question about using transistors for clipping devices

[plexi12000]

just wondering if some are "better" or more desirable for dist. pedals than others?  if so, what characteristics make them "better"?

can i compare data sheets to help decide which ones to try?  or are they more or less "all the same"?  Thank you--

[antonis]

IMO, the first (and most significant) characteristic is that, although you wire them as a diode, they still remain transistors..

Furthermore, you have to "play" with specific types (BJT, JFET, MosFet, UJT) and/or combination with maybe some small capacitors in parallel..

I know that it's not an answer but it's quite the same like asking for different type of diodes..
(and I don't only focus on threshold voltage..)

[GibsonGM]

It's a good question, worth thinking about.   What happens when you clip a signal in any silicon device (transistor, opamp (which is mostly fet-based), diode...) is....clipping       

If you look on a scope, you'd see that the signal clips faster than with other methods of clipping....tubes tend to give a more rounded knee, indicating the creation of even-order harmonics.   The fast-clip of transistors = more harsh sounding 3rd, 5th harmonics and beyond....which is why ppl often don't like transistor (or opamp....) clipping.   And why ppl discuss using caps to try to somewhat tame these 'artifacts'.   

Even-order distortion products, those nice harmonics, make your sound fat, pleasing, and just work well with what our ears like to hear.  The odd-order harmonics are grating, fizzy, fuzzy....useful, but not 'pleasant' to many. Using the 2 types together can give some really awesome tone!   

To give some more info re. the original question...different transistors no doubt have different internal capacitances and such, so you might notice some mild differences in those harmonics ("The Fizz", so to speak).  I doubt it would be a moment of epiphany, but still...yeah, you'd end up tailoring capacitances to make the distortion sound better, undoubtedly.   

Someone like PRR would have more info re. the actual property differences of BJTs and how they affect what harmonics will be generated, on the level of a physicist   Distortion IS generated a little differently given different parameters.

Me?  I'd say "they can be a little different, but not worth doing research to see if you can TELL what they'll sound like..try a few, trim some caps, see what you get..." 

[Kipper4]

You could do worse than checking this out

http://www.muzique.com/lab/zenmos.htm

Right now my favorite is 2x 3mm yellow leds in a feedback loop of an LM833 op amp. as seen in the tonefreak Severe

Currant exprements also include the Shaka5 (on the breadboard) awaiting the mosfets.

Breadboard a screamer and swap out the differant methods of clipping seen at AMZ.
low gain Germaniums(not suitable for a fuzz) also have a differant characteristic.
Check out the OCD too
Lots of possibilitys here. Have fun.
Rich

[R.G.]

Have you read Distortion 101 at Geofex.com?

[tca]

just wondering if some are "better" or more desirable for dist. pedals than others?  if so, what characteristics make them "better"?

can i compare data sheets to help decide which ones to try?  or are they more or less "all the same"?  Thank you--

You will need to define "better" better.

From my own experience you really need to build (read breadboard) to see how each device sounds when clipping in a real implementation. The same holds for sim programs. Although learning to read datasheets and play with software is a good thing, but note that not all datasheets are well written and sims programs work nicely in linear regimes not so great when the math model starts to fail.

Cheers.

[ashcat_lt]

If you look on a scope, you'd see that the signal clips faster than with other methods of clipping....tubes tend to give a more rounded knee, indicating the creation of even-order harmonics.   The fast-clip of transistors = more harsh sounding 3rd, 5th harmonics and beyond....which is why ppl often don't like transistor (or opamp....) clipping.   And why ppl discuss using caps to try to somewhat tame these 'artifacts'.

I don't believe this to be true.  What causes odd-order harmonics in symmetry.  If the positive swings and the negative swings are both clipped at the same "distance" from 0, and the turnover at either end is the same "curviness", then you get odd harmonics.  As one side starts to clip differently from the other (less symmetry in the transfer function), more even order harmonics are introduced.

The sharpness of the corner has more to do with how many higher order harmonics come out.  Sharper corner = faster changes = more high frequency harmonics.

[teemuk]

The fast-clip of transistors = more harsh sounding 3rd, 5th harmonics and beyond....which is why ppl often don't like transistor (or opamp....) clipping.   And why ppl discuss using caps to try to somewhat tame these 'artifacts'.

Then again, Tech 21 for instance, probably made a small fortune with pedals based on opamp overdrive.


And overall I think you are making way too narrow generalisations.... transistor...? Ok.... In what circuit architecture? Differential amp is different than an emitter follower, which is different than a SRPP, which different than a common emitter amp, which is different than a cascode.... and so on. "Transistor" can mean millions of different variations and not all them "hard clip"... Heck, many clip "softer" than the generic 12AX7 common cathode amp.

