An idea for variable x-over / pulse-width - can this work?

[Mark Hammer]

I don't know how many of you have ever worked with crossover distortion, but its an interesting beast.  You can hear what it sounds like by digging up Youtubes of the ZVex Machine pedal.  The Boss HM-2 also uses some crossover distortion, though not as overtly as the Machine does.

Unlike the diode-based clipping we all know and love, sticking diodes in series with the signal essentially "clips the sides" of the waveform.  That is, the signal doesn't pass until it reaches some threshold voltage, at which point the diode lets it through, and the signal instantly shoots up, instead of having a brief rise time like any stringplucked into vibration normally would.  Once the signal starts to die out, it gets choked off again at the tail end.  The "chopping" at the onset and offset of the waveform yields harmonic content, though of a different type than more traditional amplitude clipping does.

Since diodes conduct, based on the voltage hitting them, it is possible to vary the conductivity of diodes to audio signals by adding a DC bias voltage to the audio signal.  Jack Orman provides an example or two of this in action with his "warp controls" ( http://www.muzique.com/lab/warp.htm ).

In any event, it occurred to me that if a person had a variable DC bias source to add in with the audio, then one could have variable crossover distortion and effectively vary the duty cycle as a modulation effect.  I know it is NOT duty cycle, strictly speaking, since it would not be applied to individual pulses, but the net effect would be similar, in that the onset and decay of plucked/strummed notes/chords could be passed or chopped by series diodes in tandem with a DC bias.

I have to confess that I have not put this to the test as yet, but in theory it seems like a plausible idea.  One op-amp for an LFO, one opamp for a gain stage to bring the signal up, and perhaps a back-to-back pair of green LEDS to block/pass the signal.  Maybe an additional op-amp stage for gain recovery and a little tone-shaping.

It could be a whole new approach to modulation sounds,  Sound do-able?

[blackieNYC]

Very intriguing.  Looking for stuff like this.  What couldn't use an LFO?
I seem to recall a series vs. shunting diode circuit with a blend in between. Sounds simple enough.  I thought maybe AMZ but I don't see any crossover circuits at that AMZ link.  Except the op amp FB loop I suppose.  That must give a crossover waveform, no?

[psychedelicfish]

Except the op amp FB loop I suppose.  That must give a crossover waveform, no?

It might give something that resembles crossover distortion on the -input of the op amp, but the output signal is clipped similarly to diodes to ground. Basically, as I'm sure you'll be aware, the voltage gain of an op amp stage is the ratio of the resistor across the output and -input and the resistor from the -input to (AC) ground. When you stick some diodes across the first resistor, when the voltage on the output increases and the diodes' threshold voltage is reached, they begin to conduct and look like a low value resistor. The the total resistance from the output to the -input is now far less, meaning the gain of the op amp is reduced.

[Mark Hammer]

Well, don't confuse diodes in the feedback loop and the biasing of them, with diodes in series with the output of an op-amp.

I think I need to draw this out for folks to think with.  But in the interim, think of it this way:
- signal gets boosted/pre-amped
- diodes impose conduction threshold that only part of the boosted signal exceeds, and is permitted to pass
- LFO provides varying bias voltage to diodes to mimic variable threshold and permit greater portions of the signal to pass

[mth5044]

If I understand correctly, the diodes are the employer, the incoming signals are the employees. They get prepped by some kind of training; college, vo-tech, a preamp, supposedly to get them ready for a job. The employer only allows a certain caliber above a threshold through to.. Employment I guess.

Now, the LFO at the top of its game is like if your Dad is the CEO of the company. Add that to the signals resume and it automatically gets put through by the employer. The LFO when cycling down or up would be previous experience. It may or may not be enough to help a lower caliber employee in.

I blame whatever I just wrote on my fever and debilitating nausea. That sounds like a lot of fun to experiment with. I really don't know much about anything, but I guess it would be some kind of stuttering trem that depended a lot on how hard and when in the LFO sweep you picked the string.

[psychedelicfish]

How about envelope controlled X-over distortion?

[Mark Hammer]

Sure, that can work, I suppose.  Although one wonders if envelope-controlled distortion is actually noticeable as such.  Listen to this - http://www.youtube.com/watch?v=pk5POcGAUjY - and ask yourself if you could discern momentary changes in X-over distortion that corresponded to your picking.  To my mind, it would yield something that you couldn't plan for deliberately, the way you can plan around an auto-filter, or even an envelope-controlled tremolo.  Fading in and out of this kind of "pulse-width", via LFO, might be something more serviceable.  I'm just guessing though, since none of us have ever heard either of these two options.

[MrStab]

i tried an experiment with 2 compressors fed by splitting the signal with an also-experimental variable crossover circuit a while back, but it was kinda a disaster.  you might've actually posted on that thread, Mark.

i'm still keen to try and get my head around this concept, though - especially as i'm trying to come up with ideas for my own distortion at the moment. i'm currently thinking of splitting the signal to two clipping opamps - so one knob for high gain and one for low gain, as per an AMZ article (but with feedback loop clipping) - and i'm wondering whether or not it'd be cool to allow the crossover frequency to vary. i think i have a very rough idea of what you're describing, but i'll need to absorb & ponder a bit more. i'll try to stay in the loop and offer any sub-novice ideas i might have.

