Clipping and Transistors for overdrive

[geo]

Okay, so the trend seems to be that folks like slower-clipping transistors like Ge fellows to the Si guys 'cause they clip slower and it sounds more tube-like, is this right? Is there any demand for fast-switching transistors in guitar effects pedals? (e.g. super-hard clipping)

[newey]

geo-

Hello and Welcome to G-Nutz2! Although I see that asmith has already welcomed you in the other thread.

Is there any demand for fast-switching transistors in guitar effects pedals? (e.g. super-hard clipping)

Why, are you sitting on a shipping container full of 'em or something? ;D ;D

Seriously, people have tried just about everything at one time or another. I seem to recall we had some discussion of Si vs Ge diodes a while back. I'll have to look in the archives a bit.

[geo]

I'm working in a graphene lab and I might have an idea for epitaxially attaching graphene to hexagonal-Boron Nitride, which should theoretically make it a gapped semiconductor, so you could make FETs out of it. Since it's two-dimensional with ridiculously high mobilities, I'd expect it to clip substantially faster than anything that exists today.

Based on what I know so far, that makes me think it'd sound terrible. But if there's some use for this kind of thing in audio electronics, this is your device.

[flateric]

Isn't there a problem with Ge transistors being very inconsistent so you have to check through a load first and tweak bias? Newer Si-based ones will give you predictable, consistent results. I've not heard much tinal variation in all the opverdrive boxes I've put together, not compared to swapping out the diodes and trying different combinations. For me that gives huge difference in the qualities of the clipping, but not the transistors.

[geo]

Ge transistors' performance varies with temperature, and they're far more inconsistent to begin with. They do clip softer and slower though, which gives a preferable tone if you get a good transistors.

Still interested in what use there is for super-fast clipping transistors in audio though. (Mostly for fuzz/drive/dist)

[ijustwannastrat]

I say give her hell..... If you make something amazing, can I name it?



(It's going to be called KIIMH Drive, BTW)

[thetragichero]

sounds good to me
located in the us or...?

[geo]

KIIMH Drive? Huh? I was just gonna call it a hex-drive 'cause graphene's a 2D hexagonal lattice of carbon and you're stacking it on hexagonal Boron-Nitride.

Is what located in the US? Physics lab I'm in? Yeah, we're in NJ. Still have to see if this procedure works for stacking the graphene and boron-nitride right, nobody's managed yet. Even if we do, it's gonna make for one crappy sounding transistor unless you can find a circuit where faster-clipping would sound better.

Just bear in mind having this transistor to begin with means first verifying a theoretical model of physics, so it's far from a given that I'm gonna be able to make this. (And even if it gets made it's gonna be a homebrew type thing where the exact properties are gonna vary a bunch by device since there's gonna be some variation from the theoretical model 'cause the model doesn't account for lattice mismatch at all and there're sure to be other issues.)

Find me a circuit where a fast-clipping transistor won't sound like garbage and I'll see if I can build a stompbox and upload a recording if and when I manage to manufacture the device.

[jcgss77]

Nov 22, 2011 23:59:00 GMT -5 geo said:

KIIMH Drive? Huh? I was just gonna call it a hex-drive 'cause graphene's a 2D hexagonal lattice of carbon and you're stacking it on hexagonal Boron-Nitride.

Hey geo, in order to understand what that is, you need to do a forum search of "kyle is in my head". You will then understand this name. And you will laugh.

Nov 22, 2011 23:59:00 GMT -5 geo said:

Is there any demand for fast-switching transistors in guitar effects pedals? (e.g. super-hard clipping)

Yes. I demand it!!! When do we start on a circuit?

I say, if you are going to make a new type of transistor, do it! Nothing like having something that no one else has, that doesn't sound like anything else out there! I like it, very Nutz!

[geo]

Allow me to answer your enthusiasm with a platitude:

[cynical1]

I'll see your platitude...and raise you...



Remember what they said about air medals...