[GibsonGM]

Uh oh...pick on GM time?        Try getting nice even-order harmonics out of a transistor as you force it to clip.  Now see how easy it is to do it with a 12AX7.  Generalization?  A little, but good rule of thumb. 

My explanation re. transistor CLIPPING was geared to the OP, who is just getting into DIY electronics.  Yes, of course biasing and how hard you hit a device affects its clipping characteristics - but what I said was generally true, unless you want to split some pretty fine hairs.  Some pedals do use opamp clipping, of course - and if you LIKE harsh, fuzzy odd-order stuff, cool.  You can do that with a tube, too, if you hit it hard enough.     

Yep, CE is different than CB etc, or SRPP.   But - BJTs don't 'behave' that well in clipping, no matter what configuration you are putting them in (as I am taught, anyway).   Clipping is clipping...it's even order harmonics vs. hard.   Again, unless 'the literature' is wrong?  Don't transistors enter clipping suddenly, and generate a lot of odd order hash?   *shrug*

I didn't say it sucked...and I'm not comparing FETs (which have SOME tube-like clipping chars.) and BJTs - just Si BJTs.   

**Edit for clarity**  How 'bout this: any given transistor is going to enter clipping very suddenly, which generates odd harmonics, which are known to sound 'harsh'.  SOME even harmonics are generated, but not as much as would be produced by a 'properly' biased tube (a hard clipped tube also begins to generate more odd harmonics).  This is a fact of the physical world.  They enter (and exit) clipping suddenly. 

Different transistors may sound different due to different proportions of these harmonics, and may also have different clipping thresholds.   By adjusting signal levels and surrounding circuitry, you can make many different transistors behave quite similarly to other transistors. 

Beyond this, I have no opinion - whether a 2N3904 would "sound better" than a 5088...no opinion - to me, they clip fast and hard....trying to 'lightly' clip them results in a lot of irritating hash, in my humble opinion.   That's qualitative vs. quantitative.

[ashcat_lt]

Nope, still not exactly right.  Again, whether you get odd or even harmonics has nothing to do with how quickly the curve turns over, but it's symmetry around the origin, and that's a period.  How quickly it turns over tells us something about the proportion of higher harmonics to lower, but not whether they will be even or odd. 

I do agree that a faster turnover (symmetrical or not) does make it sound "harsh" and "fizzy", but this is because there are more higher harmonics produced, and again has nothing to do with even vs odd.  I have, in fact, heard some pretty smooth fuzz come out of a clipped opamp.  It comes from controlling the frequencies going into (less high frequencies  in = less high harmonic "fizz" out) and out of (cut down the high harmonics that have been generated) the opamp.  The same can be true of diode clippers. 

The big thing with tubes - and transistors actually, I think - is that the curve is in fact different at either end.  Going into saturation the curve is not rounded the same as when it's going into cutoff, and therefor the transfer function is not symmetrical, and therefor you get some more even-order harmonics, which is supposed to be more "musical", but if there are enough of them high enough in the spectrum, it'll still sound harsh and fizzy.

Edit - course the OP didn't really specify what circuit he was talking about.  I guess I thought we were talking about using transistors in place of diodes in a standard clipping arrangement, in which case it kind of really is just a diode.  The characteristics of that clipping will be noticeably different from running a transistor gain stage out of headroom.

[karbomusic]

whether you get odd or even harmonics has nothing to do with how quickly the curve turns over, but it's symmetry around the origin, and that's a period. 

I could agree with this based on experimentation. I might also add that having an AWG with lots of control over the waveform and running that into a scope can be a bit of an eye opener on how waveforms create their resulting harmonics aka distortion. This also takes some of the mystery/mojo away. In other words, there is the waveform one wishes to create (even if they don't realize that's what they are doing), then there is how one might go about achieving that by modifying a guitar signal with electronic parts.

I have a arbiter fuzz around here somewhere that I can tweak the waveform via resistors to about 66% symmetry resulting in the following harmonic structure: 1,2,4,5,7,8,10,11

[GibsonGM]

Well, sure, I agree with most of what you say, Ashcat - you're certainly no dummy   Just that I have always been led to believe that transistors clip rather abruptly, and "harshly" (that word again, without an objective definition).  And that you can't control this transition the way we do with tubes...even asymmetrical clipping with tubes has "The Knee", where you generate "mellow" (2nd order; octave) tones.  Transistors don't show 'the knee' quite as well - less knee.   Merlin Blencowe discusses this in his book, which as been something of a 'holy grail' to me.   

And my 'education' (admittedly, not in a school of engineering) has led me to believe that the abrupt changeover from fairly linear to distortion = 3rd order with lack of 2nd order.    Don't have the math to explain why that would be...I'd like to know more about why "how quickly it turns over...not whether they will be even or odd..".   If it doesn't, then what does?    Surly not just the assymetry?  Since we could have one side clip, the other not clip, and still have even order harmonics, no?

