Two-stage amplifier distortion?

[joshb]

I was listening to all the distortion plugins in Logic and came across the Distortion II plugin which is meant to emulate Leslie amplifier distortion. One of the options is:

Growl: Emulates a two-stage tube amplifier similar to the type found in a Leslie 122 speaker cabinet, which is often used with the Hammond B3 organ.

Any idea what that would mean? I'm familiar with the typical atan/tanh/cos types of wave shapers, but how would the emulation of a two-stage (I'm guessing that means push-pull) amplifier be different and emulated?

[kvaca]

I was listening to all the distortion plugins in Logic and came across the Distortion II plugin which is meant to emulate Leslie amplifier distortion. One of the options is:

Growl: Emulates a two-stage tube amplifier similar to the type found in a Leslie 122 speaker cabinet, which is often used with the Hammond B3 organ.

Any idea what that would mean? I'm familiar with the typical atan/tanh/cos types of wave shapers, but how would the emulation of a two-stage (I'm guessing that means push-pull) amplifier be different and emulated?

title seems confusing - you mean two state or two stage?

btw: bear in mind that leslie distortion is also created via set of two very different speakers, so its remotely similar to multiband /=two band/ distortion

btw2: pushpull desin is probably the most common tube amp design, you usually find it in guitar amps, too

[Hideki]

It has two stages of amplification. FIrst the 12AU7 differential amp, then the two 6550 connected as a push-pull power amp.

[joshb]

title seems confusing - you mean two state or two stage?

btw: bear in mind that leslie distortion is also created via set of two very different speakers, so its remotely similar to multiband /=two band/ distortion

btw2: pushpull desin is probably the most common tube amp design, you usually find it in guitar amps, too

My apologies, typo...I meant "two-stage".

What I'm asking is if tube overdrive is typically emulated with gain and soft clipping (overly simplified), what would be the difference in emulating a push-pull amplifier?

It just seemed curious to me that they would make that differentiation in the documentation, and trying to guess what the code difference would be.

Thanks.

[iiphii]

For modeling a push-pull power amplifier stage, you would have to feed the input signal to two identical tube models (for instance waveshaping, with a signal dependent DC-offset) in parallel. The phase of the input in the other stage is opposite to the other one. The phase of the output of the opposite phase signal from the tube model is reversed again (multiplied by -1) before summing the to the parallel output. Check out A Review of Digital Techniques for Modeling Vacuum-Tube Guitar Amplifiers, Pakarinen J (2009) from Google Scholar. Figure 7 shows a model like this and it is discussed in the text as well.

[joshb]

This is very interesting, but of course the phrase "signal dependent DC-offset" seems a bit mysterious. Meaning, something like an envelope follower? So the amount of DC offset is based on overall volume?

[iiphii]

I've tried calculating a weighted moving average for n amount of previous input samples, ie. a simple FIR-filter.

[Jafo]

It's based on overall volume, past volume, outgoing volume, incoming volume, frequency, voltage, current, temperature and a whole lot of other factors, all of which interact in unexpectedly complex ways with each other. Real circuits are bloody messy, especially when distortion is involved; hit a tube or a diode (perhaps in a feedback loop) and it responds to how hard you've been playing. Simple waveshapers and polynomials like tanh() and x**3/3, on the other hand, work ok, but they're superficial; their harmonic structures are kinda flat and static. In a tube, you'll get more and higher-order harmonics with how hard you play; tanh() just gives you the same basic sound (until you start hitting ±1.0). Most tellingly, the attack and sustain of a tube offer different sounds: bright and sparkly at first, mellowing out to warm and soft. "Do You Feel Like We Do?" indeed!

Blencowe's books, Pakarinen's dissertation, Doidic's patent, Serafini's Simulanalog papers, Koren's equations... you're in for a world of Googly adventure! ("Adventure" in its classical meaning of "somebody else having a hell of a rough time a thousand miles away" )

Scientists come and go, but Ampere's name is always current.

[Miles1981]

For valves, it's a little bit more complicated than atan/tanh/cos. But it gets "easy" if you implement them with the DK-method and some good implementation of the valve equations (from the papers by Ivan Cohen). Just really costly (not yet sure if people really want them one day in ATK or additional explanations on my blog on modelling them with NR).
Clearly you can't just use a simple waveshaper for these, as the sound would be quite different from the reality. But indeed, a push-pull is "just" the sum of two stages in phase opposition.

[joshb]

This is all great stuff, guys. Thanks so much.

I'm also compelled to ask: Is there anything else that might help my simulation sound more like a Leslie amp distortion than a guitar amp distortion? Obviously, filtering seems to be a very big part of the whole sound as well.