Happy Trails

Cynical One

[ashcatlt]

Honestly I think you'll find that other factors have more to do with the sound of this box than the "speed" with which the transistor clips. Most signals don't really sit in that "knee" region for very long.

Pre- and post-filtering and biasing to set the clipping symmetry will be where you get the most variation. Even tubes sound like crap without filtering.

That's not to say you shouldn't follow through with your experiment. It might turn out to be the next big thing.

[jcgss77]

Since jFets are going the way of the dodo, I think you should definitely fry together some of these. If you have the ability, the patience, and the Nutziness do it. jfets are my personal favorite transistor, and I have a bunch of j201's and I am buying some n5454 soon hopefully. They are great for hi-gain circuits.

[geo]

Alright, I manage to fabricate any I'll be sure to hit you guys up with the specs and have you hash out a couple circuits for me to stick 'em in.

[JohnH]

Nov 22, 2011 6:58:09 GMT -5 geo said:

I'm working in a graphene lab and I might have an idea for epitaxially attaching graphene to hexagonal-Boron Nitride, which should theoretically make it a gapped semiconductor, so you could make FETs out of it. Since it's two-dimensional with ridiculously high mobilities, I'd expect it to clip substantially faster than anything that exists today.

Based on what I know so far, that makes me think it'd sound terrible. But if there's some use for this kind of thing in audio electronics, this is your device.

Great! I'm right behind you, and I understood every word of that up until the word 'epitaxially'.
(I looked up Graphene:
Graphene is most easily visualized as an atomic-scale chicken wire made of carbon atoms and their bonds. The crystalline or "flake" form of graphite consists of many graphene sheets stacked together. wikipedea
And I know what a flake is, I used to live in California.)

But basically, is this gonna be like a JFEt in that it is controlled by a gate voltage? If so, if you can plug your little monster into a circuit where there is a resistor on the drain and you sweep the gate voltage from - something to + something, and plot a graph of the resulting drain voltage, that may suggest what it might be good for.

John

[geo]

John:

Epitaxially means the crystal lattices are aligned. (So there should be a certain stacking between h-BN and graphene to induce an asymmetry in the graphene lattice and thereby inducing a band gap.) Epitaxial usually refers to growth of crystals on each other (like how they grow silicon chips) so it would be more accurate if I had said aligned.

This should be a FET that is controlled by a gate voltage. If and when it's fabricated you can bet your bottom dollar I'll have plots to characterize it and the actual measured band gap.

[jcgss77]

So are you actually making these? How do you do it?

[geo]

You exfoliate graphene from highly-oriented pyrolitic graphite and place it on a wafer of silicon with a thin insulating layer of silicon dioxide. You print gold leads around the graphene sample and wire-bond to them. You can pass current through these leads and the silicon works as a gate. As far as mounting the hexagonal Boron-Nitride on the graphene you just have to do that about a billion times and hope it lines up properly or you come up with a more clever way to make sure the lattices are aligned. After that you'll wanna do some kind of packaging or sealing (possibly just sandwich the whole thing with h-BN without aligning it to the lattice) so the sample doesn't degrade over time, or vacuum seal it in a rubber cap somehow.

[geo]

So we're not looking into aligning it with h-BN to open up a band gap at present, but if anyone thinks they can find a use for it (resistance is still changed with gate voltage), I've got:
DOS v Carrier Density
Conductance v Gate Voltage
dI/dV v Sample Bias
Conductance v (Gate Voltage - Drain Voltage)

Is there any way to upload pictures in these posts?

[andy]

I use photobucket- once a photo is uploaded you can click on the link, and paste it here. If you use the one beginning it will post the picture itself in your reply.

I'm afraid I have no idea what the rest of your post just there refers to. However, the reason people like a slower clip is the ability to inhabit the areas between clean and fully clipped, particularly the tipping point where it's sort of clean, sort of dirty. That is the typically dynamic blues/classic rock area of things.

