What is different about bias tremolo?

[Mark Hammer]

I may well have asked this question before and simply forgotten.  But it is a question worth asking now and then.

The sound of tube amp tremolo where the tremolo is created by modulating the bias on the power tubes is a really delicious sound.  While it's not the sort of thing that jumps out at you the way clean-vs-overdrive does, it doesn't take long to be able to identify it by sound.  But I've never really understood just what it is that makes it sound nicer to our ears.  There are some nuances that are hard to pin down.

Is it simply the modulating waveform? (i.e., such that a suitably shaped LFO could do the same thing modulating an OTA, FET or photocell?)
Is it the way the initial modulating waveform is, itself, altered by being tube-based?
Does the fact that the output stage is being modulated result in some sort of change in the frequency content? (i.e., the frequency response of the power tubes changes as they are biased up and down)

Finally, we are talking about a straight-up full-spectrum modulation, and not the dual-band countermodulation one sees in some older tube amps (and apparently some newer pedals).

[armdnrdy]

I don't have an answer but...
I have a link:
http://www.lynx.bc.ca/~jc/tremoloCircuits.html

[Mark Hammer]

Thanks.  I always learn something from J.C.  Unfortunately, while he shows how to implement it on numerous amps that don't presently have it, I still don't know what makes it sound so pleasant to us.

I'm game for any %^&*amamie theory.

[armdnrdy]

Scroll about 3/4 of the way down the page...to 'The Bias Modulation design'
I think that the answer might be waiting there. 

http://www.300guitars.com/articles/the-ins-and-outs-of-tremolo/

[Mark Hammer]

Okay, getting closer.  Thanks for that.  John Frondelli frequently posts over at MEF.

I assume that if (mild) crossover distortion in the power stage is part of it, that bias tremolo is confined to Class A/B (push-pull) designs.  Correct?

But it can't be JUST crossover distortion, the modulation sounds....different.

[Transmogrifox]

Yeah, has to be a balanced push/pull design with relatively well-matched tubes, else you get pumping in the speakers.

One might "stompboxify" this effect with a pair of 12AX7's wired as a differential pair (think Gilbert cell, which essentially this is).

Pulling ideas out of air, I would think the effect is very complex where each element of of the sound contributes something to the net effect:
Frequency response
Distortion
Amplitude (gain)

If I were digitally modeling this, I would start with the gain characteristic and see how much it sounds like the real thing, then move on from there.  My guess is the modulated crossover distortion effect is more like the "spice" of the whole thing. 

No doubt the difference in the shape of the modulation is anything but subtle. LFO-to-gain is working on an entirely different principle, so my hunch is you could get the "flavor" of it by playing with LFO shapes.  Intuition tells me the emulation would need to be much more "smart" than only LFO to be able to fool yourself in a blind test, but if I was working for a company like Behringer it would be good enough for a $30 dsp stompbox emulator.

After all that I don't think I said anything you haven't already deduced yourself, but it makes me think it's one of those things that would be fun to sit down and work out in a notebook some day...combined with lots of time with an oscilloscope. 

Thanks Mark, your posts always pose intriguing questions to ponder.

[Eb7+9]

well hello again Mark and all,

I just posted a video to show what happens when bias gets down-modulated on a 12ax7 diff-pair

https://youtu.be/IaETOhkY3S0

a bit easier to scroll across the SIM'd waveform than using scope and signal gen. on a live circuit ...
it also provides an opportunity to show my high-accuracy 12ax7 algebraic modeling in action

---

with photo-elements: either they're used to provide variable loading from a current source, or used in conjunction with a fixed resistor to form a variable voltage divider ... same as turning a gain pot up and down // ie., is linear and so produces no wave-shaping

in this case, the aesthetically "interesting" aspect will depend mainly on shape of LFO waveform ...
and response of light elements (bulb and cell) to LFO waveform

---

in the bias-modulated case, gain elements are brought closer and closer to cut-off ...
but as we approach cut-off one side of the waveform starts to clip "softly", thus leading to waveform distortion

in this case, the aesthetically "interesting" aspect will combine both LFO waveform shape and circuit response to that shape, AND the distortion effects (read harmonics) ... so, it's a slightly variable harmonics circuit // maybe a bit more interesting to the ear ... at least, slightly more complex

the reason why this works (ie., not too objectionable) in an instrument context is that the distortion occurs when the signal is small
so, it's relatively masked ... if it were the other way around (somehow) it's doubtful the circuit would have been used

---

note, the 12ax7 simulation can be used to infer what happens in a push-pull power stage where bias-modulating Tremolo is applied similarly ...
the reason behind this, as the (high-Z/low-Z) plate characteristics (ie., Pentode vs. Triode) play a negligible role here

note, years ago PRR put out a similar (12au7 based) limiter circuit that could be used for low-ratio leveling ...

[Mark Hammer]

Thanks, and "hi!".

