Guitar amplifier - Current-drive & Soft-clipping, combining both

[TinkeringSteve]

This guy here describes (among others) two principles:
http://education.lenardaudio.com/en/13_guitar_amps_3.html

1) The current-driving of a speaker, which is supposedly closer to what old tube guitar amplifiers did vs. what usual solid state amplifiers do. I.e. using for neg. feedback, the voltage over a shunt resistor, which decreases as the speaker impedance increases, i.e. the would-be loss in output is counteracted.
2) And soft clipping (the circuit with the two output BJTs + emitter resistors).

So to better simulate the behavior of a tube amp (small wattage, like 1...5 W or so, in my case), I'd like to implement both concepts.
But in a way, they seem a bit contradictory to me, in that his example with the two transistors, it looks to me like the output gets lower when the impedance of the speaker rises - a thing that was to be avoided, as the goal of the current drive.
Yet apparently, tube amps have both.
So I guess it's a matter of scale, and balancing?

Could I implement something like that with some low Watt audio amplifier IC + those 2 (presumably bigger) BJTs at the end?
Or in a different way?

I actually tried modelling something in LTspice, with an opamp - and the 2 transisors behind it, driving the speaker, and feeding the voltage over the speaker shunt resistor way back to the inverting opamp input.
Then I fiddled with some attenuation/amplification of that feedback, but I didn't get it to work, as far as I can tell
=> See attached LTspice file. (yeah, I'm just some tinkering monkey who doesn't know what he's doing, not an EE, no pretense here )

[2N3055]

This guy here describes (among others) two principles:
http://education.lenardaudio.com/en/13_guitar_amps_3.html

1) The current-driving of a speaker, which is supposedly closer to what old tube guitar amplifiers did vs. what usual solid state amplifiers do. I.e. using for neg. feedback, the voltage over a shunt resistor, which decreases as the speaker impedance increases, i.e. the would-be loss in output is counteracted.
2) And soft clipping (the circuit with the two output BJTs + emitter resistors).

So to better simulate the behavior of a tube amp (small wattage, like 1...5 W or so, in my case), I'd like to implement both concepts.
But in a way, they seem a bit contradictory to me, in that his example with the two transistors, it looks to me like the output gets lower when the impedance of the speaker rises - a thing that was to be avoided, as the goal of the current drive.
Yet apparently, tube amps have both.
So I guess it's a matter of scale, and balancing?

Could I implement something like that with some low Watt audio amplifier IC + those 2 (presumably bigger) BJTs at the end?
Or in a different way?

I actually tried modelling something in LTspice, with an opamp - and the 2 transisors behind it, driving the speaker, and feeding the voltage over the speaker shunt resistor way back to the inverting opamp input.
Then I fiddled with some attenuation/amplification of that feedback, but I didn't get it to work, as far as I can tell
=> See attached LTspice file. (yeah, I'm just some tinkering monkey who doesn't know what he's doing, not an EE, no pretense here )

That page is so full factual errors and wrong conclusions that you should just forget about it.
Both valve and SS amp of classic design are voltage sources. Valve amps have nonlinear transformer element and higher output impedance, resulting in lower damping factor. Also, valves and bipolar transistors amplify based on different principles, and have different transfer characteristics, that influence how they distort as signal gets through the amp. That includes difference in how they go into limiting when you push them to the edge.
All that is reason for different sound. In the end, nowadays, SS amps are SUPERIOR in every characteristic. If you wish tube sound, you make preamp that has tube (or signal shaping that resembles tube) and do all tube distortion there, and end up with SS power amp..

Better find some good books on amplifiers (Douglas Self for instance) and dig in...  Good luck!

[TinkeringSteve]

That page is so full factual errors and wrong conclusions that you should just forget about it.
Both valve and SS amp of classic design are voltage sources. Valve amps have nonlinear transformer element and higher output impedance, resulting in lower damping factor. Also, valves and bipolar transistors amplify based on different principles, and have different transfer characteristics, that influence how they distort as signal gets through the amp. That includes difference in how they go into limiting when you push them to the edge.
All that is reason for different sound. In the end, nowadays, SS amps are SUPERIOR in every characteristic. If you wish tube sound, you make preamp that has tube (or signal shaping that resembles tube) and do all tube distortion there, and end up with SS power amp..