And yes, the question is written from a perspective of not knowing that we'd need more to answer it properly!  Taken on face, it looks like "a diode is a diode is a diode"....

Sometimes I also LIKE opamp or BJT clipping, and it IS smooth, but has some other components I don't associate with, say, tubes NOT totally insanely (square wave) over driven.  When in the mood for a massively clipped square wave, something like that does the trick  

[teemuk]

The big thing with tubes - and transistors actually, I think - is that the curve is in fact different at either end.  Going into saturation the curve is not rounded the same as when it's going into cutoff, and therefor the transfer function is not symmetrical, and therefor you get some more even-order harmonics, which is supposed to be more "musical", but if there are enough of them high enough in the spectrum, it'll still sound harsh and fizzy.

My point is, we can't really discuss "transistors" or "tubes" because in practice we always have a CIRCUIT.

Yes, the transfer curves of devices vary. But transfer curve of a -tube- like 6L6 isn't exactly identical to transfer curve of another tube like 12AX7. Similar differences you find from semiconductors as well. In both cases some are "softer" than others (e.g. pentode vs. triode vs. beam tetrode).

All these curves apply to certain test conditions and basically highlight that performance is actually rather non-linear given wide range of possible operating conditions set for the device. In other words, the performance is in the end evaluated on circuit basis, and the circuit can change the -final- transfer curve dramatically (e.g. use or non-use of negative feedback, constant current loading, etc.)

It's rather pointless to discuss "tube" vs. "transistor" when in practice we have millions of circuit variations in which to utilise those devices. Many tend to perform and behave quite differently than others. There is no single "tube" or "transistor" tone.

[teemuk]

Then there's the thing that waveforms "harmonic pattern" is just one thing.

It was discovered pretty quickly after publishing of Hamm's studies that his observations on tubes vs. transistors were somewhat flawed. A transistor stage can be designed to produce some "ideal" pattern of harmonic distortion, no big deal in that, really.

Such stage as is, however, usually just sounds extremely "sterile" and unfascinating.

Why? Because it merely produces a harmonic pattern of distortion in which specific harmonics just boringly increase or decrease in relation to magnitude of overdrive.

But then it was realized that a typical "tube amp" is almost never a -single- gain stage. Usually we have more than two stages, which are capacitively or inductively coupled. When we drive such stage with an asymmetric input signal the asymmetry gradually shifts DC operating point and dynamically alters symmetry vs. asymmetry of clipping.

This produces a harmonic pattern where dominance of either even or odd harmonics "swirls" in interaction with overall overdrive and picking dynamics. Nice. Again no big deal to replicate such performance with transistors.

Then we have huge effect of pre and post distortion EQs, in varying forms of guitar tone, "fingers", amp "voicing", EQ dials, speaker+cab, etc. They all have major impact on how distortion sounds. IMO, they have bigger impact than how the signal even clips or distorts assuming magnitude of such is about equal.

So we get to the point where it's not just about a single "ideal" harmonic pattern of how the circuit should distort, it's about a widely interactive and dynamic pattern, which must be tweaked according to preferences prevailing in certain applications. Then, it's also about the entire overall EQ'ing of the amp, both pre and post virtually every distorting stage. This "voicing" is likewise tweaked according to preferences prevailing in certain applications.

[GibsonGM]

Great description, Teemuk....EQ rules   We've all heard purely transistor pedals - some DSP - that sound ungodly good, so....

Who is this "Hamm", and is there a good resource you can recommend about harmonic content vs. 'quality', such as you're alluding to?  I'm interested in learning more about the interaction of nonlinear devices and what we perceive as "good", if such a resource is out there.  And of device behaviors, to better understand this. 

Short of knowing that I could fire up a tube amp and a transistor amp, grab the scope, and prove some of this to myself (which I can't right now since packing to move), I can't think of any other sources for *good* information about this topic.

[R.G.]

Russell O. Hamm's AES paper at the 1972 convention: "Tubes Versus Transistors: Is There an Audible Difference?"

http://milbert.com/Files/articles/TvsT/tstxt.pdf

We're getting off into the weeds with the discussion, guys. The OP wanted to know whether one transistor was better as a clipper than another. This is much like asking can one horse run faster than another - the answer is obviously, yes. He also wants to know why. The devil is in the details.

We should probably take the tubes vs transistors versus harmonic structure versus time variance versus pre and post EQ to a different thread. We're confusing the OP and the other mass of readers who don't post, there continually being about 100-200 guests here compared to 10-20 members.

As an aside, Hamm proved his point. Is there an audible difference? In some circumstances, clearly yes. Whether this is like a blinking red sign or not is part of the neverending audio discussionblather.