for positive and negative half-cycles is what makes it "asymmetrical".  The description of "smoother" is only warranted when one half cycle is just hot enough to be clipped, while the other is not.  As you might expect, that doesn't happen at EVERY setting, or with EVERY pick stroke on EVERY string.

It is also the case that what people DO to produce the asymmetry also happens to increase the clipping threshold which will naturally decrease the amount of clipping obtainable overall.  That's a bit like turning down your guitar volume and saying it makes the pedal "smoother sounding".  The pedal is every bit as capable of savage squaring of the signal as before.  It's just that there is a mismatch created between what the pedal circuit needs to receive to provide that squaring, and what you're feeding it.

[Hanglow]

I have a rotary switch on mine that does a number of combinations (I did it without really knowing what was what at the time) but it worked well

1: Normal
2: Asymmetrical with extra si diode on one side
3: Asymmetrical with one si diode on one side, Ge on another
4: LED's - symmetrical
5: LED and Si - asymmetrical
6: none (i think)

And I second the search feature suggestion, I put that selector switch on because it has been discussed at length here and there was a diagram for the switch that someone posted a while ago.

[johngreene]

Hi, I recently modded my TS-9 to 808 specs. I have become interested in the difference between aysemetrical and symeterical clipping. The way I understand it is the aysemetrical asymmetrical s a "smoother" distortion compared to symmetrical, is that correct?

Common misconception.  If I cranked the gain on ANY pedal set up for "asymmetrical" clipping, and fed that pedal with a heavily boosted signal, you'd have a pure square wave for each half cycle (pos and neg peaks), except that one half cycle would be a higher amplitude than the other.  Not exactly "smoother", just a different tone.

Depends on what your definition of 'smoother' is.

This is not true with a TS-808, diodes in the feedback loop type pedal. And because the opamp is not ideal, it cannot produce a perfect square wave. It will have a finite slew rate. So, given that you have some 'slope' to the 'rise and fall' of the waveform and a 'rounded' top and bottom, when you use assymetrical 'clipping' you introduce more even order harmonics as the waveform starts looking like even order harmonic distortion. More even harmonics = more 'smoother' sound.

--john

[stm]

I have a rotary switch on mine that does a number of combinations (I did it without really knowing what was what at the time) but it worked well

1: Normal
2: Asymmetrical with extra si diode on one side
3: Asymmetrical with one si diode on one side, Ge on another
4: LED's - symmetrical
5: LED and Si - asymmetrical
6: none (i think)

And I second the search feature suggestion, I put that selector switch on because it has been discussed at length here and there was a diagram for the switch that someone posted a while ago.

This illustrates why the "which diode combination is best for clipping?" questions keep appearing: apparently there is plenty of people that have experimented but no definitive conclusion as to which actually sounds best or which resembles better a particular target sound.

[dachshund]

I was looking at this the other day, nothing new, just a good consolidated article.
http://www.muzique.com/lab/tclip.htm

[Mark Hammer]

Depends on what your definition of 'smoother' is.

This is not true with a TS-808, diodes in the feedback loop type pedal. And because the opamp is not ideal, it cannot produce a perfect square wave. It will have a finite slew rate. So, given that you have some 'slope' to the 'rise and fall' of the waveform and a 'rounded' top and bottom, when you use assymetrical 'clipping' you introduce more even order harmonics as the waveform starts looking like even order harmonic distortion. More even harmonics = more 'smoother' sound.

--john

I'm going to respectfully contest some of this by noting that the guitar is not a signal generator with constant level output.  The end result is that as the signal amplitude rapidly declines following plucking/picking of the string, the application of two different clipping thresholds - one to each half-cycle - means that we are essentially looking at a scenario where the likelihood of any wave-peak being clipping (perhaps more accurately, the proportion of all wave-peaks on that side of the midpoint) changes over time.  With the threshold being significantly higher for one half cycle than for the other (and presumably the builder has arranged this), clipping will disappear pronto for one half cycle and linger longer for the other.  I suppose the result of that can be described in terms of different orders of harmonics, but the basis of it is that one half cycle is clipped noticeably more than the other. 

More importantly, one is not adding more harmonics of a certain type.  Rather harmonics of some types are showing up less often and, even when they do, disappearing from the output signal much faster than others.  Again, remember that if the signal is low enough amplitude, relative to the clipping threshold dictated by the diodes, all the engineered asymmetry in the world will go entirely unheard because neither half cycle is being clipped at any point in the lifespan of the note.

What has certainly never been explored here by any of us, is the point at which measurable asymmetry becomes inaudible, even by golden ears.  Imagine the scenario where both half cycles have the same clipping threshold; say 2v p-to-p, or 1v per side.  Slowly, but surely, the clipping threshold of one half cycle gets reduced, we'll say by 50mv steps.  So, 1v vs 950mv, 1v vs 900mv, 1v vs 850mv, etc etc.  What is the smallest amount of asymmetry needed to actually hear a timbral difference?