A quick clipping device would presumably do away with that 'hinterland' and jump from pristine clean to mega dirty at a given point. Perhaps a 'dynamic channel switch' pedal is the way to think. One which means that if you play softer you have a clean clear sound, and if you play harder, full on metal distortion. The key would be to make the clipping point within the dynamic range of the guitar- perhaps controls for output volume and tone, and another for sensitivity to fine tune it to your guitar or desired clipping point. It might be possible to set it so that a soft strum is clean, and a hard one is full on dirt, or perhaps the volume dial on the guitar would be clean up till '9', then filth when tweaked round to '10'.

The other consideration might be the tonal characteristics of this new component. Let's just assume it sounds like the most aggressive screaming metal distortion anyone has ever heard. It may be that the way it clips is neither here nor there, but used purely in it's clipped state it might be a new trend in heavy guitar (which would be a welcome change from the incessant Mesa through Vintage 30's thing we've had for so long!)... or whatever it happens to sound like, but I'm getting ahead of myself there. The 'dynamic response channel switch' effect is the main point here.

[sumgai]

geo,

I understood what you said.

---

andy;

All this talk about hard and/or soft clipping, and how slow or fast a device is, that's all a red herring. Not to berate you, but it comes from boutique pedal makers that have no formal/classical education in electrical (or electronics) theory. Sad to say, but there it is. And of course, once crap gets leaked out to the web, it self-perpetuates until it ends up on Snopes, except for technical stuff and Marketing Feldergarb.

But that's not why we're here today. Today's lesson is gonna be some homework for you. I want you to investigate two things: "slew rate" and "saturation". When you've absorbed their meaning, come back and report to us what effect those two parameters have on clipping.

Please.

HTH





sumgai

[andy]

Thanks Sumagai!

"Better to keep your mouth closed and be thought a fool than to open it and remove all doubt"

Despite having displayed the profundity of my misunderstanding of the function of transistors (which I will read up on, though it's been a while since I handed any homework in on time!) I think the main misunderstanding was with speed. I was thinking of how sharply the end product would transition from a clean to a strongly distorted signal with an increase in input (hard vs. soft knee). From a cursory glance at a page on slew rate i tseems we are talking about a more literal version of reaction time, and something much further above my head at this point than I might otherwise admit!

Still, I'm glad you pulled me up straight away- as you said, the internet is full of secondhand nonsense and I may just have been giving it a helping hand there...

[jcgss77]

Cool, we get to try these babies out! And a collaborative effort on a guitarnutz2 original design! Or a couple...

[sumgai]

andy,

Well, I think a better term for what you describe would be "the transition time" between clean and dirty (be it somewhat dirty or full-on hella dirty). The speed of that transition is under user control, all other things being done up right proper. (IOW, under good design practices.)

Which then says, if the user has more-or-less full control, then why do we worry so much about what type of transistor (or IC) is used to accomplish the mission? Well, there are several answers, but it boils down to this: Ge devices tend to break into the distortion region more gently, compared to Si devices. Moreover, they tend to drop back out of that region with less "urging" on the part of the user, or the rest of the circuit. I'm saying here that Ge stuff can be easier to control, provided that all other elements are up to snuff. Plus, it's usually cheaper, due to simpler circuitry, but in this day and age, sometimes just using a buzzword (Look! It's Germanium! Just like Jimi used!!) is enough to drive the market price well out of reasonableness.

That doesn't mean that Silicon can't do the job, only that to accomplish the mission, we need to spend more time on the design, and pay closer attention to the component values (tolerance ranges need to be closer, etc.). Additionally, Si may need a higher power supply voltage to get the job done as well as the Ge unit - that's a possible deal-breaker for some folks. (Expense, space available, etc.)

Sadly, Ge devices are rarely made these days that have any kind of quality control... sort of like 50 years ago, only higher priced.   Hence, a designer can come up with a humdinger of a plan, but building multiple copies that all comport with that design is very cost-ineffective - it's labor intensive, to say the least. (The official name of the game is test, test, and test some more. Rinse and repeat as necessary. Sigh.) All of which goes to explain why we usually find Silicon in most effects pedals or amplifiers, and why solid-state device designers like geo are looking for alternatives that might imitate the exhibited properties of Germanium, yet have much greater quality control, and hopefully be cost effective for designers/builders.