Better find some good books on amplifiers (Douglas Self for instance) and dig in...  Good luck!

Thanks for your effort in writing this, but it seems clear you're not much into guitar amps.
This is a completely different arena from HIFI, very different goals, and hence, probably different ways that typical circuits work.
The reasons you gave for the difference in sound barely scratches the surface of what makes tube _guitar_ amps sound the way they do. At least that's what I gathered from this document, among others:
https://peavey.com/monitor/pvpapers/Chapter3.pdf

(the interesting stuff for here, starts ~ at page 3, transformers)

And that guy designed a bunch of famous tube amps, he knows what he is talking about.
As one can glean from this document, there are several things you would call defects, which are, nonetheless, essential in producing the typical tube guitar amp sound.
It's goal is NOT perfect reproduction of the (not really pleasant) electromechanically picked up steel string sound, but to *shape* a certain sound from the raw material coming out of the pickups.
Not entirely dissimilar to the Hammond organ, whose numerous "defects" have become loved by musicians and listeners, and are now emulated in software replicas - and simpler replicas missing those "defects" sound dull / like toys

[2N3055]

That page is so full factual errors and wrong conclusions that you should just forget about it.
Both valve and SS amp of classic design are voltage sources. Valve amps have nonlinear transformer element and higher output impedance, resulting in lower damping factor. Also, valves and bipolar transistors amplify based on different principles, and have different transfer characteristics, that influence how they distort as signal gets through the amp. That includes difference in how they go into limiting when you push them to the edge.
All that is reason for different sound. In the end, nowadays, SS amps are SUPERIOR in every characteristic. If you wish tube sound, you make preamp that has tube (or signal shaping that resembles tube) and do all tube distortion there, and end up with SS power amp..

Better find some good books on amplifiers (Douglas Self for instance) and dig in...  Good luck!

Thanks for your effort in writing this, but it seems clear you're not much into guitar amps.
This is a completely different arena from HIFI, very different goals, and hence, probably different ways that typical circuits work.
The reasons you gave for the difference in sound barely scratches the surface of what makes tube _guitar_ amps sound the way they do. At least that's what I gathered from this document, among others:
https://peavey.com/monitor/pvpapers/Chapter3.pdf

(the interesting stuff for here, starts ~ at page 3, transformers)

And that guy designed a bunch of famous tube amps, he knows what he is talking about.
As one can glean from this document, there are several things you would call defects, which are, nonetheless, essential in producing the typical tube guitar amp sound.
It's goal is NOT perfect reproduction of the (not really pleasant) electromechanically picked up steel string sound, but to *shape* a certain sound from the raw material coming out of the pickups.
Not entirely dissimilar to the Hammond organ, whose numerous "defects" have become loved by musicians and listeners, and are now emulated in software replicas - and simpler replicas missing those "defects" sound dull / like toys

And that Peavey paper says nothing I didn't say. I don't car if that guy made famous amps. Technically stuff he says doesn't make sense. If he made something that sounds good it was trough experimentation without really understanding theory behind the process. Good for him and good for people that love sound of stuff he made. But he shouldn't teach theory. For example, emitter resistors (called emitter degeneration resistors) will in fact create more local negative feedback, decrease amplification factor (sensitivity of amplifier) and if there is global negative feedback, it will actually decrease output impedance of amplifier because it will linearize output stage. Less clipping will come from fact that you will get less power output at same preamp drive....
There are limiters that you install between preamp and power amp that have soft clipping. both protecting amp and making it sound better.
Speaking of Peavay, they call that tech DDT...

And I used to be musician. I know that guitar amplifier doesn't sound like a HiFi amp. I was telling you that modern way to do it is to do all the sound shaping in preamp, and basically run power amp in HiFi mode because that way you have more repeatable results and more repeatable sound.

But you, of course make amps whatever way you like. It's your stuff and your fun..

All the best on your journey..