[PRR]

> 2 compressors fed by splitting the signal with an also-experimental variable crossover

There's several audio things called "crossover".

IIRC, you were compressing Highs and Lows differently, using a "crossover Filter" to separate the two ends of the audio.

Mark's talk of diodes tells me he is thinking on "distortion where wave Crosses-Over from negative to positive".

Run a drywall screw through the back of your guitar so the sharp tip hits the center of a string. Back off a hair. Pluck. Screw up until the point just barely hangs-up the string as it crosses through the center of its vibration.

It is a classic FLAW in push-pull amplifiers. Q101 pulls the speaker one way. Q102 pulls the speaker the other way. In simple/crude designs, Q101 goes "off" before Q102 goes "on" and vice-versa. There is a "drag" or "dead zone" in the center of the output wave. Distortion on small signals.

In my experience, crossover distortion is un-natural and should NOT be tolerated.

However as Mark says it is a useful sub-culture of guitar distortion.

And continuing to screw with Mark's "DC bias" thought: if my drywall screw is slightly off of the center of the string vibration there will be an asymmetry. I'm having trouble picturing what that would sound like, or if it could have any musical use, because I'm so allergic to crossover distortion that my mind recoils.

> think I need to draw this out

Yes.

Also thinking that stuff-it-and-see(hear) may be quicker than talking and thinking. You have boosters, you have diodes, a resistor, and probably a pot to give variable DC bias. No, it won't be perfect the first time, but a listen and look may be real enlightening.

[MrStab]

scepticism and insights noted, Paul. compressing highs & lows separately was pretty much what i was trying to do with the compressor thing (i think i made a handful of errors that snowballed into needless over-complication).

guess i've got too vague an understanding (or none at all) of the different meanings of "crossover". i'll keep reading and see if i absorb anything, even if it doesn't turn out to be useful (in my case).

[jblack547]

The main crossover distortion I know about is the kind that Class B push pull amps create, exactly as noted earlier. The distortion is created when one transistor shuts off and the other turns on at the wrong times. This type of circuit is used for current gain and has very low output impedance. The transistors are just apposing emitter followers.

Almost all of our vintage design guitar amps use Class B on the output to drive speakers. Having said that, in a small signal application the distortion could get interesting. Definitely wouldn't sound like the clipping we are used to hearing. You could replace the biasing resistors with an LDR and drive the led with the LFO to control the bias. I know I've heard crossover distortion before but I don't remember what it sounds like. We were always trying to get rid of it. Nasty if I remember right.

This is a decent explanation of class B operation. In one of the schematics, they show the distortion.
http://www.electronics-tutorials.ws/amplifier/amp_6.html

[Mark Hammer]

Mark's talk of diodes tells me he is thinking on "distortion where wave Crosses-Over from negative to positive".

Run a drywall screw through the back of your guitar so the sharp tip hits the center of a string. Back off a hair. Pluck. Screw up until the point just barely hangs-up the string as it crosses through the center of its vibration.

It is a classic FLAW in push-pull amplifiers. Q101 pulls the speaker one way. Q102 pulls the speaker the other way. In simple/crude designs, Q101 goes "off" before Q102 goes "on" and vice-versa. There is a "drag" or "dead zone" in the center of the output wave. Distortion on small signals.

In my experience, crossover distortion is un-natural and should NOT be tolerated.

However as Mark says it is a useful sub-culture of guitar distortion.

And continuing to screw with Mark's "DC bias" thought: if my drywall screw is slightly off of the center of the string vibration there will be an asymmetry. I'm having trouble picturing what that would sound like, or if it could have any musical use, because I'm so allergic to crossover distortion that my mind recoils.

> think I need to draw this out

Yes.

Also thinking that stuff-it-and-see(hear) may be quicker than talking and thinking. You have boosters, you have diodes, a resistor, and probably a pot to give variable DC bias. No, it won't be perfect the first time, but a listen and look may be real enlightening.

As the header indicates, I'm thinking of it as a quick and dirty way of imposing a sort of pulse-width modulation on an otherwise unconverted unperverted guitar signal.  I suppose in some respects it would be a tiny bit like one of those aliasers that use a clock and variable-rate S&H to "chop" an audio signal into smaller slices.  Only in this case, varying the DC bias would vary how much of the audio signal's lower-amplitude dips you would be allowed to hear.  That's why I say a "sort of" PWM.  True PWM would be amplitude independent.  This thing would be sort of like a modulated gate, with interesting sonic effects arising.