As you can probably guess, I applaud this effort. ;D



You've done well on your homework so far!   Keep reading, spend at least a few moments on saturation, because that's the official name for 'the region where distortion occurs'. Once you get a handle on this, a lot of things will start to become clear in the context of distortion units, or even unwanted distortion from amplifiers that are misbehaving.

HTH





sumgai

[andy]

Ok, my understanding of saturation before reading up would be, simplified, that if one were put a sine wave through a device, there would come a point were the output capability of that device would be surpassed, so while the input signal could still be increased, the peaks of that wave won't increase anymore- or to put it bluntly, the wave will be 'squared off'. That sine wave will behave increasingly as (in fact, increasingly become) a square wave.

Now this is clearly where the confusion has come in for me- the purported difference between transistor gear and valve gear is that the transistor stuff will saturate (which I was calling clip- same thing?) rather sharply (or 'quickly'), much like chopping the tops of the wave with a band saw. Valve gear supposedly tends to round the edges of the 'cut' a little more, rather like removing those tops with a belt sander. Sound even close?

I'm sure I could have found some diagrams to save all those descriptions there, but that would have involved looking the subject up straight away, and taken some of the fun out of this... ;D

[sumgai]

andy,

No diagrams or charts will be harmed in the writing of this post!

You have it right, and most of us call it clipping, because that's exactly what's happening to the signal - the peaks are being clipped off, thus making the resultant waveform look more and more "square".

But now note, I purposely did not include "valve" (really tube*) gear for just the reason you included - there's a misunderstanding prevalent in the musician world that says what you said, tubes being "more rounded" or "softer".

Sorry, but strictly speaking, that's not due to the tube, but to the magnetic transformer placed between the tube(s) and the speaker(s). That transformer has its own set of characteristics that make it ideal for our purposes, and among them is the fact that it takes a non-trivial amount of time for a signal cause a build-up of flux, which of course is then transferred to the other coil, and thence back into a signal. On both sides (primary and secondary), that time is far longer than what a tube or a transistor can switch from off to full, and back again. In point of fact, a tube may not be clipping at all, yet the distortion that one hears may be due solely to the transformer itself! But it's usually a combination of both. (Why? Because designers/manufacturers are loath to over-design a transformer just to ensure a signal that's either totally clean, or that any distortion is due solely to tube-clipping. Weight and cost are the big deal-breakers here.)

But enough of that, back to saturation.

The whole premise here is, how can I control the saturation of a device? In short, the lower the voltage levels involved, the easier it is to control. In Silicon, there is a certain "tolerance" for incoming signals that near the rail voltage (slang for the power supply voltage). Loosely translated, that means the input signal will tend to remain clean even though its peaks may be with a few percent of the rail voltage. Compared to Germanium, that's at least an order of magnitude better.... in Ge devices, they've been known to start clipping when the signal is still several percent below the rail, even more than 10% below!

Thus, if we can control the input signal such that it stays within that tolerance range, or only rarely exceeds it, then we can get a pretty nice (desirable) tone out of the circuit. There are many ways to do this, a full discussion of which far exceeds the mandate of this Forum. As per dictum, Google is your friend.

SUMMARY:
Germanium good, Silicon not quite as good..... with one glaring exception. If it weren't for Germanium's inconsistancy in production, you'd probably've never heard of Silicon. At least not in the context of musicians and gear.

HTH





sumgai



* You do understand that "valves" are switches, and that transistors are just as much a switch as any tube ever made, right? Tubes are called that because of their shape, not their function. I rest my case on the fact that only the British use the word "valves", the rest of the world uses "tubes" or their local language equivalent.

[yakkmeister]

Yakkmeister's Rant Time^TM

* You do understand that "valves" are switches, and that transistors are just as much a switch as any tube ever made, right? Tubes are called that because of their shape, not their function. I rest my case on the fact that only the British use the word "valves", the rest of the world uses "tubes" or their local language equivalent.