[TinkeringSteve]

And I used to be musician. I know that guitar amplifier doesn't sound like a HiFi amp. I was telling you that modern way to do it is to do all the sound shaping in preamp, and basically run power amp in HiFi mode because that way you have more repeatable results and more repeatable sound.

OK, then I misread what you wrote.
I know that most of distortion these days happens in the preamp section.
But still, there are a lot of guitarists who, nonetheless, think that they only get the real good sound if they also crank up the amp so that the power amp begins to go into nonlinear territory.
Which could be explained by what's mentioned in the Hartley Peavey paper I linked to: Because the dynamically fluctuating (degree of) output transformer saturation, and power supply sagging, would kick in, to play a role in shaping the sound.
And this is not something (that I was aware of) that guitar preamps model.
In this light, it makes even more sens that there is now the trend for amp makers to throw one low-watt tube amp on the market after another, from 1W to 5W, ever more popular. Because if you scale everything down, you can crank up the amp, get the effects I just mentioned, without your neighbors declaring war, or your ear drums busting.

That is my hypothesis from the superficial understanding I have of these things, anyway.

About the negative feedback by the emitter transistors - yes I noticed, that was what I meant in the OP about those two goals seeming somewhat contradictory (driving with current, and soft clipping).

Then again, if tube guitar amps really were voltage sources - would it not be ok to operate them with no speaker connected?
Manufacturers usually say "warranty void if operated without speaker" - and, again, in my superficial ideas of how this works - I'm not claiming I know it all - I'm just confused - it would seem a voltage source would not care about that - very high connected resistance (i.e. nothing at all connected), there simply won't happen anything. But a current source, on the other hand, would try to drive harder when there is a very high resistance (nothing connected), and the output transformer might not like that?

[2N3055]

Valve amplifiers are voltage sources, just bad ones, with rather high output impedance..
That means that voltage on output will different when loaded /unloaded. Also with very high inductances  in transformer's windings it is prone to voltage spikes that can damage something. With load on output it will be more "civilized" in that regard.

Mathematically, ideal current source is very large voltage source that in series has very large resistor. That is how current sources are specified, by their internal resistance (in real world you will have complex impedance really) and compliance voltage.

Valve amps are actually, because of their intermediate output impedance, a sort of both bad voltage and bad current sources. But they predominantly behave as voltage sources, as internal impedance is not that high.

I'm aware of guitar player mentality. Suffered it for very long time. Good ones (studio musician level) today (and years ago too) mostly shape sound in pedals, sound processors and preamps. They like it that way, because that way they hand you cable from DI box and they know exactly what they are getting. They use amps only as monitors...Or they use stage monitors..

It is same thing as with other audiofoolery. There are pros that use what they use, there are "normal" users who buy some decent Yamaha receiver and spend money and time on listening music. And there are magic crystallisation universe aligned cable people that buy stuff costing 100000 USD that supposedly make music special, but all they do is buy crap and discuss it with other people with too much money, and they don't even listen the music most of the time. Just brag how much money they spent.

Guitar players will not brag how much they spent, it's not their thing.. They will go about hov something feels and makes them feel. Most of them are not technical at all, and don't understand how things work and that you can achieve something in many different ways. They usually stumble upon button combination that gives some sound they like and then insist they want that amps and those buttons.

They are special sort. I have many friends from that world and love them dearly, but I always make fun of them when they start with mumbo jumbo...


 

[TinkeringSteve]

Thanks for the explanations.
Yeah, well, I know that there is some voodoo in the world of musicians. Those guitar amp things, though, have been demonstrated to me. It's not totally voodoo. I'm not saying that it's categorically impossible to get the same results in different ways which may be more efficient.
But it was the case until very recently, though. 10 Years ago, amp modellers produced brittle, broken, absolutely horrible sounds for what musicians call "high gain sounds", i.e. multi stage distortion / compression etc, in the rather peculiar way that tube amps design for the heavier kind of music do. The digital modellers anyway - some analog modelling circuits, like the Peavey ones, sounded quite good already. This has improved recently, but it's not surprizing that not everyone is jumping aboard immediately.
And I don't want to buy a Kemper profiling amp, I just want to build my little, for the start anyway, minimalistic amp, modelling some of the aspects of such tube amps.