I guess the next question in line is whether this aspect can be accomplished in the analog solid-state domain.  I have a Line 6 Tap Tremolo that has a nice bias tremolo mode, but that's all digital modelling.  Is there something we could do with "mere mortal" parts, omitting tubes?

[Groovenut]

Thanks, and "hi!".

I guess the next question in line is whether this aspect can be accomplished in the analog solid-state domain.  I have a Line 6 Tap Tremolo that has a nice bias tremolo mode, but that's all digital modelling.  Is there something we could do with "mere mortal" parts, omitting tubes?

Maybe use a pair of LND150s in place of the power tubes in a Fender tremolo circuit? They are depletion mode devices like tubes but are high voltage (500V) so I dont know how well they would behave at 9VDC.

Just a thought....

[armdnrdy]

Here's some info on the LND150:

http://www.hawestv.com/amp_projects/amp_solid_tube/fetpreamp_200v1.htm#depletion_fets

[pinkjimiphoton]

hey mark,
i think part of the esoteric sound you speak of is simply the slower response time of the tubes to the modulation... as they "fade" they fade in and out differently from solid state. maybe you can adapt a valve situation to do the same, with a 12ax7...

but i'm thinking a 47 bulb like in a fender amp and a photo resistor may get ya in the ballpark, modulate the power to the incandescent lamp and maybe it will get you a similar effect?

did you ever try jon patton's harmonic tremolo? i have a board awaiting population still.... doing more guitars these days than FX. good luck bro!!

[tubegeek]

Here's one that wobbles the bias on a preamp tube, so, no, it doesn't have to act on a push-pull power stage (follow the connection marked "C" coming out of the tremolo section):



I actually built up one of these, pretty much a Champ with tremolo, nice little project, a friend found me an AM/FM radio in the trash that had almost everything I needed to build it. I hadn't ever built an amp with tremolo before, so I wanted to try it and this was just the ticket.

I got some tips from Randall Aiken's phase shift oscillator article here: http://www.aikenamps.com/index.php/designing-phase-shift-oscillators-for-tremolo-circuits and using his info, I embiggened the caps (to slow down the trem) and used a dual 1M speed adjustment pot in place of a single for a wider range of control.

My gut feeling is that the bias-vary tremolo is putting the tubes towards and away from cutoff in a juicy nonlinear fashion - not that the bias waveform is distorted so much, although I'm sure it probably IS, but that the tube under its sway is responding quite nonlinearly to the wiggle while the audio signal overlays it.

[Transmogrifox]

Yes, the bias modulation certainly works single-ended as a gain-control element, but the DC "pump" component has to be adequately filtered so that it doesn't become audible, or worse, damage speakers.

Here are some ideas.  Looks like a JFET circuit is promising with a little bit of fiddling:
http://cackleberrypines.net/transmogrifox/TransmogriNotes/BiasTremolo/

For the lazy, here's the JFET proof of concept simulation schematic:


Here's the modulation shape:

The distortion looks similar to 12AX7 implementation (see link above)


Interesting thing is the waveshape looks similar to the rEAgenerated tremolo toward max depth.  I wonder how EA/rEA trems compare to this sound.

At moderate depths the EA and rEA sound/behave about the same, and simulation waveforms look about the same.  The rEA gets interesting toward max depth in the same way as this implementation...only it's more square.  That aspect can be fine-tuned by application of a constant bias against the control FET gate.

rEAgenerated trem shape under similar depth settings:

[idiot savant]

Not exactly the same but you can also modulate pentodes quite well by wiggling the screen grid. I did a one tube pentode trem that way.

[ashcat_lt]

I just posted a video to show what happens when bias gets down-modulated on a 12ax7 diff-pair

https://youtu.be/IaETOhkY3S0

I think there's something wrong there.  You call it a "differential pair", but it's really just a parallel pair of tubes.  Both are getting the same signal, in the same polarity.  You need the phase splitter and the differential mixer (usually done in the OT) to actually see what would happen in an A/B power stage. 

This does, however, pretty much show what happens to each tube in the power amp - they get slammed into the upper rail and more and more of the (currently, nominally) negative going input peaks get clipped off, and eventually this cutoff crosses through what we might call the 0 point, leaving only the very tips of the positive peaks and then eventually just going into full saturation (or is that cutoff, whatever, the voltage doesn't change).  When you then reverse polarity of the output of one of these (doesn't actually matter which) and then combine the two together, you will see that it is basically "ripping the waveform apart from the inside".  Should end up with a whole lot of crossover distortion, just exactly like misbiasing the output tubes would be expected to do.

[tubegeek]

the DC "pump" component has to be adequately filtered so that it doesn't become audible, or worse, damage speakers.

Note how the 1st triode section is coupled to the volume control with the .005/470K/.005 "tee" filter circuit - I think that's exactly why the extra components are there and the coupling caps are small values: to provide additional LF rolloff for removing the trem from the output.