They're called "valves" here in Aus too
Since electron tube valves are switches and valve valves are also switches (for fluids) it makes just as much sense to call valves valves since "tube" is only kinda what it looks like - bulb, bottle or can would be more accurate, depending on the valve.
If we take the concept of "named according to appearance" then transistors should not be called transistors but "black thingies" or "3-legged Giggerwhatzit" or local language equivalent (please note, this does not include those naming conventions that describe the package, like DIMM, MOSFET, etc since valves also have these - octal, B7G (heptal), noval, duodecar etc but just the overall 'thing')
Note that it's most likely that valves were named valves after John Ambrose Fleming, inventor of the first practical valve - the 'Fleming Valve' - while consulting to the Marconi Company. Thus calling them "tubes" fails to pay homage to the guy who really got things working. It's like Watts, Ohms, Coulombs and Henries (except those were discoveries, not inventions ...)

Join me next time for another exciting aside from Yakkmeister's Rant Time^TM!

[ashcatlt]

RG Keene has this theory about using multiple stages, each just barely clipping the signal, so that each stage stays pretty much in that "knee" region. I don't know that he's ever really come up with "finished product" though. Guitar signals are wildly dymanic and pretty unpredictable. Even if you built such a thing for one specific guitar/pickup/player, it's tough to get it to react the same across the neck and up the fretboard. It's an interesting idea, but difficult to implement.

sg touched on a common misconception there, too. Many people have the idea that tubes in general are "slower" than their solid state counterparts. In fact, the slew rate of a tube compares pretty favorably even to modern opamps, and some can even be faster.

What does slew rate mean in audio terms? Basically a slower slew will distort high frequencies a d transient peaks before (and more than) similarly sized lower frequencies. It's supposed to be part of the special sound of a vintage Rat pedal that the original opamp/compensation cap combination led to a very slow slew. It's actually kind of tough to find current production opamps which can be slowed down that far.

I just love the fact that what we tend to love about certain devices is the way they fail to perform their intended functions!

[sumgai]

Jun 11, 2012 9:03:02 GMT -5 yakkmeister said:

They're called "valves" here in Aus too

The Queen's on your money, you're still British in all things that count.^@   (copyright G. House, 2008)

[sumgai]

ash,

In general, slew rate is meant to be a measure of the performance of a single active device, and not a circuit of several components. IOW, we can tell what's happening inside the case of that device, and use that data to inform our circuit design decisions.

Adding a cap does not in and of itself change the slew rate of a device, but of course it will have an effect on our perceptions of that circuit's performance. For instance, a 741 op amp with a slew rate of ½V/µs (very slow) will still have that slew rate internally, even though we've added a cap outside to affect the tonal range of the signal - the cap simply doesn't "know" about slew rate, it merely impedes the signal from passing through, the lower the frequency goes. (Check out digital timing circuits for more info on this factor.)

R.G. probably isn't the first guy on the block to determine that it's easier to control the overall tonal palette by using several stages, but he's certainly given it some rather thorough consideration. I know for a fact that at least one member here has been experimenting with that very factoid, and has come up with some pleasing results (at least to him!).

BTW, I'm sure we all know this, but I'll bring it up anyways.....

---

All,

Eric Johnson has since recanted, but for a long time he was famed for hankering after that perfect tone. He was on record as stating that for him, his distortion unit sounded best when the battery was putting out only 7 vDC or so. Right there, a whole passel of Tone Nazi's ran off into the sunset, yelling that low batteries were good... After all Eric Johnson said so!! All that really did was prove my point - as the incoming signal's peak voltage approaches the rail (power supply voltage, in this case a battery), the tendency becomes greater for it to break into the saturation region, thereby becoming distorted. There's no law that says you have to maintain a full 9 vDC or whatever, you can certainly experiment here, just like you change components to experiment with gain, etc.

HTH





sumgai