I did find circuits for the typical preamps of that sort. So I just wanted to also make a power amp which interacts with a guitar speaker, which is not a full range HIFI speaker, but typically one with a peculiar response, in the typical way that such tube amps interacted with it - but with solid state circuits, and for low wattage.
Which is why I was intrigued by those pages linked earlier.

I have found a circuit which uses a TDA 2003, configured for current-driving (and not attempting soft clipping).
The soft clipping thing... well I figured it would be a natural seeming way of protecting the stuff from blowing the speaker or my ears, in a not abrupt emergency shutdown way or how ever one would normally do such things. Nnot with a hard threshold, and actually in a way that a musician can playfully interact with it, like they have been doing for decades.

But I guess I would be fine with some simpler but effective kind of protection circuitry, if the preamp circuits by themselves sound good enough.
What would such overloading protection look like?
Or is using a TDA 2003 by itself protection enough? It has some internal protection, other than the 2002. And I have some laying around. (not from clipping and bombarding the speaker with too much higher spectrum end energy, though? Which the soft clipping was supposed to solve - I know you use a HIFI amp only maybe 1/2 cranked up, but that's not what you do with a guitar amp, it's just weird, especially if you have a 2...5W one or so )
Maybe the 30W speaker can stomach a hard clipped 5W signal, but my ears, I'm not so sure.

I wonder whether the circuit on that black page, where you said "too many erros, forget about it" would be a solution...
where he had a high valued electrolytic cap (ripple filter), followed by a big resistor, then another big cap. When there is an increased  power draw due to e.g. clipped signal, the caps after the resistor would be emptied quickly (I imagine, if all is dimensioned correctly) and the resistor would limit how fast they get filled again, so it would also limit the power that can be output for a longer while.
Seems primitive, but would that do that job?

[2N3055]

TDA 2003 is cheap chip. Get it and make an experiment. If you tinker long enough you might get sound that you like.
If you can find really old valve radio with single EL84, you can use that to make very sweet sounding guitar amp.. All you need is one ECC83 (double triode) and EL85 (power pentode) to make 5-6W amp that will sound very similar to VOX AC30 (if you use it with Celestion speaker,  of course) just not so loud...

But if you want to do this more seriously, you need to UNDERSTAND how and why it works..

So back to books. Douglas Self is a good author, that explains how it works. By explaining how to make almost perfect HiFi amp, he explains what is purpose of what part of the circuit and how it affects sound.
So you know what should you tweak to achieve something.

There is also :
High Performance Audio Power Amplifiers (Ben Duncan-Ed.Newnes)

And many others.

That Lenard site, all he does is deliberately making power rails to sag when loaded. He's not making it clip softer, but actually harder.

When classic complementary AB class output stage is driven hard, it is clipping from basically two reasons. One is from the fact that it is not rail-to-rail topology, so as it goes closer to rail voltages, it will start to be very nonlinear, ie. it will start to loose amplification, and will start to round of top of the signal. And if you go harder, you will eventually reach point when transistors will get fully saturated (open) at the top of the signal, and will basically connect power rail to output as good as they can. And that is a HARD clipping point, because you hit hard limit of your PSU.
All that guy does is making that happening faster and worse, and making that happen at lower power that it would normally. You can achieve exactly same effect if you get normal amplifier, put power rail to smaller voltage to begin with , and connect 5 Ohm resistor in series with speaker...

Soft clipping could be achieved by giving power amp full voltage to keep it in clean power regime, and connecting DDT type limier in front of it, to make sure it never goes over power where it starts to clip, and in limiter you shape how you clip signal.  Like you said well, Peavey does it right. 
I will forever remember their little Bandit 65 for guitar, and TNT 130 and Mark III head for bass...

There are many schematics for legendary amps out there. Study them.
And don't be shy to experiment. Try what that Lenard guy says, and try what I said. Measure and listen.
And then read books from people that really know what are they doing...