I've toyed with deliberate x-over distortion in a few forms.  As noted earlier in the thread, the Boss HM-2 Heavy Metal pedal uses a back-to-back (b2b) pair of germaniums in series with the signal to add some "side-clipping" to the peak-clipping, as well as act as a sort of crude noise control.  (Owners of these can install a useful mod of an SPST to simply bridge a straight wire contact over these diodes to yield a slightly different tone.)  The Zvex Machine uses two cascaded stages of x-over distortion to roduce a rather severe sound that seems to share much in common with the Ampeg Scrambler at its absolute worst.  I also added some variable x-over to a Gretsch Controfuzz that I cloned.  The stock unit has a 220k fixed resistor before the blend control.  I added a b2b pair of GEs, and a 250k variable resistance in parallel with that 220k, so I could blend in varying amounts of x-over along with the peak-clipped distortion.  It sounds real nice on bridge pickups.

If one listens very carefully to a sustained chord or note, processed by an X-over distortion unit, however, you can hear an interestng modulation of the tone, as the strings starts to settle down and more of the signal begins to fall below the forward voltage.  My idea was to deliberately vary that sub/supra-threshold amount by cyclically varying the DC bias on the diodes.

Here's a schematic of a PAiA lowpass filter from the 70's.  It's really just two RC pairs (47k/.005uf) forming a 2-pole lowpass without any resonance adjustment.  The diodes provide a path to ground for the caps, and their conductance is vared by the DC bias applied to them via the control voltages (ADSR, joystick, LFO, etc.).  Apply the same  logic to some diodes in series with the signal (instead of going to ground):

[R.G.]

There is a similar idea that is used in some radio communications circuits. I believe it was called a "corer" in analogy to a device for cutting the core out of an apple. I believe it's used for "blanking" signal below some threshold.

It's an analog computation circuit that uses opamps and comparators. Conceptually, using just the positive-going half of the signal, it works like this:

A comparator compares a reference voltage to the signal as it rises positive. When the signal is below the reference voltage, the output of the comparator is zero. When the signal rises above the reference, the comparator output goes positive. The comparator output is used to gate the signal into an analog computation circuit. The opamp computation circuit has an output that is zero when the comparator signal is zero, and is equal to the difference between the reference voltage and the signal when the comparator signal is high.
The negative going side has similar but opposite action.

In operation, the circuit detects when the signal is above the reference, sits only that portion above the reference down onto 0V. In practice, this is what you would get with series diodes which had ideal turn on and a variable forward drop equal to the reference voltage. What is different here is that the reference voltage is just that - a control voltage. It can be adjusted either staticly or dynamically.

Like many analog calculation circuits, it's a bit more complicated than using side effects of simple parts, but it gives you tidy, accurate control of what happens.

[Mark Hammer]

The part that has me stymied is what sort of current I should be feeding to the node where the LFO and audio signal sum at the "input" to the diodes.  I'm assuming that the LFO doesn't really need to have a P-2-P output of more than maybe a volt or two.  But I gather there would need to be some current limiting resistor between that output and the summing node.

Or am I way off?

[R.G.]

Not way off, but you might want to consider a current mirror to do the current driving, not a control voltage through a resistor. Again, more complex, but better control.

[blackieNYC]

>

Run a drywall screw through the back of your guitar so the sharp tip hits the center of a string. Back off a hair. Pluck. Screw up until the point just barely hangs-up the string as it crosses through the center of its vibration.

And continuing to screw with Mark's "DC bias" thought: if my drywall screw is slightly off of the center of the string vibration there will be an asymmetry. I'm having trouble picturing what that would sound like, or if it could have any musical use, because I'm so allergic to crossover distortion that my mind recoils.

Small bear doesn't sell drywall screws, so I used a sheet metal screw.  Before a which-screw/nail-sounds-better debate starts up, I should say that I regret taking on the project in the first place

[thehallofshields]

Mark,

Did anything ever come of this?

I imagine you're going to get some of the messy-ness of the Green-Ringer or worse.

[Mark Hammer]

I don't know if you ever saw the 1980's Michael Keaton comedy called "Night Shift" ( http://www.imdb.com/title/tt0084412/ ).  Keaton's character is an ambitious "ideas man" who, as straight-man Henry Winkler's city morgue co-worker, gradually ruins his life, pausing every few minutes to record another bright idea into his little pocket dictaphone.  There are times when I feel like Keaton's character; spewing all sorts of ideas off the top of my head, and forgetting that I need to follow up on them.

So, no, I didn't play with it....but it deserves playing with.

[armdnrdy]

I don't know if you ever saw the 1980's Michael Keaton comedy called "Night Shift" ( http://www.imdb.com/title/tt0084412/ ).  Keaton's character is an ambitious "ideas man" who, as straight-man Henry Winkler's city morgue co-worker, gradually ruins his life, pausing every few minutes to record another bright idea into his little pocket dictaphone.  There are times when I feel like Keaton's character; spewing all sorts of ideas off the top of my head, and forgetting that I need to follow up on them.

So, no, I didn't play with it....but it deserves playing with.

I love this movie! Have it on DVD.

"Note to self....finish the PAiA Hyperflange."   

[Mark Hammer]

 

My very own white whale!!