Let's complicate it even further.  Once we've established the minimum threshold discrepancy needed to hear "asymmetry" (and this assumes we're doing everything needed to maintain amplitude constancy post-clip) using a steady state signal from a tone generator, let's switch to an impulse-type signal such as that of a plucked guitar string, and ask what the minimum discrepancy in clipping threshold might be for someone to hear a timbral difference.

Of course that's sort of a trick question.  Why?  Because if one half-cycle is clipping and the other isn't AT ALL, that ought to be easy to notice.  And if one side is clipping a bit more/more-often than the other, that should be harder to notice.  So, how much asymmetry you need to hear asymmetry will depend on how hard the signal is being clipped overall.  And clearly the sorts of controls you'd need to build into such a testing process are generally not going to be available in real-world pedals.

I suppose all of this is rather academic, innit?  If a person likes the way a pedal sounds by adding or subtracting or changing a diode here or there, then that's all that really matters.  And, as you note, what may apply to the diodes-to-ground context doesn't translate flawlessly to the diodes-in-a-feedback-loop context.  My concern is simply that novices not set themselves up for having unrealistic expectations by either describing or explaining things to themselves inaccurately.

[johngreene]

While it is true a guitar signal is not a signal generator with a constant level output, it is a composition of mostly harmonically related parts. The same 'ideas' apply that apply to a single tone, they just vary with time. Using a single tone for the initial analysis/understanding of a circuit's function/behavior is a completely valid approach IMHO. The transfer characteristic of the circuit is the same for both.

For sake of argument I'll state the following:

The lowest gain of a diode feedback circuit can be less than 10 and still achieve, symmetrical, clipping for standard signal diodes. I don't think anyone that is considering asymmetry is thinking about running the 'drive' all the way at the minimum. So, I'll assume that for all practical purposes, the waveform will be clipped 'all the time'.

Actually you ARE adding harmonics of a certain type. The nature of 'limiting' is a 'harmonic producer'. This can be very easily demonstrated with a single tone but we won't go there.

Finally, I have done a lot (a whole lot) of experimenting with asymmetrical clipping. I find I don't like it much. By increasing the even order harmonics through asymmetry, the audible result is a sound that just sounds 'cleaner'. I suspect this is because you are adding the evens at the expense of 'losing' the odds. Since guitar strings already contain a significant amount of even harmonics, you aren't adding much until you add a lot. It does take a lot of asymmetry for me to hear what I would call a noticeable difference though. If you are listening to the pedal alone, in a quiet room, etc. then it is possible to notice smaller offsets, but I also think they sound 'less' and not 'more'. Now, run this even harmonic rich signal into another symmetrical clipper...... different story.

I have also noticed that everyone claims the asymmetry makes for a more 'tube-like' sound but this would only hold true for single ended (and mostly) preamp tubes. The output, push-pull, clip very symmetrically from all the tests I have done. Now if I could just invent a pedal that would result in a hard, symmetrically clipped waveform when it actually hit the speakers of the amplifier, I would have something....

--john

[tonefreak]

This illustrates why the "which diode combination is best for clipping?" questions keep appearing: apparently there is plenty of people that have experimented but no definitive conclusion as to which actually sounds best or which resembles better a particular target sound.

That is because tone is subjective.  I've tried ICs with TS808 circuits that a friend would swear is "the" chip.  I throw it in and end up putting back my JRC4558D... I just liked it better.

Same with clipping... IMO, there is no "best" configuration.  I think the point is to experiment what has been presented here in DIY and make that detemination for yourself.  You are the ultimate judge of your sound.

[WGTP]

I was just messing with this last night.  Here are some suggestions to make it really a Screamer.  For diodes use a single 2N7000 mosfet.  This will have a clipping threshold of .6v in one direction and 2v. in the other.  Change the 4.7K to 1K and the .047uf to .1uf or .2uf for more bass.  OR if you want to "Ratify" your Screamer in addition to the 1K/.1uf "leg" add a 10K/.2uf "leg" for some serious Rat Screaming.   

[N Orlo]

Ok, thanks for the information... actually I did use the search but all the topics I found were about actually performing the mod. I was more interested in how it changed the timbre of the sound. Interesting info nonetheless, I assumed it would make it smoother b/c on a website I found talking about sound waves and such it made a reference to aysmmetrical clipping as being more tube-like, hence smoother/warmer/whatever (when compared to solid state). And like you said, there are pages of information about this topic, more than enough so that trying to find what i was looking for was enough to make my head spin(or maybe im just blind  ) so i asked.

Thanks,
Nick