I had a friend that had Fender Reverb with bad output stage. Output transformer pulled in humidity and went bad.  That was many years ago, no way I could get parts in my country..
And he needed an amp, he had gigs to play.
So over the weekend I threw together a simple amp based on quasi-complementary 2N3055 AB class design, and neatly arranged that, and transformer for that inside the box. That thing had open back, you could see additional electronics.  I simply connected input signal that went to old power amp and played a little with feedback on power amp to get right amplification. I also had to tweak feedback in preamp because there was overall feedback for output etc etc.. I don't remember all the details, long time ago.

Bottom line is, it worked, and worked so well he kept it for years. No more vonky 6V6 in their old sockets that he had to reseat after transport..
It was a what we would call today a hybrid, borne out of necessity. But it worked well. Tremolo didn't, but he didn't care for that.
And everybody was praising "out of this world sound that only valves can make, not this transistor crap". LOL. We had quite a few laughs about that.

[Audioguru again]

I like the sounds produced by an acoustic guitar when live or played through a hifi sound system.
Electric guitars play odd noises created by their fuzz pedals and old fashioned very distorted toob amplifiers. Their speakers also produce odd sounds.
I think that most electric guitar players are deaf to the high levels of distortion harmonics they make.

[dom0]

Valve amplifiers are voltage sources, just bad ones, with rather high output impedance..

Current and voltage sources are the same thing (thanks to Thevenin), both in theory and practice (see: LDOs, R2R output stages). So saying that "the output of a tube amp looks more like a current source" is wrong isn't really correct either.

Valve amps are actually, because of their intermediate output impedance, a sort of both bad voltage and bad current sources. But they predominantly behave as voltage sources, as internal impedance is not that high.

... that hits the nail on its head.

[magic]

What a silly argument

If you want to produce the effect of a speaker driven from hig source impedance, driving the speaker from high impedance is actually the easiest way of getting there. Emulating it in the preamp means complex opamp filters or DSP and it only works right if you get your model right.
And in audio, convincing your users that you got it right could easily become a task of its own, even harder than getting it right in the first place.
Even Douglas Self would readily admit it, gritting his teeth

[magic]

There is indeed some errors in your article.

I have to nitpick: neither current, nor force, nor power going into the speaker should be kept constant with frequency for accurate reproduction. It's a common myth, it's wrong. And possibly neither voltage, to the horror of all the damping factor fanboys.

The impedance peak at low frequencies is caused by mechanical resonance. It means that once the cone has been put into vibration at the right frequency, it keeps vibrating at that frequency with very low energy input. That's why its impedance increases (no current wants to go into it) and that's why power into the speaker decreases. And this is the right thing to happen, in a HiFi application. Trying to push constant power at resonant frequency will cause bass boost. That being said, you aren't doing HiFi and if you want to emulate a tube amp you gotta do like tube amps do.

OTOH, the rise of impedance at high frequency is caused by parasitic inductance of the voice coil. It also causes lower power to be delivered into the speaker and this time it results in legitimate treble rolloff. Current drive actually makes more sense in such conditions.

I'm not entirely convinced by that emitter degeneration trick. And even less if they only increased it from 0.22Ω to 0.47Ω. Problem is, any loss on emitter resistors is almost perfectly compensated out by the opamp. The better the opamp (more open loop gain, higher gain bandwidth product) the less difference those resistors make.

I would rather be looking into combining soft clipping feedback networks with the constant current circuit from your article. Or go into the "low overall feedback" rabbit hole, which eliminates the problem of global feedback "correcting" output stage sag.

Also, your design oscillates like mad

[2N3055]

@magic,

Who are you arguing with and about what ?

[magic]

First with the idea of emulating tube output stage frequency response in small signal circuitry, later with the article linked in the OP.

[2N3055]

First with the idea of emulating tube output stage frequency response in small signal circuitry, later with the article linked in the OP.

Aha, OK thanks!

[SiliconWizard]

Yeah. Emulating a tube power amp stage with just a tiny tube preamp stage just doesn't work. Not even saying about emulating it with no tube at all. There are many effects, not just on clipping (as some naive approaches assume), but also on frequency response, which are not even fixed, but dynamic.

Whether one particular person doesn't like the sound of tube amps doesn't matter here. It's just a matter of taste indeed. It just so happens that many, probably most guiitar players do. Of course tube amps distort the sound all over the place in several, and intricated ways. That's extremely tough to emulate. The point is not whether this technically makes sense. We're not listening to music with audio analyzers. (Which is incidentally why "HiFi" is still such a hot, and often controversial topic.)

I have never really come across pure solid-state approaches that came close. Digital emulations - some now are pretty good, but still. Not quite there either, and the "good" ones are pretty complex.

Not sure the constant current output stage would approach this much better. I'd guess this could rather be some kind of output stage that would adapt smoothly between a voltage source and current source, effectively adapting its output impedance dynamically.

[TinkeringSteve]

There is indeed some errors in your article.

I have to nitpick: neither current, nor force, nor power going into the speaker should be kept constant with frequency for accurate reproduction. It's a common myth, it's wrong. And possibly neither voltage, to the horror of all the damping factor fanboys.

Interestingly, this dude wrote an entire book about current-driving speakers for, different than in my guitar tinkering, lower distortion:
https://www.current-drive.info/

I have not read the book or much of the text, but frmo what I remember having glanced over, only certain speaker types, which lend themselves to it (i.e. those seem to exist or could be made), are apparently recommended.

[SiliconWizard]

I remember an old project in Elektor (early nineties), "the current amp". It was designed to directly drive ribbon speakers, which have extremely low impedance, and are more commonly driven through transformers (which obviously are costly and introduce a significant amount of distortion.)

[magic]

Interestingly, this dude wrote an entire book about current-driving speakers for, different than in my guitar tinkering, lower distortion:
https://www.current-drive.info/

I have not read the book or much of the text, but frmo what I remember having glanced over, only certain speaker types, which lend themselves to it (i.e. those seem to exist or could be made), are apparently recommended.

There is a surprising lack of any decent information on the workings of speakers and microphones on the Internet. I also haven't read any trustworthy books on the topic because I don't even know where to look for them, probably not in the local book store or library. So I am actually quite clueless.

But I can say that much:

Firstly, I wouldn't necessarily trust somebody who starts with saying that everything done in the 100 years history of audio is wrong. Chances are that the author just doesn't fully understand what is being done in audio.

Secondly, there is a common misconception among current drive fanboys that current is what determines the speaker's movement. Case in point:

According to laws of physics, electric current is that which in a speaker driver effects diaphragm acceleration, which in turn produces sound pressure. Yet all power amplifiers strive, often tooth and nail, to control the voltage at the loudspeaker terminals, which only indirectly affects the current flowing in the voice coil.

By the well known Lorentz law of magnetic force, force exerted on the cone by the flow of current is proportional to the magnitude of current. Fair enough, school level physics. It is also a well known fact in the world of electric DC motors. Acceleration is force divided by mass, another known fact.
But what those people invariably miss is that this isn't the only force acting on the cone. At low frequencies, force exerted by the suspension becomes relevant too. At the frequency of (mechanical) resonance, it is indeed the only force required to keep accelerating and decelerating the cone's mass back and forth, like a pendulum. In theory at least, because in practice the cone is damped by frictional losses.
Current drive is blind to that and pushes the same amount of force, increasing the swing until frictional losses rise sufficiently to absorb the excess force.

That being said, voltage drive has its own problems due to non-mechanical and non-linear impedances which occur in the voice coil, such as inductance or thermally induced resistance drift (supposedly a real thing in some circumstances). If you can build a transducer which doesn't suffer from resonance at its operating frequencies, current drive may indeed reduce distortion and power to you. But that's an "if".

[radioactive]

I haven't really played around with the electronics for guitar much, but I will say that Line6 has got the tube amp, effects, and speaker cabs down to an art in DSP for at least 11 years now.  At least when it comes to direct line-in recording from their devices.  Maybe not as good through an external amplifier/cabinet. 

[TinkeringSteve]

I had one of the Line6 things over 10 years ago, and while the lightly overdriven sounds for blues and what not were ok, the more you went into distorted territory, the more brittle or muddy, depending on emphasis in input spectrum, the so called modelling sounded.
Let's say I was less than impressed

I recently heard the Kemper Profiler, though. That thing is not cheap per se, for the territory it covers, though...
That, now, was impressive.